All OpAmp mic design (no FET at first stage)

[rogs]

Rogs does it yet again!
Have you considered opa1655 because of it's even lower noise? I forgot to include it in my last Mouser order

I did look at the OPA1655 and the noise figure seems to be even lower than the 1641 ..... They don't always make reading data sheet noise figures very easy, and often quote different parameters! .
Whether it would make much difference in most 'real world' applications, I'm not sure?
Certainly it's ambient noise that dominates in the sort of environments I use my mics....

It doesn't state on the data sheet front page bullet points that the 1655 amp is unity gain stable. (They make a special point of mentioning that in the OPA164* data sheet ) It may well be of course?.... Just needs to be checked for unity gain buffer use!

The 1655 draws twice as much current at the 1641 .... still not that much, but it's not a plus point.
The input impedance is a lot lower on the 1655 - only about 100M in differential mode. That may be a bit low?

The input capacitance is also better on the 164* series according to this comment from a TI engineer posting on a different forum:
"The OPA1641 would be slightly better. It has the most stable input capacitance vs. common mode voltage that I've measured on any op amp. When we developed the OPA1656, this was also a major design target for us, and the designer and myself spent quite awhile in the lab with previous CMOS audio op amps (specifically OPA1652 and OPA1688) hunting for sources of input capacitance variation. The end result was slightly better than the previous parts, but we still met some limitations from the CMOS process itself. The dielectrically isolated JFETs on the input of the OPA1641 are superior in this regard."
Looks like the genuine JFET input of the 164* series may win out there?

Probably all just marginal differences in reality of course?

 

[soliloqueen]

a lower current version of the 1641 does exist in the OPA145 (it's specified for different use cases, but electrically that's more or less what it is), but it's 50% more expensive. not really much use for it considering it has slightly higher noise and distortion and phantom power already supplies enough power for 2 channels of the 164x, but i thought i'd share

 

[rogs]

a lower current version of the 1641 does exist in the OPA145 (it's specified for different use cases, but electrically that's more or less what it is), but it's 50% more expensive. not really much use for it considering it has slightly higher noise and distortion and phantom power already supplies enough power for 2 channels of the 164x, but i thought i'd share

My original OPIC multi-pattern mic used 3 x 164* op-amps .. (1 x 1642 and 1 x 1641). That was to enable it have a level adjustment, to allow matching of the output levels of wayward dual sided capsules....
That did of course drop the available phantom power volts available to power the devices, which reduced headroom as a result.
It was also fractionally more noisy - altough that wasn't really too much a of a problem, in practice.

When I did the OPIC.45 multipattern version that was designed to use one of my Arienne Audio 'flat 47' capsules, so I didn't need to worry about balancing output levels, and could resort to using a single OPA1641.

I think the next version might use the OPIC 45 circuitry, plus a transformer coupled output.
That might well produce the best mic I've done yet?......

• Single OPA1641 for minimum noise and maximum headroom.
• 'Flat 47' capsule for a lovely smooth FR, and simple pattern control.
• Transformer coupled output for a little extra noise free gain - and a respectable CMRR.

I don't actually need any more mics of course... I just can't resist experimenting!

 

[tk@halmi]

With the help of an opamp, ironically, it is also possible to build auto biasing (servo augmented) discrete JFET amplifiers, and eliminate the need for manual tuning.

You still need to know the target device parameter range, and set sensible values for source and drain resistors, but you would not have to tune for individual 2N4416 devices, for example.

 

[rogs]

With the help of an opamp, ironically, it is also possible to build auto biasing (servo augmented) discrete JFET amplifiers, and eliminate the need for manual tuning.

You still need to know the target device parameter range, and set sensible values for source and drain resistors, but you would not have to tune for individual 2N4416 devices, for example.

Would the auto bias seek out lowest distortion - or symmetrical clipping?. AFAIK those two parameters are not often (ever?) coincident.....
If op-amp designers are able to reduce the noise floor of JFET op-amps by just a couple more dB, we could probably dispense with discrete JFETs altogether, at the front end of condenser mic impedance converters ?....

 

[Mr. Moscode]

I was looking at the recent thread about driving transformers from op-amp outputs, and I thought - I've not yet tried an OPIC version with transformer coupled output?....
So - I dug out a old Neutrik NTE1 transformer (reasonable price and quality) and added a transformer to the 'standard' OPIC circuit output:

View attachment 140047

Seems to work pretty well !

I replaced the capsule with a 68pF capacitor, and applied a test signal from my NTI Minirator generator, (which outputs a pretty decent signal).
Working into a Sound Devices USB Pre pre-amp, the response is within 0.5dB from 40Hz to 20KHz, and only c. 1.5dB down at 20Hz. ... Pretty impressive for such a tiny transformer!
Distortion figures yet to be finalised, but it sounds OK - and looks pretty good on a 'scope...
Turns a single sided OPIC output into a differential one (so some extra noise free gain!) and makes for a pretty impressive CMRR.
Still experimenting with the value of the resistor in series with the op-amp output (only included because I've had problems in the past with op-amps driving purely reactive loads).... I suspect the value is not that critical?

As usual, I've made notes along the way - if only to remind me in the future what I've done -- my memory is not what it was!

