DIY RF Condenser Mics

[global-records]

Version without T2, and a simple way to bias FET, if it has not been offered before.

 

[rogs]

I don't think you'll get a useful mic without the the high 'Q' noise free gain the second inductor adds to the assembly.
I always encountered way too much noise, when trying to rectify the signal from the bridge directly.
As regards the FET biasing. T2 secondary essentially grounds the FET gate, so it's automatically 'self biased' to the optimum setting for this dual infinite impedance detector / differential audio driver role....

 

[Icantthinkofaname]

This is a pretty long thread so I'm gonna ask, has anyone managed to make a good small diaphragm DIY RF mic?

 

[rogs]

This is a pretty long thread so I'm gonna ask, has anyone managed to make a good small diaphragm DIY RF mic?

Probably unlikely --- at least one using the inductors used for the main project. SDC capsules will tend to have lower capacitive values than LDC, so you're likely to need alternative inductors.
I've not found any suitable ones 'off the shelf 'so far -- someone else might be luckier there?...

Of course there is the option to make your own inductors, but that just takes us right back to the start of this thread......

 

[global-records]

I don't think you'll get a useful mic without the the high 'Q' noise free gain the second inductor adds to the assembly.
I always encountered way too much noise, when trying to rectify the signal from the bridge directly.
As regards the FET biasing. T2 secondary essentially grounds the FET gate, so it's automatically 'self biased' to the optimum setting for this dual infinite impedance detector / differential audio driver role....

The terminal inductor here works just the same in resonance with T1 and capacitors. I can't describe the math exactly yet, but the fact is that it can only have one best value. And you can see how the current in it jumps sharply.

When I simulated a circuit with two transformers, the first thought was - it's strange that the primary and secondary T2 have almost identical reactive currents of microampere values. I tried to pick up various inductors and the transformation coefficient. Either I was looking badly, but there was not much improvement in the output. The impedance of the capsule and capacitor connection is 50-100 ohms? (according to the simulation).

As for "bias", maybe I called it wrong, I understand that this happens automatically. And my addition allows to adjust the voltage on the jfet regardless of the scope of the RF signal. If necessary.

I appreciate your practical experience, because several people have made microphones according to this project in the form that exists now and it works well.

My experimental 4.4MHz layout with a high-capacity capsule and self-wound coils works, but not yet as I would like. It's too early to show the result. It looks like I swung too hard. 280 picofarad ~ 10 microns the capsule gap obviously turned out to be too damped, which is audible as a suppression of high frequencies. Perhaps because of this, the sensitivity on the contrary fell, and did not increase, as one might expect.

You once mentioned that you managed to achieve a minimum unbalance of the bridge 200mV p-p. What do you think is the main deterrent? The difference in the inductance of the secondary windings? (by the way, what are the typical values on your TOKO/Spectrum?). Have you selected the most accurate ones for installation? The oscillator?

 

[rogs]

I look forward to seeing (and hopefully hearing! ) the results of your experiments, in due course.
It'll be interesting to see what you discover, in your quest to improve the inductors required for best results.

No, I've not selected inductors, they do seem to be reasonably consistent though - especially the newer 5u3HH type.
Those do behave slightly differently from the earlier 5u3H type, with the optimum oscillator frequency limits changing slightly
I've not seen any other data, other than the published sheet on this page: SPECTRUM COMMUNICATIONS

The only useful TOKO coil I have found (with a centre tapped secondary ) was their type KACSK3894A, but it's only 2.6uH, so I didn't find it very useful for use with 65pF capsules...

I've not found any need - so far - to modify the 'natural' Vp self bias point of the FET. The modulation amplitude is minimal compared to the RF carrier level, which of course sets the bias, as it turns the FET on and off every half cycle.
I did have one example where the RF amplitude on the gate measured at around 200mV - although as the scope lead loaded the test point by a few picofarads, it may not have been that accurate a figure? That mic produced some of my best results - at least regarding noise output.
But it was almost balanced and, as I indicated earlier, you don't want the audio modulating the capsule value both sides of the actual balance point...

