Electret Microphone Stereo PreAmp - Strange Behaviour

[hgg]

Hello guys,

I am trying to make a stereo microphone preamp to use with my pc, based on the following schematic
I found on the net.


Schematic

I wired up on the breadboard the first channel and it works perfectly!  I am using the very fast LM4562
opamp which made all the difference.  I've tried the LM358N which is useless with a lot of noise and
terrible sound, then the TL072 which way better and finally the LM4562 which is perfect even with the
very cheap electret capsules I am using.  (They will be upgraded soon).

Then, I tried to wire on the breadboard the second channel as well and a strange behaviour occurred.
(Strange, because I cannot explain it...)  Below is the actual circuit.  The other channel is mirrored and
the only difference is the green 10uF cap to ground which is from a different brand . 


Breadboard

If I connect a signal generator common to both inputs the amplification works fine with a flat
frequency response from 300Hz to 20kHz.  (When I connect the signal gen, I remove the electret
biasing 10k resistors.)


20Khz

No signal degradation or attenuation up to 20kHz


500Hz

Same signal level down to 500Hz with only some ramping but that's a different story.
(Result of high pass filtering? any ideas?)

So everything ok with the signal generator.
The problem now occurs when I connect the microphones on the circuit.  Specifically, the first channel
(left on the photo) always works ok, but when I connect the second microphone here is what happens:
(Both multi turn pots are set to zero ohms for the highest amplification)


Oscillation

I get a huge 3V oscillation on the second channel!  When I start increasing the resistance of the second
channel pot the oscillation stops at around 520 Ohm.  After that, the circuit looks that it works ok but
actually the second channel amplifier attenuates high frequency signals.  The first channel works fine.


Vowel sound

Low frequency vowel sound is not attenuated.


Hissing sound

... but a higher frequency hissing sound is attenuated quite a bit.
(Left 2nd channel is yellow and Right is blue)

(The other thing that I observed is that the second channel is a lot noisier and if I touch the 1k5
resistor that goes to ground I get a lot of noise.  Nothing happens if I do the same to the first
channel.)


I have tried to swap microphone capsules, use different ones, swap the op-amp with different types,
swap the green capacitor, but the same thing happens with consistency all the time.   (I've powered
the circuit from a battery with the same results.)

So, can anybody think of a reason for this circuit behaviour?
Any help is appreciated.

Thank you.
George.

[SeanB]

Place the 47uF decoupling capacitor right across the IC power pins, not dangling somewhere on the board. It must be right at the pins, with short leads.

[hgg]

It made a difference but a very small one.  Signal attenuation of the second channel is a little
bit lower but the problem continues.

[rob77]

i would try to build it on a piece of veroboard or protoboard (socket the opamp of course), breadboards are sometimes unpredictable.

[hgg]

I know, I have had bad experiences with breadboards previously... but this time it does not looks that the
problem is from the breadboard because the stereo circuit works perfectly when I connect the signal
generator.

[ciccio]

I don't see the feedback caps (2 pF on the schematic) in the breadboard.
My experience is that in parallel with 33 Kohms the capacitor should be at least 22 pF (for suitable HF response).
In professional audio "standard" feedback values are 22 Kohm //100 pF...
You do not need a 2.5 MHz bandwidth (33 Kohm // 2 pF).
Uncompensated op-amps are evil in audio, even if the data sheets do not show them.
Best regards

[rob77]

Considering an average breadboard, you have approx 2.5pF between every 2 adjacent rows. - that's why i suggested to use a different build technique.

[kg4arn]

I agree with the others. The breadboard is not the place for the 4562 with its 55 MHz GBW.  It's going to be very very difficult to stabilize the cicuit.

[hgg]

Ciccio, you are right they are missing and this is because when I tried the suggested 2pf ceramics
they made no difference.  After your suggestion I've tried two 33pF caps and they made a lot of
difference.  The noise problem and the oscillation stopped completely!  The amplification increased
slightly.

Only thing that remains is that I still cannot match the frequency response of the two channels
but its way better than before!  Maybe the breadboard indeed interferes with the circuit here.   
At least I can order the new microphone capsules now...   

Thank you guys for your help!

[joseph.anand]

I have actually "breadboarded" 4 channel microphone preamplifiers using 4 dual opamps (the second opamp in each was used to adjust levels), but none with such a high GBW. However you have basically taken a design for a mono amplifier and replicated it for multiple channels. 

Consider doing the following: Instead of the 10K resistor Use two 4.7K resistors (Say Ra, Rb) in series with one terminal of Ra connected to Vcc, the other end of Rb connected to the microphone. Now connect a 10-25uF cap from the series junction of Ra and Rb to GND.

Alternatively source the biasing for each of the microphones from a voltage divider network (separate ones for each microphones) and ensure that you use a 10-25uF cap across each of the lower resistors in the  network. I'm too lazy/busy to draw a schematic, but I think you should be able to get a picture of what I'm trying to say.


ps: I'm unable to see your schematic or any of your posted images clearly, and clicking on them takes me to the homepage of postimage.

