All OpAmp mic design (no FET at first stage)

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I've used both the OPA1641 and 1642 in my various projects here: OPIC Impedance Converter
Unlike the design linked to in the 'instructable' above, I don't regulate the supply to the op amp, and get a little extra headroom as a result.
Using the single amp 1641 draws half the current, so that helps a bit as well!
The audio output is single sided - although the output is still impedance balanced - so there is no additional noise introduced by the resistors in the inverting second amp used for a differential audio output.
I've found the various configurations I've tried all work pretty well .. low distortion and low noise....
@rogs I tried dropping you a DM, bot i'm not allowed for some reason.

I just built your opic circuit on a perfboard for an extra Akg electret i have. Just as i was finishing the build i realized the capsule is coupled to half rail instead of ground on one end. The only reason i can think of is not to present half rail dc voltage across the capsule which might interfere with polarization charge within capsule. Is this the only reason, or am i missing something?
 
@rogs I tried dropping you a DM, bot i'm not allowed for some reason.

I just built your opic circuit on a perfboard for an extra Akg electret i have. Just as i was finishing the build i realized the capsule is coupled to half rail instead of ground on one end. The only reason i can think of is not to present half rail dc voltage across the capsule which might interfere with polarization charge within capsule. Is this the only reason, or am i missing something?
The main reason for biasing the input that way was to try get away with only using a single 1G resistor, and keeping the circuit as simple as possible!
It will depend on the polarisation of the electret capsule whether that bias voltage is additive or subtractive....and that will of course affect the sensitivity a little.
Again, how much will depend on what the value of the 'baked in' polarisation voltage is, for any individual capsule.

When it came to using the same circuit with an LDC capsule - but without a voltage multiplier - then the loss of some 12v out of 48V could well be significant, so I included a second 1 G resistor to allow the op-amp to be referenced to 'half rail', and the capsule to be referenced to ground. That works pretty well.

This op-amp has quite a low noise figure, so even with only 48v as a polarisation voltage it's possible to get pretty good results. Especially with low noise transformerless mic preamps -- like those found in many of todays' budget audio interfaces.
It's a simple circuit. No FET bias worries, low distortion, decent headroom, and pretty low noise. The audio is single sided, but the line is still passively impedance balanced to help optimise the CMRR.
No reason that input configuration - with the capsule referenced to ground - shouldn't be applied to a FETless electret as well, as far as I can see, if the extra sensitivity is important?

I find it works pretty well.
(Some notes on the input with the extra 1G resistor here: https://www.jp137.com/lts/OPIC42.pdf )
 
Last edited:
The main reason for biasing the input that way was to try get away with only using a single 1G resistor, and keeping the circuit as simple as possible!
It will depend on the polarisation of the electret capsule whether that bias voltage is additive or subtractive....and that will of course affect the sensitivity a little.
Again, how much will depend on what the value of the 'baked in' polarisation voltage is, for any individual capsule.

When it came to using the same circuit with an LDC capsule - but without a voltage multiplier - then the loss of some 12v out of 48V could well be significant, so I included a second 1 G resistor to allow the op-amp to be referenced to 'half rail', and the capsule to be referenced to ground. That works pretty well.

This op-amp has quite a low noise figure, so even with only 48v as a polarisation voltage it's possible to get pretty good results. Especially with low noise transformerless mic preamps -- like those found in many of todays' budget audio interfaces.
It's a simple circuit. No FET bias worries, low distortion, decent headroom, and pretty low noise. The audio is single sided, but the line is still passively impedance balanced to help optimise the CMRR.
No reason that input configuration - with the capsule referenced to ground - shouldn't be applied to a FETless electret as well, as far as I can see, if the extra sensitivity is important?

I find it works pretty well.
(Some notes on the input with the extra 1G resistor here: http://www.jp137.com/lts/OPIC48.pdf )
The only reason i asked is capsule shielding situation i have here, because of the dimensions i just have to connect one side of the capsule to the ground. So i was just wondering if there was something more i wasn't seeing. Awesome, elegant and simple circuit! I'll use it for kick drum mic because of the high headroom.
 
