View source for AlphaCom 5.xx - 7.xx - Release Notes

[[Category: Release notes]]
Next Release - [[AlphaCom 8.xx - Release Notes]]

This document provides the release notes for AlphaCom 7 with incremental bug fix releases. The release notes for AlphaCom 7 describe new features, improvements and issues fixed after initial release.

'''Software in production:'''	AMC 08.01<br />
'''Software released date:'''	2000-03-01<br />
'''Note:'''  Additional features available in 08.20<br />
'''Note 1:'''  We sometimes do bugfixes in older versions while working with a new version.  You can’t read the list version by version, and always assume that a correction is included in the next.  Be aware of high release numbers, e.g. 06.05 vs. 07.01.  Check version dates, and also comments for each version.<br />
'''Note 2:'''  For each software version the NVRAM version is listed. If the NVRAM version is different, the AMC board must be coldstarted, and then you must do a SendAll from AlphaPro to restore the configuration.<br />
<br />
=AlphaCom 5.xx to 7.xx Release Notes=

==AMC 07.63 (1999-12-13)==
 Release: Official, available on request
 NVRAM version 07.20

===Functional changes===
----
'''Load control can be disabled:'''<br />
Load control algorithm introduced in AMC 07.11 can be disabled. (We try to find out if the load control algorithm is a bit conservative, giving too low throughput.) This option must be activated from the TST console on the master, use the “nvram” command, set module_profile.st_profile [0].volume_level to an non-zero value.<br />
<br />
<br />
<br />
==AMC 07.62 (1999-11-30)==
 Release: Official, available on request
 NVRAM version 07.20

===Errors corrected===
----
'''Load NVRAM from flash backup if NVRAM is corrupted, more improvements:'''<br />
Observation: AMC with bad battery may not restart after a power off. Problem was that after the NVRAM programming was loaded from the flash, a error -message was logged before the state-part of NVRAM was autoloaded. If the state-part was corrupted, this operation could give a processor reset. Corrected.<br />

'''AMC realtime-clock driver tries to restart clock if stopped:'''<br />
AMC realtime clock may stop if the AMC battery is bad. The SW checks the status of the realtime clock and try to restart the clock if it as stopped.<br />
<br />
<br />
<br />
==AMC 07.61 (1999-08-12)==
 Release:  Official, for selected customers 
 NVRAM version 07.20

===Errors corrected===
----
'''Wrong line number in line error reports:'''<br />
ER 1471: Observation in AMC 07.60: A line error is reported to a display station as port number 65536.  ToneTest skip reports port no as 998.  Corrected. <br />

'''Callback on mail from remote exchange ends in private ringing:'''<br />
ER 1475. Observation in AMC 07.60. Corrected. <br />
<br />
<br />
<br />
==AMC 07.60 (1999-06-29)==
 Release:  Official, for selected customers 
 NVRAM version 07.20

===Errors corrected===
----
'''Paging 44 does not work:'''<br />
ER 1455: Observation: Paging using 44 during conversation does not work in versions AMC 07.11 - 07.12. (This is due to a bug in the  “optimize-for-time” compiler option.)  Corrected. <br />

'''Load NVRAM from flash backup if NVRAM is corrupted, improvements:'''<br />
If NVRAM data are loaded from flash at startup, a few checks on the NVRAM data are added to make sure that the exchange can complete the start up using these data. If the test fails, autoload is performed.<br />

'''Priority mail events not reported:'''<br />
ER 1461, 1462. Observation in AMX X07.60 versions: the event "Priority mail in idle" is never reported off. The event Priority Mail is not reported at all. Corrected.<br />

'''Open Duplex Conference mixer in board position 20 gives reset:'''<br />
... at startup. ER 1468. Corrected.<br />

'''Various adjustments to format of texts written to log-port:'''<br />
All reports does now print month before day.  Adjustments to station-line-error report.<br />
<br />
<br />
<br />
==AMC X07.60 (1999-06-11)==
 Release:  Official, for selected customers 
 NVRAM version 07.20

===Functional Changes===
----
'''SET_DIRNO_TEXT (119) data protocol message / action added:'''<br />
Makes it possible to change the text of a directory number from data-protocol or event handler. Format : 
 SET_DIRNO_TEXT( NET_OBJ_REF /*Directory number */, TEXT16                /* New text */ )
Example of use from event handler: $set_dirno_txt L8201 'Georg'
The function was made for SEG for a Silent Alarm application.<br />

'''Load NVRAM from flash backup if NVRAM is corrupted:'''<br />
If the magic number at the start of the NVRAM is incorrect, AMC will at start-up check the flash-backup. If the backup is valid and has the correct NVRAM-version, AMC will load the NVRAM from the flash backup instead of doing a autoload. <br />
It has been a few reports on that the NVRAM is autoloaded for no reason. This may be because of EMC-problems, bad battery or SW-errors, we don’t now yet. This feature should remove most of the problems if this happens.<br />
<br />
<br />
<br />
==AMC X07.60 (1999-06-10)==
 Release:  Official, for selected customers 
 NVRAM version 07.20

===Functional Changes===
----
'''EXT_MAIL and Q_ELEM_ADDED dataprotocol messages adjusted:'''<br />
The outgoing EXT_MAIL(48) and Q_ELEM_ADDED(75) dataprotocol messages are adusted. The mail-reference parameter is now put in LOCAL_ENTITY format, not LOCAL_DIRNO. The change was requested by Cetec. <br />

'''Support for duplex station in open-duplex conference:'''<br />
If a station has the “open duplex station” property flag set from AlphaPro, the station is allowed to enter a open-duplex conference without lifting the handset. <br /><br />

===Errors corrected===
----
'''Reset related to groupcall camp-on queue on Gong:'''<br />
ER 1456: Observation: In multimodule systems having several non-overlapping groups covering more than 2 modules, the system resets in some situations. The frequency of resets is higher if the gongtone has long a duration (e.d door bell feature). Corrected.<br />
The problem occurs if one group call is active, and a second non-overlapping group call is in camp-on on the gong.   If the first one is answered, and the waiting group has members in 3 modules or more, connection of audio will cause reset due to timeslot inconcistency.<br />
<br />
<br />
<br />
==AMC X07.60 (1999-06-04)==
 Release:  Official, for selected customers 
 NVRAM version 07.20

===Functional Changes===
----
'''CONN_REQUEST dataprotocol message works with AlphaNet:'''<br />
The CONN_REQUEST (10) dataprotocol message can now be used to set up conversation between two stations in different exchanges in a AlphaNet. (Earlier versions of AMC supported only CONN_REQUEST-call between stations in the same exchange. <br /><br />

===Errors corrected===
----
'''Exchange does not start if Intercard Cage data link is disabled:'''<br />
This error was introduced in AMC X 07.60, in conjunction with the flow control. The error is now corrected.<br />
<br />
<br />
<br />
==AMC X07.60 (1999-06-01)==
 Release:  Official, for selected customers.  MultiConference 
 Main new feature: MultiConference
 NVRAM version 07.20

===Functional Changes===
----
'''Faster exchange boot:'''<br />
Some optimizations of the runtime system, plus some adjustmenst to basic station handling has reduced time from reset until all stations in the master module is available to come 50 seconds.  Slaves still are somewhat slower due to all serial communication that must take place.<br />

'''General exchange speed and reliability improvements:'''<br />
Same modifications as AMC 07.11.  Traffic capacity increase of some 40% for bank & finance high traffic systems.<br />

'''MultiConference System:'''<br />
The Simplex Conference feature has been extended to allow a MultiConference, which allows one station to be member of (listen to) up to 8 Simplex Conferences at one time, and speak into one of them at a time.<br />
The MultiConference requires a MCS station to use all features.  A MCS station is just a mechanical variant of the CRM4 station.  The station has DAKs which allows operation like this:  press, speak while holding, release to diconnect.  The DAKs has LEDs indicating current channel allocation, and which channel has audio.<br />
As you can see below, most features can be set up using a standard station, so this functionality has a number of other applications also 

'''AGA “board use” MultiConference Monitor:'''<br />
An AGA channel (two SBI channels) can be set up as MultiConference Monitor.  This means that up to 6 Simplex Conferences are mixed, and the sum is available as a 0dB audio signal out of the AGA board.  8 monitors are available total, while 6 is easiest to wire in using a ASLT filter and conncetion board where the conferences are avaiable as ab-wires.<br />
For AGA boards in board positions 21 and 22 (multiprogram AGA positions), the MultiConference mix will be available in the corresponding programs also, allowing standard stations to listen to multiple conferences.<br />

'''Features / dir.nos for MultiConference system:'''
 7980 "MCSActivate"	feature=10 	id=0/1
 7981 "MCSPriority"	feature=10  	id=2
 7984 "MCSDeActAll"	feature=10  	id=3
 7982 "MCS Select"	feature=11
 7983 "MCSTalkBack"	feature=1
All these features are implemented as regular features with additional parameters.  They can be dialed from a keypad for testing, but must be allocated to DAK keys to be useful for a customer.

*TalkBack: Parameters are <AGA board/channel/subchannel> <conference><br />The configuration of the key controls AGA hardware. Pressing the key while listening to a MultiConference feeds the conference.
*Select: Parameter is <conference> <TalkBack DAK><br />Reprograms the TalkBack key’s <conference> parameter.
*Activate: Parameter is <TalkBack DAK>, <TalkBack DAK>, <TalkBack DAK>,<br />Pressing Activate enters a mode where TalkBack keys are toggled in and out of it’s pre-selected conference. When toggled in, audio is heard immediately. Terminate by pressing Activate once more.
*Priority: Parameter is <TalkBack DAK>.<br />Switches OFF all conference monitoring, then forces ON the selected channel.
*De-activate all: Parameter: <none><br />Variant of priority, i.e. switches off everything.

'''Events for Simplex Conference:'''<br />
New events:

*Simplex_Conference_Member: Occurs when a station joins or leaves a conference.  Related_to = Conference.<br />Sub-event 0 = ASLT member, sub-event 1 = AGA member.
*Simplex_Conference_Audio: Occurs when someone feeds the conference (but not driven by station M-key in this sw version).