PDF notes here: https://www.jp137.com/lts/OPIC.TX.pdf

You could change R5 & 6 to 1M ea which could make C4 smaller, say 10ufd to lower the freq rejection of the PS. The rise in voltage, currently 15V, for the op amp would not be much, though. You may be able to make R3 & 4 smaller , say 4.7K ea., which could help damp the OPT and give the OPA more headroom.

It's also a candidate for adding gain in the NFB loop.

I'd like to see square wave response at 10Khz or a freq sweep to unearth the resonance(s) or ringing F of the OPT under different terminations. A snubber may be in order.

 

[Mr. Moscode]

Would the auto bias seek out lowest distortion - or symmetrical clipping?. AFAIK those two parameters are not often (ever?) coincident.....
If op-amp designers are able to reduce the noise floor of JFET op-amps by just a couple more dB, we could probably dispense with discrete JFETs altogether, at the front end of condenser mic impedance converters ?....

Downside might be the input pin of the op amp has to be off the board do avoid the effects of humidity, dirt accumulation and problems in seaside salt air environments. I've seen it in lower Z circuits before.

 

[rogs]

You could change R5 & 6 to 1M ea which could make C4 smaller, say 10ufd to lower the freq rejection of the PS. The rise in voltage, currently 15V, for the op amp would not be much, though. You may be able to make R3 & 4 smaller , say 4.7K ea., which could help damp the OPT and give the OPA more headroom.

It's also a candidate for adding gain in the NFB loop.

I'd like to see square wave response at 10Khz or a freq sweep to unearth the resonance(s) or ringing F of the OPT under different terminations. A snubber may be in order.

Like all my stripboard projects, I simply offer ideas for other hobbyists to experiment with...
That's where stripboard comes in useful -- Easy to modify - oh, and cheap!
Be interested to hear of the results of any changes you make -- if you decide to experiment?..

In the meantime, I shall endeavour to see if I can measure anything useful regarding any non linearities introduced by the transformer, into different preamp inputs....

 

[tk@halmi]

Would the auto bias seek out lowest distortion - or symmetrical clipping?. AFAIK those two parameters are not often (ever?) coincident.....

The servo will force the JFET source voltage you preset. This can ensure some symmetry... is all it can do.

 

[Mr. Moscode]

The servo will force the JFET source voltage you preset. This can ensure some symmetry... is all it can do.

You can find the sweet-spot, i.e. symmetrical clipping by adjusting ratio of R5 & 6 and not waste power on a servo. Once found it should stay that way - UNLESS - there is DC leakage around the capsule gigohm Rs.

Also I would think you may want C11 from the earlier circuit to avoid subsonics from saturating the OPT. Without the OPT you'd have some low freq extension.

Have fun.

 

[Fenton Bresler]

Greetings all,
I'm reading about headroom in this thread but I'm struggling to see how much is needed here. How large of a supply does the opamp really need? Am I very much mistaken or isn't a capsule putting out tens of mVs? I've seen this circuit with a 12V zener on the supply. Isn't 12V major overkill? I'm guessing I'm missing something crucial here, I would very much appreciate if someone could help me understand.
Thanks in advance.

 

[ccaudle]

Am I very much mistaken or isn't a capsule putting out tens of mVs?

A separate thread by forum member @kingkorg found that a condenser capsule could generate 22V p-p when close to a drum.

 

[tk@halmi]

A separate thread by forum member @kingkorg found that a condenser capsule could generate 22V p-p when close to a drum.

Those are peak transients. Don't feed that to your DAW without limiting.

 

[Khron]

Those are peak transients. Don't feed that to your DAW without limiting.

... But surely you'd want that limited, as opposed to "clipped within the mic's circuitry", wouldn't you?

 

[kingkorg]

Those are peak transients. Don't feed that to your DAW without limiting.

This was exactly my intention, so i could control them in the daw. Not to mention it helps tremendously with triggering if necessary. So why not?

 

[Fenton Bresler]

Genuine question, how do you get +26dBu into your DAW?

 

[rogs]

Genuine question, how do you get +26dBu into your DAW?

As a 32 bit float signal....
Actually +26dBu is a tiny bit high, but my Zoom F3 recorder - which only records in a 32 bit float format - has a max line input signal level of +24dBu (34v p-p )

(Some useful details about the 32 bit float format here: https://www.sounddevices.com/32-bit-float-files-explained)

 

[kingkorg]

Genuine question, how do you get +26dBu into your DAW?

I was using M-Audio 2626 for testing, which has pad on the inputs, no issues whatsoever.

The other option i used was Schoeps circuit with just 10V of polarization voltage. Multiple ways to skin the cat. The point would be, condenser capsules can put up with way more we give them credit for. And since there's a lot of discussion regarding microphone transient response, this info is quite useful. Even with dynamic mics at this position i was getting several volts p/p.

 

[soliloqueen]

It's funny because vintage circuits change performance on high transient sources like drums a lot and for a long time I saw people on here being like "vintage circuits can't affect sound that much because they only affect much higher levels than a capsule would output" and then Kingkorg proved that capsules are actually really hot in terms of voltage. Of course tube and transformer microphones sound different on drums. The transients are huge! They're distorting!

 

[tk@halmi]

Genuine question, how do you get +26dBu into your DAW?

Most AD converters in audio interfaces these days run on 3.3, 2.5 or most recently on 1.8 volts. And convert nicely slightly less.

No point in bringing 26dB snare into the DAW either. We are long past tape days when we had to fight high noise floor.