asterisk users - Mar 2005 - Major problems with TDM400 and specific telephones: suggestions?

> Attached to the bottom of this e-mail is an edited version of an e-mail I 
> originally wrote to Digium tech support regarding Ouch and Power alarm 
> errors I have been receiving on my TDM400.  It contains a great deal of 
> detail regarding my setup.  In the end, I have found that one of the 5 
> phones I'm trying to make work with Asterisk is contributing to the 
> generation of these errors.
> 
> The phone in question is what I would consider to be a good-quality GE 
> two-line cordless telephone.  Digium's guess is that it is
"putting power
> on the telephone line and the card doesn't like that."  They have
given me
> zero solution other than to use a different telephone.
> 
> If this were a $25 garbage telephone I could understand.  Or, if *any* 
> other device had problems with it, I could understand.  But this was a 
> reasonably expensive, seemingly reasonably high quality telephone.  It is 
> also a telephone that I have used quite successfully not only on standard 
> POTS lines, but also on a variety of ISDN NT-1's with zero problems.  I
> don't mean that I've used this model.  I've used this *phone*
on at least
> 4 different brands and models of NT-1's, several different POTS lines
and
> even an SPA-2000.  Not one bit of problem.  Yet the TDM400 card just 
> chokes itself with Power alarm and Ouch errors.
> 
> Does anyone have any idea of what I can do to try to correct this?  Is 
> there some sort of filter or adapter that I can use to condition the line 
> for the TDM400 FXS modules?  I'm handy with a soldering iron:  if
you've
> got an idea for a circuit, I'm game.  I'm going to try
experimenting with
> some caps and coils.  Anyone been down this road yet?
> 
> As an aside, why is it that just about *any* other device with an analog 
> interface you can buy today more robust than the TDM cards?  I've used 
> countless different ISDN NT-1's without problems, from $100 cheapo
models
> to $1000 high-end devices and tons in between and none have had problems 
> like this.  Now there's a ton of SIP gateway devices.  They don't
seem to
> have these issues.  Why do the TDM cards?  And most importantly, can an 
> end user do anything about this?
> 
If you draw a schematic of the fxo module on the TDM card, its almost
exactly like the tech note schematic for the Silicon Labs chipset.
First guess is that was the starting point for whoever built the
card and modules for digium.

It appears on the surface that whoever did the design and layout
is not an industry leader in professional hardware design, but there
is a lot of room for opinion in that statement. Part of that stems
from a missed circuit board trace on the E/F model, noisy reset
line between the pci controller chip (Tigerjet) and the fxo modules,
pci controller looses its ID, etc.

I've improved the stability of my card by adding a capacitor on the
reset line. Hasn't taken a hit in over two weeks.

There is also major differences when comparing external gateway
products to the TDM card, and most of those difference involve 
the controls inherient to dedicated microprocessors (or controllers)
on the gateways, verses asterisk's approach of relying on the host
processor for everything (and it's associated uncontrolled/unknown
pci motherboard structure).

On top of all that, the drivers for the card (as of right now) are
the bare minimum functions needed to make the card function. The
drivers have never been extended to preload the chipset's registers
as documented in the SI tech notes.

If you want to play around with the card, download the pdf files
from www.silabs.com for the chipset on the card. Then take a look
at zaptel/fxstest.c to dump the registers to better understand
how the chipset registers are loaded. You'll need to complile that
code as a standalone app and run it when asterisk is not loaded.

Trying to follow the code path for a functional TDM card is not
to be taken lightly. Code is scattered across multiple drivers
and buried in asterisk modules. Even those that consider themselves
good asterisk developers stay way from this one.

Hello!

Attached to the bottom of this e-mail is an edited version of an e-mail I 
originally wrote to Digium tech support regarding Ouch and Power alarm 
errors I have been receiving on my TDM400.  It contains a great deal of 
detail regarding my setup.  In the end, I have found that one of the 5 
phones I'm trying to make work with Asterisk is contributing to the 
generation of these errors.

The phone in question is what I would consider to be a good-quality GE 
two-line cordless telephone.  Digium's guess is that it is "putting
power
on the telephone line and the card doesn't like that."  They have given
me
zero solution other than to use a different telephone.