'''Data protocol message: AUDIO_DETECTED ($VOX):'''<br />
Data message which signals to the conference that audio is present.  Can be generated e.g from RCI connected to an Audio Detection Board, or a radio system’s squelch output.
For use in action strings, use abbreviation $VOX.<br />

'''OWN action command:'''<br />
Event Handler action command which can replace the %1 and %2 station references.  Useful from RCI to look up a stations phys.no and name, or in Event Handler to reference a CRM4’s DAK keys.<br />

'''%1.DAK action string parameter:'''<br />
This parameter looks up a DAK string in in a station’s DAK tables, and returns the key number.  Intended for use together with the IND action which controls LEDs on CRM4 / MCS stations, so that LED indication goes automatically to the relevant DAK, and if DAK configuration is changed, indication will follow.<br />

'''Simple Link Layer to AlphaCom data protocol:'''<br />
Readable ASCII formatted data protocol suited for debuging with a standard ASCII terminal.  Full AlphaCom data protocol available.<br />
To simplify operator interface, symbolic names can be used for commands.<br />
Type $dial_digits L101 L102 to connect 101 to 102.<br />
Network layer is also available, so @2 $dial_digits L101 L102 means “send the command to node 2 and execute it there”.<br />

'''Event Handler Actions are full data protocol:'''<br />
The symbolic form of the Simple Link layer is also possible to send from the Event Handler’s action strings.  This means that RCIs and Events now can do virtually all exchange features, including controlling other exchanges in AlphaNet.
Errors Corrected<br />

'''Reset due to Hot-swap of AGA:'''<br />
ER 1358:  If you take out an AGA board in pos. 20 or 21 for service while the exchange is running, the exchange will reset.
Corrected.  Probably introduced in 07.20 with the AGA Board Use functionality.<br />
<br />
<br />
<br />
==AMC 07.33 (1999-08-27)==
 Release:  Official, for selected customers (bugfix)
 NVRAM version 07.20

===Error Corrected===
----
'''Stento Multidrop Linklayer - Slave mode: No retransmitts:'''<br />
Observation: The Stento Multildrop linklayer in slavemode (developed for the Taipei underground system): Frames are not retransmitted even if ACK is missing.  Correction:	OK.<br />
<br />
<br />
<br />
==AMC 07.32 (1999-03-25)==
 Release:  Official, for selected customers (bugfix)
 NVRAM version 07.20

===Functional Changes===
----
'''Open duplex conference dataprotocol message added:'''<br />
If a external device includes a station into a open duplex conference (830x), the AlphaCom will now return a message when the handset is lifted and the station is connected to the conference-audio.  The new message is:
 OD_CONF_ENTERED 118 (0x0076)
 NET_OBJ_REF   /* Station which enters conference /
 NET_OBJ_REF   /* Which conference /
<br />

===Error Corrected===
----
'''Data activation of Call request transfer to remote station fails:'''<br />
Observation:	E.R. 1424: Dataprotocol message SET_REQ_TRANSF to a remote node+dirno  does not work.  No display update, and no transfer of mails. This works if the transfer is activated from the station. Correction:	OK.<br />
<br />
<br />
<br />
==AMC 07.31 (1999-03-10)==
 Release:  Official, for selected customers (bugfix)
 NVRAM version 07.20

===Error Corrected===
----
'''MPP with free text does not work:'''<br />
Observation:	E.R. 1430: When sending a MPP free text message to radio pager from eventhandling/RCI, nothing happens.  
Correction:	OK. The AlphaCom added a control character to the text in the ESPA message, which made the paging reject the message.<br />

'''Tone test fails in slave module:'''<br />
Observation:	E.R. 1406: The Station line Tone Test generated tones on stations in slaves, but reported all as faulty. 
Correction:	OK<br />

'''#SERVICE* work only one time after reset if not #TST ON*:'''<br />
Observation:	#SERVICE* works only once if #TST ON* has not been used first.
Correction:	OK.<br />

'''Optimised some CPU wasting:'''<br />
Observation:	Exchange feels slow in high traffic.
Correction:	11%  of the AMC CPU capacity was used to scan the 6 APC-RCI’s and do some interrupt loadcontrol. This load has been reduced to 0.1%, freeing up CPU-capacity for other useful work. <br /><br />

'''Known Bugs and/or Workarounds:'''<br />
'''Reset when Hot-plugging AGA board:'''<br />
Observation:	Removing a AGA board when the exchange is operating may cause reset.  This is especially a problem during commissioning, as the AGA releases resources according to NVRAM, which can have been changed by AlphaPro since the last exchange start-up.<br />

'''Hotline DAK 8 activated always when going on-hook:'''<br />
Observation:	Hotline will often be configured to start after 5 seconds, i.e. in many cases the hotline DAK 7 will NOT be activated.  The hotline on-hook action DAK 8 will ALWAYS be activated though.  As this is intended e.g. for CancelCallRequest, redundant messages should do no harm.<br />
<br />
<br />
<br />
==AMC 07.30 (1999-01-18)==
 Release:  Official, for selected customers.  New functions.
 Includes AMC 07.08 and 07.09 corrections.
 NVRAM version 07.20

===Functional Changes===
----
'''Feature added: Disconnect own station from ongoing groupcall:'''<br />
If you press C-key or dial 7884 on a station which is receiving a groupcall, the sound from the groupcall will disappear from this station. Limitations:

*The station will not be free to receive a new groupcall before the first groupcall is disconnected totally.
*EVENT/RCO control  is not turned off before the groupcall is disconnected totally.

The disconnect-from-groupcall feature has feature-number 53. Feature 53 is not autoloaded into class of service 1. C-pres in Idle disconnects ongoing groupcall if feature 53 is included in the ClassOfService for this station. (C-pres in conversation just disconnects the conversation as before).<br />
Disconnect-from-groupcall is also available as a data protocol message (116).<br />

'''Logging of line errors to remote exchange and/or external device:'''<br />
AlphaCom can be programmed with two addresses which will receive error report messages. If the receiver is an AlphaCom exchange,  it will print the error report on the system log port. The format of the printout has been adjusted to fit a report on a single line.  <br />
The error reports are sent using a new data protocol message with well defined parameters, STATUS_REPORT(114). External devices like a PC can use this message in a logging application. <br />
AMC 07.30 can only route reports regarding station line errors to remote devices. Support for routing reports for boards, audio lines etc will come in AMC 08.00. Current versions of AlphaPro can not program the forward addresses. Until AlphaPro 08.00 arrives, use the nvram-command on the TST-port.<br />
ex_profile.line_monitoring.report1_addr<br />
ex_profile.line_monitoring.report2_addr <br />

'''LOG action - Print to system log from Event Handler / RCI:'''<br />
You can program something like LOG “Door B4 opens” as an action for a RCI or a system event. When this the condition is triggered, the text “Door B4 opens” is printed on the system log port, preceded by a timestamp. This feature can be used to set up simple call logging applications. <br />
The printout is rerouted in the same way as that line-error reports. The printout is sent via the new message LOG_STRING (117). <br />

'''Insert name of event source in action (%1.NAM()):'''<br />
In the event handling the code %1.NAM() inserts the display text name of the “owner” of the event. %2.NAM() inserts the name of the “related-to” dir.no of the event.  This is particularly useful with the LOG command described above, but can also be used for other purposes, like sending mail to pocket pagers or stations. <br />
The name inserted is 12 characters wide, right padded with space.  %2.NAM() only works if %1 and %2 is in the same exchange. <br />

'''MPC protocol input commands: >, <, - _:'''<br />
Support for the following  MPC protocol input commands added: 
 > aaaa (M-pres)
 < aaaa (M-release)
 - aaaa (Offhook)
 _ aaaa (Onhook).
MPC protocol is intended for legacy applications. New applications should use the AlphaCom protocol. <br />

'''Check if M-key pressed when entering program conference:'''<br />
If the M-key is pressed when a station is included into a conference, the station starts feeding the conference. Requested for an evacuation application where conference-setup and M-key use is controlled by external equipment.<br />
 
'''Telephones on ATL are not required to press M in groupcall:'''<br />
If a telephone connected to a ATL board starts a groupcall, the microphone is connected to the groupcall immediately after the gong tone. <br />
(Tip: It is also possible to program M-pres on * (DAK-1)  and M-release on # (DAK-2). This will allow the telephone to set absence, call forwarding etc. without having to wait for timeout.) <br />

'''Simple conversation setup from data protocol:'''<br />
A parameter has been added to the CONN_REQUEST (10) message, which can specify that AlphaCom shall not require acknowledge messages to set-up and disconnection. This makes it possible for a simple “fire and forget” application to set up conversations in AlphaCom. <br />

'''Fast startup / reset of external device (RIO):'''<br />
If AMC 07.30 is used together with RIO 02.10 SW, the AlphaCom can detect a RIO reset and restore the RIO state within a few seconds. Earlier versions of AMC has to rely on a several minute timeout to detect reset of RIO. The AlphaCom can also detect a new RIO added to the bus within a few seconds.<br />
AMC now polls devices programmed in NVRAM all the time, also when AMC thinks the RIO is missing. This allows the RIO to inform AMC of its presence. (Requires ack-timeout to be set to 200ms or lower.)<br />

'''TST port “rack” command display SW version of boards:'''<br />
Useful for debugging.<br /><br />

===Error Corrected===
----
'''Improved errorhandling on ISO1745 linklayer (RIO communication):'''<br />
The ISO1745 linklayer (used for RIO communication) implementation  is improved to handle noise on the databus in a better way. The probability of  loosing messages is reduced. What has been done: 

*Extra characters received due to noise are ignored.
*Retransmits message if a select sequence fails.
*Faster timeout on missing RIO activity (previously 10 seconds, now 1 second).

RIO 02.10 SW should be used to improve the RIO error-handling. <br />

'''Global groupcall blocked by missing exchange:'''<br />
Observation:	E.R. 1339: If an exchange that is member of a global group goes down, the global group call becomes busymarked after a couple of minutes.   The group call becomes available again when the faulty exchange comes up again.
Correction:	OK<br />

'''Reset when SendAll from AlphaPro:'''<br />
Observation:	E.R. 1414: The Exchange resets during SendAll from AlphaPro, just after the directory numbers has been sent.  This happens if many directory numbers has been reprogrammed, but all display texts still has the autoloaded value “Station nnn”.  This condition caused the sorting of the namelist to take 22 seconds, which in turn caused the watchdog to trig exchange reset. 
Correction:	OK (sorting algorithm slightly adjusted)<br />

'''OpenDuplex conference via AlphaNet had closed microphone:'''<br />
Observation:	E.R. 1410: Dial into an exchange via AlphaNet, and join a local conference by dialing 83xx.  The conference sound is heard, but the microphone is not switched on.
Correction:	OK <br />

'''Hot plug in of standby AMC board reset active AMC board:'''<br />
Observation:	ER 1386. (Hot standby requires modifications to both AMC and APC HW)
Correction:	OK <br />

'''AMC board in standby mode has confusing LED indication:'''<br />
Observation:	AMC in standby mode blinks green like the active AMC, how do you know which is the active one?
Correction:	AMC in standby mode is steady green<br />

'''AMC standby instability when used with the multidrop linklayer:'''<br />
Observation:	Taipei railroad project: If the data link to the active AMC board failed, the system reset it self several times before it settled down with the AMC on the working datalink being master.
Correction:	OK<br />

'''Leap year handling wrong:'''<br />
Observation:	E.R. 1421: 29 days in February every year if you adjust clock sometime between 1 jan and 28 feb. Otherwise 28 days in February, even if is actually a leap year.  
A consequence is that wakeup (7883) does not work from evening feb 28  to morning  mar 1. <br />
Correction:	OK  (After SW upgrade you have to set the AlphaCom clock once to set the leapyear counter right. Use AlphaPro or station (785) to set the clock.) <br />