I have tended to find that the more the bridge is unbalanced - creating a greater RF carrier amplitude - the more the noise.
I haven't as yet done any tests to see how those two parameters correlate precisely.

As we have discussed much earlier in the thread, these 'off the shelf' inductors are not perfectly suited for optimum results.
I shall be fascinated to see how your 'simulations' turn out in practice....

 

[RuudNL]

Just some information about microphones based on the RF principle:
In the 60's, the technical department of Dutch National Radio (NRU, Nederlandse Radio Unie) developed their own microphones based on the RF idea. The schematic they used was super simple! They work on 12 volts.
I still have one of a similar microphone (MO432) in my own collection and even after 60 years it is still functioning and sounding good.
I have the complete article about the microphone development available, but (unfortunately) it is in Dutch...

 

[Pariah Zero]

Well, I've built two of these boards (and have three more PCB's + all the parts), so I'm fairly excited to give this project a go. My only disappointments so far:

* I didn't have capsules beforehand, so I had no measured capacitance values for C4. As one does, I forgot to order capacitors in the expected ballpark for C4, so now I'm stuck waiting for shipping.
* I ordered some single-sided K67 style capsules, and received double sided ones. For once, fate is unfair in my favor! I'm pretty sure I'm moving those capsules to other mics in the future, though. My own OPA Alice PCB's can use 'em.

Much like has been discussed in the thread: The math for getting this design multi pattern is well beyond me; it happens when you don't have to flex that muscle in a couple decades.

All said, however, I'm OK with that - if this turns out half as good as I'm reading about, it's going to be pretty neat.

 

[Pariah Zero]

I've fully assembled the mic, and have been able to tune it for maximum output - sadly, at least with the first capsule, I've been disappointed with the results - there just isn't any sensitivity to speak of - even after tuning it, I can have my preamp turned all the way up and while I get intelligible speech, there's still quite a bit of noise.

Obviously that capsule isn't the best fit, whatever it is. Could somebody with more knowledge than I about the principle of RF mics help me out here: I get the feeling that putting a dual-sided capsule in may not "just work", even if the rear diaphragm and backplate are wired together, because RF likes to make my head hurt (thinking about it)

 

[rogs]

To start with I would suggest simply using one side of the capsule, and make a cardioid pattern mic.

What is the measured capacitive value of each side of the capsule?...... and what value of C4 did you choose to match that value?....

As we have discussed earlier (and there is a comment on the schematic notes) there is a fairly limited range of capacitive capsule values that will work well with these particular inductors.
It may well be that a much greater range of capsule values could work well with alternative inductors.
So far I 've not found any, but other folk may have had more luck..... and may post their suggestions on this thread at some point?

There are some very knowledgeable members in this group who have much better academic understanding than I do of the maths involved in sorting out suitable inductor parameters...

 

[rogs]

Here's a 'raw' (unprocessed) audio sample to give an idea of the kind of signal/ noise ratio you might expect from a typical RF.AMX mic.

https://tinyurl.com/RF-AMX10-test
That was made using one of these capsules: https://www.aliexpress.com/item/1005003678385151.html
The PCB was fitted with a 10MHz crystal.
C4 in this case was 47pF and 33pF capacitors in parallel (c.80pF).

 

[Pariah Zero]

What is the measured capacitive value of each side of the capsule?...... and what value of C4 did you choose to match that value?...

The capsule currently in it is single-sided, and had a capacitance value that I measured at around 80-90 pF - I have two of the same make/model meter, and I'm at least getting consistent readings between the two.

As a result, I used a single 68 pF cap, taking the small amount of space and the instructions to use the 'closest but smaller' to mean 'single' cap size. I think I can parallel a 68 pF and 10 pF easily enough, and re-tune. I'll give that a go this evening and report back.

I'll also measure my other double-sided capsules, and see what they come out to.

