[Asterisk-Users] Re: [Asterisk-Users,
Andrew] Will Echo problems EVER be solved, I'm scared
Rich Adamson
radamson at routers.com
Thu Aug 25 04:33:22 MST 2005
I'll jump in here with one comment. I worked with an individual in Canada
that could not get rid of the echo (some time ago with an x100p). As a
very experienced telephony engineer and two years of asterisk experience,
I logged into his system and tried many many changes without impacting
the problem. He swapped his motherboard from a rather current one to an
older slower P3. Echo disappeared. Same distro, same * code, same card.
We had no idea what the root cause was but guessed it had something to
do with the pci bus and/or interrupt structure that we could not quantify
with anything that we had available at the time. Would have been very
interesting to have had the original motherboard in a captive lab
environment to jack with.
------------------------
> Thank you for the suggestions Andrew. I have not come across some of them
> before and will give them a shot. Based on my reading, changing the
> motherboard should have minimal impact unless that motherboard and the
> TDM400P don't get along (aka. IRQ sharing). I have disabled everything that
> is not needed and I do not believe I have any IRQ problems and I am NEVER
> wrong ;). Calls are crisp and clear. . No snap, crackle, pop. It would
> be a beautiful thing if not for the echo.
>
> To get the RX/TX levels, run "ztmonitor 1 -vv", dial a telco 1004hz 0dbm
> test phone # and set the quantitative RX number to around 14500. With 2
> lines (which I don't have) you test the TX level by looping out to the other
> PSTN. Without a second line you do the simple ztmonitor test for the TX
> levels.
> http://www.voip-info.org/tiki-index.php?page=Asterisk+zapata+gain+adjustment
>
> I was MOST DEFINITELY NOT wildly changing settings. It would require a
> whole book to explain properly what I did but the end result was that I
> pretty much covered every possible combination of settings. I have read the
> white papers and EVERYTHING else I could find on the web to determine the
> most logical and proper way to go about this. I was NOT approaching this
> like some back yard six pack scientist.
>
> I made a mistake when I used the word "levels" to describe what fxotune
> does. The bottom line is that it did not change anything.
>
> My settings are pretty much default except where I stated otherwise.
> Network is a Linksys WRT54g (ie. Switch). Asterisk server on port 1,
> GXP2000 on port 2, 9133i on port 3.
>
> I have NO echo between SIP phones!
>
> #cat /proc/pci
> Bus 0, device 0, function 0:
> Host bridge: Intel Corp. 82815 815 Chipset Host Bridge and Memory
> Controller Hub (rev 2).
> Prefetchable 32 bit memory at 0xd0000000 [0xd3ffffff].
> Bus 0, device 1, function 0:
> PCI bridge: Intel Corp. 82815 815 Chipset AGP Bridge (rev 2).
> Master Capable. Latency=32. Min Gnt=12.
> Bus 0, device 30, function 0:
> PCI bridge: Intel Corp. 82801BA/CA/DB/EB PCI Bridge (rev 1).
> Master Capable. No bursts. Min Gnt=6.
> Bus 0, device 31, function 0:
> ISA bridge: Intel Corp. 82801BA ISA Bridge (LPC) (rev 1).
> Bus 0, device 31, function 1:
> IDE interface: Intel Corp. 82801BA IDE U100 (rev 1).
> I/O at 0xf000 [0xf00f].
> Bus 0, device 31, function 3:
> SMBus: Intel Corp. 82801BA/BAM SMBus (rev 1).
> IRQ 11.
> I/O at 0x5000 [0x500f].
> Bus 1, device 0, function 0:
> VGA compatible controller: ATI Technologies Inc Rage 128 RF/SG AGP (rev
> 0).
> IRQ 10.
> Master Capable. Latency=32. Min Gnt=8.
> Prefetchable 32 bit memory at 0xd4000000 [0xd7ffffff].
> I/O at 0x9000 [0x90ff].
> Non-prefetchable 32 bit memory at 0xd9000000 [0xd9003fff].
> Bus 2, device 1, function 0:
> Unknown mass storage controller: PCI device 1095:3124 (CMD Technology
> Inc) (rev 1).
> IRQ 11.
> Master Capable. Latency=32.
> Non-prefetchable 64 bit memory at 0xdb008000 [0xdb00807f].
> Non-prefetchable 64 bit memory at 0xdb000000 [0xdb007fff].
> I/O at 0xa000 [0xa00f].
> Bus 2, device 2, function 0:
> Communication controller: Tiger Jet Network Inc. Intel 537 (rev 0).
> IRQ 5.
> Master Capable. Latency=32. Min Gnt=1.Max Lat=128.
> I/O at 0xa400 [0xa4ff].
> Non-prefetchable 32 bit memory at 0xdb009000 [0xdb009fff].
> Bus 2, device 4, function 0:
> Ethernet controller: Realtek Semiconductor Co., Ltd.
> RTL-8139/8139C/8139C+ (rev 16).
> IRQ 10.
> Master Capable. Latency=32. Min Gnt=32.Max Lat=64.
> I/O at 0xa800 [0xa8ff].
> Non-prefetchable 32 bit memory at 0xdb00a000 [0xdb00a0ff].