'''Partial remote number on DAK did not work towards TouchLine:'''<br />
Observation:	Program a partial remote number plus digit to complete the remote number on a DAK.  The remote exchange is a TouchLine.  The connection in the remote exchange fails.  Example: "4" is a partial remote number, "400" is the dir.no. of a station in the remote exchange. Programming "I4 P00" on a DAK does not work. 
Correction:	OK (This is really a bug in TouchLine TAC, but AMC 07.30 goes around the problem)<br />

'''Reset when global groupcall is overridden by priority:'''<br />
Observation: E.R. 1418:  A global groupcall is started from a multimodule exchange.  The  groupcall is disconnected by a higher priority groupcall in the exchange of  the initiator. This leads to exchange reset. 
Correction:	OK <br />

'''Reset when busy override of digit 0 pressed in conversation:'''<br />
Observation: E.R. 1420: Station A is in conversation with station B. Station A is pressing digit 0 (probably for mic-mute).  Station C dials to A, gets busy and uses busy-override (MM or 58).  The conversation is not set up, B is not disconnected.  If  station C presses C-key,  subsequent conversations to or from A will fail.  If  station C presses digit 5  before pressing C-key, the exchange will reset after 10 seconds. 
Correction:	OK <br />
<br />
<br />
<br />
==AMC 07.21 (1998-10-28)==
 Release:  Official, for selected customers (bugfix)
 Additions for Silent Alarm application.  Bugfix, incl. AMC 07.07 corrections.
 NVRAM version 07.20

===Functional Changes===
----
'''New feature:  Event trigger without tone:'''<br />
New feature 52 (displayed as “reserved” in AlphaPro 07.20) behaves identical to feature 85 “Event Trigger”, but has no cancel tone.  
Application:  Silent Alarm from a station, where a push-button does a DAK operation activating an event setting up alarm to other stations, and no sound is heard in the initiating station.<br />

'''Default speaker is allowed to dial:'''<br />
Previously, the default speaker station of a simplex conference had “dead” keyboard.  Someone else had to cancel the conference, and thus restore normal operation of the station.  When used in an office environment in a Silent Alarm application, the station starting an alarm could not stop it.
Now, the default speaker is allowed to do normal dialing incl. switching off conference.  When returning to idle it becomes default speaker again if the conference is still active.<br />

'''Additions to 7873, feature “Toggle feature COS 1”:'''<br />
Dir.no 7873, feature 37 “Toggle feature”, moves a feature in or out of COS 1, depending on the previous state.  The feature is autoloaded as 37/0, if you create it with AlphaPro it becomes 37/1 by default.. New functionality:  

*- A dir.no with feature 37/2 will always add this feature to COS 1.
*- A dir.no with feature 37/3 will always remove this feature from COS 1.

The idea for the regular functionality is that a service man can temporarily allow certain “programming from station” features, so that station configuration can be changed from the station itself while walking around in a building.<br />
The new functionality is used in a Silent Alarm application, where the simplex conference feature is made accessible in COS1 (most stations) when the alarm starts, and conference is made inaccessible when the conference is terminated (controlled by Event Handler actions).  This solves an eavesdropping problem as office stations are set up as default speaker if someone activate the conference any time.<br /><br />

===Error Corrected===
----
'''Reset when answering CallBack due to inconsistent USER / STATION:'''<br />
Observation:	ER 1387: The exchange will reset if someone dials 70 +8 to answer a CallBack from a station which has a user which in turn has no station (i.e. inconsistent internal tables).
Correction:	The software does a consistency check, and the Callback will be aborted in this case (must be deleted from the display using 70 + 0).
''NOTE:''   AlphaPro does not handle the reference from Station to User correctly if one changes (or deletes) the physical number of a User. <br />

'''National characters in display data mapped twice in AlphaNet:'''<br />
Observation:	ER 1389: The A station in an AlphaNet conversation lost some special national characters (converted to basic English A-Z 7-bit characters).  For Hebrew, all characters were garbled by this mapping.  This error was introduced in 07.20.
Corrected.<br />
<br />
<br />
<br />
==AMC 07.20 (1998-10-15)==
 Release:  Official, for selected customers.  New functions.
 A number of features required by sold projects (e.g. Taipei) has been implemented.
 Second release within the AlphaCom/8 project. 
 This version contains all bugfixes done on 07.03 to 07.06.
 Use AlphaPro 07.20 to activate new features.
 NVRAM version 07.20

===Functional Changes===
----
'''Support for the ATLB board:'''<br />
ATLB is a line board which uses standard analog DTMF telephones.

*General idea:  Telephone shall be able to use most exchange features. Normal dialing from DTMF telephones. Note that features requiring M-key does not work.  This includes GroupCall (M to speak) and features like CallForwarding (M to accept activation). Also:  A telephone can NOT receive audio with the handset on, so audio program, group calls and simplex conferences are meaningless (even using a loudspeaking telephone, which also just waits for ringing when its not in use).
*Board type: The AlphaCom detects the ATLB automatically, and there is no setting in AlphaPro where the user must define “telephone”.  The board type automatically modifies a few features: - The ATLB is always “private”, i.e. it will always ring even when called from a station with private break-through priority. - A new “ring” command is sent to the line board as part of private ringing on/off.
*DAKs on the * and # keys. The * is DAK 1, # is DAK 2. If you can find a telephone with keys labelled A, B, C, D (dials 4 unusual DTMF tones) these are handled as DAK 3, 4, 5, 6. The DAKs can be programmed from the telephone itself using 784.
*Hotline: Hotline activates DAK 7 when off-hook, and DAK 8 when on-hook. See separate description below.
*R-key for digits during connection: The AlphaCom works like a PABX in the sense that DTMF reception during speech is not possible.  You must press the R-button to get dialing tone, then you can dial one or more digits. Features like DoorOpening (R6) and Inquiry (R2cccc) are possible.  Note that during Inquiry, you must toggle back and forth using R2 every time!
*Answer Call Request with handset off: As a telephone operates with handset off (i.e. busymarked), the operation of CallRequest has been tuned: - The telephone receives voice message & ringing while in queue - The guard can answer back and connect directly
*Line monitoring: It’s possible to use a resistor across the telephone wires close to the telephone.  Thus a wire break can be detected and reported as line error.

'''Open duplex conference for 4 participants:'''<br />
We now have a “handset” conference where all participants can speak and listen at the same time.  It’s possible to have 20 conferences with their own properties like dir.no and priority.  An open duplex conference works much like a generalized conversation, i.e. the station is busy and the microphone is always on, and the display shows a menu for digits during connection.<br />
Basic requirements / restrictions are:

*- Participants must be off-hook (handset or headset). Intercom stations (ASLT) and telephones (ATLB) can be used.
*- Participants only in the master module (among the first 138 subs).
*- AGA board is required as conference hardware. Each participant uses one AGA channel (out of 16), i.e. one AGA board can support 4 conferences with 4 stations each.   Channels are allocated dynamically, i.e. one AGA board can support 8 conferences with just 2 stations each.  These is no one-to-one relation between conference no and AGA channels. AGA channels are reserved as conference mixers in the new “Board Use” table in AlphaPro.

Activation/operation:

*- 8300-8320 dir.nos (feature 88) for 20 conferences.  Each station can join a conference by dialling 83xx.  The station rings until the handset is lifted.  The station leaves the conference by hanging up.
*- 74/75 for remote setup.One station can force up to 4 other stations into conference by dialling 75 + group + 83xx.  Each station rings until the handset is picked up, or times out after 30 seconds.  The conference itself has no indication of ringing stations, i.e. the present participants can talk while waiting for additional stations.  Stations can be forced out of conference by dialling 75 + group + 8300.
*- 56 during conf.  (feature 89) for adding new participant.  A conference participant can drop out of the conference temporarily to add another station by dialing 56 + dir.no.  The dialing station returns to the conference immediately, while the called station starts to ring.
*- Master cancels all.  Normally each participant can leave the conference without disturbing the others.   An optional “master” station can be configured, and has the effect that when this station leaves the conference, the whole conference is terminated.  This feature is intended for control room applications where conference is primarily used for conversation where Call Requests can be answered by adding a third station, and when the guard hangs up the others need not talk anymore.

Data protocol support:  It’s possible to add/remove stations to/from conferences via data protocol.  In addition, if a station leaves the conference by hanging up, a data message is sent to the node/device that initiated the conference.<br />

'''Support for the CRM 4 station:'''<br />
The CRM4 is a Control Room station lime the CRM3 (4-line display), but with up to 96 DAK keys each with 2 LEDs.  Best suited for “fixed” installations like prisons with few guard stations and many subs.<br />
Main features are:

*- 100 DAKs total. DAKs are programmed in the exchange and can do what normal DAKs do.
*- LED/buzzer control via Event Handler and new action IND via display data. This means that LEDs operate just like lamps on RCOs, using the repertoire of the Event Handler.

Note:  Very limited number of CRM4s per exchange both due to DAK storage and Event Handler action lines.  Please contact Product Support in Trondheim for assistance.<br />

'''Tone Test of station:'''<br />
Extension of standard line monitoring.

*- Tests complete audio path via line board, wires, station speaker and microphone. Errors are reported as mail/log/event, but stations are not “down”
*- Automatic start once per hour during low traffic periods (30 minutes on 550 subs). Brief report on log port, listing only new line audio errors
*- Manual start 7885 (feature 90) (6 minutes on 550 stations). Complete system status on log port, incl. line wire errors, intended for service man to get system status before repair, and to verify system after repair.

As part of this, the log messages have been normalized for the various error types.<br />

'''Group restrictions:'''<br />
The AlphaCom has an access control mechanism called ClassOfService which works per feature.  But many features like Conferences and GroupCall have many “individs”, and it’s desirable to give access to some, and restrict others. We now have the ability to protect each directory number by using one group for access, and one for restriction:

*- If COS NOT OK:  Allow access from group nn.
*- If COS OK:  No calls from group mm.
*- Transfer, search etc. are restricted in addition to basic dialling.

'''Handset off hook too long reported as error:'''<br />
Handset off is a problem both with TouchLine stations (w/o display) and telephones.  TouchLine station off-hook can’t be separated from line break, while on a telephone off-hook is close to wire short.To handle these situations, the exchange now has a timer running while the handset is off and no feature is active.  The timer is autoloaded to 60 minutes, but can be set to much shorter value using TST NVRAM editor in security systems.  When the timer times out, a normal error indication mail/log/event is generated, but the station is not taken out of operation.