That was made using one of these capsules: https://www.aliexpress.com/item/1005003678385151.html

I've ordered four of 'em. (I needed some capsules anyway...)

Over the weekend, my son had a broken slot car - the contact (made of wire braid) wore off. So, I soldered a new one back on. My son wanted to me to show him how to solder, and, having nothing else readily available, I pulled out one of the three remaining RF-AMX10 boards, and we started soldering in the resistors.

 

[rogs]

........ As a result, I used a single 68 pF cap, taking the small amount of space and the instructions to use the 'closest but smaller' to mean 'single' cap size........

The schematic notes do include a comment that 2 capacitors in parallel may be required (bottom line of the notes - in brackets).
I maybe need to make that more obvious? -- it can be quite important!

I should perhaps repeat that this project is very much an experimental one , and not a completed 'hobby kit' style project.
You will probably have noticed that there are some component locations on the PCB that are not used.
Khron had very patiently changed the board layout several times as the project progressed, and different aspects of the project were discussed and modified. The current layout has proved to be very useful -- and without Khron's input the project may well have foundered a couple of hundred posts ago!
Some project details are still subject to discussion and change, so it remains very much an ongoing experiment.
I'm always fascinated to learn what others discover and suggest as it continues....

 

[global-records]

I think I found the reason for the failures with the use of a homemade capsule. The surface resistance of the metallized film that I used to build it is about 10 ohms/square, besides aluminum has an oxide film. This is a lot for this type of scheme. This not only leads to an uncompensated by selection of C4 imbalance, but also to serious degradation of the output signal. But when setting up the inductors, the resonance peak will still remain, and the frequency does not change.
I'm not saying, but are there defective factory capsules? Not good enough contact to the diaphragm, metallization resistance/thickness, material (gold or not). It may even affect the type of capsule, center/edge-termination. Well, the length and diameter of the wires in the end.

 

[Pariah Zero]

The schematic notes do include a comment that 2 capacitors in parallel may be required (bottom line of the notes - in brackets).
I maybe need to make that more obvious? -- it can be quite important!

I should perhaps repeat that this project is very much an experimental one , and not a completed 'hobby kit' style project.
You will probably have noticed that there are some component locations on the PCB that are not used.
Khron had very patiently changed the board layout several times as the project progressed, and different aspects of the project were discussed and modified. The current layout has proved to be very useful -- and without Khron's input the project may well have foundered a couple of hundred posts ago!
Some project details are still subject to discussion and change, so it remains very much an ongoing experiment.
I'm always fascinated to learn what others discover and suggest as it continues....

I'm aware it's experimental; I've read (or tried to read?) most of the posts. The first step is replication, of course.

Rest assured: I'm in it for the tinkering. I'm not an amateur musician, not a podcaster, and my Rode NT1 will more than suit my professional videoconferencing needs.

I figured there'd be future PCB spins as it gets refined.

I'm not sure if I missed or "ignored" that instruction. (My ability to measure the capsule's capacitance is arguably a bigger problem - definitely at least 10 pF of error. I'll close that gap tomorrow, hopefully.)

The thing that actually confused me: The R12 'link' resistor. I spent far longer than I should have expecting an internet URL for a specialty resistor, not realizing 'link' was a term for a 0 Ω resistor.

 

[Pariah Zero]

Update from the (slightly) upgraded metrology department: I upgraded from a multimeter that could read 0.01nF (±5.0% reading + 7 digits), and even then, it wasn't that repeatable - varying by as much as ±0.02 nF

I moved to an el cheapo μcontroller based component identifier that can measure down to a tenth of a pF, with a few percent error. It's not ideal, but it produces repeatable results.

I definitely had the wrong value for C4: Instead of the ~70-90 pF I measured with the multimeter, the new device measures at 49.5-49.9 μF - repeatedly.

That means the 68 pF C4 cap had to go, and be replaced with a 47 µF cap. Unfortunately, after replacing it and re-tuning, performance didn't change much - I wouldn't be surprised if it's somewhat similar to the issue shot saw last year, where it's sensitive to some unknown property in the capsule.

