DIY RF Condenser Mics

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I've now built and tried out several of these mics, using different capsules and circuit variations.
I've decided that the 2mA version I described above is probably the best option for most situations...

After all, this type of mic really comes into its own outdoors, and is therefore more likely to be used with portable recorders.
It's then that current drain can be important.
OK, the noise floor is slightly higher, but is still way below any outdoor ambient noise level I've managed to record  so far ...

I've redone the schematic and parts list to allow for different capsule types, and have posted a combined document here:

http://www.jp137.com/lts/RF.AMX10.v.2.3.2mA.schematic.and.parts.list.08.20.pdf

(There is a link to that document on the website Schematics page as well)

I'm not sure just how important this kind of mic is for indoor use? - but it has been interesting to listen to the same capsule, when used with  both  a 'standard' Hi-Z Schoeps type circuit and with this Lo-Z RF bias circuit.
It certainly sounds a little different - although it's difficult to qualify 'different'. ....Slightly less 'HF edge' seems to sum it up.
Understandable perhaps, when you change the  bias on the capsule from c.60V DC, to around 2 or 3 volts of RF AC bias.....
 
I have a selection of MKH mics I use indoors all the time.  They are definitely among the quietest mics I own, and the most detailed. 

I suspect a lot of what I and others like about them has to do with the capsule design and the corrective EQ approach.  Is lack of an RF specific capsule a bottleneck here, the last 5%?  Particularly, I'm unaware of any push-pull capsule design as found in the MKH mics of the last 35 years, they sound quite a bit clearer and cleaner than the previous non-PP MKH types. 

I see 25pfd mentioned as the substitution cap value with the earlier MKH models for alignment.  It's been mentioned by someone who repairs those mics in America that a capsule that presents as shorted can be resurrected with a couple of good raps on a table(!). 

I would think it worth exploring SDC variants, and identifying capsules that are more 'in the direction of' the specs found in the Sennheisers, both LDC and SDC.   

Historically, the obscure AKG RF mic is interesting, it's covered in a BBC technical paper along with the MKH 404 (I think).  I find no mentions of that mic online. 

I'm tempted to try your circuit with a Neumann TLM67 I have, since I have a spare set of Neumann metalwork....but I owe someone a report on a U87 build first! 
 
EmRR said:
I would think it worth exploring SDC variants, and identifying capsules that are more 'in the direction of' the specs found in the Sennheisers, both LDC and SDC. 

So far, I have only experimented with a variety of cheap Chinese 34mm capsules -- both centre and edge terminated.
The chance to try out something more appropriate to the concept -- like the Sennheiser 'push pull' capsules -- would be a fascinating next step to try. 
Sadly, the prices of actual Sennheiser capsules are WAY above my hobby budget, and I've not seen any others of that type offered. 
I'm guessing Rode must have their own 'RF' capsules for their range - for example - but I've never seen any offered for sale? (Understandably).

It has been interesting to note how 'standard' cheap capsules seem to react differently when biased with a few volts of RF - rather then 60 or 70 V of DC.....
Although I don't have test equipment sophisticated enough to be able to measure those differences objectively.

There is quite a variation in the sensitivity of the different types of capsule - with the 'K67' types seemingly more sensitive than the K47 types.  That presumably has something to do with membrane tension .
Sadly, discovering which cheap Chinese capsules are built with lower tension membranes seems to be largely a matter of luck... The published 'specs' usually appear to be just a bunch of numbers picked out by the marketing guys....
All enjoyable experiments though !

EmRR said:
I'm tempted to try your circuit with a Neumann TLM67 I have, since I have a spare set of Neumann metalwork....
I'll look forward to reading about what you discover -- when you get the chance to try that...
 
I've just had the chance to try out an alternative inductor for this project.

The same UK supplier (Spectrum Communications) has introduced a new range of  'High Stability' IF cans (see here: http://www.spectrumcomms.co.uk/Components.htm?LMCL=si5NNn ) .
The version used for this project has the part no: '5u3HH' (as opposed to the original '5u3H' part no. - (which is still available)

The newer version works as well as the original -- in fact, I suspect slightly better....
It seems easier to 'tune' accurately -- although I've yet to have any actual figures on the differences.

Although the 'Q' is listed as being less, the practical results seem to be very similar to the 5u3H coils.

And it is the high 'Q' of the inductor assembly in this project that makes all the difference! :)

 
I've only just realised that I've had 'tunnel vision' about one aspect of this project.....
I had sort of become fixated on the idea of maximising the oscillator output amplitude, to create the best results.
In reality, the dominant source of system noise is from the oscillator, so reducing the amplitude also reduces that noise.
In addition, high oscillator amplitude increases the crystal drive levels unhelpfully -  and of course increases overall current drain.
It also - with certain capsules - tends to create an overly 'hot' mic.

So I've made a couple of simple resistor value changes to reduce the oscillator amplitude.
New version (v.4.1) here: www.amx.jp137.com

End results...