'''Hotline:'''<br />
New station option, selected per station:  When the handset is lifted and the user does not start dialing within a configurable period, the DAK 7 is invoked automatically.  When the handset is replaced, DAK 8 is invoked. This is intended for security type stations, where lifting the handset does a CallRequest, and replacing it does a CancelCallRequest (if DAK 8 programmed).  You can program any dir.no on the DAK, e.g. conversation setup, audio program selection, join simplex conference, etc.The period is autoloaded to 5 seconds, which allows mixed dialing/hotline use of stations.  For pure CallRequest systems you can set the time to 0.<br />

'''Ringing in Call Requester Mode:'''<br />
When a station does a Call Request (enters the mail/display queue of another station), it’s placed in “Call Requester Mode”.  The station first receives a voice message (if ASVP board present), then it will hear a ringing tone, and the LED will flash.  The ringing tone is new - previously there were only silence after the voice message. Note that the station leaves CallRequesterMode as soon as the station is used incoming or outgoing.  This may cause switch-off of queue indication even if the station is still in the queue.  If the call button is pressed again, indication will be switched on again.<br />

'''Board use table:'''<br />
The AGA board over time has become a very useful exchange component.  It’s no longer possible to set “default behaviour” just from the board position.  To solve this, a new table called “Board use” has been introduced. Initially this table is used just to assign new functionality, which is:

*- Conference audio mixer for open duplex conference
*- AGA port as station (“audio terminal”). The table is autoloaded to “default use”, which means that the old rules apply, i.e. determined by board position.

Note that AlphaPro has no knowledge of which board actually sits in each board position in the exchange.  This configuration only works for AGA boards!<br />

'''AGA port as station:'''<br />
To simplify installations with many 0 dB audio lines, which previously required ASLTs + FBSARs, we now can use AGA ports as stations.  All audio operations are possible, including use as default member in conferences. An AGA station of course has no signalling capacity, but it’s possible to use e.g. the TouchLine data port to do dialing on behalf of the AGA station, or you can use EventHandling actions. (Typical Engineering solutions).<br />

'''Support for Duplex Station:'''
There is a new property “Duplex station” which can be activated per station. Conversation 

*- between two “duplex stations”
*- between one “duplex station” and a lifted handset is not duplex controlled by the exchange (i.e. open duplex even if the station is electrically on-hook).

This is intended for stations with built-in duplex controller, i.e. using the upcoming DSP-based “finance station kit” or new industrial stations.<br />

'''New station type “Interguard”:'''<br />
This works identical to a master station, but tells the exchange that this is a display station electronics.  This station will come up faster, and error situations are resolved more quickly and more correctly as there is no need to do alternative tests for non-display stations.  The setting is useful for installations where line monitoring is important.<br />

'''Conversation timeout at 1 hour 45 minutes:'''<br />
As part of the enhanced line monitoring, there is now a long timeout on conversations to assert that problem stations will be tested and errors reported. If this is a problem, you can configure the time using TST NVRAM editor:<br />
ex_profile.timeouts.max_conv_timer   (value in 100 ms unit)<br />
ex_profile.flags.conv_lim_enabled (boolean 0 or 1)<br />

'''Addition to the EventHandler:'''<br />

*New action IND which controls CRM4 LEDs via display data messages. Documented in AlphaPro help + CRM4 manual.
*New parameter %1.PHY which supplies the physical number of a station. Allows use of all actions which has physical no as address (in 07.0x we offered only DIR to be used e.g. in CCTV protocols)
*New parameter %3.CHG(ON_text,OFF_text) which returns either ON_text or OFF_tex depending on the transition value of the event. This reduces the number of lines required in the event handler by 50% e.g. for CRM4 operations.
*Easier representation of some common ASCII codes. \r for \x0D (return), \l for \x0A (linefeed)

'''DAK can be assigned AreaCodes and RemoteNumbers:'''<br />
You can now dial on your own station
 784, AreaCode + LocalNo + M,  DAK-key
 784, RemoteNo,  DAK-key
 784, Prefix + RemainingDigits + M,  DAK-key
In previous versions this had to be entered in AlphaPro.<br />

'''100 DAKs per station possible:'''<br />
You can, somewhat manual, establish 100 DAK keys for a few stations.  The DAK string interpreter will handle a line starting with “>” as a “pointer” to another station’s 10 DAKs.  Example:  You have a CRM4 on physical 200.  Find some non-existing stations, say physical 501 to 510.  Now you program 200’s 
DAK 1 to “>501”<br />
DAK 2 to “>502”<br />
…<br />
DAK 10 to “>510”<br />
Now you can program the keys from the station itself using 784, or you must use AlphaPro on physical 501 for the 1-10 DAKs, go to 502 for the 11-20 DAKs and so on. The feature is intended for the CRM4 station, which can have up to 100 DAKs, but can even be used e.g. with substations.  On substations you would re-direct only the first entry e.g. to “>550” on a lot of stations, and then you can change the configuration on 550 and it takes effect for all substations without further programming (or you could “route” the substations via a physical station where the user could dynamically change the configuration with 784!)<br />

'''DAK function on digit keys:'''<br />
It’s now possible to use a station’s digit keys as DAKs 1-10.  This is very handy both for telephones and industrial stations. Note that the current implementation switches ALL digit keys to DAK.  There is no way to do normal dialing from a station in this mode! Note that AlphaPro 07.20 does not support this feature, which means that it must be turned on using TST NVRAM editor (support planned for AlphaPro 08.00).  Please contact Product Support in Trondheim for details.<br />

'''AlphaNet audio line test can be switched off:'''<br />
When using 2 mbps multiplexers with AlphaNet, we frequently have problems with the periodic tone test causing lines to become busymarked. The tone test is autoloaded to use 200 ms periods for measuring silence, then tone.  MUXes seems to introduce transmission delays due to voice coding and level adjustments (AGC).  This means that the measured difference (tone - silence) is too small after 200 ms.<br />
Possible actions:

*Increase test time
*Stop test entirely. Set ex_profile.timeouts.iech_test_timeout = 0xffff.

A more thorough description is available as an Application Note, please contact Product Support in Trondheim.<br />

'''Hot standby feature for AMC:'''<br />
Two AMC boards can be plugged in simultaneously.  Which board controls the module is determined by a microcontroller on APC. This feature was delivered on an Engineering project and is not available as a standard product.  Please contact Product Support in Trondheim for further information.<br />

'''AlphaNet multidrop linklayer:'''<br />
A version of AlphaNet where the exchanges are connected on a 4-wire 485 bus for signaling, using node 1 as bus master (audio must still be wired point-to-point).  This is not a standard product - it was developed for an engineering project used in a low traffic railway application.<br /><br />

===Errors Corrected===
----
'''DAK string playback unreliable in high traffic:'''<br />
Observation:	Some long DAK strings, e.g. to activate Call Forwarding, may fail during high traffic peaks.  The station stops in the middle of the sequence, and the display asks for more input.  
Correction:	The code was re-written to playback one DAK a time for the whole exchange rather than concurrent for all stations (primarily to save RAM).  As a side effect, DAK timing problems disappeared!<br />

'''Multiple RCI/Event actions:  action after MST/MGR fails:'''<br />
Observation:	If the action string contains several commands (separated with semicolon or <RETURN>), commands following an missing optional parameter will not be performed  (MST and MGR has optional priority and g).  
Typical example is error reporting where you want to do the traditional mail to station followed by mail to pager or setting an RCO.
Corrected as side-effect of Event Handler extensions.<br />

'''ASVP not available in autonom slave:'''<br />
Observation:	A slave module which lost contact with the master and therefore restarted in autonom mode could not make use of an ASVP board.
Corrected.<br />

'''Double-M to activate priority failed frequently:'''<br />
Observation:	Very difficult to make a “good” double M sequence.
Correction:	Time window for the second M increased.<br />

'''Reset when direct paging + feature 47 with answer 98x + 8:'''<br />
Observation:	Activate wireless paging by dialing a direct paging number, then 47 + 66x.  When this is answered on another station using 98x the display indicates a 66x message sent from “60 Info”, and when pressing 8 to connect the exchange reset.<br />

'''Reset when MGR mail to group w/o delete two times:'''<br />
Observation:	The MGR action without “g” option (i.e. group delete) failed the second time and caused exchange reset.
Corrected.<br />

'''Cancel group call where members went down:'''<br />
Observation:	If a line board goes down during a group call, the group call resources are not released.
Corrected.<br />

'''Reset due to >128 characters in MPC port:'''<br />
Observation:	If the exchange received more than 128 characters on the MPC input before <RETURN>, the exchange would reset.
Corrected.<br />

'''Display data messages was “nationalized” unconditionally:'''<br />
Observation:	Data commands sending bytestrings to TouchLine displays was corrupted.
Correction:	In AlphaNet, display texts are transported in 8-bit international format.  When sent to a display station, the text is localized according to the language of the station by converting 8-bit characters to 7-bit, and a few country specific characters are mapped to ASCII 0-7 for the LVCD display’s RAM character generator.  This translation was performed on ALL bytestrings sent to displays, also non-text binary messages.  Translation now for selected display commands only.<br />

'''AlphaCom data protocol Q_ELEM_ADDED destroyed character in text:'''<br />
Observation:  	The first character of a mail text was corrupted in the data output message Q_ELEM_ADDED.  Error introduced in 07.10.
Corrected.<br />

'''Serial port 0 did not work on RS422:'''<br />
Observation:	Port 0 is used in slave modules, normally on RS232.  On an engineering project we found that the RS422 interface did not work.
Corrected.<br />
<br />
<br />
<br />
==AMC X07.10 (1998-08-05)==
 Release:  Bugfix, limited distribution

===Errors Corrected===
----
'''TouchLine data protocol: 4-digit dir.no fails:'''<br />
Observation:	4 digit numbers in the input commands of the TouchLine protocol does not work.
Corrected.<br />
<br />
<br />
<br />
==AMC X07.10 (1998-05-05)==
 Release:  Some new features to selected customers
 Released to selected subsidiaries for test.  
 The version was found so stable that it was allowed as a standard release for customers (the 789 X indication was not removed).
 First release within the AlphaCom/8 project. 
 Developed from AMC 07.02 - does not contain later bugfixes
 RIO/ISO 1745 problems: Same corrections as in AMC 07.03
 Use AlphaPro 07.0x - there are no new configurable features.
 NVRAM version 07.10

===Functional Changes===
----
'''TouchLine Data Protocol input commands:'''<br />
Implemented commands:
 H 	can dial any number (handled as "dial digits" internally)
 I 	is connect two stations, politely
 J 	is connect two stations, with priority
 Q 	is cancel, implemented as C-key
 X 	is re-transmit 
 ? 	is software version

'''Transfer of call request to station/group in remote exchange:'''<br />
Transfer of call request can now go to 

*- local station (as before)
*- local group
*- remote number (the remote number may be a single station or a group in the remote exchange).

