In search for Ultra-high quality electret mic interface circuit

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
It is astonishing while searching all over the internet I am finding so many magnificent DIY loudspeaker projects being measured by so cheap mics. DIY'ers often have no idea that many interesting aspects of transducers are buried down below noise and non-linearities of cheap microphones and acompanying electronics. On the other side, there is small, obscured world of serious professionals possessing extremely sophisticated and expensive measuring gear and elusive knowledge, who are not willing to tell an of their secrets. Being as always too-much curious, I started to dig myself into the rabbit hole into the world of quality measurement microphones.

I found great electret capsules capable of over 105 dB of clean dynamic range, where noise floor is near ten times better than with broadly known Panasonic WM-61A capsule. What is more, these capsules have 3-terminal output so internal FET can be reconfigured for greatest linearity at high SPL, up to 130 dB SPL.

Many of the designs of microphone preamps over the internet are designed for dynamic or ribbon mic's, not for electret. What is more, these are often spoved with audiophoolish myth and being designed as 'musical' not as linear.

It is not the case for me. I am searching for top-performance and simple circuit which will interface electret capsules' FET outputs and typical mic preamp, for example E-MU 0404 USB which I own. I want to build microphone with extreme linearity up to 130 dB SPL for near-field THD measurements. I am searching for most linear and lowest-noise circuit with gain in range of 6-20dB, not more. I would like to put it inside the mic, right after the capsule.

I filtered-out three basic routes:

1. Linkwitz-modded capsule with interface or just a capsule giving you an access to FET's source in order to bias it in a different way. My capsules are capable of this 'by default' Appropriate circuit is published here:
System Test

2. Instrumentation amplifier interface.
For convenience I would like to use THAT1512 IC which will serve as an input stage and second, differential amplifier stage will be this one, inside of E-MU. THAT1512 is widely known as best integrated solution out there novadays. The way how it can work with 3-terminal or Linkwitz-modded capsules is still an unknown for me...
http://www.thatcorp.com/datashts/THAT_1510-1512_Datasheet.pdf page 7


3. Transimpendance amplifier interface.
I have spotted this interesting approach in OPA1678 datasheet and it immediately gained my interest. It promises lowest noises but exact idea how it works with 3-terminal capsules is unknown for me.
http://www.ti.com/lit/ds/symlink/opa1678.pdf page 24

Here is explanation and simulations:
http://www.ti.com/lit/ug/tidu765/tidu765.pdf

Here you are very appealing example of such a circuit:
Low noise Mini Electret Microphone PreAmplifier - Electronics-Lab

Another interesting approach is here, but I doubt if noise performance and linearity can rival with examples above:
https://www.directap.com/Microphones

Another one is here, it is called Wallin preamp, this is from 1999:
http://orca.st.usm.edu/~jmneal/preamp1.htm

But hey, there is year 2018, c'mon with something newer!
I would like to ask you for some guidance, which circuit topology can give best results with decent, modern electret capsules with 3-terminal output.
 
Last edited:
...where noise floor is near ten times better than with broadly known Panasonic WM-61A capsule.

That would be impressive, any documentation?


Nice link, thanks, but the WM61A is now unavailable.
A friend on the "Mike-Builders" forum says they found none of the replacements as consistent, I am not sure what the current recommended replacement is.
So if Windforce really has found a quieter and quality replacement that will be excellent.

Best wishes
David
 
Hey did anyone read this article of Samuel Groner "Reducing Transformerless Microphone Preamplifier Noise at Low Gain Settings"? Title sounds very promising but I cannot afford to spend 33 USD only to read the things I quite don't understand... I also seek for low-gain and low noise circuit. My knowledge in the electronics is limited and I really would like to ask some one to help in building decent circuit around this EM173 capsule. I can design, make and even populate PCB SMD board but I don't understand secrets of analog low-level electronics.
 
The easy way to improve linearity is to raise the supply voltage.
As the 2SK123 is a 20V jfet, a 15 V supply is appropriate. Aim for about 8V across the FET, so RL=(15-8)/0.3 ie 22k pull up.
This will give you high sensitivity and decent high SPL handling

OK but how it will compare in terms of linearity with '2-wire Linkwitz mod'?