If this were a $25 garbage telephone I could understand.  Or, if *any* 
other device had problems with it, I could understand.  But this was a 
reasonably expensive, seemingly reasonably high quality telephone.  It is 
also a telephone that I have used quite successfully not only on standard 
POTS lines, but also on a variety of ISDN NT-1's with zero problems.  I 
don't mean that I've used this model.  I've used this *phone* on at
least
4 different brands and models of NT-1's, several different POTS lines and 
even an SPA-2000.  Not one bit of problem.  Yet the TDM400 card just 
chokes itself with Power alarm and Ouch errors.

Does anyone have any idea of what I can do to try to correct this?  Is 
there some sort of filter or adapter that I can use to condition the line 
for the TDM400 FXS modules?  I'm handy with a soldering iron:  if you've
got an idea for a circuit, I'm game.  I'm going to try experimenting
with
some caps and coils.  Anyone been down this road yet?

As an aside, why is it that just about *any* other device with an analog 
interface you can buy today more robust than the TDM cards?  I've used 
countless different ISDN NT-1's without problems, from $100 cheapo models 
to $1000 high-end devices and tons in between and none have had problems 
like this.  Now there's a ton of SIP gateway devices.  They don't seem
to
have these issues.  Why do the TDM cards?  And most importantly, can an 
end user do anything about this?

Tim Massey





Hello!

I have been struggling to get a TDM400 card working for some time now.  I 
have a TDM400 with 3 FXS modules and 1 FXO module.  Right now, the FXO is 
installed but not connected to anything, and the FXS are connected to a 
number of telephones, including an inexpensive Lucent standalone analog 
phone, a 2-line AT&T standalone analog phone, a 2-line GE analog 
speakerphone with AC adapter, an inexpensive VTech cordless telephone with 
AC adapter, and a 2-line GE cordless phone, also with AC adapter.  I am 
testing the board right now just by calling back and forth internally from 
extension to extension.

As long as I do not have the GE cordless phone plugged in, there seems to 
be no problems so far.  At least, I do not believe I have yet seen the 
problem when the GE is not plugged in.  Also, if I do not pick up the GE 
cordless phone, I do not get any errors.  However, if I take the GE 
cordless phone off hook or put the phone back on hook, I will get "Ouch, 
part reset, quickly restoring reality (#)" errors, as well as "Power
alarm
on module #, resetting!" errors.  This does not happen every time I pick 
up or hang up the phone, but after a relatively short number of times 
(say, under 20 at the very most) I will get one or both of the errors.

It seems that the "Ouch" errors are indexed from zero, and the
"Power
alarm" errors are indexed from one.  The port that generates these errors 
seems to vary.  While the port that the GE cordless phone is plugged into 
seems to appear a decent amount, it is far from consistent.  For example, 
right now the cordless phone is in port 1.  However, I've gotten two Ouch 
errors, one for 0 and one for 1, and I've gotten three Power alarm errors, 
two for module 2 and one for module 1.

After an error, channels on the the board often become very staticy.  I 
believe that this only happens after a Power alarm, not after an Ouch 
error.  In fact, I am pretty (but not completely) sure that after an Ouch 
error the board (or at least individual channels that are reset) clears 
up.  Also, not just the channel that have had an error is affected: 
sometimes (but not always) all of the channels are affected.  Sometimes 
the dialtone can be heard through the static, sometimes not.  Even when 
the dialtone can be heard, it does not respond to DTMF tones.  This varies 
from channel to channel:  for example, right now, the dialtone can be 
heard through the static on one channel, but the other two have louder 
static, and I cannot hear a dialtone.  I don't think the static is 
drowning out the dialtone:  I think it's plain not there.  Also, if I 
leave those channels off hook, I do not get a busy signal.  Or at least, I 
don't hear a busy signal...

Exiting Asterisk does not affect this.  The static stays the same.  Of 
course, with Asterisk exited, there is no dialtone on any channel, but the 
static remains.  Restarting Asterisk causes the dialtone to return on 
those channels that had dialtone before, but the static remains.

Bouncing the wctdm driver *does*, however, restore operation to all 
channels, at least for a little while.

