Alarm panels and how they dial DTMF digits - pots migration from DMS to Metaswitch issue
Matthew Yaklin wrote:
An example would be the fire alarm panel should dial 877-386-4923. Yet we collect the digits on the Metaswitch as 877-38923. That is it. The 6 and 4 are missing
...and then later wrote...
[...] and no matter what we tried those two digits in my previous example were always missing.
You wrote this here, but elsewhere what you wrote seems to suggest the symptoms seem more "random" than this (sometimes all digits are correctly decoded and passed on, sometimes different digits or different numbers of digits are missing, etc.). If the mode of failure really is as consistent as you suggest here, though (or even if it happens the majority of time), have you considered recording the audio coming out of one of the fire alarm panels while it's trying to dial, so that you can take a spare Adtran, put it on the bench, configure it identically to one that you have in the field, and feed it a recording of the fire alarm panel dialing out, in order to reproduce the issue? If you can repro outside of production, then you can test & iterate much more quickly (as well as at your own convenience) than when you are forced to test on an actual customer installation. -- Nathan
Matthew Yaklin wrote:
An example would be the fire alarm panel should dial 877-386-4923. Yet we collect the digits on the Metaswitch as 877-38923. That is it. The 6 and 4 are missing
...and then later wrote...
[...] and no matter what we tried those two digits in my previous example were always missing.
You wrote this here, but elsewhere what you wrote seems to suggest the symptoms seem more "random" than this (sometimes all digits are correctly decoded and passed on, sometimes different digits or different numbers of digits are missing, etc.). ----- Since I have been working with several different pots lines to several different alarm panels they all have a similar problem of collecting DTMF digits reliably before any SIP is involved. In the case of the one I was working today (gr303 sub) I can pull up all of its call attempts for the last two days in SAS and yes.. it was always missing the same two digits when it dialed a specific number. Very reliable failure situation. ----- On other pots lines the failure would be more random. The common denominator is that if you looked at what it dialed you could see the pattern of digits missing. Perhaps in one call example it would be missing 3 digits. On the next it would dial the same number and you would be missing 4 digits. But the same number was attempted to be dialed each time. If you knew the number it was supposed to dial to reach the central office you can see the pattern which was not obvious to us at first because we had to ask the alarm panel guy what it was supposed to dial. They contain multiple numbers in most cases. If the mode of failure really is as consistent as you suggest here, though (or even if it happens the majority of time), have you considered recording the audio coming out of one of the fire alarm panels while it's trying to dial, so that you can take a spare Adtran, put it on the bench, configure it identically to one that you have in the field, and feed it a recording of the fire alarm panel dialing out, in order to reproduce the issue? If you can repro outside of production, then you can test & iterate much more quickly (as well as at your own convenience) than when you are forced to test on an actual customer installation. ----- I have not considered that. I work remotely with all of this gear and I would have to guide a tech to do such a thing. I don't think a butt set and a tape recorder would be sufficient to record the audio which is about all I could possibly scratch up without spending money. I will freely admit I never had to go to such lengths to debug this type of problem. I was hoping someone came across this problem before and would say "hey Matt, alarm panels have this setting called XYZ. You need to change it to this. Or slow down the dialing settings which is available in every alarm panel I touched". -- Nathan
On 2/13/23 17:46, Nathan Anderson via VoiceOps wrote:
You wrote this here, but elsewhere what you wrote seems to suggest the symptoms seem more "random" than this (sometimes all digits are correctly decoded and passed on, sometimes different digits or different numbers of digits are missing, etc.).
If the mode of failure really is as consistent as you suggest here, though (or even if it happens the majority of time), have you considered recording the audio coming out of one of the fire alarm panels while it's trying to dial, so that you can take a spare Adtran, put it on the bench, configure it identically to one that you have in the field, and feed it a recording of the fire alarm panel dialing out, in order to reproduce the issue? If you can repro outside of production, then you can test & iterate much more quickly (as well as at your own convenience) than when you are forced to test on an actual customer installation.
I'll second this and add that not only would this give you an opportunity to bench test things, but it would also let you easily rule out certain settings like impedance and levels since you can quickly vary them on both ends. Just seeing if it does or does not reproduce reliably based on playback in both the original and lab environment would be useful. Another thing nobody's mentioned is trying to get one of these fire panels for yourself. If they're as old as you're implying, they may well be readily available cheap on the secondary market. If you can get one, not only can you lab IT up, too, but you can also get your hands on all the various configurables it may or may not have that your customer is being less than cooperative with. -- Brandon Martin Mothic Technologies 317-565-1357 x7000
participants (3)
-
lists.voiceops＠monmotha.net -
myaklin＠firstlight.net -
nathana＠fsr.com