All OpAmp mic design (no FET at first stage)

[Cqwet Dbdfte]

0.001% should be -100dB THD+N, and could be audible depending on its harmonic content. Even orders may be inaudible
I would think most mikes and speakers will not meet that -100dB , and certainly not my ears.
Percent of full scale can also be misleading if not the full scale is used.
I have not seen much of weighing the harmonics against their annoyance factor.
The -100dB number may disguise other more annoying factors, odd order, monotonic noise, intermods, who knows.
Tube amps humble %THD numbers vs. some solid state types would make people cover their ears.
Some people prefer single ended no feedback power amps, with objectively much worse THD.

 

[Matt Syson]

You also have to consider the performance figures for whichever op amp you chose when faced with the circuit you are using it in. Using such a high value feedback resistor (or should it be considered as a biassing resistor only? ) may indicate the original design and op amp uses a topology internal where parts of the usual attributes are bypassed like using the offset input terminals of a LM301 or a NE5534 to input signal from a pair of J FETS (for example) essentially using only the output drive capability an internal voltage gain. This makes basic function of a circuit much more OP amp specific.

 

[Kwinn]

@rogs

I would like to build a microphone using your excellent OPIC circuit. I have a few questions that I hope you can help clarify:

Just like kingkorg, I plan to utilize an external PSU to supply +36V for the OPA1641. Based on your guidance, I will exclude R4 and R5 from the setup. Additionally, I have access to two external +18V supplies. Could I use these instead of a single +36V PSU, and would this allow me to eliminate R6 and R7 as well? I'm curious about any potential benefits (such as sonic improvements) or drawbacks (like possible adverse effects due to slight rail voltage discrepancies) that might arise from using a dedicated supply for each rail.

My objective is to build a microphone with substantial headroom. Would it be safe to operate at +36V, the maximum specified voltage, or should I consider a slight safety margin, perhaps +35V or even +34V? Are there any adverse effects associated with running the device at full +36V?

Thank you very much for your assistance. I look forward to your insights.

 

[rogs]

@rogs

I would like to build a microphone using your excellent OPIC circuit. I have a few questions that I hope you can help clarify:

Just like kingkorg, I plan to utilize an external PSU to supply +36V for the OPA1641. Based on your guidance, I will exclude R4 and R5 from the setup. Additionally, I have access to two external +18V supplies. Could I use these instead of a single +36V PSU, and would this allow me to eliminate R6 and R7 as well? I'm curious about any potential benefits (such as sonic improvements) or drawbacks (like possible adverse effects due to slight rail voltage discrepancies) that might arise from using a dedicated supply for each rail.

My objective is to build a microphone with substantial headroom. Would it be safe to operate at +36V, the maximum specified voltage, or should I consider a slight safety margin, perhaps +35V or even +34V? Are there any adverse effects associated with running the device at full +36V?

Thank you very much for your assistance. I look forward to your insights.

I have not tried using an external supply myself, so I can't really comment on any of the specific differences that may occur. kingkorg maybe able to help on that one?....
It is a bit of a specialised area -- I'm presuming you will be connecting to a line input, rather than a mic input? .. I can't think of any mic premap that scould cope with a max input signal up to to +20dB -- or even more !

Regarding the max supply. The OPA1641 data sheet lists the max supply as 36v (or +/-18v). That is an operational limit.
The absolute limit is 40v, so 36v should be fine.
If you do decide to go for +/-18 v remember to reference the 1G resistor to ground - which would be centre point around which the signal is referenced, with a split supply.
If you do decide to go for a 36V rail, then the 'half rail voltage' that the input 1G resistor will be referenced around +18V. You may need to up the polarisation voltage for the capsule, to maintain a resonable DC potential acorss the capsule.
That or use an AC coupled input configuration - perhaps like the one I've suggested in THIS version.....

 

[Kwinn]

Thanks! This helps a lot.
Yes, I thought about using the line input of my AD converter when facing high SPL. The unit is capable of +30dB.

 

[Glenn Wardle]

Having looked at input stage suggestions over the many years this forum has chatted, the recent choice has invariably been OPA164x as it's very low current and really high input Z, but it's not the quietest (5.1nV/RtHz), so what about OPA165x? (2.9nV/RtHz)- I know at 3.9mA you couldn't run more than two off standard P48 - but that would suffice. Input Z is not quite so high as OP164x, but if it's wired as a unity gain buffer then input Z is just differential, which comes in at 6 x 10^9 Ohms. If you can live with the current, it looks like a really good choice. Has anyone tried this with a large capsule?

 

[rogs]

Having looked at input stage suggestions over the many years this forum has chatted, the recent choice has invariably been OPA164x as it's very low current and really high input Z, but it's not the quietest (5.1nV/RtHz), so what about OPA165x? (2.9nV/RtHz)- I know at 3.9mA you couldn't run more than two off standard P48 - but that would suffice. Input Z is not quite so high as OP164x, but if it's wired as a unity gain buffer then input Z is just differential, which comes in at 6 x 10^9 Ohms. If you can live with the current, it looks like a really good choice. Has anyone tried this with a large capsule?

Problem i, the two noise figures are quoted differently on the Texas data sheets:

OPA164* -- 'Low Noise: 5.1 nV/√Hz at 1 kHz'
OPA165* --' Voltage noise: 2.9 nV/√Hz at 10 kHz'

So - slightly different parameter names, and at different frequencies (1KHz and 10KHz)
How important that is, I'm afraid I don't know? ...I must defer to those who understand how noise figures are derived !

