theory questions relating to mic preamps

[timtimtim]

A couple of theory questions mainly directed at MSH since he's the man with the in-depth knowledge of these things round here (if he would be so kind)

The inamp THAT1512 is specified as 1nV/root Hz noise at gain 1000, and 34nV at gain 1. Would I be correct in assuming it will be 3.2nV at gain 100, and 10.5nV at gain 10? If so there's obviously no point in using it for a condensor mic with a high output, 5534 will be much quieter at low gains.

Why is balanced wiring really so much better than normal? If you twist the signal and the gnd return wires together you will get just the same cancellation of interference pick-up as you get with the two twisted antiphase wires of balanced wiring.

 

[mshilarious]

A couple of theory questions mainly directed at MSH since he's the man with the in-depth knowledge of these things round here (if he would be so kind)

The inamp THAT1512 is specified as 1nV/root Hz noise at gain 1000, and 34nV at gain 1. Would I be correct in assuming it will be 3.2nV at gain 100, and 10.5nV at gain 10?

You'd need to check the datasheet; it won't be linear. Inamps are two-stage amps; the input stage of the THAT is really quiet; the output stage less so. Until the input noise is given enough gain to get over the output noise, you'll pretty much have the output noise (divided by gain for EIN). Once you get above the output noise, then you just have the input noise. So figure with a gain of 4 (12dB) you have EIN of 9nV or so; that should be quiet enough for most condensers.

If so there's obviously no point in using it for a condensor mic with a high output, 5534 will be much quieter at low gains.

There's really no point in worrying much about preamp noise with a high-output condenser mic at all, so yes 5534 is fine. Really, you only need quieter for very low sensitivity mics on quiet sources. Personally, in that situation I usually just use a more sensitive mic . . .

Why is balanced wiring really so much better than normal?

Yes. When you need it, you need it. If you are worried about noise performance, you would thus also be worried about interference. Unbalanced mic-level wiring is not feasible in my studio, and I don't even like unbalanced line-level signals as I can measure the interference there too.

If you twist the signal and the gnd return wires together you will get just the same cancellation of interference pick-up as you get with the two twisted antiphase wires of balanced wiring.

No, twisted pair is a method to ensure that the interference signal on each lead is exactly the same so that the common mode rejection of a balanced amp is ideal. It does no good to twist a pair of unbalanced leads.

http://www.bluejeanscable.com/articles/balanced.htm

If you are wed to the inamp topology (I rather like it myself), you can roll-your-own with 5532 + 5534 for low noise at low gain--or pick an even quieter dual opamp for the input (5534 does make a pretty good output driver). Just be sure to have very closely matched resistors.

 

[Minion]

I can"t really help you with your first question but the second I think I can help with .....

Why is ballanced better than unballanced Twisted pair ??

Well , twisted pair only offers EMI rejection but can still pick up other sources of noise like possibly RFI and Power supply noise and also crosstalk from other adjacent cables ......

Ballanced differential amps which are what mic preamps are , compare the signals on both Hot and Cold signals and rejects anything that isn"t common to both so any induced noise is rejected , also ballanced lines offer 6db more headroom than unballanced lines ......

Cheers

 

[timtimtim]

You'd need to check the datasheet; it won't be linear. Inamps are two-stage amps; the input stage of the THAT is really quiet; the output stage less so. Until the input noise is given enough gain to get over the output noise, you'll pretty much have the output noise (divided by gain for EIN). Once you get above the output noise, then you just have the input noise. So figure with a gain of 4 (12dB) you have EIN of 9nV or so; that should be quiet enough for most condensers.





So with a gain of 16 it would be EIN of 2.25nV?

With 5534 its 3.5nV and I've measured the noise in a narrow band around 10kHz and I reckon the amp is adding 1.6dB to the total noise (amp + mic) at that frequency, which isn't a lot but if I can improve it just that little bit more by using a THAT then it might be worth doing?
Or would it be better to use two lower noise opamps? After my experience with AD8099 I'm wary, but would any of these do: LT1028, ISL28190, ISL53190, TSH300? Or any others? (must run at 9v)

 

[timtimtim]

I can"t really help you with your first question but the second I think I can help with .....

Why is ballanced better than unballanced Twisted pair ??

Well , twisted pair only offers EMI rejection but can still pick up other sources of noise like possibly RFI and Power supply noise and also crosstalk from other adjacent cables ......

Ballanced differential amps which are what mic preamps are , compare the signals on both Hot and Cold signals and rejects anything that isn"t common to both so any induced noise is rejected , also ballanced lines offer 6db more headroom than unballanced lines ......