• Less overall current drain - now around 2mA
• Reduced crystal drive  - to around 75uW
• Reduces system noise by around 4dB
• Reduces sensitivity to around the same level as using the same capsule, with a 60V polarising voltage and added to a Schoeps style circuit.

Haven't found any downsides -- so far! :)
 
I have been trying to read some of the background literature and am trying to understand Arends "DE CONDENSATOR MICROFOON MET HALFGELEIDERS _ OV 442". I have OCRd the paper and used Google Translate to make a first stab at the English. Perhaps someone who is fluent in Dutch, English and electronics terminology could cast an eye over the attached document and try to improve the translation. I think I have got the Dutch part right, manually correcting errors from the OCR process, but Google translate still make difficult reading of some of the passages.

I tried to attach the document as a Word document but that is not allowed in the forum. The pdf version does open in Word for editing, without too much distortion.

Incidentally, the oldest reference I have found to using rf biassing for condenser mics is the paper by J. J. Zaalberg van Zelst in 1947 (there is a version of it in English: Circuit for Condenser Microphones with Low Noise Level). That paper references earlier works; but I think they are not related specifically to rf biassed condenser microphones.
 

Attachments

  • Arends - Condenser Microphone using Semiconductors OV 442.pdf
    858.8 KB · Views: 51
Fascinating to finally see and read both Arends and the Zaalberg van Zeist notes - in English!
Thanks for that...

Baxandall refers to the work of both on the last page of his notes here: http://www.jp137.com/lts/Baxandall.RF.mic.pdf

He points out how it seems that those projects - and his - had followed similar thought processes - even though neither had any knowledge of the other's existence!

As Baxandall points out -- and I can confirm -- 'tuning' the bridge assembly makes a big difference to the performance of this type of circuit.... 
 
I have been measuring some new capsules prior to putting them in some new rf microphones I am building. The K67 capsules measure 47-52 pF, but the C12 double sided capsules measure 62 pF for each diaphragm, both sides and both capsules measure the same - far less than Rogs suggested as typical. It seems it is important to check the actual capacitance in order to get the right value for C4.

My capsules mostly came from WGTCentre (2 × K67 single sided, centre terminated, and 2 x C12, double sided, edge terminated). The other two K67s were ultra-cheapies from eBay, $13 each.

I think I will use the 10 MHz options with 56 pF for C4 for all these capsules, based on the measured capsule capacitance (C4 47 pF for the odd capsule).

Rogs, did you make a switched-pattern version like you suggested a couple of years ago? Or just the fig-8 version you reported? I will certainly make one rf fig-8. I may make a more conventional high-Z multi-pattern mic with the second C12 capsule.
 
Last edited:
Rogs, did you make a switched-pattern version like you suggested a couple of years ago? Or just the fig-8 version you reported? I will certainly make one rf fig-8. I may make a more conventional high-Z multi-pattern mic with the second C12 capsule.
No, I've never tried a switched- pattern version. I decided that it's probably a step too far, in that the tuning between different versions probably couldn't be restricted to capacitor changes, and would require a change in inductor core slug positions as well.
Bit of a non starter I suspect .. although I'd be pleased to be proved wrong! :)

I 've not tried an omni version yet. Might be interesting, in that with both capsules connected in parallel the effective capacitance of the capsule will be doubled. So dramatic changes in the required value for C4, I suspect?

I have built a figure of 8 version using a double sided RayKing edge terminated capsule. (photo attached)
Even though both sides of the capsule measured as having the same capacitive value (around 65pF -- so a 10MHz version built) the actual output level from each side was different - by around 4dB. That needed to be corrected by adding a small (10pF) capacitor across the 'hot' side.
That brought the 2 levels to within 1dB. So a small cap fitted into either of the vacant C3 or C4 locations, as required.
Another thing to be wary of with figure of 8 versions is listening 'live' through headphones. Each side will sound quite different, as a result of the polarity of acoustic 'leakage'.....Can be very misleading !

As you have noticed, there can be quite a range of capacitive values for different capsules. I have only had a couple of edge terminated capsules where the measured value was as high as 90pF. They have given me my best results so far. For a couple of reasons, I suspect...
• They use 3 micron mylar and seem to have quite a low tensioned membrane (They collapse if you put more than around 50V DC across them!)
• They measure around 90pF, which makes them ideal for use with an 8MHz oscillator , which allows the inductors to have a slightly higher 'Q'
• They have a higher sensitivity, which makes the effective signal to noise ratio slightly better ( less preamp gain required).

Most K.67 style capsules seem to come in around 60/65pF, and the couple of K.47s I have tried seem slightly lower.
The K.47s are also a lot less sensitive (by around 8dB).

The dominant noise source from this mic seems to be generated by the oscillator - at around 12dB - so using a high sensitivity capsule gives the best results. 'A' Weighting doesn't help the noise figure either. Most of the noise seems to be mid range white noise.

As Sennheiser comment in their technical notes on RF mics, they are ideally suited to low tension capsules . Sadly, membrane tension is not one of the parameters usually quoted in Chinese capsule specs!.

