DIY balanced low noise and distortion connection of electret condenser microphone to audio interface

[Mr. Haelscheir]

This is a follow-up to https://www.audiosciencereview.com/...hones-with-motu-m2-and-rew.49384/post-1783062 (post #6) where I identified limitations with the off-the-shelf RØDE VXLR Pro and VXLR+ adapters for converting 48V Phantom Power to plug-in-power: the VXLR Pro incurs low noise, but suffers from increasing third-order harmonic distortion in the bass from the transformer or whatever it is using to convert from unbalanced to balanced, while the VXLR+ while having excellent distortion performance incurs some kind of common noise hump, whether or not this is being picked up internally by the MOTU M2 when in unbalanced operation or is coming from the use of Zener diodes to do the conversion. I have also been picking up significant mains noise and its respective harmonics which is mitigated with shielding, though said shielding was implemented on different capsules for which I had been encountering too much second-order distortion perhaps due to my not supplying the optimum bias voltage and load impedance.

I lately came across the "Using a separate bias supply with a split load resistance" section under https://diymics.com/how-to-use-elec...balanced-input-that-provides-phantom-power-pp which showed a fairly simple way to get a balanced signal out of an electret condenser microphone (ECM) without a transformer while providing the correct bias voltage and load impedance. Given this, I am looking for feedback on the following wiring schematic for my upcoming in-ear microphones:

1. I intend on connecting the 0.5*R_L (in this case, 1.1 kΩ) resistors as close as possible to the capsule, maybe even as precision SMD resistors soldered directly to the back of the 8 mm or 9.7 mm capsule so it fits nearly within the custom in-ear mount.
2. Regarding shielding, I've seen https://www.audiosciencereview.com/...o-wire-an-electret-microphone-properly.26912/, whereby this would necessarily be following Configuration #3 since I cannot connect the ECM's ground terminal directly to ground in this split load resistance design. I might be able to short the ECM capsule's case to the shielding.
3. For ergonomics, I would have a thinner (in order to minimize compromising the seal of on-head headphone measurements) balanced and shielded wire running from the in-ear mic to a support necklace or strap to which I would mount the bias voltage rechargeable batteries (in this case, 3 V). Here, I would like feedback on whether shielding is necessary for the DC connection from the rechargeable battery around my neck back up to the in-ear mic; if anything, I might as well use a shield to transmit the ground connection for the battery. I am also supposing that it wouldn't be a good idea to have the DC connection bundled within the same shielding as the balanced pair.
4. From the support necklace/strap, I would then splice the thinner balanced and shielded connections with a regular XLR run, say, about 6' in length to reach my audio interface and accommodate rotating on a swivel chair when taking HRTF measurements.
5. Within the male XLR connector's housing, I would have inline electrolytic capacitors to isolate the DC bias from the audio interface for which the 48V Phantom Power will be kept OFF.

 

[AnalogSteph]

Re: 2, please check whether the capsule case isn't tied to the source (-) terminal already. It certainly doesn't make much sense to tie it to cable shield.

Re: 1 and 3, I would consider going with a 9 V block while upping the resistors accordingly (3k3 then). Should help linearity and the wire gauge required for the battery cable if nothing else, not to mention that I'd trust contact on the annoying terminals more than the average AA/AAA holder. Battery voltage should be blocked with some capacitance, I'd give it 10-100 nF at the ends of both resistors and a few dozen µF electrolytic at the battery. Especially if the battery cable is unshielded, you'll probably want to give the mic side another few hundred pF in parallel.

Are you aware of this gentleman's approach to linearizing a capsule like that by cascoding?

John Conover: Using the Panasonic WM61A as a Measurement Microphone

Though I don't think the circuit is easily adapted to the balanced approach, and you may not want any more complications than there already are.

BTW, have you considered a more "Schoeps-y" circuit, is is that what one of the Rode adapters already does?

Basic FET Circuits for Mics

 

[Mr. Haelscheir]

Re: 2, please check whether the capsule case isn't tied to the source (-) terminal already. It certainly doesn't make much sense to tie it to cable shield.