Cheers

I know balanced is better but I'm just trying to see why. Any interference picked up by twisted balanced wires will cancel, but surely that is so for an unbalanced twisted pair too because not only EMI but also RFI will be picked up equaly by the signal wire and the ground wire and so will cancel?

 

[mshilarious]

So with a gain of 16 it would be EIN of 2.25nV?

*In theory* if you blend 1nV * gain + 34nV, you can solve with this equation:

nVtotal = ((1nV * gain) ^ 2 + 34nV ^ 2) ^ 0.5

EIN = nVtotal / gain

So for gain of 16, that would yield 37nV and EIN of 2.3nV

That said, real life is a bit more complicated. First, you have to consider the noise spectrum; above 10Hz or certainly 100Hz an amp is going to have white noise across the audio spectrum. Mics do not; they have 1/f noise from the capsule until that noise drops below the noise from the FET (which is white). So solving the required noise of a mic amp is not necessarily straightforward. Certainly the mic will be (much) noisier below 500Hz or so.

It's often easiest to measure the mic amp noise with 150 ohm termination, and measure the mic noise with a capacitor replacing the capsule. Knowing both of those, you can subtract the amp noise to derive the mic noise.

With 5534 its 3.5nV and I've measured the noise in a narrow band around 10kHz and I reckon the amp is adding 1.6dB to the total noise (amp + mic) at that frequency, which isn't a lot but if I can improve it just that little bit more by using a THAT then it might be worth doing?

10kHz sounds about right for an average mic/mic amp combo. But is 1.6dB worth chasing after? Human sensitivity to noise drops quite a lot above 10kHz. If it were 3-6kHz, I would say yeah go after it, but only if you are using enough gain at the mic amp to matter.

Or would it be better to use two lower noise opamps? After my experience with AD8099 I'm wary, but would any of these do: LT1028, ISL28190, ISL53190, TSH300? Or any others? (must run at 9v)

The only one I've tried that I can think of is ADA4841-2, that's my latest favorite.

 

[mshilarious]

but surely that is so for an unbalanced twisted pair too because not only EMI but also RFI will be picked up equaly by the signal wire and the ground wire and so will cancel?

No. Interference is picked up by both wires, but the interference on the grounded lead is shunted to ground, so the interference voltage will be different. And even if it wasn't, you'd still need a balanced input amp to cancel the common mode signal.

Try building a mic cable that shorts pin 3 to 1 and see for yourself.

 

[timtimtim]

*In theory* if you blend 1nV * gain + 34nV, you can solve with this equation:

nVtotal = ((1nV * gain) ^ 2 + 34nV ^ 2) ^ 0.5

EIN = nVtotal / gain

So for gain of 16, that would yield 37nV and EIN of 2.3nV

That said, real life is a bit more complicated. First, you have to consider the noise spectrum; above 10Hz or certainly 100Hz an amp is going to have white noise across the audio spectrum. Mics do not; they have 1/f noise from the capsule until that noise drops below the noise from the FET (which is white). So solving the required noise of a mic amp is not necessarily straightforward. Certainly the mic will be (much) noisier below 500Hz or so.

It's often easiest to measure the mic amp noise with 150 ohm termination, and measure the mic noise with a capacitor replacing the capsule. Knowing both of those, you can subtract the amp noise to derive the mic noise.



10kHz sounds about right for an average mic/mic amp combo. But is 1.6dB worth chasing after? Human sensitivity to noise drops quite a lot above 10kHz. If it were 3-6kHz, I would say yeah go after it, but only if you are using enough gain at the mic amp to matter.




The only one I've tried that I can think of is ADA4841-2, that's my latest favorite.

ADA4841-2 looks excellent. I think I'll go for it. Thanks MSH

 

[timtimtim]

No. Interference is picked up by both wires, but the interference on the grounded lead is shunted to ground, so the interference voltage will be different. And even if it wasn't, you'd still need a balanced input amp to cancel the common mode signal.

Try building a mic cable that shorts pin 3 to 1 and see for yourself.

Of course how obvious, why didn't I see that?

 

[commanda]

Ballanced differential amps which are what mic preamps are , compare the signals on both Hot and Cold signals and rejects anything that isn"t common to both so any induced noise is rejected , also ballanced lines offer 6db more headroom than unballanced lines ......

Cheers

Balanced differential amps which are what mic preamps are , compare the signals on both Hot and Cold signals and rejects anything that IS common to both so any induced noise is rejected.

CMRR (common mode rejection ratio).

Amanda