[SOLVED] Echo canceller Schematic

Status
Not open for further replies.

zentos

Member level 1
Joined
Apr 7, 2014
Messages
33
Helped
0
Reputation
0
Reaction score
0
Trophy points
6
Activity points
278
I'm working for a design which i can couple regular telephone audio to GSM phone and eliminate echo. Because of the sidetone from the telephone GSM side has mass echo sound. Does anybody knows echo canceller schematic which is not advanced. advices are really appreciate,

Thanks.
 

Because of the delay and variable frequency response and level of the coupled signal, I can't see an easy way. For telephones there is a transformer to couple in a bit of anti phase speech into the earpiece, but it is all rough and ready. Modern phones with their loose coupling to the ear and mouth, will suffer from this more then old fashioned telephones, in fact I am suprised that this has not been taken care of in the GSM phone.
Frank
 


Thanks chunky. I have seen anti-sidetone schematic here **broken link removed**. For me it is very hard to find 600 ohms impedance audio transformer in my area. Not sure about how many turns should I use to make 600 ohms 1:1 transformer. Cant wait for long time if I order from internet. I tried with varies filters (High pass- lowpass, ets..) and able to cancel echo but voice quality like a hell.

If anyone knows how to eliminate sidetone from a telephone, really appreciate.

Thanks.
 

The sidetone in a phone is needed so you hear low loud or soft you are talking. Sidetone has nothing to do with echo produced by an inferior two-wire network.
Major telephone carriers already use echo-cancellation so why do you need one in your phone?
 


Its has a effect as when coupling both audio. If there is small delay time it becomes a echo. Because person calling from PSTN receive his voice again to his speaker. If there is no sidetone there is nothing to come back. Latency does not matter when there is no side tone.
 

Most telephone calls to and from North America have telephone companies that use echo cancellers so that a person talking into a telephone does not hear his own voice delayed as an echo. The echo occurs because the two-wire telephone line impedance is not perfectly matched at both ends. The sound received at the distant end does not come out the earphone and go back into the microphone like in a speakerphone.

I worked with telephone conferencing equipment that had speakerphone equipment that used a microphone and a loudspeaker. If both ends of a long distance telephone call has this system then the sound goes around and around making acoustical feedback howling or severe echoes. A simple speakerphone uses switching to pass only one direction that has the loudest sounds which is called "voice switching". But then frequently voices are blocked by background sounds at the other end or by echoes. A digital echo canceller was "trained" of the echoes in the room and cancels received sounds from being transmitted back. Speakerphones and tele-conferencing equipment have no sidetone.
 

I have found this after googled. Seems Hybrid circuits can eliminate sidetone.


Anybody has any opinions. Will make this tomorrow and see how can be fix this.
 

I do not know why you think that sidetone creates echoes. Echoes are delayed sounds caused by a long distance reflection of sounds. Sidetone is your own voice AT EXACTLY THE SAME TIME without any delay.
 

Hybrid circuits of the sort you posted are good for cencelling sidetone if and only if the impedance of the line is known reasonably well, usually at least one arm of the bridge is made variable and complex, typically a variable resistance and switched shunt caps.

A Hybrid will help if the 'echo' is really excessive sidetone on the POTS side of your interface, but can do nothing for delayed echos caused by the GSM side (Which is surely where your problem is), correcting this really needs adaptive echo cancellers which pretty much means serious dsp.

Most phone switch hardware that has to deal with GSM includes such processing on board, but it is mostly trade secret and these are patent infested waters, there may however be some code in asterisk or one of the softphone projects that you could port to a blackfin or such.

73 Dan.
 

I do not know why you think that sidetone creates echoes. Echoes are delayed sounds caused by a long distance reflection of sounds. Sidetone is your own voice AT EXACTLY THE SAME TIME without any delay.

My project is voip to gsm gateway. In this case voip signals out only to a telephone line (Skype phone adapter). So connected regular phone makes sidetone. When connecting audio to gsm it makes echo. Hope you understand.



In this diagram transmitter only shows triangle. Is that op amp or just simple input?
 

I agree with Dan Mills that a hybrid reduces excessive sidetone but has no effect on echo caused by delay and reflection from the distant end.
 

I agree with Dan Mills that a hybrid reduces excessive sidetone but has no effect on echo caused by delay and reflection from the distant end.

I still believe sidetone makes echo as I tested with android phone as a voip device and there is no echo.

Found same issue here https://electronics.stackexchange.com/questions/27421/hybrid-phone-echo-cancellation-circuit

Can't say exactly 100% this is the issue. Once tested with hybrid circuit will update.

Kindly tell someone is that triangle (transmitter) , Op amp or just simple audio input?
 

I agree with Dan Mills that a hybrid reduces excessive sidetone but has no effect on echo caused by delay and reflection from the distant end.

You said that sidetone doent effect for echo in this case. This diagram shows clearly where echo begins.



if there is no sidetone, echo does not exist.

- - - Updated - - -

Kindly someone tell me how to wind 600:600 ohm audio transformer. I wind 60 ohm DC resistance 1:1 transformer with 45AWG 600 cm long for primary and same for the secondary . but it did not work with mentioned schematic. Any advice?
 
Last edited:

You did use a proper transformer core (Es and Is)? I think its too low an inductance. Apply your signal generator to one winding, put a 600 ohm resistor with a CRO or level measuring voltmeter across it and swing the frequency from 20 Hz to 20 KHZ to see what the response is.
Frank
 

You said that sidetone doent effect for echo in this case. This diagram shows clearly where echo begins.
if there is no sidetone, echo does not exist.
The diagram shows audio coupling from the earphone to the microphone that occurs in a speakerphone but not in a telephone. Sidetone is the opposite, producing coupling from the microphone to the earphone.
Speakerphones already are made with simple voice-switching (transmit and receive at different times) or modern built-in echo cancellation (Polycom). The echo canceller prevents sounds received from the distant end and picked up by the speakerphone microphone from being transmitted back to the distant end as an echo.
 

Attachments

  • Polycom speakerphone.png
    121.1 KB · Views: 180


Frank thanks for the reply. I did mentioned 600 ohms dc resistance and its not. its 60 ohms DC resistance.
I use 5mm x5mm 10mm height core to wind 45AWG 600 cm which gives me 60 ohm DC resistance. Frankly I'm not much knowledgeable about transformers. I have seen most of 600 ohms transformers say 600 ohms impedance and DC resistance around 40-60 ohms. Do I need wind 600 DC resistance? Also I found 120 Ohms DC resistance 1:1 dial-up modem transformers. Can I use that transformer for this schematic? replies are highly appreciate.

- - - Updated - - -


Thanks Audioguru for the reply. Telephone and GSM audio couple with wires (Used some capacitors and resistance). So Telephone speakers and GSM speakers are simply audio sources.
 

Why you are opted to use a transformer when an active circuit can do the trick, only if input/output isolation is in sight.
 

Why you are opted to use a transformer when an active circuit can do the trick, only if input/output isolation is in sight.


Kindly tell me is that transformer required for mentioned schematic (hybrid). What actually do this transformer in this schematic?
Its 1:1 and gives the same output as input.
 


Last edited:

Status
Not open for further replies.

Similar threads

Cookies are required to use this site. You must accept them to continue using the site. Learn more…