3
\$\begingroup\$

I'm trying to design a preamp for this electret mic.

I kind of pieced this schematic up from something I have found on the web.

Schematic

Does that look ok? I'm going to sample the output with an ADC (internal in an MCU), and run an FFT on it.

The opamp is a rail to rail op amp, it doesn't have low voltage offset though, should I care about it? The IC has 4 op amps on it, should I add another stage or two and lower each stage's gain ? Or should I use the rest as active filters ?

Do I need any filtering or anything like that?

What is the best way to set the gain?

I was thinking of using a potentiometer for R7.

Here is an ac analysis from LTspice:

AC Analysis

Here are the LTSpice files for the simulation.

\$\endgroup\$
  • \$\begingroup\$ I'm not sure of your simulation, that circuit should go up to 80dB in its central bandwidth. And low offset (low noise) op amp are a must for such high gains. \$\endgroup\$ – Vladimir Cravero May 22 '14 at 21:46
  • 1
    \$\begingroup\$ I'd recommend AC coupling the second op amp. Capacitor to input and resistor network just like the first op amp. \$\endgroup\$ – WhatRoughBeast May 22 '14 at 23:32
  • \$\begingroup\$ @WhatRoughBeast . +1 that does help with the offset voltage. Good thinking \$\endgroup\$ – Marla May 23 '14 at 0:01
4
\$\begingroup\$

Generally, okay!

  1. The DC gain of each stage is one, so a few mV of offset voltage is of no importance at all.

  2. The gain seems a bit on the high side unless it's some kind of hidden microphone. Remember that the V/Pa sensitivity is always quoted for RMS 1 volt, and you'll start to get clipping around 3.2V peak-to-peak.

  3. You might want to put an RC filter in the bias circuit of the electret. Something like a few hundred ohms and 100uF to ground.

  4. Pot for R7- fine, you might want to switch it with the series cap and ground the pot, it will work. You could use something like a 100K rheostat-connected-pot with a 4.7K in series. A log 'A' taper (volume control) pot connected as a divider between the two stages would be better. You'd have to do something like AC-couple the wiper to the second stage with a resistor pair as in the first stage to maintain the bias at 1.65V.

  5. If you're sampling at 50-100kHz you should not need any filtering, otherwise you might want to consider an anti-aliasing low-pass filter with a cutoff below half your sampling frequency.

  6. The LMV324 at least is low voltage and has no crossover distortion but it's pretty noisy. Even with perfect shielding and no power supply noise, I'd expect about 0.25V p-p noise at the ADC input with the given gain, over a 20kHz bandwidth.

Edit: Schematic of capacitively coupled gain pot below. C1 (maybe 100n) is not strictly necessary, but it avoids wasting DC current in the pot. C2 could be 33n as in the input coupling.

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
  • \$\begingroup\$ Does the lmv324 have a high enough slew rate for full output at 20khz? \$\endgroup\$ – Bitrex May 23 '14 at 3:50
  • \$\begingroup\$ @Bitrex good for about 100kHz with 3.3V supply according to the data sheet fig 24. \$\endgroup\$ – Spehro Pefhany May 23 '14 at 4:01
  • \$\begingroup\$ Thanks, I know very little of microphones and op amps. I really didn't (still don't) know how much gain I really need. I used this spice model of a mic I found, and found out that a conversation from 1m away is 60dBSPL, so i plugged those numbers in the model. The mic will be about 30cm away from some speakers which play music. Should I lower the gain ? Could you explain about the AC coupling of the pot you mentioned please ? Lastly, I could use a different op amp (this one is easily available for me). Any suggestions ? (I'll be searching for them in ebay most likely). \$\endgroup\$ – Mike May 23 '14 at 6:55
  • \$\begingroup\$ Maybe start with gain about 1/3 or so of what you have. I'll edit to add the pot schematic. You could see if the noise is okay for you, or do a parametric search.. noise voltage for that part is 46nV/sqrt(Hz), so 1/5 or 1/10 of that would be good, and not too high noise current (pA/sqrt(Hz)). \$\endgroup\$ – Spehro Pefhany May 23 '14 at 10:07
  • \$\begingroup\$ Would it be a good idea to use two different op amps, a low noise for the input stage, and a rail to rail for the 2nd stage ? \$\endgroup\$ – Mike May 23 '14 at 11:41
1
\$\begingroup\$

Your circuit and simulation look reasonable. Provided that the model for Op-Amp is accurate. I did download your LT Spice files and view. A cascade amplifier of gain nearly 5000 will yield results that might require shielding of the amplifier.

  1. Your amplifier seems to be amplifying out to 100 kHz. Might want to roll off that gain a bit with a follow up filter (as you said you have extra units in the package).

  2. You have very high gain here, and anything over about 700 uV from the electret microphone will start clipping (distortion) at the rails.

  3. Your resistance values look good for balancing Op-Amp currents, but Offset voltage (as indicated in comment above) could trip you up.

Using a potentiometer in place of R7 to adjust gain works. Go for it. Build it, and then enjoy seeing the differences between a simulator program and real life.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.