This version has no automatic protection against circular transfers, so you should only use it in situations where the call hierarchy is simple (mails will be discarded after 16 jumps).  Note also that if you use parallel servers (a group of stations), and one of them transfers it’s call requests, the system is not able to find and cancel entries in all other queues when a call is answered. Transfer affects both Call Request (622, 623, 624, 625) 
and Cancel Call Request (626). (Hopcounter added to  SEND_MAIL, SEND_TXT_MAIL and CANCEL_MAIL data protocol messages)<br />

'''Answer to global Group Call:'''<br />
It’s now possible to have global Group Call with Answer (previously, only global Announcement).  Use the normal answer code for local group call, e.g. 99.
Note:  AlphaPro 07.01 don’t allow configuration of answer (“reply”) in the screen for global groups.  But answer is autoloaded TRUE, i.e. all global groups have answer.  If this causes a problem, it’s possible to disable it with the TST console.<br />

'''Feature for rebuilding global group tables: Dir.no 7879, feature 51:'''<br />
AlphaNet global groups are configured for each node in AlphaPro, so that each node knows which groups it belongs to.  All nodes broadcast this information to all other nodes so that when a global group is started, the initiating node knows which exchanges are members.  This dynamic info is stored in NVRAM, i.e. it survives a reset.   The global group membership is broadcast after exchange reset, plus once every 24 hours.<br />
During system commissioning it’s easy to come into situations where one ore more nodes are not connected when broadcasts occur, and it’s impossible to reset remote nodes.  Also, if a node is removed from a network in normal operation, the remaining exchanges must remove this node from it’s global group tables, else the “missing exchange” timer of 3 seconds always delay groupcall setup.<br />
To solve these problems a new feature 51, dir.no 7879, has been introduced.  (Not autoloaded, you must create it in AlphaPro to use it).   It erases it’s own global group tables, and broadcasts a message to all other exchanges to do the same.  Then, each node broadcasts its own membership to all others.  Each node waits 0.2 seconds times own node number before broadcasting, which means that the complete operation worst-case takes approximately 1 minute.<br />
Data protocol:  BUILD_GLOB_GRP message implemented.<br />

'''Faster conversation setup when known B-number in AlphaNet:'''<br />
When you are dialing an area code or an exchange prefix in AlphaNet, a speech connection is established, then further dialing is sent to the remote exchange.  When you release the last digit, the remote end opens up the audio path in duplex.  The user listens to the remote exchange’s connection tone.<br />
There are now several situations where the complete number in the remote exchange is known before AlphaNet set-up is started:  Remote numbers, Call Back, Answer Global Group.  In this case it’s the local exchange which waits until the user releases the last digit, and then the full AlphaNet audio path setup takes place, causing an observable longer connection tone.<br />
The AlphaNet protocol now includes the remote exchange’s local number in the initial audio path setup, thus speeding up the remote end, and reducing the need for messages being sent between the two exchanges.  The user observes faster response, and the AlphaNet itself is will behave better in high traffic situations.  <br />
The change in protocol adapts itself to nodes running older software versions.<br /><br />

===Errors Corrected===
----
'''Partial remote number on DAK does not work:'''<br />
Observation:	ER.1313: Program a partial remote number plus digit to complete the remote number on a DAK. The connection in the remote  exchange fails. Example: "4" is a partial remote number, "400" is the dir.no. of a station in the remote exchange. Programming "I4 P00" on a DAK does not work. 
Correction:	OK<br />

'''Impossible to login AlphaPro on a recently used local node:'''<br />
Observation: 	ER.1315: Connect AlphaPro cable to AlphaCom A, connect via AlphaNet to AlphaCom B.  Move cable to AlphaCom B. It is impossible to connect via AlphaNet to AlphaCom A before two hours is gone.
Correction: 	The local node stays in transparent mode for 2 hours (the 2 hour time-out applies to all connections, so if you log on to a remote node and then do nothing for two hours, you will have to log on again).  Now AlphaCom accepts new connect in this state. <br />
Known downside of fix: If two AlphaPro PC's operates in a AlphaNet, one AlphaPro may snatch a connection from the other AlphaPro.<br />

'''Remote RESET cause error message in AlphaPro:'''<br />
Observation:	ER.1292:  If you use AlphaPro to log on to a remote exchange, and then press the “RESET” button in the WACS communication program, you will get an error message "Reset failed (wp:8)" indicating that WACS has lost contact with the remote exchange.  
Correction:	The remote exchange will actually reset some 15 seconds after you push the “RESET” button in WACS.  The acknowledge message to AlphaPro was lost during the reset.   Now reset is delayed one second, to allow the acknowledge message to be returned.<br />
NOTE:  After a reset, the remote exchange assumes local AlphaPro again.  You must therefore log in again, either by stopping/starting the communication program again, or select your local exchange, then the remote exchange again in AlphaPro.  If you fail to log on again, all operations like “Send Changes” or a new “Reset” fails with various error messages, all including (wp:8) which means lost contact.<br />

'''Trace of data protocol sometimes lost messages:'''<br />
Observation: 	Data protocol trace on TST-port ("dt"-command): If messages arrives from several ports simultaneously, messages may be missed on the trace. The trace also adds significant delay to the operation of the data-protocol. 
Correction: 	Trace printout now handled by sending copy of messages to a low-priority process. This removes problems due to reentrancy and blocking of application or linklayer due to terminal printing.<br />

'''C-key after area code hangs the station for 30 seconds:'''<br />
Observation:	ER 1320: Pressing C-key immediately after having dialed a area code may sometimes cause the station to hang with a busy tone in 30 seconds.  After 30 seconds the station and the used resources are released.
Correction:	OK <br />
<br />
<br />
<br />
==AMC 07.18 (2000-05-11)==
 Release: Official, available on request

===Errors Corrected===
----
'''Exchange failed to start even with Master Flash:'''<br />
ER 1643:  Master-flash feature developed in version 07.16 to ensure AMC operation even if AMC battery fails did not work properly.  If shorted battery, AMC cycles between red indicator and HW reset.  (A stack overflow error bombs the restart-timestamp-log, so that the 4-resets during 10 last minutes does not work.) 
Corrected.  After power-up with shorted battery, AMC resets 4 times rapidly, and then comes up OK. <br />
<br />
<br />
<br />
==AMC 07.17 (2000-02-17)==
 Release: Official, available on request

===Errors Corrected===
----
'''Lockup of station in high traffic:'''<br />
ER 1630:  If a complete dir.no is dialled, and then C-key is pressed before the feature set-up is complete, the station would lock up until exchange reset.  This can occur only in very high traffic where station key-presses are buffered until AMC can process them. Corrected.<br />
<br />
<br />
<br />
==AMC 07.16 (2000-01-31)==
 Release: Official, available on request
 Improved handling of a partially corrupt NVRAM. 
 Developed for safety-critical evacuation applications. 
 These improvements also appears in AMC 08.01.
 NVRAM version 07.01. 
 The NVRAM-state format has changed, so all state-information will be lost when upgrading from 07.15 or earlier to 07.16.
 (Mail, Absence, transfer, wakeup, program, conference...)

===Functional changes===
----
'''7821 Master Flash Backup:'''<br />
The new master flash-backup feature can be used in safety-critical applications. A master flash backup is made by dialling 7821 on the Super User. The NVRAM is restored from the Master FLASH every time the exchange reset. If you use this backup option, do remember to make a backup to Master FLASH every time you change the configuration of the system. <br />
If the backup to Master FLASH has been used and the exchange makes 4 resets within 10 minutes the NVRAM-state is deleted (Mail, Absence, transfer, wakeup, program, conference...). If the directory number 7821 is missing, it can be created, using feature = 41 and Paramter-id = 2. 

'''Mail and error datastructures improved:'''<br />
The datastructures in NVRAM-state for mail and SW-errors are made less vulnerable for errors in the data. Memory pointers has been replaced by indexes which can be range-checked. The mail-liststructures is now redundant. At startup, the consistency of the mail-liststructures is checked. If inconsistent, the lists are reconstructed. TST "err" - command user interface changed

===Errors Corrected===
----
'''Reset when listing errors:'''<br />
ER. 1441:  If a error report has more than 80 characters in one of the lines, the exchange resets when it is printed on the TST port. Correced. <br />
<br />
<br />
<br />
==AMC 07.15 (1999-12-13)==
 Release: Official, available on request
 NVRAM version 07.01

===Functional changes===
----
'''Load control can be disabled:'''<br />
Loadcontrol algorithm introduced in AMC 07.11 can be disabled. (We try to find out if the loadcontrol algorithm is a bit conservative, giving too low troughput.) This option must be activated from the TST console on the master, use the “nvram” command, set .module_profile.st_profile [0].volume_level to an non-zero value. <br />
<br />
<br />
<br />
==AMC 07.14 (1999-11-11)==
 Release: Official, available on request
 NVRAM version 07.01

===Errors corrected===
----
'''RS 422 not working:'''<br />
ER 1477: Observation: Serial ports could not be used in RS422 mode. This was an error in AMC 07.11 - 07.13.
Corrected.<br />

'''Reset when releasing M-key i conference:'''<br />
ER 1478: Observation: Exchange reset when releasing M in any single-module simplex confernce. This was an error i AMC 07.13.
Corrected.<br />

'''Load NVRAM from flash backup if NVRAM is corrupted, more improvements:'''<br />
Observation: AMC with bad battery may not restart after a power off. Problem was that after the NVRAM programming was loaded from the flash, a error -message was logged before the state-part of NVRAM was autoloaded. If the state-part was corrupted, this operation could give a processor reset. 
Corrected.<br />
<br />
<br />
<br />
==AMC 07.13 (1999-06-29)==
 Release: Official, sent to selected customers
 NVRAM version 07.01

===Errors corrected===
----
'''Load NVRAM from flash backup if NVRAM is corrupted, improvements:'''<br />
If NVRAM data are loaded from flash at startup, a few checks on the NVRAM data are added to make sure that the exchange can complete the start up using these data. If the test failes, autoload is performed.<br />
<br />
<br />
<br />
==AMC 07.13 X 990617 (1999-06-17)==
 Release: Experimental, sent to selected customer
 NVRAM version 07.01

===Errors corrected===
----
'''Paging 44 does not work:'''<br />
ER 1455: Observation: Paging using 44 during conversation does not work in versions AMC 07.11 - 07.12. (This is due to a bug in the  “optimize-for-time” compiler option.) Corrected. <br />
<br />
<br />
<br />
==AMC 07.13 X 990609 (1999-06-09)==
 Release: Experimental, sent to Matmor, Israel 
 Correction for a exchange resetting daily at a hospital. (Previous fix did not work.)
 NVRAM version 07.01

===Errors corrected===
----
'''Reset related to groupcall camp-on queue on Gong:'''<br />
ER 1456: Observation: In multimodule systems having several non-overlapping groups covering more than 2 modules, the system resets in some situations. The frequency of resets is higher if the gongtone has long a duration (e.d door bell feature).
Corrected.  The problem occurs if one group call is active, and a second non-overlapping group call is in camp-on on the gong.   If the first one is answered, and the waiting group has members in 3 modules or more, connection of audio will cause reset due to timeslot inconcistency.<br />
<br />
<br />
<br />
==AMC 07.12 (1999-06-04)==
 Release: Official, available on request
 NVRAM version 07.01

===Functional changes===
----
'''Load NVRAM from flash backup if NVRAM is corrupted:'''<br />
If the magic number at the start of the NVRAM is incorrect, AMC will at start-up check the flash-backup. If the backup is valid and has the correct NVRAM-version, AMC will load the NVRAM from the flash backup instead of doing a autoload. 
It has been a few reports on that the NVRAM is autoloaded for no reason. This may be because of EMC-problems, bad battery or SW-errors, we don’t now yet. This feature should remove most of the problems if this happens. <br /><br />