The main reason for biasing the input that way was to try get away with only using a single 1G resistor, and keeping the circuit as simple as possible!
It will depend on the polarisation of the electret capsule whether that bias voltage is additive or subtractive....and that will of course affect the sensitivity a little.
Again, how much will depend on what the value of the 'baked in' polarisation voltage is, for any individual capsule.

When it came to using the same circuit with an LDC capsule - but without a voltage multiplier - then the loss of some 12v out of 48V could well be significant, so I included a second 1 G resistor to allow the op-amp to be referenced to 'half rail', and the capsule to be referenced to ground. That works pretty well.

This op-amp has quite a low noise figure, so even with only 48v as a polarisation voltage it's possible to get pretty good results. Especially with low noise transformerless mic preamps -- like those found in many of todays' budget audio interfaces.
It's a simple circuit. No FET bias worries, low distortion, decent headroom, and pretty low noise. The audio is single sided, but the line is still passively impedance balanced to help optimise the CMRR.
No reason that input configuration - with the capsule referenced to ground - shouldn't be applied to a FETless electret as well, as far as I can see, if the extra sensitivity is important?

I find it works pretty well.
(Some notes on the input with the extra 1G resistor here: http://www.jp137.com/lts/OPIC48.pdf )
Just tested the mic, with that 3$ electret from Ali. The thing is in the noise range of my NT2-a. Which is pretty amazing for an opamp and a 3$ elwctret capsule 👏👏👏
 
An all OpAmp mic design has benefits and drawbacks.

Benefits:
Ultra low THD in the output stage
High output level
Very high power supply rejection

Drawbacks:
Excessive white noise (5nV/sqrt(Hz) versus 2,5nV/sqrt(Hz) in a good JFET design)
Excessive current shot noise (2..3 times more than in a single JFET stage)
High input capacitance versus a well designed JFET stage

The designer has to be aware of these facts...
 
An all OpAmp mic design has benefits and drawbacks.

Benefits:
Ultra low THD in the output stage
High output level
Very high power supply rejection

Drawbacks:
Excessive white noise (5nV/sqrt(Hz) versus 2,5nV/sqrt(Hz) in a good JFET design)
Excessive current shot noise (2..3 times more than in a single JFET stage)
High input capacitance versus a well designed JFET stage

The designer has to be aware of these facts...
My primary reason for trying an op-amp in this kind of project was simplicity.
An ideal project for a novice DIY mic builder, for example?
I have tried various 'Schoeps style' clones in the past and had quite acceptable results... So I wondered just how an op-amp 'front end' would compare, and again found the results quite acceptable.

I'm sure there are ways of improving the nuances of very high spec impedance converter circuitry, but whether these make very much difference outside the scientific - or very high end audio - world, I have my doubts.

The input capacitance of the OPA1641 is around 5.5pF - (fairly linearly over the entire range, it would appear from fig 36 on the data sheet (see HERE ) ) - That does not seem an excessive figure?
In this configuration, there are no additional resistors in the signal path, so that helps minimise further noise generation.
Depending on the noise figure of the following mic preamp - which will need to provide any additional gain required of course - I've found that in the environments I'm using my mics, the external ambient noise always dominates.
So noise figures lower then the typical 5nV/sqrt(Hz) quoted for this device probably wouldn't make any significant difference in many 'real world' usage situations...

As you say, there are ways of improving aspects of impedance converter circuitry, but I would suggest that we are well into the realms of the 'law of diminishing returns' there.

I would suggest that this simple (and cheap!) circuit 'punches above its weight', as they say?..... I've certainly found it to be quite useful.
 
Last edited:
My primary reason for trying an op-amp in this kind of project was simplicity.
An ideal project for a novice DIY mic builder, for example?
I have tried various 'Schoeps style' clones in the past and had quite acceptable results... So I wondered just how an op-amp 'front end' would compare, and again found the results quite acceptable.

I'm sure there are ways of improving the nuances of very high spec impedance converter circuitry, but whether these make very much difference outside the scientific - or very high end audio - world, I have my doubts.

