Questios about designing with opamps

[ddietz]

Hi everyone, as the title says I have a few questions regarding designin with opamps. In specific, I'm trying to better understand their use in mic preamps. For these questions, let's assume no phantom power and a dynamic mic like the shure 57 with a 300-500 impedance and output voltage of 1.6mV max.

I'd like to use the OPA2134 as reference for my questions.

Let's say I want to use the dual ic for a single channel with the first half having the balanced input into the in+ an in- respectively. The output then goes to pin 6, the second in+ for a second gain stage. An example schematic can be found at www.namad.ee/micros/micamp1.gif

How is the first and second gain stage calculated? The author claims 20dB and 40dB respectively but I don't get how.

Next question, I've read that mic pres should have an input impedance 5-6 times the mics output impedance. Since my mic is 300-500, that means my amp range should be 1k5 to 3k ohms. Yes? If that's the case, that impedance should be used to ground at both in- and in+, correct?

I've tried a simpler setup using the mic unbalanced and simply connected to one in+ of a opa2134, a 22x gain (66dB right?) and the output right to my interface for recording. It actually sounded pretty darned good, better than the tascam us800 pre and with lower noise. I feel like more gain is needed though. Will this change if i switch to using balanced and differential input. Basically I don know if using unbalanced by connecting the mics - and ground pins kills some power, adds noise, etc. Assume all this happens at the pre, located at the interface.

Would anyone be willing to give a quick description of biasing input voltages and such. I've seen it mentioned but never explained. In reality, while sure building a Green mic pre or Taos is the way to go for ultimate performance, I was amazed at how good a simple circuit sounded compared to want I'm currently using for recording. Honestly, I don't really have the cash for a dedicated Green pre at the moment.

Thanks all!

[SoNic_real_one]

22x gain is NOT 66dB. Is 27dB...

Use: gain [dB] = 20 log(Amplification [ratio])

[ddietz]

Thanks Sonic, I had forgotten how to convert that! Learned it recently but guess it did not stick

[sofaspud]

The gain is set by the feedback resistance. For the first stage, it's the resistors marked with an asterisk. The 10k and 1k form a 10:1 ratio; 10x is 20dB. For the second stage it's the 1k R10 and the 100k pot. 100x is 40dB. The pot is a gain adjust.
Next question... well, your math is correct. R4 & R5 in your example are 10k; that would probably work well for you also. I'm not sure 3k would necessarily be better. I just don't know.
22x (roughly 27dB) gain isn't much for a mic. Your example has a more practical 1000x (60dB). Balanced lines are used in low-noise apps because they utilize common-mode rejection. The noise is present on both lines (balanced) so it gets rejected by the circuit (an op amp amplifies the difference between its inputs).
Download a copy of TI's Op Amps For Everyone. It is pretty thorough in describing the basics of op amps.

[ddietz]

Thanks Sofaspud, appreciate the feedback.

I believe I understand now about the feedback. Previously I though the denominator for the feedback ratio was everything until you hit a ground. I assumed that one simy had to sum all the resistors. Guess it is always just the two resistors, the one between inverting and output, over the resistor closes to he inverting junction. Is that right?

Thanks for the link, I'll definatley check it out.

Another question though, any suggestions for how to do each gain stage assuming a max final gain of 60dB (1000x)? Best to do equal gains or small the big, etc. Though, given the amount of gain needed and the noise of parts, I'm guessing two chips should be used.

[sofaspud]

No, the gain calculations are little more involved than that. They are different for different configurations, eg inverting, non-inverting, buffer, differential. I wanted to stick with the example and not overwhelm you needlessly.
Do check out the link, and TI's site. They have lots of free and excellent op amp pdfs.
I think the example is a good design. There are other issues that come into play when you set op amp gain too high or too low. A 10x stage followed by the adjustable gain stage is a good compromise between performance and parts count/complexity IMO. There is no reason to have equal 32x gains for each stage.

[ddietz]

Thanks again for your response. Now let us see if I am getting any of this as reading through the TI Design pdf has brought up a few questions for me.

First I'd like to point out that the my link above to the example has a typo in it. It should be http://www.nomad.ee/micros/micamp1.gif and I will link it here for convenience. Note it is not mine and credit is as shown on the image.



From the TI Design pdf I see I have two different circuit types in the example. Stage 1 with the in+ and in- is a differential setup, which is useful in limiting source/line noise as it provides CMRR. Gain calculations for that are G= V1[-pin]-V2[+pin](Rf/Rg). I think the input voltage would be the same at each pin, but inverted, so the gain equation becomes -V - V(Rf/Rg) or -2V(Rf/Rg). Since V is small this is just -2(10/1) or -20. Gain is then 20log(-20)= ~26dB. Correct?

Stage 2 is inverted. The TI OPamps of Everyone pdf shows gain is G=-(Rf/Rg). IN the example above, max gain is then -100, GdB= 20 log(-100) = 40. Right? The negative part confuses me.

Pretending all the above is right, and we ignore the phantom power/diode I the above, what can I do to maximize performance and minimize noise using a OPA2134 or similar?

Thanks!

[sofaspud]

In this case, the negative sign denotes the signal is inverted. If you remember where inversion occurs, you don't need to include the sign in the formulae.
There are entire books on maximizing op amp performance. I think the most important things you could do are match the asterisk resistors and use a clean power supply. It looks like one DC blocking cap is labelled 4.7 and the other 7.4. An obvious typo.

[cbdb]

The common way to reduce noise in a opamp mic pre is to add a JFET LTP (drain coupled pair ?) at the input. The dominant noise source of almost all amps with voltage gain is the first stage.

[ddietz]

Thanks sofa and cbdb!

I figured that about the sign since I got the right answer by ignoring it

With a JFET input stage, that is similar to the ESR project 66 (sorry, hard to link on my phone). Right? Any starting links where I could read more into how that sort of thing functions?

Thanks again all. I threw together a quick perfboard version of the linked circuit using a virtual ground and 9v battery with the 2134. I ignored all the phantom power and diode stuff. Sounded better noise wise than the built in pres on my tascam interface. That's with the spagetti of allegator hook up wires. I think it needs a high pass filter. It is really clear but a little to clear on high frequencies with an acoustic guitar.