Re: 1 and 3, I would consider going with a 9 V block while upping the resistors accordingly (3k3 then). Should help linearity and the wire gauge required for the battery cable if nothing else, not to mention that I'd trust contact on the annoying terminals more than the average AA/AAA holder. Battery voltage should be blocked with some capacitance, I'd give it 10-100 nF at the ends of both resistors and a few dozen µF electrolytic at the battery. Especially if the battery cable is unshielded, you'll probably want to give the mic side another few hundred pF in parallel.

Are you aware of this gentleman's approach to linearizing a capsule like that by cascoding?

John Conover: Using the Panasonic WM61A as a Measurement Microphone

Though I don't think the circuit is easily adapted to the balanced approach, and you may not want any more complications than there already are.

BTW, have you considered a more "Schoeps-y" circuit, is is that what one of the Rode adapters already does?

Basic FET Circuits for Mics

Thank you for the feedback.

The kind of capsule I am looking into is https://docs.google.com/gview?url=https://api.puiaudio.com/filename/AOM-5024L-HD-R.pdf&embedded=true.

1. I wouldn't know whether the case is already tied to the source terminal. The capsule already has capacitances between the terminals (I'm not sure if these are symbolic representations of the device's capacitance or actual internal capacitor devices). By "the ends of the resistors", do you mean the left, right, or both ends in my diagram? Are those suggested capacitors bridging those nodes to ground? And for the battery, I guess the capacitor would effectively be bridging its terminals.
2. Said capsule has a nominal Vs of 3 V and a maximum Vs of 10 V. Are you saying that operating at an upper Vs of 9 V while upping the resistors to maintain the same current draw would improve linearity and allow getting away with a lower gauge?
3. Regarding https://www.johncon.com/john/wm61a/, I guess it wouldn't be practical to fit the additional circuity all within the in-ear mic, and I would imagine it would not be as useful if the external JFET had to be situated all the way amid my support necklace/strap insofar as I want the balanced driving to start as close to the microphone as possible. Regarding linearity, I had already gotten exceptional distortion results with https://www.farnell.com/datasheets/1653468.pdf when driven by the VXLR+ adapter, the same adapter yielding high second-order distortion out of https://docs.google.com/gview?url=h...ame/TOM-1537L-HD-LW100-B-R.pdf&embedded=true; I want to maintain or exceed the performance of the former while at least getting rid of the noise hump seen in https://www.audiosciencereview.com/...hones-with-motu-m2-and-rew.49384/post-1783062 (post #6).
4. I suppose my schematic based on the following schematic from https://diymics.com/how-to-use-elec...balanced-input-that-provides-phantom-power-pp is similar to the lower schematic within https://audioimprov.com/AudioImprov/Mics/Entries/2015/4/23_Basic_FET_Microphone_Circuits.html where the FET is already within the capsule, except that I am solely using a battery for the red power supply parts, and I presume the green output stage is already implemented by my MOTU M2's mic preamp, else I would vie that an upstream preamp closer to the capsule shouldn't be necessary for this length of run. Are you saying the VXLR Pro is similar in implementation to that Schoeps-y circuit? It is advertised as being transformer-balanced which is what I believe to be responsible for the third-order bass distortion in Figure 1 of https://www.audiosciencereview.com/...hones-with-motu-m2-and-rew.49384/post-1783062 (post #6).

 

[Mr. Haelscheir]

In https://diymics.com/how-to-use-electret-condenser-microphones/, it says, "In order to obtain the benefits of a balanced connection, the two load resistors must be located as close to the microphone element as possible, so that the cable carries fully balanced signals." Does anyone know why this proximity to the capsule is needed? The load resistors would have a much higher resistance than the long wires, and since electrically, the wires would comprise the same node regardless of where the load resistors are located, wouldn't the long wires still be fully balanced even if the load resistors were to instead be located within the custom adapter at the audio interface's end? I understand that the balanced pair would have parasitic inductance and capacitance, but is there a difference between having that between the capsule and the load resistors versus having it between the load resistors and the audio interface? I don't want to go through the trouble of fitting the load resistors behind or near the in-ear mic capsule if it isn't necessary.

 

[AnalogSteph]

Does anyone know why this proximity to the capsule is needed? The load resistors would have a much higher resistance than the long wires, and since electrically, the wires would comprise the same node regardless of where the load resistors are located, wouldn't the long wires still be fully balanced even if the load resistors were to instead be located within the custom adapter at the audio interface's end?

