Twin Servo Mic preamp - oscillation at maximum gain

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

elskardio

Well-known member
Joined
Sep 29, 2010
Messages
563
Location
Montreal - Canada
Hi Guys,

I'm working on a twin servo mic preamp based on the 1997 Jensen's schematic.

The preamp works perfectly (and sounds really good) but at maximum gain I get oscillation and the audio fades... I changed R6 & R12 to 68R to lower the maximum gain range and it did the trick. But should I be able to get that extra dB or two? Anyone experienced with that circuit and can share some insights?

Thanks

Here's the schematic:
Jensen-Twin-Servo-1997.jpg

 
Not this particular circuit,  but did you use a ground plane? How is the power supply decoupling, you want capacitors as close to the opamps as possible.

Some schematic questions.  Why is there not a load isolator,  like X1, at the output of A2? The opamps feedback network is unusual too,  as it creates a variable pole dependent on the gain.
 
Look at layout... High gain should be more stable wrt op amp, but positive feedback crosstalk of output to input will be worse at higher gain.

Look at layout.

Did I mention "Look at layout"?

JR 
 
Everything that the other guys said 😃
But you should def have a load isolator on the output of the 2nd amp.  Going to 68r from 43r1 for maximum gain on both amps isn't the end of the world though, looks like you lose about 8dB gain across both amps.  So, about 60dB from the 990's plus another 6dB from the input transformer.  Should be a nice quiet amp too with those low feedback impedances👍
Good luck
 
If your dual pot is connected with wires (it isn't placed on the PCB), try to increase the EM  isolation between pot's decks using screened and grounded cables for each stage.  Also, you can try to increase C5 and C6 capacitors, but check then the frequency response at  high end and max gain. Do you have any picture of preamp's interior?
 
More specifically to what others have said, try making the ground for R3, R6 and the transformer screen BLK separate. Meaning connect those 3 grounds together but isolated from the other grounds and then run a separate ground wire back to the filter cap ground of the power supply.

Note that if you're doing this on a breadboard, you might have trouble getting it to work. You might have to skip straight to the PCB stage.
 
> positive feedback crosstalk of output to input will be worse at higher gain. Look at layout. Did I mention "Look at layout"?
 

Attachments

  • TwinServo-oscillate.gif
    TwinServo-oscillate.gif
    48.8 KB · Views: 117
The circuit should be 5-6 inches from input to output. Looping the circuit in a U form without a full shield can form a GREAT OSCILLATOR .The transformer leads need to be twisted. I also would add a local RC to each of A1, A2 & A23 on the PS volt lines. Typ decoupling of 10 to 100 ohms and 47uf to 100uf and .1uf bypass on each leg are good practice. The GROUND from R3, R6, R9 & C2 should connect to A1 COM and R12, R14 & C4 grounds should connect to A2 COM. 
Like all other said check for LAYOUT, LAYOUT & LAYOUT. Look @ PRR dwg.
 
Hi Guys,

thanks for your replies and suggestions  ;)

Turns out the potentiometer was the problem... to fit the front panel (500 series) I have to mount the pot sideways. I'm using a ground plane and the pot touching the pcb was causing the oscillation. I attached wires to the pot and I can now use the full gain.

Now I must figure out how to mount the pot sideways where it belongs without the oscillation problem. I'll start by removing the ground plane underneath. Maybe add a small piece of mu-metal between the pot and the pcb?

Cheers
Gabriel
 
elskardio said:
Turns out the potentiometer was the problem... to fit the front panel (500 series) I have to mount the pot sideways. I'm using a ground plane and the pot touching the pcb was causing the oscillation. I attached wires to the pot and I can now use the full gain.
The pot is not the problem. The problem is what is stated in the schematic notes and what other have been saying which is that the problem is the layout. Specifically, the traces on the PCB going to / from the pot have to be layed out in a very specific way. But how exactly is somewhat moot unless you're going to redo your PCB. So if you want to just finagle the PCB you have, you should just leave the pot soldered to the board but cut the traces with an xacto knife and then tack shielded cable in it's place. But you need to use two conductor with shield and incorporate R6 (or R12) into the assembly so that you can use the shield as ground and then the two conductors can be used for send / return. And the "send" should be mounted as close to R4 (or R11) as possible and the "return" conductor should be connected as close to -IN socket as possible.
 
squarewave said:
The pot is not the problem. The problem is what is stated in the schematic notes and what other have been saying which is that the problem is the layout. Specifically, the traces on the PCB going to / from the pot

I know the problem is not the pot itself... but probably the too long traces and bad ground layout  ;)

I'll start working on a PCB revision and I'll try to follow everyone suggestions.