The capsule I've used so far was the cheapest capsule from WGTcenter - the pictures are accurate for its appearance, and I don't feel it a great loss to have to replace it (other than time). Its whole purpose was a cheap capsule I wouldn't feel bad about destroying while learning to mount & solder a 34mm capsule.

So, I'll put this into a holding pattern while I wait for international shipping on the other capsules - I don't have any edge-terminated capsules and I'd like to give 'em a try in my OPA Alice mics too.

 

[rogs]

........ the new device measures at 49.5-49.9 μF - repeatedly.

That means the 68 pF C4 cap had to go, and be replaced with a 47 µF cap. Unfortunately, after replacing it and re-tuning, performance didn't change much - I wouldn't be surprised if it's somewhat similar to the issue shot saw last year, where it's sensitive to some unknown property in the capsule.............

I'm afraid a capsule with that low a capacitive value is not likely to perform very well in this case.
The capsule is in series with C4, so for a capsule value of 49pF in series with a 47pF cap fitted for C4 you are going to get an effective value of around 24pF as the load across T1 secondary.
Using a 10MHz crystal, that would ideally need a inductor with a value around 10.5uH. Significantly above the max 8.2uH figure posted for the Spectrum inductors.
So a much lower 'Q' than is ideal. Lower sensitivity - and thus more noise.

T2 and C8 adds to the loading across T1 secondary - although the maths required to calculate exactly how much extra 'loading' have so far proved elusive.

To date, I have had my best results with capsules above 65pF in value.
Edge terminated capsules tend to have a higher sensitivity, which is also useful for this type of project.
Centre terminated 'K.67' style capsules tend to be more effective than 'K.47' types - again because of the higher sensitivity.
It does seem to be a bit of a lottery, when it comes to buying these cheap Chinese capsules though......

 

[Pariah Zero]

I'm afraid a capsule with that low a capacitive value is not likely to perform very well in this case.
The capsule is in series with C4, so for a capsule value of 49pF in series with a 47pF cap fitted for C4 you are going to get an effective value of around 24pF as the load across T1 secondary.
Using a 10MHz crystal, that would ideally need a inductor with a value around 10.5uH. Significantly above the max 8.2uH figure posted for the Spectrum inductors.
So a much lower 'Q' than is ideal. Lower sensitivity - and thus more noise.

T2 and C8 adds to the loading across T1 secondary - although the maths required to calculate exactly how much extra 'loading' have so far proved elusive.

To date, I have had my best results with capsules above 65pF in value.
Edge terminated capsules tend to have a higher sensitivity, which is also useful for this type of project.
Centre terminated 'K.67' style capsules tend to be more effective than 'K.47' types - again because of the higher sensitivity.
It does seem to be a bit of a lottery, when it comes to buying these cheap Chinese capsules though......

Then it's time to wait for shipping. That's not a huge loss, as it's time to learn more.

 

[reega]

just built the board and tried it out. Oscillator is working at 10Mhz. Capsule capacitance is 58pF. C4 I've tried 40pF and 55pF. Get absolutely nothing. What could be wrong? When i look at signals, I'm getting a sawtooth between ground and one of xlr wires, on the other I'm getting small noise signal.

So, some stuff is working, orher stuff is not? One of T1/2 are off?

Here are some pics.

Also, pic at jfet input..

 

[rogs]

The coils look very different from the Spectrum ones I've seen. All those I've had so far have a matt finish, and a small ferrite core slug .
They have either the ident '5R3' or more recently '5U3H' printed on them. (Photo attached)
The ones in your photo look very different? ... Have you found an alternative supplier -- and are the specs the same?

The capsule value you've described is a bit low for use with 5.3uH coils. C4 is likely to perform better with 55pF fitted than with 40pF although, as I say, you might have problems getting a high enough 'Q' from the tuned circuits with that value capsule.