The input capacitance of the OPA1641 is around 5.5pF - (fairly linearly over the entire range, it would appear from fig 36 on the data sheet (see HERE ) ) - That does not seem an excessive figure?
In this configuration, there are no additional resistors in the signal path, so that helps minimise further noise generation.
Depending on the noise figure of the following mic preamp - which will need to provide any additional gain required of course - I've found that in the environments I'm using my mics, the external ambient noise always dominates.
So noise figures lower then the typical 5nV/sqrt(Hz) quoted for this device probably wouldn't make any significant difference in many 'real world' usage situations...

As you say, there are ways of improving aspects of impedance converter circuitry, but I would suggest that we are well into the realms of the 'law of diminishing returns' there.

I would suggest that this simple (and cheap!) circuit 'punches above its weight', as they say?..... I've certainly found it to be quite useful.
Yes, of course, simplicity rules :) It was not my intention to run you down. For hi cap hi output capsules (for example 80 pF, -34 dBV) there is really no necessity to think it over. But if you are using an sdc capsule (35 pF and -40 dBV) things change significantly...
 
The main reason for biasing the input that way was to try get away with only using a single 1G resistor, and keeping the circuit as simple as possible!
It will depend on the polarisation of the electret capsule whether that bias voltage is additive or subtractive....and that will of course affect the sensitivity a little.
Again, how much will depend on what the value of the 'baked in' polarisation voltage is, for any individual capsule.

When it came to using the same circuit with an LDC capsule - but without a voltage multiplier - then the loss of some 12v out of 48V could well be significant, so I included a second 1 G resistor to allow the op-amp to be referenced to 'half rail', and the capsule to be referenced to ground. That works pretty well.

This op-amp has quite a low noise figure, so even with only 48v as a polarisation voltage it's possible to get pretty good results. Especially with low noise transformerless mic preamps -- like those found in many of todays' budget audio interfaces.
It's a simple circuit. No FET bias worries, low distortion, decent headroom, and pretty low noise. The audio is single sided, but the line is still passively impedance balanced to help optimise the CMRR.
No reason that input configuration - with the capsule referenced to ground - shouldn't be applied to a FETless electret as well, as far as I can see, if the extra sensitivity is important?

I find it works pretty well.
(Some notes on the input with the extra 1G resistor here: http://www.jp137.com/lts/OPIC48.pdf )
Just wanted to say i built the opic48 with TLC070 and it works like a charm. I get some squealing on power up and shut down, it could be i messed something up underway, i'll check it. I wanted to ask why 10n for c11 instead of usual 1n?
 
Last edited:
Just wanted to say i built the opic48 with TLC070 and it works like a charm. I get some squealing on power up and shut down, it could be i messed something up underway, i'll check it.

Try the circuit without the capsule bias oscillator? Or if that's not used, i'm not dure where to start...
 
Just wanted to say i built the opic48 with TLC070 and it works like a charm. I get some squealing on power up and shut down, it could be i messed something up underway, i'll check it. I wanted to ask why 10n for c11 instead of usual 1n?
I think you'll need to increase the value of R4 and R5 when using a TLC070 with the existing OPIC48 circuit.
R4 because the max DC supply for a TLC070 is listed as16V, rather than the max of 36 V shown for the OPA1641.....
The value of R5 will need to match the new R4 value, to help keep the line impedance balanced.
A quick look seems to suggest that the TLC070 has a slightly higher noise figure as well - although the difference is probably not that critical.
So - a higher noise figure and reduced headroom, compared to using the OPA1641 - unless I'm missing something here?

C11 as 10nF simply because it was one less different value of capacitor to source (C7 and C8 were already 10nF)
I calculated that the HPF effect working into a 1G input impedance would reduce the gain at 20Hz by 0.07dB using a 1nF, and by 0.007dB with a 10nF.
Largely academic I felt .... Again, unless I'm missing something else?
 
Just wanted to report i am just done building yet another @rogs opic48 but with a twist, as i mentioned in another thread.

I wanted to utilize the opamp in it's full glory and supply it with stable 36v for maximum headroom. Pretty much overkill by all means, but i have reasons for this. I deleted R4 and supplied 36V through 5pin XLR from a simple 2$ xl6019 DC-DC boost converter.

Not sure if removing R4 does anything to unbalance the signal, but i haven't noticed any penalty in noise. No extra noise coming from the DC-DC converter either. Even simpler one can use 9v batteries in series...