Thanks
 
I don't know anything about 500 series. But you should post an image of your layout with clearly labeled parts because AFAIK no one in this thread has explained exactly how to layout things to minimize noise and the possibility of oscillation. For example, certain low level / high impedance / high gain related traces should have returns right next to or under them to cancel currents / minimize loop areas.
 
squarewave said:
More specifically to what others have said, try making the ground for R3, R6 and the transformer screen BLK separate. Meaning connect those 3 grounds together but isolated from the other grounds and then run a separate ground wire back to the filter cap ground of the power supply.
What?
R3, R6, R8 & R12 should be returned to COM of the 1st DOA, and the output xfmr black return wire to COM of the 2nd DOA. COM is the junction of the two 0.1uFdecoupling caps inside the 990.
Ground follows signal.
Returning these sensitive reference points to a dirty ground is a receipe for noise.

EDITED cause fingers faster than brains
 
abbey road d enfer said:
What?
R3, R6, R8 & R12 should be returned to COM of the 1st DOA, and the xfmr black wire to COM of the 2nd DOA. COM is the junction of the two 0.1uFdecoupling caps inside the 990.
Ground follows signal.
Returning these sensitive reference points to a dirty ground is a receipe for noise.
Respectfully no. As you state COM is just supply bypass caps so you DONT want those referenced to signal if you can help it.

But you and I have had this conversation before so I feel it's redundant.
 
squarewave said:
Respectfully no. As you state COM is just supply bypass caps so you DONT want those referenced to signal if you can help it.
Replace the "ground" connections with the actual resistance they present and you'll know why I wouldn't reference anything to the "filter cap ground of the power supply."

But you and I have had this conversation before so I feel it's redundant.
  Agreeing to disagree is not constructive. Having diverging opinions on a same subject from two knowledgeable persons is disturbing for other members.
 
abbey road d enfer said:
Replace the "ground" connections with the actual resistance they present and you'll know why I wouldn't reference anything to the "filter cap ground of the power supply."
  Agreeing to disagree is not constructive. Having diverging opinions on a same subject from two knowledgeable persons is disturbing for other members.
You guys scaring the children again?

JR
 
abbey road d enfer said:
Replace the "ground" connections with the actual resistance they present and you'll know why I wouldn't reference anything to the "filter cap ground of the power supply."
That's not helping me see what you're saying. Just explain what loop of components / grounds would be created that would cause noise / instability.

This issue makes me think of those ground plane inductance videos boji posted where the return follows the trace over ~1KHz even though there's a more direct path back to the supply. Those videos explain why low impedance bypass caps need to be right next to the source and not at the power supply.

So taking that into consideration you are right about R6 returning to A1 COM because most of the output current of A1 is coming form FB and not LOAD. So I have to revise my original statement to try running R3, R6, A1 COM and BLK on a separate ground. But there's no reason for A2 LOAD to be connected to any grounds of A1.

Also, with a careful layout it should not be necessary to use a separate ground wire. I was only suggesting that because the poster was experiencing oscillation so decoupling A2 LOAD from A1 +IN could have helped.

But still, in this particular circuit, I believe it might actually be better to use a separate ground for R3, R6, A1 COM and BLK because A2 LOAD throws a lot of current around so there might be an advantage to giving the parts on the input (R3, R6, A1 COM and BLK) a separate path to supply ground.

abbey road d enfer said:
Agreeing to disagree is not constructive. Having diverging opinions on a same subject from two knowledgeable persons is disturbing for other members.
If you think I'm saying something incorrect then by all means go ahead and say so. I'm a big boy. I could be wr-wr-wrong. I have been wrong before and when I realize I am I admit it. I just don't understand what you're saying in this case. You say connect transformer BLK to A2 COM? That makes zero sense to me. BLK is just for shunting common mode noise to ground. Capacitively coupling the +80dB output of A2 LOAD to A1 +IN through the transformer shield does not seem like a good idea to me.
 
One of the things I really like about electronic design is the objective performance metrics that are generally not open to debate or subjective opinions. That said people do still find things to disagree about.

You are both adults so figure this out... or not.

I do not care enough to wade in.  The OP said he has it working, and the application note seems to contain adequate general advice.

JR 
 

Latest posts

Back
Top