[Asterisk-Users] Asterisk Maintenance

[Jim Van Meggelen]

asterisk-users-bounces at lists.digium.com wrote:
> Chris Glover wrote:
>> On Wed, 8 Dec 2004, Michael Welter wrote:
>> 
>>> I went on a service call yesterday to find an asterisk system with a
>>> T100P card on a Qwest PRI and a TDM40B card connected to fax
>>> machines. The TDM40B LEDs were not lit, and the system did not
>>> respond to keyboard input. However, calls were being processed for
>>> the PRI and 7960 phones. 
>>> 
>>> I replaced the TDM40B card with a new one, and the system now seems
>>> to be ok. But I'm wondering, why would the LEDs go off?  Why would
>>> keyboard input fail? 
>>> 
>> 
>> Hi,
>> 
>> Cant help with the TDM card, but I have seen PCs get affected by
>> mains bumps. I used to have a PC where the keyboard controller would
>> crash but I was still able to telnet into the PC and reset it.
>> 
>> Check the mains supply, there might be something with a big inductive
>> load, that causes some ripples that are ending up on the DC outputs
>> of your PCs PSU. 
>> 
>> HTH
>> 
>> Chris
>> 
> This was a situation where I had to recover the root password, so I
> wasn't able to ssh into the box.  I had to power-down the
> box, recover
> the password, and then reboot.
> 
> Also, the box is on a UPS, so I'm assuming the AC power is generated
> from the battery and is a perfect sine wave.  

That is probably not a safe assumption. The unfortunate fact is: all a
UPS can be certain to do is provide battery back up. It MIGHT do more;
it SHOULD do more; but to qualify as a UPS, it only has to provide
battery back up. If clean power is the goal, what you need is a
_power-conditioned_ UPS, or just a power conditioner (with no UPS).

I have seen UPS units which actually _add_ noise to the circuit, and,
when running on battery, they actually provide a reversing-polarity DC
wave, stepping from DC+110V to 0V to DC-110V, sixty times per second:

+110V ______        ______        ______          
            |      |      |      |      |      |  
            |      |      |      |      |      |  
            |      |      |      |      |      |  
0V----------|------|------|------|------|------|  
            |      |      |      |      |      |  
            |      |      |      |      |      |  
            |      |      |      |      |      |  
-110V       |______|      |______|      |______|  


Another term I've seen used is pseudo-sine wave, which does something
similar to the above:

+110V ___                ___                ___                 
         |              |   |              |   |              | 
+55V     |__          __|   |__          __|   |__          __| 
            |        |         |        |         |        |    
0V----------|--------|---------|--------|---------|--------|----
            |__    __|         |__    __|         |__    __|    
-55V           |  |               |  |               |  |       
               |__|               |__|               |__|       
-110V

Either one is a non-sinusoidal AC circuit with really nasty harmonic
issues.

Some people say the quality of the AC doesn't matter because the
switching-type power supplies in modern computers are immune to noise.
But I believe in the garbage in garbage out theory, so I always make
sure to pay a bit more for the conditioned UPS units. OneAC and Powervar
have some decent ones, and APC seems to keep busy (stay away from their
Back-UPS line - I had them in the lab for analysis a few years ago and
they pump out some really nasty looking AC. The Smart-UPS line is OK,
though).

> But that begs the
> question:  does a UPS system connect the mains to the output,
> or is the
> input power used to charge the battery, and the battery used
> to generate
> the output power?

It seems to be common these days to have a relay that switches you to
battery when the AC fails. This is supposed to happen fast enough that
the caps in your power supply won't drain before the batteries kick in.
Kinda disturbing in a sense, but this is very common. I think you're
going to tend to pay a lot of money for an online UPS system.

Personally, I think what's more important to be aware of is power
conditioning. It'll give you a completely re-generated, clean AC signal
and ground reference. 

These folks have some thought-provoking wite papers:
http://www.powervar.com/english/research/treference.asp

What I've learned to do when buying a UPS is watch out for terms that
some manufactuers use to lure you into thinking their UPS does more than
it really does. If you want properly conditioned AC, terms like
"protection" should be viewed with suspicion. Not that protection is a
bad thing, but if they avoid the word "conditioned" it's for a reason -
they ain't providing it.

We tend to take power and grounding for granted. We shouldn't.

I have often wondered if some of the problems folks have with the TDM
cards could be corrected with some power quality engineering. It's one
of those things that provides measureable benefits if you have enough
statistics, but it's hard to sell prevention. I have been asked
questions like "why did you make us spend thousands of dollars cleaning
up the power? This system hasn't had any of the problems you said we
might. What a waste of money!"

A very strange science, is power and grounding. I've seen guys with
years and years of experience, all experts, have huge debates over the
proper way to ground a system.