Noisy audio preamp circuit, how to reduce hiss?

[Curi0sity]

Hello all,

I am designing a microphone preamp with some odd conditions. The microphone is dynamic. The impedance of the microphone is around 4 to 8 ohms (very odd mic, M-87/AIC). The goal is to interface the mic with a PC sound card which is designed to operate with an electret type. I have designed a circuit utilizing an op amp and an audio transformer. The circuit provides the desired gain and interfaces with the sound card appropriately, however the circuit has an annoying noise floor that I am trying to get rid of.

The circuit features both high and low pass filters so it should eliminate any noise outside of the audible range which it seems to do when I put it on a scope. I also understand that a transformer will inherently pick up noise. However even with the transformer removed from the circuit there is still a white noise present. I have done a lot of reading and reduced my resistance values to do my best at eliminating resistor noise. I cannot quiet this thing down. I'm hoping you geniuses can guide me in the right direction.

Some specs on the circuit;
The op amp is a OPA2134PA
VCC is +9vdc (currently a battery)
I have added bypass capacitors to the op amp and the 'power supply'
The gain is set to 201 (the mic is very low level)

Here is the circuit

Hopefully I didn't forget any pertinent information.

Thank you for looking! any advice is appreciate.

 

[Hesch]

However even with the transformer removed from the circuit there is still a white noise present.

Try using an amplifier with lower noise, for example:

http://www.analog.com/en/products/a...amplifiers-greaterthanequalto-12v/ad8599.html

Also use low noise resistors. ( Google ).

 

[davenn]

dunno why type of resistors you have used ?
if carbon, change to metal film ... less noisy.Dave

 

[tfr000]

What do you hear with nothing attached to the sound card input?
You say the sound card wants an electret mic? Impedance in the ~1k-2k range. Your circuit is probably pretty low output impedance, although the 4.7uF would add some at low freq.
What kind of caps did you use?

 

[Averagesupernova]

Typically a low Z mic will have a lower voltage output than a higher Z. So, right off the bat you are that much closer to the noise floor. Unless I missed it, I don't see the spec of your transformer. Isolate the noise by starting at the mic and short circuiting the input. Move closer to the sound card input to try to determine where you are picking up the noise. Obviously be careful to prevent damage to the op-amp.

 

[meBigGuy]

It is a bizarre microphone. More like a speaker-as-mic. You can search for speaker-as-mic.

1. Use an ultralow noise amplifier as above or http://www.linear.com/product/LT1115
2. Use metal film resistors
3. Need good Power supply bypassing (electrolytic or tantalum with a ceramic) and good layout (small summing junction, etc)
4. Some series impedance in your power supply line before the amp would be a good idea

If you power it from a battery you can determine whether you are getting supply noise.

You need a bypass capacitor on the supply to the 10K resistors with a small resistor feeding it. Currently any hiss on your supply goes right into the amp (well, it's divided a bit).
Something like 10 ohms into 10uF in parallel with 0.1uf ceramic and that feeds your 10K resistor.

You might be better off with a well bypassed dual supply so you didn't need the bias resistors.

Reducing the gain resistors might help, but I'd look carefully at the opamp specs for the best noise performance.

Mic Spec for frequency Response is: 200 – 6000 Hz, so roll it off more. 17Khz is too high for that mic.

I'm really surprised on how few transformer options there are and how expensive they can be. More transformer gain would be a winner, if you can find/afford it.

 

[Curi0sity]

Davenn - I did use carbon film resistors, just stock from the electrical prototyping bin. I will order some low noise resistors. I didn't realize resistors could cause so much noise (I suppose I could calculate it...) Thanks

TFR000 - With nothing connected to the sound card input the output is silent. With the circuit connected and no audio transformer I hear white noise and some RF interference. With the Transformer and mic connected to the circuit I hear ALOT of 60hz buzz and ALOT of RF interference. (This is weird because the RF should be filtered out). The caps are polypropylene. (I did try ceramic and it didn't make any difference)

Averagesupernova - Woops, forgot to post the transformer I'm using. It's a Tamura MET-35
see; http://www.tamuracorp.com/clientuploads/pdfs/engineeringdocs/MET-35.pdf

meBigGuy -
1. I will check into the other amps and get some ordered.
2. I will use metal film resistors
3. I am currently using a 9V battery with a 100uF electrolytic and a 0.1uf ceramic as bypass just to ensure clean power. The circuit is still noisy so it's not from power.
4. I don't understand what adding series impedance to the power supply does. Would you mind elaborating on this a bit so I can read about it?

I didn't think about adding a bypass cap to my voltage divider, thanks for this I will try it.

My goal is to drive this with a single supply, but I will try a dual supply design and if it's significantly quieter I will go this route.