===Errors corrected===
----
'''Exchange does not start if Intercard Cage data link is disabled:'''<br />
This error was introduced in AMC 07.11, in conjunction with the flow control. The error is now corrected.<br />
<br />
<br />
<br />
==AMC 07.11 X 990528 (1999-05-28)==
 Release: Experimental, sent to Matmor, Israel 
 Workaround for a exchange resetting dayly at a hospital. 
 Resets seem to come when a groupcall is started, and some data in a software module(CP) has been corrupted. 
 This fix to AMC 07.11 cleans up these data-strucures before starting groupcall. 
 Some more error-reporting is added which may assist in finding the real cause of the problem. 
 NVRAM version 07.01
<br />
<br />
<br />
==AMC 07.11 (1999-05-21)==
 Release: Official, available on request
 Load control introduced in order to prevent resets due to high traffic.
 Some optimisations enable AMC to handle more traffic. 
 Note that 07.11 is the successor of 07.09.  X07.10 was branched off a long time ago, and is now replaced by 07.60.
 NVRAM version 07.01

==Functional / non functional changes==
----
'''Load based flowcontrol introduced to prevent reset due to overload:'''<br />
The master monitors its own CPU load, if overloaded, it tells low-level drivers in all modules to slow down message passing to the master. This prevents uncontrolled buildup of queues in the master, which caused exchange reset in previous versions. The queues in the low level drivers can now fill up if traffic is too high over time. User events like key presses etc is discarded in a controlled way if the low-level queue is full. Statistics is shown in the TST- “wd_stats” command.<br />
Note: In a multi module system this requires SW upgrade of the slave-AMC’s in order to work.<br />

'''Load dependant discarding of display messages option added:'''<br />
Building and sending display-messages accounts for about 20% of the AMC CPU usage during conversation setup and disconnection. An programmable option is added, so that AMC discards ordinary display messages during period of overload. If this option is activated, the capacity to process calls is increased, but display updates is lost during periods of high traffic. This option must be activated from the TST console, use the “nvram” command, set .module_profile.st_profile [0].ext_feat = 1.

'''Recompiled AMC SW with “optimize-for-time” compiler option:'''<br />
Using the optimize option of the C-compiler increases the callhanding capasity of AMC by 30%. The EPROM code size is also reduced.<br />

'''C-press removes groupcall or audio-program, not both:'''<br />
If the station pressing C-key has access to feature 53 (disconnect groupcall) and the station is member of an active groupcall, the groupcall is disconnected, but the audio program is not disconnected. Otherwise the C-pres idle works as in earlier versions. <br />
Limitation: If a station is member of a active groupcall and it is listening to a audio program, pressing C-key twice will not remove the audio program. This makes the C-key function more useable for Morgan Stanley. Using C-key to disconnect groupcalls saves 90% on CPU usage compared to dialling a two digit directiory number to disconnect the groupcall. <br />

'''Increased timeout for M-key in conference from 10 to 108 minutes:'''<br />
Only applies to station with no-display-ABCD-interface. There is a timeout on use of M-key in order to detect a change from no-display to display station. If the timeout expires, the station-line-interface is reset. This timeout is now increased when the station is in conference. This was required by an evacuation application where conference-setup and M-key use is controlled by external equipment (Postgiro).<br />
<br />
<br />
<br />
==AMC X 07.09 (1999-03-01)==
 Release: Experimental
 To be used for debugging at a customer site (Chase London)
 NVRAM version 07.01

===Functional / non functional changes===
----
'''More error traps added:'''<br />
This version also trap MC68000 CPU  exceptions and AMX operating system errors, so that they are logged in the NVRAM error buffers.<br />
<br />
<br />
<br />
==AMC X 07.09 (1999-01-29)==
 Release: Experimental
 To be used for debugging at a customer site (Troster)
 NVRAM version 07.01

===Functional / non functional changes===
----
'''“Core-dump” before reset due to disaster-errors:'''<br />
If a disaster-type of error occurs, a dump of vital system information is printed to the TST port before reset.  Information reported is: 

*- message-headers of all messages in TST-queues
*- state-variable of all TST-FSM’s in process AICE
*- information on stack-usage of all TST-tasks
*- the number of memory buffers used
*- watch-dog statistics
*- the number of AMX-messages in queue on data-linklayer
*- the stack usage of the data-linklayer

The core-dump only appears if the TST-port is enabled (#SERVICE* - port 0, #TST ON* - port 7). The core dump information is not stored in NVRAM, a PC logging to file has to be connected when the reset occurs in order to save the information. The “core-dump” can be taken also during normal operation by typing “core” in TST.<br />

'''Two-layered Watchdog gives application one minute before reset:'''<br />
A high priority process (sync) now triggers the watchdog every 5 s. A signal KICK_DOG is propagated through most processes every 5 s (like before).  If a KICK_DOG has not got through the system during the last minute, sync calls error_amc with disaster. <br />
Result: A error message is logged in NVRAM, and core-information is printed on TST if timeout.  The time the application is allowed to be busy before reset is increased from 10s to 60s. The roundtrip delay of KICK_DOG can be printed on TST-port<br />

'''Optimised some CPU wasting:'''
11% of the AMC CPU capacity was used to scan the 6 APC-RCI’s and do some interrupt loadcontrol. This load has been reduced to 0.1%, freeing up CPU-capacity for other useful work. <br />

'''Backspace key can be used when typing commands at TST-port:'''<br />
Previously the backspace character would be buffered along with the typed characters, and fouled up the input!<br /><br />

===Error Corrected===
----
'''Reset when TST message buffer allocation failure calls error_amc/disaster:'''<br />
Observation:	Reset during high traffic. Correction:	Improved logging of reset cause. If TST runs out of message buffers, an error-message is logged in NVRAM, and core-information is dumped before reset.  (Previously reset was taken without logging the error. A stack-overflow error had also to be fixed to get the core dump out.)<br />

'''#SERVICE* worked only one time:'''<br />
Observation:	ER.1415: The #SERVICE* command only worked once after reset. Corrected:	The #SERVICE* now also works immediately, previously there was a 0-10s delay before the TST-prompt appeared. <br />

'''A stack-overflow error has been corrected, stacksize increased:'''<br />
(Module dt.c. We don’t know if this error could cause problems in AMC 07.09. ). The stack-sizes of the main process is increased by 1000 bytes, because inspection revealed that the margins was not too good. <br />

'''Reset when typing on TST-port when error-message is printed:'''<br />
Observation:	If the TST console was used interactively when a message was printed, a reset could occur. Corrected. Observe that output is skipped in this situation, but the hangup is avoided. <br />
<br />
<br />
<br />
==AMC 07.09 (1999-01-13)==
 Release: Official, standard for production
 NVRAM version 07.01

===Functional Changes===
----
'''C-press disconnect own station from ongoing groupcall:'''<br />
This is an extension to the disconnect-from-groupcall feature introduced in AMC 07.08.  C-press in Idle disconnects ongoing groupcall if feature 53 is included in the ClassOfService for this station. (C-pres in conversation just disconnects the conversation as before).  Feature 53 is removed from autoload of class of service 1. Directory number 7884 autoloaded for feature 53.<br />

'''Check if M-key pressed when entering program conference:'''<br />
If the M-key is pressed when a station is included into a conference, the station starts feeding the conference. Requested for an evacuation application where conference-setup and M-key use is controlled by external equipment. <br />

'''Fast  first-time startup of external device (RIO):'''<br />
Additional improvements to the fast RIO startup algorithm in AMC 07.08: The AlphaCom can detect a new RIO added to the bus within a few seconds. (Requires RIO 02.10 SW). <br /> 
Technical description: AMC now polls devices programmed in NVRAM all the time, also when AMC thinks the RIO is missing. This allows the RIO to inform AMC of its presence. (Requires ack-timeout to be set to 200ms or lower.)<br />

'''TST Warnings:'''<br />
Added some “ignore-statements” to suppress a number of uninteresting TST-warnings. More interesting information may be available if future problems has to be debugged.  <br /><br />

===Error Corrected===
----
'''Partial remote number on DAK does not work towards TouchLine:'''<br />
Observation:	Program a partial remote number plus digit to complete the remote number on a DAK.  The remote exchange is a TouchLine. The connection in the remote  exchange fails. Example: "4" is a partial remote number, "400" is the dir.no. of a station in the remote exchange.  Programming "I4 P00" on a DAK does not work. 
Correction:	OK (This is really a bug in TouchLine TAC, but AMC 07.09 goes around the problem)<br />

'''Reset when global groupcall is overridden by priority:'''
Observation:	E.R. 1418: A global groupcall is started from a multimodule exchange.  The  groupcall is disconnected by a higher priority groupcall in the exchange of  the initiator.  This leads to exchange reset. 
Correction:	OK <br />

'''Reset when busy override of digit 0 pressed in conversation:'''<br />
Observation:	E.R. 1420: Station A is in conversation with station B. Station A is pressing digit 0 (probably for mic-mute).  Station C dials to A, gets busy and uses busy-override (MM or 58).  The conversation is not set up, B is not disconnected.  If station C presses C-key, subsequent conversations to or from A will fail. If station C presses digit 5  before pressing C-key, the exchange will reset after 10 seconds. 
Correction:	OK <br />

'''Leap year handling wrong:'''<br />
Observation:	E.R. 1421: 29 days in February every year if you adjust clock sometime between 1 Jan and 28 Feb. Otherwise 28 days in February, even if is actually a leap year. A consequence is that wakeup (7883) does not work from evening Feb. 28 to morning Mars 1. 
Correction:	OK  (After SW upgrade you have to set the AlphaCom clock once to set the leap year counter right. Use AlphaPro or station (785) to set the clock.) <br />
<br />
<br />
<br />
==AMC 07.08 (1998-12-11)==
 Release: Official, not standard
 NVRAM version 07.01

===Functional Changes===
----
'''Feature added: Disconnect own station from ongoing groupcall:'''<br />
To use this feature a directory number with feature number 53 has to be programmed from AlphaPro.  If you dial this number on a station which is receiving a groupcall, the sound from the groupcall will disappear from this station. Limitations:

*The station will not be free to receive a new groupcall before the first groupcall is disconnected totally.
*EVENT/RCO control is not turned off before the groupcall is disconnected totally.

