[Asterisk-Users] PCI Problems

Andrew Kohlsmith akohlsmith-asterisk at benshaw.com
Thu May 25 11:09:45 MST 2006


On Thursday 25 May 2006 13:06, Sean Cook wrote:
> There are no known compatibility issues (IRQ, IO etc) with ANY Sangoma
> hardware and ANY make/brand of PC/server- NONE

Not exactly.  Their hardware and drivers play nicely far more often than the 
older Digium boards, but I have personally been bitten by their S518 not 
playing nice with their A101u in a Dell P3 system.  The S518's driver would 
cause very clear and repeatable audio "chirping" on the T1.  Replace the 
A101u with a T100P and no issues whatsoever.

This was about 18 months ago.  Drivers have changed since then, and this may 
no longer be an issue. That particular system has been EOLd so it's a 
complete non-issue to me.

I've mentioned this to Sangoma in passing, but seeing as it is a fairly 
irregular case, I don't expect any fuss to be made of it.  Digium's cards are 
historically much finickier (is that a word?) but I know that their rev2 
TE405/410 and latest TDM400 carrier is significantly better than their older 
stuff in terms of compatibility.

Technically speaking, both Sangoma and Digium use the *identical* Xilinx 
Spartan II FPGA for their PCI interface (on the A104 and TE405/410).  I feel 
fairly confident to also say that they will be using the same PCI VHDL 
"block" to interface that part to the PCI bus.  So what this comes down to, 
by and large, is the drivers.

There have been some hardware PCI interop issues with Digium's stuff, but I 
know for a fact that these have been fixed in their rev2 hardware.

Also technically speaking (but not PCI-speaking), Sangoma's multiport cards 
can do some things that the Digium cards cannot.  Specifically, Sangoma's 
cards have no issues whatsoever with having their spans synchronized to 
completely different clocking sources.  This is achieved by using a 
single-port framer for each span.  Digium's framer is a single chip that 
supports four spans, and one of the framer's limitations is that all spans of 
the same technology (T1/E1) must share a clocking source.  (I've dug around 
in the framer's datasheet several months ago for a separate nefarious 
project, which is why I know this.)  For MOST people this isn't an issue, but 
it comes up now and again.

Sangoma's single, dual and quadspan cards can also fit in a half-height PCI 
slot.  Digium's quadspan cannot, and I believe that their dualspan cannot do 
this, either.  For some people this is an issue.

Sangoma's echo cancellation option board also seems to be far "heftier" than 
Digium's VPM.  I've got an A104d that replaced a TE406 and completely 
eradicated all of my echo troubles.

I like Digium's products, and I like Sangoma's products.  I'm not endorsing 
one over the other in this post at all; I'm merely pointing out that your 
"none" and "never" are strong words that no vendor can meet.  Sangoma's 
response to oddball hardware has been MUCH more customer-oriented than 
Digium's, at least historically speaking.  I believe that Digium's made 
*significant* improvements in their customer relations over the past little 
while, but by and large the products I buy from both vendors "just works" and 
I hardly ever have to call support.

-A.



More information about the asterisk-users mailing list