You've got a point. The mic capsule is a 2-terminal floating source much like a phono cartridge or something. As long as it has as little parasitic capacitance to its surroundings as possible (I would expect the case of a 2-terminal electret to be tied to its ground terminal, so much more opportunities for parasitic coupling there), it should by pretty balanced.

It's a fairly high-impedance affair either way, of course. Your connecting cable needs to be:
* thin, flexible
* shielded twisted pair
* of modest length in order to not let capacitance get out of hand (I suppose you're not after much more than 1.5 m or so, similar to PC headsets)
* non-microphonic

That's quite the list, but there's probably some kind of thin microphone cable available from the usual suspecs that should fit the bill.

At least all the supporting circuitry could go into a single box, without the need for an extra battery cable like you had originally planned.

 

[Mr. Haelscheir]

You've got a point. The mic capsule is a 2-terminal floating source much like a phono cartridge or something. As long as it has as little parasitic capacitance to its surroundings as possible (I would expect the case of a 2-terminal electret to be tied to its ground terminal, so much more opportunities for parasitic coupling there), it should by pretty balanced.

It's a fairly high-impedance affair either way, of course. Your connecting cable needs to be:
* thin, flexible
* shielded twisted pair
* of modest length in order to not let capacitance get out of hand (I suppose you're not after much more than 1.5 m or so, similar to PC headsets)
* non-microphonic

That's quite the list, but there's probably some kind of thin microphone cable available from the usual suspecs that should fit the bill.

At least all the supporting circuitry could go into a single box, without the need for an extra battery cable like you had originally planned.

Yeah. For this application, since I would be at my desk with my MOTU M2 nearby, I would likely need no more than 1 m of wire. Do said parasitics mainly compromise signal level over longer runs rather than noise performance?

https://www.aliexpress.com/item/1005002732210812.html seems to be the thinnest practical at an outer diameter of 1.8 mm. I've seen a listing with an OD of 1.4 mm when there is no shielding, and my previous TOM capsules had a 1 mm OD when it was just one shielded signal wire.

Do you have any comments on https://www.audiosciencereview.com/...rophone-to-audio-interface.59256/post-2166093 (post #3), particularly 1, 2, and 4?

 

[AnalogSteph]

Do said parasitics mainly compromise signal level over longer runs rather than noise performance?

I'd mainly be thinking EMI rejection. Also, treble response would eventually suffer.

https://www.aliexpress.com/item/1005002732210812.html seems to be the thinnest practical at an outer diameter of 1.8 mm.

If you end up getting that, definitely test it for microphonics.

I was thinking maybe something like CMK 209, but that's 3 mm dia. already, or Mogami W2901, 2.16 mm dia..

Do you have any comments on https://www.audiosciencereview.com/...rophone-to-audio-interface.59256/post-2166093 (post #3), particularly 1, 2, and 4?

1. Ask your multimeter?
2. Yes. Source impedance would go up somewhat, but not dramatically so.
4. I just noticed that I didn't write what I meant to in the quote you were referring to, which may have led to confusion. It was supposed to read
BTW, have you considered a more "Schoeps-y" circuit, or is that what one of the Rode adapters already does?
So I actually don't know what exactly is going on inside the VXLR Pro.
The circuitry in the green box would have to go in your adapter, as you wouldn't do the distortion performance any favors by subjecting the contraption to the M2's rather low input impedance, although it would no doubt work.

 

[Mr. Haelscheir]

I'd mainly be thinking EMI rejection. Also, treble response would eventually suffer.

If you end up getting that, definitely test it for microphonics.

I was thinking maybe something like CMK 209, but that's 3 mm dia. already, or Mogami W2901, 2.16 mm dia..


1. Ask your multimeter?
2. Yes. Source impedance would go up somewhat, but not dramatically so.
4. I just noticed that I didn't write what I meant to in the quote you were referring to, which may have led to confusion. It was supposed to read
BTW, have you considered a more "Schoeps-y" circuit, or is that what one of the Rode adapters already does?
So I actually don't know what exactly is going on inside the VXLR Pro.
The circuitry in the green box would have to go in your adapter, as you wouldn't do the distortion performance any favors by subjecting the contraption to the M2's rather low input impedance, although it would no doubt work.