C2 and C3 should in this case either be changed for non polarized type, or take care to engage phantom and op-amp supply simultaneously.

The circuit has insanely low THD at insanely high levels. Highest i could get is 0.009% with my interface's headphone output coupled to mic input, and then my headphone out starts clipping...

Aside from wanting to use this mic as extremely close snare drum mic, i want to use it as a platform for THD measurements of varoius capsules and their inherent THD introduced by mechanical capsule properties.

So i just throw this out in case anyone finds it interesting. Also dedicated psu opens up tons of possibilities for in mic signal manipulation. I have quite a few ideas on my mind (not eq), feel free to come up with your own 😉

This is what i'd call a good transient response. Snare rimsot as hard as i could hit, the capsule is literally couple of millimeters away from the head, the dots are individual samples 48k 24bit.

Tubes get their own PSUs, why wouldn't opamps?
 

Attachments

  • 20240314_191918.jpg
    20240314_191918.jpg
    3.6 MB · Views: 1
Last edited:
Not sure if removing R4 does anything to unbalance the signal, but i haven't noticed any penalty in noise.........
In your new configuration, it would be best ro remove both R4 and R5. That will mainain the line balance, while at the same time optimising the capsule supply polarisation voltage. The 48v phantom power available at pin 3 would then essentially be unloaded from a DC point of view.
R3 and C3 will simply provide the line impedance balance.

The values of R4 and R5 at 10k were originally selected to allow for the worst case scenario - theoretically.
Assume the phantom power supply has an unlimited current capability (never going to happen in the real world!) and the maximum permitted phantom power output voltage is applied (52v).
The op-amp will draw 0.9mA from each leg. Add to that the 0.175mA drawn by the half voltage potential divider resistors R6 and R7 and essentially each leg of the phantom power will need to supply c.1mA.
That would allow the opamp to receive a 35.2v DC supply ( each phantom power 6k8 feed resistor drops 6.8v and the R4 (or R5) drops another 10v).

In practice of course most phantom power supplies do actually supply 48v or less, and will have a limited current capability.
So it would not be unreasonable to drop the value of R4 and R5 to 6k8 - maybe even slightly less? - to optimise the DC supply to the opamp.

But if you are fitting a separate DC supply, then omitting both R4 and R5 will optimise the polarisation voltage available at pin 3, and maintain the line balance.
 
Rogs,

First, thank you for all your work and for sharing it with the community. Really, greatly appreciated!

Several days ago I was discussing with @mrgrooves666 regarding some proposed adjustments to the multi-pattern circuit, so output could be a little bit higher. Here are the ideas:
- Replace 1K for a 47K on R6 (1K is acting as a 1/2 voltage divider for AC signals) (Not a good idea, as it will have an effect on fig-8 symmetry when active. See further posts on this thread.)
- Add a 1G resistor (one side to the capsule connection and the other to ground) and a 1000pf coupling cap (maybe polystyrene o polypropylene, one side to the 1G and capsule connection and the other to the signal input) this will be at each of the capsule inputs (This will prevent V/2 reaching the capsule)
- Replace OPA1641 at the output for a OPA1692 (it's an audio grade, dual unit opa) and use the additional opa unit to drive a fully balanced output (as the OPA Alice), this will increase output level by -6dB, and also this dual opa uses only 0.65 mA for each unit. 1.3mA total, less that the 1.8mA required by the OPA1641, leaving more room for the capsule polarization voltage.

I know this will increase the component count by 6 (2x1G resistor, 2x1000pf coupling caps, and 2x2k2 resistors to drive fully balanced output), but maybe this will be well justified for some of the folks that would love to have a circuit with a higher output level.

Thank you and hope you could find this ideas useful.

Regards!
 
Last edited:
Rogs,

First, thank you for all your work and for sharing it with the community. Really, greatly appreciated!

