Line level buffer: discrete or ICs based?

[anatech]

Hi George,
Pick your poison. All things in the signal path affect the sound and noise content. I tend to use diamond buffers. Those are pretty transparent, and not noisy.

-Chris

 

[OlegSh]

I am back with a little problem. Some time ago I have added LME49990 buffer (top right PCB in post #128 powered by the PSU board in same post) to the output of my R-2R volume controller. All is working nicely except if I trigger the mute relay which shorts the input of the buffer to GND I hear a click from the speakers. Without the buffer there is no click. It is not loud but it is there. Does anyone have an idea what can be wrong?

Also if I turn off the buffered volume controller before I turn off the amp I have a loud but "slow" pop as if the woofers suddenly get DC across them which decays within a second to zero. The impression is if a capacitor discharged through the speakers. My amp is AC coupled. Can it happen that the output of the buffer opamp collapses to one of the rails when supply drops below some value, e.g. minimum supply voltage?

Regards,
Oleg

 

[jean-paul]

Muting is generally done at the outputs of buffers. If you make it a fast version it will short the output at power off. You have a DC "pulse" at your outputs because the buffer outputs are going to either + or -. This phenomenon is best described as "woofer killer" and if it does not kill woofers it still is very annoying. Change your muting to the output (and definitely use a series resistor !) to avoid this problem.

 

[OlegSh]

Hi Jean-Paul,

Thanks for the explanation. I agree that muting is better done after the buffer but my early (inexperienced) planing did not include the buffer and therefore the muting relay is incorporated on the R-2R attenuator PCB

By the way, what is the best way to mute the output? Should I ground the output signal wire when disconnecting it from the buffer output or should I let it floating? I have a feeling that letting the output floating with high input impedance of the amp may result in hum or?

Regards,
Oleg

 

[jean-paul]

You mean shorting the output versus disconnecting ? Shorting to GND is best as the relay is not in the signal path that way. It will need (the already standard) series resistors of 47 to 100 Ohm that way. Those resistors serve also the purpose of protecting the opamps/buffers against capacitive loads. When shorting the outputs to GND they will limit current as well.

Member Mark Johnson has nice separate PCBs for output muting with very fast shorting at power off...

Muting must be done at the outputs whatever the practical inconveniences of implementing are. i would not know what muting at the inputs would do ?! Muting is needed to keep power on and power off pulses away from the power amps inputs as that one will like to amplify anything that arrives at its inputs

 

[OlegSh]

Thank Jean-Paul!

I'll try to change my system layout and shift the muting relay to the output of the buffer.

PS: the buffer has the output resistors already.

 

[jean-paul]

http://www.diyaudio.com/forums/powe...pcb-preamps-linestages-et-al.html#post4536744

This gentleman is clearly an engineer as he used some nice tricks to improve the working of the circuit.

 

[OlegSh]

Thanks, I did not know about Mark's muting PCBs. I'll probably make something similar. I have arduino controlling the whole thing so it must be easy to implement.

 

[jean-paul]

Mark is a nice chap and he allowed the use of his circuit in other designs. You could ask him nicely....

I would not want to depend on Arduino but would use a separate dedicated controller, be it analog or some PIC. Old fashioned, I know...

 

[OlegSh]

Dedicated separate circuit is always most reliable. I'll consider Mark's PCB but let me think about the trade offs of various possibilities first

 

[OlegSh]

I guess I found the source of audible "click" when I mute the attenuator output. When mute is triggered the output impedance of the attenuator change for a short moment from less than 5k to 10k and then to zero. All within the time it takes to switch relay contacts (~ 3 ms). The relatively high LME49990 input bias current can create input offset voltage of up to 10mV across 10k and then relay shorts it to GND which I may hear as a click. Could it be the reason? I have also LME49710 which has an order of magnitude lower input bias current. I could try it in place of LME49990 as the buffer.