OPA1612 as microphone pramplifier

[Devicet]

Hi, I'm designing an electrect microphone pramplifier that is supplied by 5V. I decided to use OPA1612 because of it's low noise input levels. I heared that I should be aware about Input current noise density - provide low impedance at input and use proper feedback resistors. Here is the circuit I have designed:



Should I remove R3 to lower input impedance and maybe use second chanel for second stage, so I could lower feedback resistors? I was planning to use second channel as 1/2 Vcc voltage buffer. The line_out should provide consumer line level (max 0.89 Vpp). As you can see I used j-fet in common drain mode for low output impedance.

 

[ccaudle]

Using a FET input amp like OPA1642 in a single stage configuration (i.e. no JFET as first stage) would be simpler and should have close to equivalent noise performance. Search for the term OPIC, similar designs have been discussed multiple times on these forums.

 

[Voyager10]

It can't work with R3 = 16K. If you look at the DC operating point of the op-amp, pin 1 can't go high enough to get pin 2 at 1/2 VCC. (If you remove R3 altogether, the DC operating point will be just fine).

 

[ntch]

What electret is that, and why do you use a jfet and a 1Gohm resistor? Isn't there a jfet in the electret body already?
Why not just go straight into the opamp, i.e. as in https://www.ti.com/lit/ug/tidu765/tidu765.pdf ?

edit: there are suggestions for opamps in the pdf, with some minimum specs.

 

[jp8]

https://www.ednasia.com/amplify-small-signals-in-low-noise-circuit-with-discrete-jfet/

 

[Devicet]

It can't work with R3 = 16K.

You're right. Will it work when R3 is pulled up to 1/2VCC? But still does this resistor make sense? It lower input impedance for OPA1612.

What electret is that, and why do you use a jfet and a 1Gohm resistor?

It is a chinese 25mm condenser microphone without FET. But I can swtich it to something better anytime. Maybe I should measure parastitic resistance and if it's around 100Mohm - 1Gohm I wouldn't put additional resistor here.

I don't want to put it directly to FET opamp. As you metioned there are better and cheaper solutions.

What about modifying my circuit, where 2SK209 is in common source mode and the Rd resisior is 960Ohm, so the output impedance of first stage isn't a lot? I would go into https://www.ti.com/lit/ug/tidu765/tidu765.pdf

The second chanel OPA1612 would have variable resistor in feedback, so I could adjust output signal level.

 

[jp8]

You're right. Will it work when R3 is pulled up to 1/2VCC? But still does this resistor make sense? It lower input impedance for OPA1612.


It is a chinese 25mm condenser microphone without FET. But I can swtich it to something better anytime. Maybe I should measure parastitic resistance and if it's around 100Mohm - 1Gohm I wouldn't put additional resistor here.
View attachment 138503View attachment 138504
I don't want to put it directly to FET opamp. As you metioned there are better and cheaper solutions.

What about modifying my circuit, where 2SK209 is in common source mode and the Rd resisior is 960Ohm, so the output impedance of first stage isn't a lot? I would go into https://www.ti.com/lit/ug/tidu765/tidu765.pdf

The second chanel OPA1612 would have variable resistor in feedback, so I could adjust output signal level.

R3 doesn't make sense. If you want to maintain this circuit concept, just remove it. But I'm afraid the whole concept of your circuit is flawed if your ultimate goal is to achieve the lowest possible noise. That's why I shared that link. R4 and the output impedance of the JFET stage already create way more noise than the OPA1612. The OPA stage also amplifies the noise of the JFET stage. At first sight, that circuit I shared does the same, but there the noise contribution of components inside the feedback loop is reduced by the feedback and noise is mainly determined by the JFET and 1G resistor only and is not amplified by the AC gain of the OPA stage. Actually, the 1G is the main contributor at lower frequencies. See Voyager10's JFET shootout thread where he proved that. You could increase that to say 2G or even 5G. But be sure to use either a teflon turret or isolating slots or guardrings and remove solder flux from the JFET area.

Btw, that capsule is probably the same one that I used recently in a mic build and it sounded quite nice. Also from the side without coloration. Another similar capsule with a mesh and white back side that I tried was just too bright sounding. I received two other 25 capsules yesterday that I'd also like to try. Maybe this weekend.

Jan

 

[ccaudle]

But still does this resistor make sense?

No, R3 serves no useful purpose in that circuit.
Do you want an inverting circuit configuration for some reason? A non-inverting configuration could be simpler and lower noise.
The gain range of the circuit as configured is very small, 40dB-45dB. Is that the intended range?

I don't want to put it directly to FET opamp.

For what reason? For this use case an OPA1642 device would be lower cost, lower noise, and lower distortion than the circuit you have shown.