>
>
> #cat /proc/interrupts
> CPU0
> 0: 1290132 XT-PIC timer
> 1: 4 XT-PIC keyboard
> 2: 0 XT-PIC cascade
> 5: 12878920 XT-PIC wctdm
> 8: 1 XT-PIC rtc
> 10: 33866 XT-PIC eth0
> 12: 41 XT-PIC PS/2 Mouse
> 14: 17345 XT-PIC ide0
> 15: 60 XT-PIC ide1
> NMI: 0
> ERR: 0
>
>
>
> -----Original Message-----
> From: Andrew Kohlsmith [mailto:akohlsmith-asterisk at benshaw.com]
> Sent: Wednesday, August 24, 2005 1:54 PM
> To: asterisk-users at lists.digium.com
> Subject: Re: [Asterisk-Users] Will Echo problems EVER be solved, I'm scared
>
> On Wednesday 24 August 2005 16:37, canuck15 wrote:
> > As others have recommended, I created a test system with the proposed
> > production parts. I bought a couple different SIP phones to try and a
> > Digium TDM01B card. I am using an older PIII 1Ghz system with
> > 815chipset (PCI Rev2.2) with 256MB for my test system. The only thing
> > that will be different on a production system is that I will be using
> > a newer chipset PC with faster processor and 512MB. Probably Intel
> > 7505, 7210, or 7211 chipsets which seem to be the most compatible with
> Asterisk.
>
> So in other words, everything will be changing on your production system.
> Not a good way to start.
>
> > My problem is that I cannot eliminate echo no matter what I try. I
> > seriously doubt that a newer chipset faster PC with more memory will
> > eliminate or even reduce my echo problems based on what I have read. I
> am
> > not about to drop more cash to try and find out. Essentially, my
> > findings are that Asterisk is NOT production capable for my
> > configuration which is via FXO and PSTN. That is probably THE most
> > common configuration so if it is not production capable like that it
> > isn't production capable period as far as I'm concerned. What a
> disappointment :(.
>
> Most of us don't have any trouble.
>
> > *Buy latest TDM400P with latest FXO module *Ensure copper connection
> > to analog telco lines and telco are not causing problems including
> > running a separate shielded line to the demarc AND having the telco
> > guy come out and test the levels, impedance etc.
>
> I'd be damn curious to know what you got out of this -- most telco guys will
> do a basic metallic check, throw on a butt-set and say "yup, I got
> dialtone."
> -- hardly a real check but that's neither here nor there. I'm also in
> Canada (1.5hrs from Toronto, ON) so I'm *really* curious who you got on the
> line to do a real line test with you. I have resorted to buying my own
> telco test equipment off ebay and using that, even though our techs here are
> excellent.
>
> > *Adjust RX/TX levels as per Asterisk Wiki using the quick Ztmonitor
> > method and by using the detailed Ztmonitor method via a Telco
> > 102milliwatt test phone #. The end result was RX=8.0, TX=-1.0. Since
> > I still have echo problems I have tried all sort of other settings without
> success.
>
> Ok good. Can you detail exactly what you did to reach these numbers? I'm
> curious.
>
> > *After ALL of the above, try every possible combination of all of the
> > following on Asterisk v1.0.9: echocancel (off, on, 128, 256, 16, 32,
> > 64), echowhenbridged (on, off), echotraining (off, on, 800), Mark 2
> > (default, aggressive, CVS head developments, bugs.digium.com patches,
> > adjust threshold level as per wiki etc. etc.)
>
> I'd posted something earlier that basically says this: Without measured,
> controlled tests, you're just pissing up a rope. Wildly changing settings
> and hoping for the best does nothing but cost you time and energy.
>
> > *Run fxotune which did not find a need to adjust the FXO levels
> > (1=0,0,0,0,0,0,0,0)
>
> fxotune doesn't adjust FXO levels, it adjusts a very simple FIR filter which
> is part of the DAA in the FXO module. IMO it helps with audio quality but
> not much with echo.
>
> > Still have echo. Aggressive mode helps a bit but then the other
> > persons voice get's cut off a lot especially when I talk and the
> > cutting in and out of the canceller is more noticeable and
> > objectionable in general than if Aggressive is turned off.
>
> Agressive mode turns the phone line into a half-duplex environment. When
> your voice energy is detected it mutes the receive audio.
>
> > I have two SIP phones. An Aastra 9133i and a Grandstream GXP2000.
> > Echo problem is the same on both phones.
>
> Do you have echo between the two phones? What about when calling out to a
> VOIP provider, dialing a DID you own that comes back in and hits the other
> phone?
>
> > Any comments and/or suggestions would be greatly appreciated as I am
> > pretty much out of ideas and ready to give up on Asterisk as a
> > suitable traditional small business phone system replacement.
>
> I haven't seen your zconfig.h nor your zaptel Makefile, and you didn't tell
> us anything about your network (network card, switch, etc.).
>
> My general advice for zaptel is to do the following:
> zaptel Makefile: underneath the comments about zconfig.h add
> KFLAGS+=-march=pentium4 (or pentium3 or pentiumpro, use the exact proc)
> CFLAGS+=-march=pentium4 (or pentium3 or pentiumpro, use the exact proc)
>
> and in zconfig.h
> - enable XLAW (optimize for small # of zap channels)
> - enable MMX
> - MARK2, no agressive mode.
>
> Whenever I've done that my echo has largely disappeared.
>
> Have you also tried flipping tip and ring going into the TDM card?
>
> -A.
More information about the asterisk-users
mailing list