[hgg]

Josheph I will indeed try your suggestions later today, thanks!

ps: I'm unable to see your schematic or any of your posted images clearly, and clicking on them takes me to the homepage of postimage.

Which browser are you using?   
Firefox displays everything fine.

[mikerj]

I agree with the others. The breadboard is not the place for the 4562 with its 55 MHz GBW.  It's going to be very very difficult to stabilize the cicuit.

I totally agree.  The biggest mystery here is why the first channel doesn't oscillate, rather than why the second channel does.

[hgg]

Oscillation solved.  It works very well with full dynamic sound and very little noise which I think
it will be taken care off by the new mic capsules.

From my little search I found that the best are:

1) Primo EM173 (Low noise and most sensitive)
2) Panasonic WM-61A (Pretty good and flat to the point they might be lying...)
    You can also apply the Linkwitz mod.
3) PUI Audio AOM-6738L-R  (Balance between price and specs)

I think I will try the PUI Audio ones. 
If someone knows better ones for a reasonable price please let me know.

Something else.
Do you know if there is a way to remove this annoying peak when turning on and off the preamp?



Thanks!

Joseph do you mean something like the following ?



What are the advantages of this biasing method?  Less noise?
Thank you.

[Zero999]

The on/off peak is due to the capacitors charging and discharging of the DC blocking capacitors. The easiest way to get rid of it is to use a relay/analogue switch and timer, so the output is only connected to the circuit, after the DC bias point has stabilised.

[hgg]

I see. 
Is that the reason why many audio amplifiers have relays?

[ciccio]

Audio amplifiers use relays for two reasons:
-the relay has a delayed on to avoid the on pulse that can damage speakers, and has an immediate off to avoid the off pulse (usually they are powered not from the main amplifier's power supply, but directly from secondary ac).
-the relay has a control circuit that allows it to switch-on only in absence of significant dc on the output terminals, to avoid damaging speakers in case of a defective amplifier channel.

[ConKbot]

Yes, for click suppression, and for general protection circuitry (which turn on/off click suppression is a part of). I.e. dc bias protection, short circuit protection.  A good plated relay contact is hard to beat in linearity and power handling for the cost.

[hgg]

I see. 
Thank you  for explanation.

[joseph.anand]

Joseph do you mean something like the following ?



What are the advantages of this biasing method?  Less noise?
Thank you.

no.

Something more like the following


Advantage is less cross channel interference. I prefer using A for my audio designs.

[hgg]

Joseph thank you very much for the schematic!
I will try it.

Isn't the 100K resistor a possible noise source?  Wouldn't the second circuit be more noise free?
(I am trying to make it as noise free as possible.  Just ordered the Primo EM172 capsules)

[hgg]

Hi,

Just a quick question on the mic preamp circuit:



In the circuit above the volume is reduced by the blue 10k log resistor while in my circuit I am using
a potentiometer in place of the red 27k resistor to reduce the gain and a fixed 10k in place of the blue pot.

I was thinking of using the same configuration as above.
Do you think that the output 2.2uF capacitor together with the blue 10k pot will work like a high pass
filter?  So when you lower the volume of the output signal you will have a reduced bass response.

Is this assumption correct?

Thank you.
George.

[Zero999]

Yes, it will be a high pass filter. The cut-off frequency can be calculated using the following formula:

F_C = 1/(2pi*RC)

Don't forger that changing the value of the red resistor will also affect the bandwidth of the circuit, as it's in series with a capacitor.

[JuKu]

> Do you think that the output 2.2uF capacitor together with the blue 10k pot will work like a high pass filter? 

Yes, as already explained.

> So when you lower the volume of the output signal you will have a reduced bass response.

No. Turning the pot does not change the resistance of the pot, assuming reasonably high impedance on the other side.

Another use for relays in audio is signal source switching. Relays don't leak signal through like electronic switches do.

[hgg]

Juku, Hero666 thank you for your replies.

Yes, it will be a high pass filter. The cut-off frequency can be calculated using the following formula:

F_C = 1/(2pi*RC)

Don't forger that changing the value of the red resistor will also affect the bandwidth of the circuit, as it's in series with a capacitor.

In what way though?  High pass filter again?

So how can you go about solving this problem?
What about if instead of the blue 10k pot I use a larger value like 100k?
Maybe this way the filtering effect will be much smaller since the resistances are greater.

[flynwill]

I see. 
Is that the reason why many audio amplifiers have relays?

Yes, exactly.

Let us know how those PUI mics work out for you.  The spec sheet isn't very encouraging
with a quoted S/N of 60 dB (Meaning a self-noise at about 34 dB SPL).  However they
also spec it at 1.5V it should get a lot quieter running of 9V like I presume you are doing.

That Panasonic capsule was the gold standard for cheap electret capsules, but it is no
longer being made.