 

[pasarski]

R3 doesn't make sense. If you want to maintain this circuit concept, just remove it. But I'm afraid the whole concept of your circuit is flawed if your ultimate goal is to achieve the lowest possible noise. That's why I shared that link. R4 and the output impedance of the JFET stage already create way more noise than the OPA1612. The OPA stage also amplifies the noise of the JFET stage. At first sight, that circuit I shared does the same, but there the noise contribution of components inside the feedback loop is reduced by the feedback and noise is mainly determined by the JFET and 1G resistor only and is not amplified by the AC gain of the OPA stage. Actually, the 1G is the main contributor at lower frequencies. See Voyager10's JFET shootout thread where he proved that. You could increase that to say 2G or even 5G. But be sure to use either a teflon turret or isolating slots or guardrings and remove solder flux from the JFET area.

Btw, that capsule is probably the same one that I used recently in a mic build and it sounded quite nice. Also from the side without coloration. Another similar capsule with a mesh and white back side that I tried was just too bright sounding. I received two other 25 capsules yesterday that I'd also like to try. Maybe this weekend.

Jan

I also think these thinner capsules with blue back sound nice. IMO the smoothest version is the one with no mesh but this kind of grate over the diaphragm.

 

[jp8]

I also think these thinner capsules with blue back sound nice. IMO the smoothest version is the one with no mesh but this kind of grate over the diaphragm.

Ah, that's good to know! That is one of the two I just received. The other one I will test, is the one in the bottom-right corner of the picture. The top-right is too bright. For my DPA4055 clone build, I'd actually want an open cardioid pattern instead of cardioid, but let's see how these will perform.

Back to the topic of low noise mic preamp design: to educate oneself on this topic, a must-read (imho) are Scott Wurcer's "Low-Noise Microphone Preamplifiers" papers from the Linear Audio Vol 1 publication, ISBN 1986464318. Available from Amazon.

Edit: now with picture of the capsules attached.

Jan

 

[Devicet]

But be sure to use either a teflon turret or isolating slots or guardrings

I'm not sure where they should be placed. All resistors are going to be smd.

Do you want an inverting circuit configuration for some reason?

No.

The gain range of the circuit as configured is very small, 40dB-45dB.

I found on Internet, that output level of electret capsule is 2-10mVpp. I took 5mVpp and read on Wiki that PC max line level is 0.894Vpp. So I have to amplify the signal 178 times =~45dB. But it seems reasonable to go into 40-60dB gain, preferably with regulation. Thanks for OPA1642 recommendation. However it is quite expensive in my country.

I should probably go into EDN's circuit with fancy feedback. At this point, I don't know how it works (feedback) and if I can place variable resistor for gain control. I see, no one uses second chanel of opamp. Do I still need it? Single chanel versions are cheaper. I see other option for opamp - OPA1662. Good to know for the future.

Did you all build low noise preamp with battery supply? I will probably have to cut off usb's ground cable to prevent ground loop, build filtration for supply section or just use battery...

 

[k brown]

What electret is that, and why do you use a jfet and a 1Gohm resistor? Isn't there a jfet in the electret body already?
Why not just go straight into the opamp, i.e. as in https://www.ti.com/lit/ug/tidu765/tidu765.pdf ?

edit: there are suggestions for opamps in the pdf, with some minimum specs.

There many, many quality electret capsules with no built-in FET. In fact most of the best ones do not.

 

[ccaudle]

No. [in response to wanting an inverting circuit]

The circuit you currently have shown is inverting. Just something to keep in mind, since generally an inverting op-amp circuit is higher noise. Maybe not much concern with just 330 Ohm input impedance (although that is a fairly heavy load for the first stage to drive, make sure you have enough bias current to not clip into that impedance).

that output level of electret capsule is 2-10mVpp.

For that particular model of capsule? At what SPL? Capsule sensitivity is usually referenced at 94 dB SPL, so keep that in mind when deciding what gain range you need. What sound source do you need to record? 94dB is relatively loud, many instruments can generate that nearby, but will be much lower farther away.

Thanks for OPA1642 recommendation. However it is quite expensive in my country.

Interesting. The single channel version OPA1641 is just under US$2 in North America, and the single channel version OPA1611 is just over US$3, so the op-amp you showed is about 50% more expensive in North America.

I see, no one uses second chanel of opamp. Do I still need it?

Did you ever need it? What is the intended use for this amplifier circuit? The circuit should be decided by the features you need.

PC max line level is 0.894Vpp

That sounds like you mean audio built in to a computer. I am not sure what the latest standards would be for that, but spending much effort for lowest noise or distortion is a bit of a waste if the other side of the connection is audio built on a computer motherboard. The audio interface will likely be the limit for the noise floor. That line level seems really low, that is only 0.3V RMS (using sine wave calibration).

I will probably have to cut off usb's ground cable to prevent ground loop

Where does USB come into the system? And cutting off a USB ground cable is generally very bad for RFI (not to mention that the cable can no longer be used to supply power, and may not even transfer data correctly if the common mode voltage difference is too high between the two devices).