Until the error occurs a single time, I can call back and forth from 
device to device (including the GE cordless) for a while without problems. 
 However, once I get a single error, I tend to get lots of errors from 
then on out until the driver is bounced.  I can get these errors by doing 
nearly anything:  picking up any phone, dialing any phone, etc.  I 
certainly do not get an error every time I do anything, but nearly any 
action has the potential for causing an error.  Also, while I believe that 
the "Ouch" errors restore the board (or at least, individual channels)
to
a working state, it does not prevent the error from happening again: 
unlike bouncing the driver the errors continue to happen quickly even 
after an "Ouch".

The GE cordless telephone is not a cheap quality (or even, for that 
matter, inexpensive) cordless telephone.  It is described as a 
business-class telephone.  It is a phone that I have used successfully 
with normal POTS lines, a number of ISDN NT-1's, and other such devices. I 
have had zero problems with it in any other case:  in fact, it's the best 
cordless phone I've ever used!  :)  I do not know why this phone should 
cause a problem with the TDM card when it does not cause a problem for 
anything else.

What about this phone could be causing this problem?  More importantly, 
what can be done to correct this?  I have loaded the driver with the 
lowpower=1 parameter, the robust=1 parameter and even both at the same 
time.  This has had no effect.  What else can I do?  I have tried plugging 
the GE cordless into each of the three FXS modules I have and I get 
problems with each of them.  I have also tried two radically different 
systems:  a Via EPIA-MII single-board computer, and an IBM Netfinity 5600 
file server.  I don't know how you can get two more different systems, yet 
both systems give me the same problems.

<Conact information and polite conclusion snipped>

Thank you,

Timothy J. Massey

On March 22, 2005 03:08 pm, tmassey@obscorp.com wrote:
> The phone in question is what I would consider to be a good-quality GE
> two-line cordless telephone.  Digium's guess is that it is
"putting power
> on the telephone line and the card doesn't like that."  They have
given me
> zero solution other than to use a different telephone.
I have a Panasonic 900MHz digital cordless phone that also causes the TDM card 
to have fits.  I've sent it to Digium to try and figure out what's going
on,
as every single other phone and fax (probably two dozen brands between the 
two) I have ever hooked up has worked just fine.  This is not a normal thing 
and it may just be that the actual POTS system is able to handle their 
particular brand of yuck.

I certainly don't blame Digium for this, but they have been more than
willing
to help me correct it, especially since I am willing to get the phone to them 
to test with since they seem to be unable to recreate it in their lab.  My 
5.whateverGHz Panasonic digital cordless phone works great, and my 900MHz 
non-digital (cheapass) cordless phone works great.
> As an aside, why is it that just about *any* other device with an analog
> interface you can buy today more robust than the TDM cards?  I've used
> countless different ISDN NT-1's without problems, from $100 cheapo
models
> to $1000 high-end devices and tons in between and none have had problems
> like this.  Now there's a ton of SIP gateway devices.  They don't
seem to
> have these issues.  Why do the TDM cards?  And most importantly, can an
> end user do anything about this?
As I said, I've hooked up countless devices to the TDM cards and this 
particuar phone is the ONLY one I've had trouble with.  It is perhaps a 
corner case in the TDM design, but as I said Mark has personally been more 
than willing to help fix this.

As an electronics designer myself, I know how unbelievably frustrating it is 
to have a customer with an issue and not be able to recreate it myself such 
that a fix can be found.

-A.

Hello.
Sorry for my bad english. I'm a french guy.

I have the same problems with siemens dect phones S100
The caller id don't work on tdm...
In France, the CID is differant than other country.
Then standard ring cadence is: 1500 3000 1500 30000 and so...



The etsi standard (used in France) say:

The short ring is called Ring Pulse Alert Signal (RP-AS) . From ETSI EN 300
659-1 V1.3.1 (2001) :
	      200ms < RP-AS duration < 300ms
	      500ms < RP-AS to FSK < 800ms
	      200ms < FSK to FSK < 500ms

Try adding 
cadence=250,1500,1500,3000,1500,3000
In zapata.conf

And use in extension.conf

exten => 200,1,Dial(Zap/1r1,20,tr)

In my case, it's ok with siemens dect

Send me your feedback.