Sensitivity is similar
S/N ratio is expressed differently so hard to compare

Strange, If I understand correctly, all manufactures use S/N ratio as related to 1Pa (94dB) reference level, A-weighted or not, this is still the same ratio...

I have been looking for an affordable measurement mic too under this thread, but a replacement for something like the Panasonic capsule seems hard to find. :(
Measurement Mic. Diy Help

They are here:

http://www.primomic.com/products/pdf/EM258.pdf

This is japan-made 6-mm electret capsule with specs substantially better than WM-61A capsules. It can be bought in singular quantities.

My luck is that I've just bought unmodified WM-61A capsule from a guy (100% sure it is original capsule). What is more I still have no clue which preamp topology from my first post will provide least noise with mentioned capsules and I still don't know whether it need to be tweaked when polarizing internal FET differently (cases of Linkwitz-modded Panasonic and 3-wire EM173 capsule).
 
Member
Joined 2004
Paid Member
The electrets and the mems mikes are rated from that 94 dB (1 PA) level for SNR. However they don't all use the same weighting. Looking at the Primo EM258 specs -32 dB sensitivity -74 dB to the noise level is -106 dBV or 5 uV noise. The next steps are the check the noise bandwidth of that measurement (A wtg) and find the equivalent - 3 dB bandwidth (approx 9500 Hz). Next step look up the equivalent resistance to 5 uV in 9500 Hz (200K Ohms). And that converts to 56 nV/rtHz. As long as i did not miss a zero one way or another (very possible) there are many opamps that are more than 6 dB quieter which means no additional noise.

The Primo EM173 noise floor is -117 dBV which translates into 1.4 uV or a thermal resistance eq. of 14K or 15 nV/rtHz. A good typical audio opamp is in the 5 nV/rtHz range so still more than 6 dB quieter. This also translates into a noise floor of 14 dB SPL A weighted. That is extremely quiet. You would likely need a vacuum chamber to get the noise low enough to test that. That is pretty close to the best available for a 1/2" microphone and this is smaller.

As for linearity I can get down to -90 db or better distortion products with the Panasonic mikes and a good electrostatic headphone. A 1" B&K plus HP electronics (the lowest distortion combination I have found) only gets a little better. In practice the mikes are not a roadblock. The ECMs can get to 40 KHz plus but do get a little ragged. The Primo's are the best available ECM's. The Panasonics are history. (I have a small stash somewhere.)
 
Hi Demian thanks, that was one great answer I was looking for! I would be glad if you could also look at the first post and tell me more about Best topologies of circuits hinted there. Which of these could I use for Primo EM173? I have several capsules, Primo's and Panasonic - I have an idea that one can be best for near-field THD / CSD (EM173 adopted for best linearity up to 135dB) and other for 1m SPL measurements (spark gap calibrated EM258 easily does 40kHz+). I think also about building the ultimate mic for nature recording so lowest noise is always welcome. As I said I am better in acoustic than in the electronics... Thanks!
 
Member
Joined 2004
Paid Member
The TI app note is your best starting point. The three termianl mike may be able to get lower noise, especially using the source as the output since its lower Z. Another approach is a full differential connection but then you have a noise burden from all the opamps.

I think the real benifit of the thre terminal connection is lower noise from having the freedom to change the bias current in the capsule. I have an example here that is running them at 2 mA. The spec sheet shows more like 500 uA. You also want to make sure that you keep the input Z of the fet as high as possible. Cascoding and bootstrapping are all positive things to do.

You could do an all discrete circuit. its an interesting option. I'll look at some ideas and get back to you.
 

Thank you for the link, Primo do seem to be the best alternative to the obsolete Panasonic.

You could do an all discrete circuit...I'll look at some ideas...

Hi Demian
The "Low Distortion Oscillator" thread has a bit of discussion on linearization of FETs.
In that thread they are used as controlled resistors but I wonder if the same principle could be applied here?
Only an undeveloped idea, but I noticed you were in that thread too so maybe you are up to speed and can see if it will work before I spend more time on it.
Or you can use it too if it looks to have promise.

Best wishes
David
 
Member
Joined 2004
Paid Member
You don't need to linearize the fet. This is a very different situation. The best case is to use the FET as an impedance converter with a really high input impedance.