Several days ago I was discussing with @mrgrooves666 regarding some proposed adjustments to the multi-pattern circuit, so output could be a little bit higher. Here are the ideas:
- Replace 1K for a 47K for R6 (1K is acting as a 1/2 voltage divider for AC signals)
- Add a 1G resistor (one side to the capsule connection and the other to ground) and a 1000pf coupling cap (maybe polystyrene o polypropylene, one side to the 1G and capsule connection and the other to the signal input) this will be at each of the capsule inputs (This will prevent V/2 reaching the capsule)
- Replace OPA1641 at the output for a OPA1692 (it's an audio grade, dual unit opa) and use the additional opa unit to drive a fully balanced output (as the OPA Alice), this will increase output level by -6dB, and also this dual opa uses only 0.65 mA for each unit. 1.3mA total, less that the 1.8mA required by the OPA1641, leaving more room for the capsule polarization voltage.

I know this will increase the component count by 6 (2x1G resistor, 2x1000pf coupling caps, and 2x2k2 resistors to drive fully balanced output), but maybe this will be well justified for some of the folks that would love to have a circuit with a higher output level.

Thank you and hope you could find this ideas useful.

Regards!
Homero - thanks for your kind comments.
I'm just a hobbyist, and I enjoy tryng out different ideas and sharing my results with groups like this one.
I do often post project notes - including the stripboard layouts - mostly to remind myself what I did at the time! :) - but the notes may also serve as a starting point for other experimenters, who may want to try out derivative ideas? --- Bit like you are suggesting here.

The reasons I chose to do the multi-pattern as I did was for two main reasons - to keep it as simple as possible, while at the same time try and keep noise levels to a minimum.
The OPA1641 is used here to serve as a differential amp (Figure of 8 - link A) and a summing amp (Omni - Link C) amp.
It also serves as an inverting amp for the cardioid pattern.
To keep the noise levels low I've chosen 1k resistors for the resistors in the signal path. ( The lower the resistor values, the lower the noise. )
That's also the reason I've kept the audio single sided.... Adding an additional inverting output - similar to Jules' Alice OPA design - adds not only the noise from an additional op-amp, but also the noise from 2 extra 2k2 resistors in the signal path.
It's not a lot, but as well as increasing the audio output by 6dB it will also increase the overall noise figure by around 3dB.

I've found that adding the extra 6dB of gain required from either a good quality mic pre-amp, or digitally in a DAW, will add less noise.

The OPA1692 does - as you point out - have a lower noise figure than the OPA1641, but not by much, and the advantage is probably negated by adding the noise from the second op-amp and it's associated resistors?

As regarding using a second 1G resistor and AC coupling it, I've only done that in my LDC version which uses Phantom Power to supply the polarisation voltage. That maximises the available voltage across the capsule, by grounding one side of the instead of referring it to the half rail.

In this multipattern version I simple measure the half rail voltage the capsule is biased at, and add that value to the output voltage selected for the capsule polarisation, by the voltage multiplier circuitry.

More recently I've been looking at creating a simpler multi-pattern version, by fitting a dual polarity voltage multiplier to the simpler LDC OPIC version which only uses a single op-amp. That should - in theory - be a lower noise version - although the differences measured on the first prototpye seem fairly minimal?
That configuration also doesn't allow for easy balancing of capsule output levels.

Experimenting with different polarisation voltages applied each capsule membrane, to create additional pattern variations is not something that I've thought about copying with this simple sum and diffence op-amp version.... An idea for the future perhaps?

There are all sorts of 'variations on a theme' of course.
I think the main thanks need to go to Jules, for his suggestion to try the OPA164* series of op-amps as an alterntive to a discrete JFET, as the front end of an impedance converter.
Like many folk, I tended to think that a balanced output automatically included differential audio , but that is not the case.
As long as the line is passively impedance balanced for CMRR purposes, there is no need for differential audio.
And that single sided audio concept is used by - among many others - companies like Rode and Neumann.
Now, whether they do it for lower noise levels - or because it's cheaper - I'll leave you to decide for yourself! :)
 
Last edited:
Here is the proposed design with the previously discussed ideas.

Once more, Rogs, I appreciate your detailed explanation regarding your design process and objectives. Of course, this revised design is aimed at achieving different outcomes (higher output) compared to your original OPIC Multi-Pattern Design, which focused on low noise and simplicity. Regardless, the concept behind your original Multi-Pattern Design is truly ingenious, thank you!

Pimped OPIC.png
 

Latest posts

Back
Top