 

[ccaudle]

and at different frequencies (1KHz and 10KHz)

There is a noise curve on the first page of the datasheet, the OPA165x devices are about 4nV/√Hz at 1kHz.
The OPA165x current noise spec is 6fA/√Hz at 1kHz, while the OPA164x datasheet lists 0.8fA/√Hz at 1kHz, so the OPA1641 should be lower noise with a high impedance source.

 

[Glenn Wardle]

It is not a 'high impedance' source. At all the frequencies of interest ( ie audio) the capsule will (or at least should) look like low z.

 

[ccaudle]

At all the frequencies of interest ( ie audio) the capsule will (or at least should) look like low z.

A 65pF capsule has an impedance of over 160k at 15kHz. I would not consider that low z.
If it was a low impedance source you would not need such a high impedance of the buffer amp.

 

[Cqwet Dbdfte]

Low current = high impedance.
Noise on semiconductors usually higher at low freq. High pass filtering may reduce this to acceptable levels.
Condenser microphones, (not electret) need charge amplifiers, like charge type accelerometers. RF type condenser mikes use some resonance differences to produce a signal.
Why avoid using a JFET for lowest possible noise? Save a couple of bucks?

 

[kingkorg]

Why avoid using a JFET for lowest possible noise? Save a couple of bucks?

Discussed to death already, op-amps have their own set of advantages over JFETs.

 

[Cqwet Dbdfte]

I did not say avoid an OP amp, great for gain, but the input could still be JFET for lowest noise. Maybe too low.

 

[rogs]

I did not say avoid an OP amp, great for gain, but the input could still be JFET for lowest noise. Maybe too low.

When you say 'too low' I'm guessing you're thinking the noise from both JFET and op-amp inputs are so far down 'in the weeds' as to not be worth worrying about for 99% of real world recordings?

The other JFET attributes seem to make life more difficult anyway...
• more complex circuitry
• bias adjustment optimised for low distortion - or symmetrical clipping - but not both together!
• less than 'rail to rail' unclipped output levels (reduced headroom).

I can see there might be some special scientific appications where absolutely lowest noise is important... but for most 'real world' applications I think using an op-amp like the OPA1641 as an impedance converter will take some beating, both for simplicity and performance? .......

 

[jp8]

Having looked at input stage suggestions over the many years this forum has chatted, the recent choice has invariably been OPA164x as it's very low current and really high input Z, but it's not the quietest (5.1nV/RtHz), so what about OPA165x? (2.9nV/RtHz)- I know at 3.9mA you couldn't run more than two off standard P48 - but that would suffice. Input Z is not quite so high as OP164x, but if it's wired as a unity gain buffer then input Z is just differential, which comes in at 6 x 10^9 Ohms. If you can live with the current, it looks like a really good choice. Has anyone tried this with a large capsule?

I considered the OPA1655 in two different charge amplifier circuits, because of its nice, small SOT23-5 package and low en, but indeed, in spoils the fun. The 2nd reason why I'll stick with the OPA1641 is that I may need a bit more current for a polarization voltage oscillator than I first anticipated. When using the OPA1655, in a worst case situation, (44V phantom, max quiescent current), the polarization voltage would no longer be the stable 60V that I had in mind.

Jan

 

[kingkorg]

After recording some drums last couple of months i tested this mic throughly, and this solution works great for the insane SPL levels reaching the capsule. I even added the pad, as the output was creeping up to the rail levels. I was getting easily above 20Vpp on the snare.

I'm not sure how good the CMMR is with R5 and the added PAD resistor(not present on the other leg), but testing with a phone, and using it in practice i haven't noticed any penalty in noise.

I left C1 in place, but i don't think it's necessary, at least not with the DC-DC board i am using.

C2 and C3 should be bipolar, but i left polarized electrolytics, i just hope i'll keep remembering to turn on the Phantom before the DC-DC board.

I'm just putting this out in case anyone finds it useful.

 

[jp8]

I'd leave C1 in, but not for the reason of ripple rejection, but for protection of the OPA. Suppose you have an already powered up external PSU and then connect it to your mic. The wire inductance, together with the 100nF capacitor, create an LC resonant circuit, which could easily generate peak voltages beyond the abolute max rating of the OPA, if you have a high power supply voltage to start with. C1 lowers the Q factor and prevents voltage swings. The same could be achieved by a resistor, though. You could calculate the required capacitor or resistor value, assuming a 1uH/meter inductance, and if you would know the circuit resistance, but you could also empirically find the required capacitor value by connecting a DSO to the power inlet and catch the voltage swings, if any.

Jan

 

[kingkorg]

I'd leave C1 in, but not for the reason of ripple rejection, but for protection of the OPA. Suppose you have an already powered up external PSU and then connect it to your mic. The wire inductance, together with the 100nF capacitor, create an LC resonant circuit, which could easily generate peak voltages beyond the abolute max rating of the OPA, if you have a high power supply voltage to start with. C1 lowers the Q factor and prevents voltage swings. The same could be achieved by a resistor, though. You could calculate the required capacitor or resistor value, assuming a 1uH/meter inductance, and if you would know the circuit resistance, but you could also empirically find the required capacitor value by connecting a DSO to the power inlet and catch the voltage swings, if any.

Jan