'''AlphaNet audio line test can be switched off:'''<br />
AlphaNet audio lines are tested by sending 500 Hz tones periodically.  On some lines, e.g. low-bandwidth MUX lines, the test can fail for various reasons.  The test can be switched off by programming NVRAM on TST-port.  <br />
Set ex_profile.timeouts.iech_test_timeout = 0xffff.<br />  
Testing of all AlphaNet lines from that exchange is then turned off.<br />

'''Fast reset of external device (RIO):'''<br />
If AMC 07.08 is used together with RIO 02.10 SW, the AlphaCom can detect a RIO reset and restore the RIO state within a few seconds. Earlier versions of AMC has to rely on a several minute timeout to detect reset of RIO.<br /><br />

===Error Corrected===
----
'''Partial remote number on DAK does not work:'''<br />
Observation:	ER.1313: Program a partial remote number plus digit to complete the remote number on a DAK. The connection in the remote  exchange fails. Example: "4" is a partial remote number, "400" is the dir.no. of a station in the remote exchange. Programming "I4 P00" on a DAK does not work. 
Correction:	OK<br />

'''Improved errorhandling on ISO1745 linklayer (RIO communication):'''<br />
The ISO1745 linklayer (used for RIO communication) implementation  is improved to handle noise on the databus in a better way. The probability of  loosing messages is reduced. What has been done: 

*Extra characters received due to noise are ignored.
*Retransmits message if a select sequence fails.
*Faster timeout on missing RIO activity (previously 10 seconds, now 1 second).

RIO 02.10 SW should be used to improve the RIO error-handling. <br />
<br />
<br />
<br />
==AMC 07.07 (1998-10-20)==
 Release:  Official, not standard
 Required for a single customer (Chase Manhattan) 
 NVRAM version 07.01

===Error Corrected===
----
'''Reset when answering CallBack due to inconsistent USER / STATION:'''<br />
Observation:	The exchange will reset if someone dials 70 +8 to answer a CallBack from a station which has a user which in turn has no station (i.e. inconsistent internal tables).
Correction:	The software does a consistency check, and the Callback will be aborted in this case (must be deleted from the display using 70 + 0).<br />
NOTE:   AlphaPro does not handle the reference from Station to User correctly if one changes (or deletes) the physical number of a User. <br />

'''Known Bugs and/or Workarounds:'''<br />
'''DAK string playback unreliable in high traffic:'''<br />
Observation:	Some long DAK strings, e.g. to activate Call Forwarding, may fail during high traffic.<br />

'''Multiple RCI/Event actions:  action after MST/MGR fails:'''<br />
Observation:	If the action string contains several commands (separated with semicolon or <RETURN>), commands following an missing optional parameter will not be performed  (MST and MGR).  
Typical example is error reporting where you want to do the traditional mail to station followed by mail to pager or setting an RCO.<br />
Workaround:	Re-arrange the sequence if possible, moving the action with optional parameter last.  Alternatively, always use an extra <SPACE> between the last required parameter and the semicolon/<RETURN>.<br />

'''Reset when many stations of type “Interguard”:'''<br />
If a 07.0x version AMC is configured with AlphaPro 07.20 or newer, it’s possible to select station types unknown to AMC 07.0x, e.g. “Interguard”.  The exchange will reset after initialising 6 such stations (green leds comes on briefly, then back to red). (The reset is caused by too much internal error logging.)<br />
Workaround:  	Use “master station” station type!<br />
<br />
<br />
<br />
==AMC 07.06 (1998-08-31)==
 Release:  Official, not standard
 Required for a single customer (Postgiro) 
 NVRAM version 07.01

===Functional Changes===
----
'''Alternative signaling of fire alarm & early warning:'''<br />
The Eltek fire alarm uses a different approach for alarm priority compared to Autronica.  Variant implemented in the ESPA 4.4.4 interface to fire alarm.<br />
<br />
<br />
<br />
==AMC 07.05 (1998-08-28)==
 Release:  Standard
 Released for standard production.
 Various corrections from AMC 07.04.  
 NVRAM version 07.01

===Errors Corrected===
----
'''Exchange unstable when AlphaPro send directory numbers:'''<br />
Observation:	If you delete a lot of numbers, AMC would very likely reset if someone was dialing numbers during the download of new number list from AlphaPro.. 
If you do not close down WACS between doing number changes there was a percentage chance that AMC would reset if some one was dialing numbers during the download.  It was a fraction of a promilla chance that AMC would never come up again without a reset.<br />
Correction:	AMC has a separate data structure which is used for fast lookup of dir.nos.  This points into the dir.no table, and will be re-built when the complete dir.no table has been transferred.  

*- Possibility for reset during number analysis now eliminated.
*- Possibility for reset in startup check-routine eliminated.

Note:  Numbers may still be unavailable during the download.<br />

'''AGA line test blocked in high node:'''<br />
Observation:	It's not possible to use the 7872 AGA line test from the node with higher node number if the AlphaNet audio line has been busymarked by normal line test.
Corrected.     <br />
<br />
<br />
<br />
==AMC 07.04==
 Release: Never released
 ... due to errors found at customer cite and corrected in AMC 07.05
 Various corrections from AMC 07.03.  
 NVRAM version 07.01

===Functional change===
'''Debug feature for stopping distribution of NVRAM to slave:'''<br />
(module_profile.st_profile[0].fea_rem_wanted > 0) disables NVRAM distribution from master to slave. <br />

'''Debug feature to stop the duplex algorithm’s consonant evaluation:'''<br />
Duplex algorithm: The high frequency evaluation can be switched off completely by setting module_profile.st_profile[0].prog_seq_no = 1. It is included in this version and later because some have difficulties with duplex switching.<br /><br />

===Errors Corrected===
----
'''Paging with freetext from RCI lost the text:'''<br />
Observation:	Sending freetext paging from RCI using the MPP action started paging, but the text did not appear in the pager’s display.
Corrected.<br />

'''First station display line continuously overwritten:'''<br />
Observation:	Occurs after busy-notify/call-request-notify received while digit pressed in menu:  The first display line of that station is continuously overwritten by the first or last menu-element.  Requires exchange reset to recover.
Corrected.     <br />

'''Reset when answering a mail from a station without directory number:'''<br />
Observation:	Answering a mail from a station without dir.no by 70 +8 causes an exchange reset.
Correction: 	Call-back to station without dir.no. within a exchange now works. Call-back to other exchange in AlphaNet still fails, but is disconnected orderly.<br />

'''Reset when Hot-pluggout of ASLT with a group-call member:'''<br />
Observation:	If you hot-pluggout an ASLT where some stations are members of an ongoing group call, the exchange will reset when the group call is disconnected. 
Correction: 	Immediately disconnect groupcall if a board with a groupcall member is removed. <br />

'''Reset when M-press without M-release in conversation:'''<br />
Observation:	The exchange is reset if several M-press without M-release. Can be a problem if M-press is inserted via data protocol.
Correction: 	The exchange will now generate a warning rather than a reset when this situation occurs.<br />
<br />
<br />
<br />
==AMC 07.03 (1998-04-14)==
 Release:  Standard, RIO performance improvements
 Released for standard production.
 Mainly corrections related to poor RIO performance. Developed from AMC 07.02
 For best results, use AMC 07.03 or newer with RIO 02.03 or newer.
 NVRAM version 07.01

===Errors Corrected===
----
'''RIO problems:  Delayed and/or lost actions:'''<br />
Observations:  Controlling pins in a RIO on a multi-drop port frequently was delayed, and worst-case lost.  
Corrections:  Caused by several problems, both in AMC and the RIO unit.

*1) ISO1745 linklayer: Fixed problem with premature time-outs (crossing

      time-out events and timer restarting)

*2) ISO1745 linklayer: Immediately polling of device when outgoing message

      to that device has been acknowledged. Improves performance.

*3) 	ISO1745 linklayer: 40ms delay between poll/select and broadcast message.

'''Station deadlock:'''<br />
Observation:  ER.1302: Station was deadlocked if B pressed C-key during connection tone
Correction:  OK.<br />
<br />
<br />
<br />
==AMC 07.02 (1998-03-27)==
 Release:  Standard, multi-module bug-fix
 Released for standard production. Developed from AMC 07.01.
 Bugfix due to reset problems in multi-module exchanges.  
 See also bugfix list for 06.05 which were done in parallel with this release.
 NVRAM version 07.01

===Functional change===
----
'''Feature to update slave NVRAM:'''<br />
Engineering feature for sending 10 bytes of NVRAM update broadcast to the slaves for a given offset.  Used to activate debug modes in slave software.<br />

'''#SERVICE* command activates TST Console on port 0:'''<br />
It’s now possible to run the TST console on port 0 (the AlphaPro port).  TST must be stopped by typing “gbye” to allow AlphaPro to use the port afterwards.  This mode is not stored in NVRAM.<br /><br />

===Errors Corrected===
----
'''Reset of slave 66 after 29 hours:'''<br />
Observation:	ER 1297:  Reset of the first slave module with minimum 29 hours in between.
Correction:	Linklayer in slave 66 may get stuck in a 2 or 3 module exchange when the master looks for a 3. or 4. slave.  Corrected.<br />

'''Reset when  MST with source station 0:'''<br />
Observation:  	The action command MST which sends a mail to a station, caused reset if the mail source was set to 0.
Correction:  	OK. <br />

'''Data protocol:  Mail problem:'''<br />
Corrected error in formatting of free-text parameter of the EXT_MAIL message.<br />
<br />
<br />
<br />
==AMC 07.01 (1998-02-27)==
 Release:  Standard. first 07.xx version
 Released for standard production.
 NVRAM version 07.01
<br />
<br />
<br />
==AMC 07.00 (1998-02-16)==
 Release:  Internal test
 Version after acceptance test corrections.  Never reached production due to bugs found in further testing.
<br />
<br />
<br />
==AMC 06.06 (1998-04-02)==
 Release:  Non-standard, RIO performance improvements
 Official release, but not released for standard production.
 RIO fix offered to 06.xx installations which does not want to upgrade to 07.03.  
 For best results, upgrade RIO to 02.03 also.
 NVRAM version 06.00

===Errors Corrected===
'''RIO problems:  Delayed and/or lost actions:'''<br />
Observations:  Controlling pins in a RIO on a multi-drop port frequently was delayed, and worst-case lost.  
Corrections:  Caused by several problems, both in AMC and the RIO unit.

*1) ISO1745 linklayer: Fixed problem with premature time-outs (crossing

      time-out events and timer restarting)

*2) ISO1745 linklayer: Immediately polling of device when outgoing message

      to that device has been acknowledged. Improves performance.

*3) 	ISO1745 linklayer: 40ms delay between poll/select and broadcast message.

