[VoIP] Berry 314A trunk test set

[Paul Wills]

It's been a long time since I messed with it but, if you trace from the 
output inwards, it's pretty easy to see where the tones come from.  The 
problem I found was a simple 40xx series CMOS chip that wasn't doing what it 
should.  It was easy enough to see with a logic probe, ie: a changing input 
without a corresponding output.  A socket and replacement IC was all it took 
to put things back in order.

>From what I hear, this seems to be a common problem with the 314A.  I ended 
up with a second unit that had some sort of "addition" to the power supply 
that might have been put there to deal with certain power surges that were 
taking out the ICs.  I did "replicate" the modification to the repaired set 
so we'll see what happens.

Since they seem to go relatively cheap on the "e" place, it might be wise to 
get a second one so you can swap cards around to help narrow down any 
trouble you find to one card.

I attached two photos of the "mod" for Jerry's benefit although I don't 
believe that it will make it over the VOIP list server.

PDW




----- Original Message ----- 
From: "Jerry Petrizze" <t2600 at sbcglobal.net>
To: "Voice Over IP Tandem for Analog Switches" <voip at ckts.info>
Sent: Sunday, November 04, 2007 4:51 PM
Subject: Re: [VoIP] Berry 314A trunk test set


>I feel the same way, why the beat note?  But if there is two parallel
> processes going on which you would expect,  is it possible that it is
> generating the same tone on both channels with just a slight time shift
> enough to create the beat.?  Beats me!!!
> Jerry
>
> ----- Original Message ----- 
> From: "Steph Kerman" <stfkerman at jps.net>
> To: "Voice Over IP Tandem for Analog Switches" <voip at ckts.info>
> Sent: Sunday, November 04, 2007 1:38 PM
> Subject: Re: [VoIP] Berry 314A trunk test set
>
>
>> Very hard to say what the problem is without seeing a schematic.  Some
>> time back Paul doped out some superficial aspects of how it generates
>> tones.  IIRC it uses a digital technique based on parallel output shift
>> registers with weighted resistor values to perform a D/A function.
>> Presumably there would have to be 2 such implementations, with analog
>> summing otherwise the sample values would have to be summed.... probably
>> beyond what they would attempt with a 6800 family CPU.
>>