1. To clarify, I meant the following part: "By "the ends of the resistors", do you mean the left, right, or both ends in my diagram? Are those suggested capacitors bridging those nodes to ground? And for the battery, I guess the capacitor would effectively be bridging its terminals." And what did you mean by "blocking" the battery voltage? As for my mention of the capsule's internal capacitances, I meant this, which either refers to actual internal capacitor devices or symbolic measured or spec'd ones:

4. To your question about the VXLR Pro, then yes, I'm pretty sure it rather uses a transformer.
Right. 2,650 Ohms input impedance is not ideal at least when your load resistances are in that ballpark. So that PNP emitter follower will help greatly increase the effective input impedance, and per that article, the local negative feedback (I don't see where said feedback is being applied if not by the 100 kOhm resistors (whether or not I will need to select different values) which supposedly rather bias the PNPs) improves the output stage's linearity; I assume that kind of linearization is separate from the linked cascoding method. I suppose the emitter load resistance would have to be selected differently if I were to use a battery.

 

[AnalogSteph]

1. To clarify, I meant the following part: "By "the ends of the resistors", do you mean the left, right, or both ends in my diagram? Are those suggested capacitors bridging those nodes to ground? And for the battery, I guess the capacitor would effectively be bridging its terminals." And what did you mean by "blocking" the battery voltage?

I think I meant bypassing when I wrote blocking, sorry for the confusion (some native language terminology creeping in). I meant that the capacitors should effectively be doing what the 100 µF does in post #3. You will notice that it effectively connects the outer ends of both resistors. I had difficulty describing it due to the nature of your original construction. If all goes into a box with just a STP to the mic, things should be easier.

 

[olieb]

Did you succeed in building in-ear mics for your liking?
I try something similar, but I use the 6mm capsules (with three terminals/wires) from PUI to have the mics at the very entrance of my ear canals for measuring my HRIR/BRIR.

First I used the a phantom power adapter similar to this

There are several variations of this phantom plug adapters and I used the "Rene Charge Amp" version, because I thought it gives the best dynamic/distortion with the simplest design. https://groups.io/g/MicBuilders/attachment/30006/1/SimpleP48

But I found quite some noise and hum when I get close to the capsule with my finger (or ear). Maybe not surprising as the capsule case is not grounded.
Shielding might help, but it is not simple to shield cables and capsule that are supposed to fit into your ear canal.

So I found this thread and made some kitchen table experiments.
On a breadboard I built the balanced circuit (common source) and on a little pcb I built the common drain circuit.

Both were connected to MOTU M2 and adjusted for level. Load resistance 2.4kΩ and voltage 7.4V and 3.7V.
I measured at about 98dBspl and 108dBspl with one side of my headphones.
Results:
- noise level is virtually identical.
- the balanced circuit is more robust though. Touching (isolation of) cables does not increase noise (hum). This does happen (a bit) with the unbalanced circuit. You can touch ground to bring this hum down again.
- distortion of the (balanced) common source circuit is higher. For H2 it is about 8-10 dB and it seems similar for H3. This will increase max SPL too.
- voltage does not make a big difference
- I changed load resistance to 4.8kΩ for the balanced circuit (breadboard) and this also did not make a big difference.
Graphs:

 

[AnalogSteph]

- the balanced circuit is more robust though. Touching (isolation of) cables does not increase noise (hum). This does happen (a bit) with the unbalanced circuit. You can touch ground to bring this hum down again.
- distortion of the (balanced) common source circuit is higher. For H2 it is about 8-10 dB and it seems similar for H3. This will increase max SPL too.

Neither is very surprising. A source follower circuit effectively is a feedback affair and as such has lower distortion than common source. You could reduce distortion even further by adding a buffer and substituting a current source for (or increasing) R_L.

Besides, the balanced circuit sees extra loading from the parallel input impedance (which generally is <R_L), which is increasing distortion. 2.4k || 2.65k ~=1.26k and 4.8k || 2.65k ~=1.71k, so the expected difference in distortion is relatively modest.

All this is why mic circuits tend to not stick to the absolute minimum of complexity if real estate and budget are not an issue.