Sound delay in i4b

[Dirk Thannhäuser]

On May 15, 2006, at 2:03 PM, Hans Petter Selasky wrote:

> On Monday 15 May 2006 12:20, Dirk Thannhäuser wrote:
>> On May 15, 2006, at 9:38 AM, Hans Petter Selasky wrote:
>>> Hi,
>>>
>>> On Sunday 14 May 2006 22:57, Dirk Thannhäuser wrote:
>>>> On May 14, 2006, at 3:18 PM, Hans Petter Selasky wrote:
>>>>> Do you have SVN installed?
>>>>
>>>> yes i have. I checked out the version which included both, the i4b
>>>> and the chan_capi. Is ist possible that my version was to old?
>>>> I just did a new checkout.
>>>> Subversion says "Checked out revision 264."
>>>
>>> That should be the latest version. It was fixed in "Revision 257",
>>> which was
>>> after that I released Version 1.6.0 in April.
>>
>> Hmm. i compiled the latest release (i4b and chan_capi) yesterday and
>> tested it with the echo test application. The delay is the same as
>> before.
>> When I call the server internal (My Phone -> ISDNCard(NT) -> Asterisk
>> (1.2.4) (isdnd not running. cards set to TE/NT mode via "isdnconfig"
>> command . The delay in the echo test application remains unchanged at
>> about 250ms to half a second. (not measured, it's just my feeling,
>> but i know the difference to the zaptel FreeBSD driver). My machine
>> is a Siemens Scenic L (1.6Ghz/512MB RAM). That schould be enough
>> power to get low latency.
>
> Sure, but that is only when the computer is not loaded. 50ms  
> interrupts were
> chosen because it will allow one to heavily load the computer  
> without loosing
> any data.
>
>> How easy would it be to play around with changing latencey. Should i
>> take a look to the chan_capi sources to get latency changed?
>>
>
> It is not very difficult to play around to get the latency down.
>
> Edit: /usr/src/sys/i4b/layer1/ihfc2/i4b_ihfc2_drv.c
>
> Change:
>
> .s_ctmt_pci        = 0x00 | 0x17, /* t50ms + trans */
>
> Into:
>
> .s_ctmt_pci        = 0x00 | 0x13, /* t25ms + trans */
>
>
> And change:
>                   /* default AUDIO limit == 100ms of data */
>
>                   /* limit fifo usage to 800 bytes(100ms)
>                    * - Assuming that about 400 bytes(50ms)
>                    * is needed for each interrupt:
>                    */
>                   f->Z_min_free -= 800;
> Into:
>
>                   /* default AUDIO limit == 100ms of data */
>
>                   /* limit fifo usage to 800 bytes(100ms)
>                    * - Assuming that about 400 bytes(50ms)
>                    * is needed for each interrupt:
>                    */
>                   f->Z_min_free -= 400;

i played around a lot with changing numbers, but i was not able to  
get latency down.
did you ever test with the asterisk echo test application? (maybe  
that latency is only here a problem) Is it possible that there is a  
bug inside the application? But on the other hand it worked fine with  
the zaptel drivers.

I think i will need more time to study sources :-)


>
>>>> Are there any "cheap" HFC-4S cards out there for testing?
>>>
>>> There exist some developer kits, but they are not so easy to get a
>>> hand on.
>>> The only place I know that sales HFC-4S are:
>>
>> A developer Kit would be the card of my choice.
>> Does Bero sell them or shoud i contact colognechip?
>
> I don't know.
>
>>
>>> http://shop.beronet.com
>>>
>>> And maybe "http://www.junghanns.net/".
>>>
>>>> I think that this also means that i am not able to run one card
>>>> (HFC-1S) in slave mode, getting the clock from the other to keep  
>>>> them
>>>> in sync without soldering.
>>>
>>> Yes.
>>>
>>>> For Voice over IP a huger delay might be ok. thats also my opinion.
>>>> but what about Faxing over ISDN. Here I think latency is an  
>>>> important
>>>> issue. (didn't test the CAPI Fax receiption, yet)
>>>
>>> Faxing works, but really you need to use line interconnect to get
>>> faxing
>>> working reliably. Or you can receive the fax in asterisk. That also
>>> works.
>>
>> Does it also work with about 250ms delay? (it exists on both sides
>> from my NT-Card to asterisk and also from the TE-Card to asterisk
>> may be that i did something wrong compiling the sources although i
>> cleaned the BSD sourcetree applied the patches again an then started
>> compiling the get a "fresh" state an no confusion with the older
>> release which was installed before.
>
> Yes, faxing works with 250ms delay.
>
> --HPS