Here is a quick effort that seems to work in simulation. J1 is the fet in the microphone. J3 is a cascode to reduce the capacitance. J2 is a Buffer to reduce the capacitance at the input and linearize the circuit. The transistors were picked almost at random since its a fresh install of LTspice and I have not imported my other models. The output will be low noise and support the full dynamic range of the microphone. You can AC couple it to an opamp or your soundcard.
 

Attachments

  • FET Microphone Buffer.PNG
    FET Microphone Buffer.PNG
    19.1 KB · Views: 591
  • FET Microphone Buffer.asc
    1.5 KB · Views: 105
Demian, many thanks for your engagement! You've picked LSK489 as random device but it seems that you did it more intentionally as it is very interesting device. I have found an application note of this:
http://www.linearsystems.com/lsdata/others/LSK489_Application_Note.pdf

There is in fact condenser microphone application described on page 12 but unfortunately without any circuit example. On the other hand it is hard to believe (to me) that your simple circuit idea can also be so utterly-linear and potentially better (?) than integrated solutions like top-performing op-amps. Shame I don't understand all of these things you post here. I am looking forward to see some more thoughts from you relating our current subject, keep going! Finally I can build several examples of mentioned circuits and compare its performance with EM173 for rest of DIY community.

EDIT:

What do you think about these ideas, they seem to be also simple:

1. DIRECT APPROACH - Microphones (on the bottom)

2. Modify a Cheap LDC Condenser Microphone: 7 Steps (with Pictures)

I have found another interesting JFET transistor for such application:

http://vakits.com/sites/default/files/J305.pdf
 
Last edited:
Member
Joined 2004
Paid Member
The J305 is primarily and RF part. Check its noise curve to make sure it breaks at a low enough frequency. J1 in my sketch in the internal fet in the capsule. The capsule is essentially a 2 or 3 pF cap so you need a very high impedance at the input of J1. if you are going to a differential input then the balanced circuit works pretty well. However you still have the miller cap compromising the HF and the lack of a buffer on the source also reducing the input Z. We don't know exactly what is in the Primo mike circuit. There are some pretty sophisticated chips for that application. Using 2K resistors and dialing up the current to 1 or 2 mA may make a difference in the noise usually lowering it.
 
You don't need to linearize the fet. This is a very different situation. The best case is to use the FET as an impedance converter with a really high input impedance.

Hi Demian,

could you perhaps expand a bit on why linearizing the FET has a subordinate importance, ie. what types of measurements is it where relatively low distortion isn't important?
Just of the top of my head I am wondering whether certain types of measurements such as step response, from which we can derive the frequency response, would still benefit from a less distorting mic preamp, no?

Regards Michael
 
Member
Joined 2004
Paid Member
Sorry. that was too quick. The linearizing of the FET for the oscillator is to make it act like a more linear voltage controlled resistor by adding a precise amoiunt of feedback. The amplifying task for the microphone is different and a resistor from the drain to the gate would cause problems, if it were even possible. The circuit I posted is an impedance converter to match the GigOhms at the input to J1 to a much lower impedance at the output. It has essentially 100% feedback and unity gain so its not difficult to get very high linearity.
 
I know this is about electret mics, but the MEMS mics sure are getting good.

Analog Devices has a slightly dated doc with some good curves showing production yields for gain and frequency response for for mems and electret. The electrets are really poor, which is why they need calibration.

The mems aren't very flat beyond 10K , but they are rock solid in terms of repeatability and also repeatability over temperature.

Now, most mems are digital out, but here's an interesting differential analog out from Invensense. This is showing a noise floor around 27 dBA and workable up to 130 dB. A bit lacking in distortion, yes (0.2% at 105 dBSPL).

Now, for recording music this would be way too bright. But as a reference mic, where you could apply a calibration curve it looks like it could be pretty interesting, especially because it could probably be a no-cal design. Every mic you purchased would work exactly like the previous.


https://www.invensense.com/wp-content/uploads/2016/02/DS-000044-ICS-40618-v1.0.pdf


MEMS Microphones, the Future for Hearing Aids | Analog Devices
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.