I did think about reducing the gain and running this output through a SECOND op amp to pick up some more gain. Would this do anything for reducing noise or would it just add gain to the noise floor? (I would imagine it would just amplify the noise floor in addition to the signal)

I will change my capacitor values to narrow the pass band. I had the circuit passing the entire audio spectrum just for testing.

Thank you everyone for your responses. I'm pretty inexperienced with analog electronics (though I do have very much experience with digital electronics). So this is all pretty new to me.

I'll make some changes to the circuit and post results. Thanks again!

 

[nsaspook]

Is the Mic and transformer input completely insulated from ground with twisted pairs for that wiring? They should be so it can help cancel common-mode noise like 60HZ buzz and induced hash but It won't help with amplifier generated noise.

 

[Jeff Rosenbury]

It's my understanding that high gain comes with high noise. It's one of the reasons to use multistage amplifiers with lower gain per stage.

Also you might consider filtering before the transformer. The symbol for it is iron core, which is non-linear (only slightly so at low power levels though). But since it is an audio transformer it may be air cored anyway. Non-linearities might move noise from outside into the pass band.

Remember any noise before the first amplifier stage gets locked into the system.

 

[Svein]

Whenever I see a schematic with general "earth" symbols, I shudder. The Signal- lead should be treated as that and connected to power ground at one specific place. It helps if you draw a resistor (value ≈0Ω) between Signal- and power ground.

 

[meBigGuy]

I misspoke. A small resistor in series feeding a capacitor to ground to filter the power. Not simply a series resistor in the power line.

 

[Averagesupernova]

A regular high Z mic will have an output impedance of 50K or so. I have seen transformers that transfer a 600 ohm impedance up to that 50K. That ratio is ABOUT what you have there. However, you are starting out with much less than that for an output Z. Wondering what would happen if you fed the op-amp directly in balanced mode. Treat the op-amp stage as somewhat of a current to voltage convertor. Run a drain wire along with the balanced mic cable.

 

[meBigGuy]

I have seen transformers that transfer a 600 ohm impedance up to that 50K.

I looked for something like that and didn't find it. I think a second transformer from 600 to 50K would be the easiest way to get noiseless voltage gain.

 

[sophiecentaur]

The source resistance can be very relevant in low noise circuits. Why not try a range of resistor values across the input and see how that affects the noise level at the output. If, as I think i read, the noise drops when the input is OC, it implies that the device needs a high source resistance.
I remember, many years ago, I connected the pins on a stereo gram pickup head wrongly and the noise level was terrible (not just the SNR). Connecting the right resistance cut the noise to well below the fridge and the gas fire noises.

 

[tfr000]

TFR000 - With nothing connected to the sound card input the output is silent. With the circuit connected and no audio transformer I hear white noise and some RF interference. With the Transformer and mic connected to the circuit I hear ALOT of 60hz buzz and ALOT of RF interference. (This is weird because the RF should be filtered out). The caps are polypropylene. (I did try ceramic and it didn't make any difference)

The caps should be fine. The fact that you are getting lots of 60hz and RF says to me that you have long, unshielded wires dangling... or you live near a radio station. Other than that, any transformer is going to have lots of wire to act like an antenna, unless maybe you use it differential mode like one of the others suggested.
It's possible you have designed on oscillator - what happens if you disconnect C1 or C2?
What if you put 1k resistance in place of C3? Maybe the opamp doesn't like big capacitive loads.

 

[meBigGuy]

my bad --- How did I miss "hear white noise and some RF interference. With the Transformer and mic connected to the circuit I hear ALOT of 60hz buzz and ALOT of RF interference". I got it my mind somehow that he was talking hiss only.

By RF interference, are you saying you hear audio from AM radio stations?

The opamp could be oscillating, among other things (like poorly bypassed supply leads, long summing junction, long input wires, etc.).
We need a full schematic with all capacitors, and a picture of the layout.
One thing is that the + input is susceptible to noise input, so needs to be as short as possible and the bias may need to be filtered.

I can only take a shotgun approach to this.

Step 1: Make the opamp with no transformer quiet when connected to a soundcard. (the 1.2K is adequate input termination).
a. Is the power supply bypassed right at the opamp with at least 0.1 ceramics
b. Is the ground system short and sweet
c. What happens if you place 1uf (or 0.1 uf) ceramic to ground on the +input? If the noise goes away, it could be getting injected through the bias resistors
(maybe add a bias supply filter as I described earlier)
d. What happens if you disconnect the battery?

When you hook the transformer up, you have a new class of issues.
You may need rf filtering in series with the mic or maybe the transformer output
You should replace the mic with an 8 ohm resistor to see if the mic is sourcing hum or noise.