'''Large group operations caused reset:'''<br />
Observation:  	Actions like MGR or Call Request to groups, activated repeatedly, would cause reset.
Correction:	Caused by watchdog, as software became 100% busy too long.  Dedicated watchdog kick now used during group operations.<br />
<br />
<br />
<br />
==AMC 06.05 (1998-03-06)==
 Release:  Standard, bug-fix multi-module, duplex, etc.
 Official release, but not in production (07.01 in production).
 Bugfix offered to 06.xx installations which does not want to upgrade to 07.0x.  
 All errors corrected here are also included in 07.02 and newer except for the noted debug options.
 NVRAM version 06.00

===Functional changes===
----
'''Debug feature for tracing data links on slaves:'''<br />
(module_profile.st_profile[0].voice_help_wanted > 0) enables trace to (port + 4)<br />

'''Engineering feature for sending NVRAM to the slaves:'''<br />
10 bytes of NVRAM sent to slave as broadcast.  NVRAM in slaves are updated slowly in the background.  If a NVRAM change is needed quickly for test, use this feature.<br />

'''Debug feature for stopping distribution of NVRAM to slave:'''<br />
(module_profile.st_profile[0].fea_rem_wanted > 0) disables NVRAM distribution from master to slave.  Not to be included in AMC 07.xx.<br />

'''Debug feature to stop the duplex algorithm’s consonant evaluation:'''<br />
Duplex algorithm: The high frequency evaluation can be switched off completely by setting module_profile.st_profile[0].prog_seq_no = 1.  Not included in 07.xx.<br /><br />

===Errors Corrected===
----
'''Station blocked if it goes down while dialing during connection:'''<br />
Observation:  	It’s possible to block a station forever if the station is dialing digit during connection, and then the station is disconnected, i.e. considered by the exchange to go out of operation.  When the station becomes OK again. it’s blocked until exchange reset.<br />

'''Slave module reset after > 29 hours:'''<br />
Observation:	ER 1297: In an exchange with 2 or 3 modules (not 4 modules), reset of the slave 66 occurs every few days.
Correction:	Linklayer in slave 66 may get stuck after a #SET MODE to non connected port (the master is looking for slave 3 and/or 4 periodically). Will lead to reset of the slave 29 hours later.  Corrected.<br />

'''TST warning:  TEXT_TOO_LONG:'''<br />
Observation:	Mail operations to station in AlphaComs loaded with Hebrew texts causes TST warning.
Correction:	Use a larger buffer.<br />

'''Duplex blocked by high frequency signals with low amplitude:'''<br />
Observation: 	Audio signals over 3 kHz, e.g. a PC fan, would lock the speech direction away from the source.  Not possible to turn the direction from the other side, even when shouting into the microphone.<br />
Correction:	Improved duplex algorithm: There is a stability test which let the current speaker continue if the absolute value of high frequency is high, which avoids fast turns in the middle of words.  It's the reason why turning against > 3000 Hz low amplitude is impossible, and this test has now been extended to also test for a timer that is started whenever there is large change in high frequencies.  This means that a continuous high frequency tone will reject channel turn only for some 100 milliseconds, causing duplex to feel "slow" but definitively working. <br />
<br />
<br />
<br />
==AMC X06.04 (1998-02-19)==
 Release:  Test only, slave module debug
 Debug SW with a echo function to provide trace of the ICC protocol. 
 Issued for Israel, ref problem report 980217-2.
 NVRAM version 06.00
<br />
<br />
<br />
==AMC X06.03 (1998-01-29)==
 Release:  Test only, ATLB board demo
 Special version with minor changes to be able to run a prototype ATLB in the exchange (very little ATLB functionality).  
 Ref. SCF.
 NVRAM version 06.00
<br />
<br />
<br />
==AMC 06.02 (1997-09-18)==
 Release:  Standard, general bug-fix 
 Official release, but not in standard production.
 NVRAM version 06.00

===Functional changes===
----
'''Call Forwarding affects Absent marking:'''<br />
Activation of Call Forwarding turns off absent marking, and Absent marking turns off Call Forwarding.  (Previously you could have both simultaneously).<br />

'''Call Forwarding not linked with Call Request Forwarding:'''<br />
Activation of Call Forwarding or Follow Me does not activate Transfer of Call Request.  (Previously linked, which made sense for guard stations using call request, but was confusing for all other stations)<br /><br />

===Errors Corrected===
----
'''Slaves resets easily due to communication noise:'''<br />
Observation:  	Slaves operating far from the master resets due to communication line noise.
Correction:	Increased number of retransmissions from 2 to 7.<br />

'''Mail from PNCI not removed:'''<br />
Observation:	If a dial-in PNCI conversation leaves mail at a station, it’s not possible to delete it by calling back.  The mail must be deleted manually with 70+0.
Correction: 	Mail is now removed when calling back to the PNCI.<br />

'''Pager blocked by simultaneous use of allocated answer codes:'''<br />
Observation:	The pager variant which allocates an answer code, which in turn in displayed in the pager, queued multiple calls for the duration of the answer timer.  In a hospital situation, this could block multiple emergency calls.
Correction:	No restriction on use of allocated answer codes.<br />

'''Busy override against busy and private bypassed busy only:'''<br />
Observation:	When using busy override against a station in private mode, which was busy right now, canceled the ongoing conversation, and then established a new conversation with private ringing.  Override had to be used a second time to break in completely.
Correction:	Connection now established directly.<br />

'''ASVP in slave module causes module reset:'''<br />
Observation:  	If a slave module contains an ASVP board, it resets continuously.
Correction:	ASVP now allowed, but it is ignored if the module comes up as autonomous slave (stand-alone mode if the master module fails).<br />

'''Several group disconnections from simplex conference caused reset:'''<br />
Observation:	If one station was member of several groups, and group disconnection (75) was performed when the station had already left a simplex program conference, the exchange was reset.
Correction:	OK.<br />

'''Feature reminder not correct for mails from RCI or data protocol:'''<br />
Observation:	A station’s feature reminder was not switched on correctly for mails coming from either RCI, or the data protocol.  Correspondingly, feature reminder was not removed if a mail was removed via data protocol.<br />

'''Data protocol:  DIAL_DIGITS ignored after first digit:'''<br />
Observation:	If multiple DIAL_DIGIT messages was received, only the first was handled correctly.
Correction:	DIAL_DIGITS from data protocol are now handled in all station states.<br />
<br />
<br />
<br />
==AMC 06.01 (1997-07-31)==
 Release:  Standard, bug-fix RIO & data protocol in general
 Official release.
 Corrections for Rikshospitalet (voice evacuation application).
 NVRAM version 06.00

===Errors Corrected===
----
'''RIO problems:  Delayed and/or lost actions:'''<br />
Observation:  	RIO applications sometimes slow or missing.
Correction:  	ISO 1745 link layer time-out problem; time-out sometimes occurred immediately.  Corrected.<br />

'''Program off with C-key has no Class Of Service:'''<br />
Observation:  	Blocking program off (dir.no 800) with COS has no effect on C-key, i.e. user can still remove program.
Correction:	COS now evaluated in C-key handling.<br />

'''Data protocol:  Minor corrections:'''<br />

*- Data protocol analysis of parameter 5 of SEND_MAIL corrected
*- Data protocol analysis, allow empty dir.no. in object-ref

<br />
<br />
<br />
==AMC 06.00 (1997-06-20)==
 Release:  Standard, featuring AlphaNet and TouchLine phase-out
 Official major release, the results of two parallel development projects:
 - AlphaCom/4, which was a general feature upgrade (TouchLine phase-out)
 - AlphaCom/5, which introduced AlphaNet for exchange interconnection
 NVRAM version 06.00

===Functional Changes===
----
'''Conversation without tones:'''<br />
The whole exchange can be set in a mode where conversations are set up and cancelled without tones.
This must be programmed in NVRAM using the TST console:<br />
.ex_profile.timeouts.th_conn_length = 0 <br />
.ex_profile.timeouts.th_disc_length = 0 <br />
<br />
<br />
<br />
==AMC 05.10 (1997-03-18)==
 Release:  Standard, RIO improvement 
 Distributed only to a few customers with RCO problems.
 NOTE:  This release does NOT contain the improved duplex from AMC 05.09!
 NVRAM format:  05.07.

===Errors corrected===
----
'''RIO actions slow/lost:'''<br />
Observation:	Both RCIs and RCOs in RIOs were delayed or lost sometimes.
Correction:

*- Retransmit select on ISO1745
*- Start Missing-Ack timer for SET_RCO-messages

<br />
<br />
<br />
==AMC 05.09 (1997-03-18)==
 Release:  Standard, duplex & RCI improvement 
 Distributed only to a few customers.
 NVRAM format:  05.07.

===Errors corrected===
----
'''Various duplex problems:'''<br />
Observation:	Duplex problems:

*- Difficult to turn after conversation start
*- Turn in the middle of words
*- Some background noise was impossible to turn against

Correction:	Adjusted startup values, improved turn requirements, improved consonant detection part.  Ref SCL / Chase Manhatten.<br />

'''Rapid changes in RCIs disturbed exchange operation:'''<br />
Observation:	RCIs starting/stopping group operations (alarms) caused problems if the inputs changed too fast.
Correction:	Made more robust to handle high RCI traffic.<br />

'''Broadcasts on ISO protocol missing when high traffic:'''<br />
Observation:	During high traffic, some broadcast messages were destroyed/lost. (RE-970303-1 AND 970303-3)
Correction:	Added flow control for outgoing ISO1745 broadcasts.<br />
<br />
<br />
<br />
==AMC 05.08 (1997-02-03)==
 Release:  Bugfix conferences, multi module 
 NVRAM format:  05.07.

===Functional changes===
----
'''Conference survives reset:'''
Simplex program conference is now restored after reset.<br /><br />

===Errors corrected===
----
'''Conference destroyed by hot plugging of ASLT:'''<br />
Observation:  	Conference does not handle that an ASLT board with a conference member is plugged out/in while the exchange is running.
Correction: 	OK.<br />

'''Slave module malfunctions occasionally:'''<br />
Observation:	Operation disturbed for stations in slave module.
Correction:	Memory over-write problem in ICC data protocol corrected.<br />
<br />
<br />
<br />
==AMC 05.07 (1996-10-12)==
 Release:  Bugfix data protocol, sound problems
<br />
<br />
<br />
==AMC 05.06 (1996-09-25)==
 Release:  Standard, selected installations
<br />
<br />
<br />
==AMC 05.01 (1996-07-15) - 05.05 (1996-09-16)==
 Release:  System test releases & bug-fixes
<br />
<br />
<br />
==AMC 05.00 (1996-06-18)==
 Release:  Test, featuring multi-module exchange
 Test release, the results of the development project AlphaCom/3, 
 which introduced a larger, multi-module exchange.
 NVRAM version 05.00
<br />
<br />
<br />
=Hardware Versions=
Only one version used.<br />

==Known problems==
'''Reset due to old battery:'''<br />
Observation:  The exchange resets with weeks in between, then more and more often until it will not start at all.
The problem is due to the battery charging from 15V.  The battery + signal is connected to a power supervisor/reset IC, and old batteries load the charging circuit so little that the battery + signal goes over +5V, and then the power supervisor IC activates reset (undocumented property of the IC).<br />
Workaround:  Solder a Zener diode 4.3 V over the battery.  Observe that this solution corrects the reset problem, but the battery may discharge somewhat faster in storage (no problem in exchange).<br />

'''AlphaCom M - Temperature Alarm:'''<br />
The series 1 AlphaCom Ms has a temperature sensor which is not calibrated properly.  The exchange will generate a Temperature Alarm too early.
Workaround:  Remove programming in AlphaPro for RCI 8 which is the temperature alarm.<br />