It is useful to try and match the value of C4 to the measured capsule value. As the JFET infinite impedance detector is always 'self biased' around the Vp 'cut off' point, I have had good results with an RF carrier of only around 200mV p-p or so. Less than is required for a conventional AM diode detector.

Remember that the capsule and C4 are in series, so that the actual capacitive value will be around half the capsule value.
However, T2 is connected across T1, and that affects the actual load of the inductor assembly.
The actual maths involved here are way beyond my understanding, but Abbey did do an approximate simulation earlier on in this thread which indicated a very selective high 'Q' point . The accurate tuning of the inductors is quite important for good results - and virtually rules out multi pattern versions, IMHO.

Ideally, you want a low tension 90+pF capsule, and a 8MHz crystal for the best results, but 60+pF capsules with 10MHz can work well - especially with high sensitivity capsules.

On final point. The polarity of the mic will depend on whether C4 is slightly larger or smaller than the capsule capacitance, not on the polarity of the capsule connections.
The easiest way of changing the polarity - if necessary - is to reverse the connections to pins 2 and 3 of the XLR.
 

Attachments

  • fig.of.8.cap.conn.2.jpg
    fig.of.8.cap.conn.2.jpg
    220.1 KB · Views: 39
No, I've never tried a switched- pattern version. I decided that it's probably a step too far, in that the tuning between different versions probably couldn't be restricted to capacitor changes, and would require a change in inductor core slug positions as well.
Bit of a non starter I suspect .. although I'd be pleased to be proved wrong! :)

I 've not tried an omni version yet. Might be interesting, in that with both capsules connected in parallel the effective capacitance of the capsule will be doubled. So dramatic changes in the required value for C4, I suspect?
...
Thank you, Rogs. Great information summarised there, including some points I had forgotten or hadn't understood, such as the polarity point. The omni version is a nice idea; maybe I'll try that.

My next ambition/dream is to try to measure diaphragm tension. It should be possible by measuring the capacitance at two very different voltages. Knowing the voltage across the capsule and the conductive area at each terminal, it should be possible to calculate the diaphragm to backplate separation at each voltage and from that, to estimate the diaphragm tension. Of course, it might all sink on the precision of measurement required and the other assumptions that would have to be made, such as the diaphragm shape at high voltage. If we assumed pistonic movement for ease of calculation but the stretched shape was actually spherical, that might swamp the measured values and make the calculation meaningless.

I'd also like to try this rf circuit with a Debenham* capsule. I have read that they can be very good on acoustic instruments, if well made. But I doubt they've ever been tried with an rf-biassed circuit. Allegedly they measure around 120 pF. Actually, I'd like to try a Debenham capsule also with a conventional high-Z circuit. I'm not sure I can bring my machining skills back up to the level required to make such high precision devices; it's been many years since I did any precision machine work but it might be fun to try. If I get this off the ground, maybe I'll start a new topic for it.

*Debenham, Robinson, Stebbings. A Stereo Condenser Microphone.
http://www.sdiy.org/oid/ldc/Debenham-StereoCondenser.pdf There is a slightly better scan available somewhere on the Internet but I can't find it just now. 3D drawings in STEP format are available here on GroupDIY, at Capsule Drawings/Designs for Large Diaphragms and at other places on the Internet.
 
Last edited:
Totally off topic, but has anyone tested NT1(black one) in high humidity conditions?

Edge terminated capsules do not really care about high humidity, and NT1's high impedance parts are covered with clear lacquer compound, so i don't really see how high humidity could be an issue.
 
Last edited:
Edge terminated capsules do not really care about high humidity, and NT1's high impedance parts are covered with clear lacquer compound, so i don't really see how high humidity could be an issue.
There is a conformal coating around the capsule terminations, but no sign of it anywhere else.
Rode make a fuss about their new NTG5 RF mic having conformal coating over all the electronics ..
Presumably they felt covering just the capsule terminations wasn't enough, where there was likely to be use in high humidity?

The NT1 HF6 capsule is sort of 'open backed' -- or at least has holes in the back, rather than a second membrane.
Whether that affects susceptibility to humidity problems, I'm not sure?

Not sure I'd take a chance using my NT1 in high humidity?... ( One reason for my building RF mics of course... so I don't have to! :) )
 

Attachments

  • NT1 cap rear and PCB.jpg
    NT1 cap rear and PCB.jpg
    102.8 KB · Views: 37
I 've not tried an omni version yet. Might be interesting, in that with both capsules connected in parallel the effective capacitance of the capsule will be doubled. So dramatic changes in the required value for C4, I suspect?

Interesting point. With my double sided C12 capsules having 62 pF each side, C4 would need to be 120 pF. Presumably I should take the opportunity to use an 8 MHz oscillator frequency.

What about C8? Should I use 68 pF for the 8 MHz frequency?

I think I won't try a multi-pattern rf mic, at least not until I get the double-diaphragm capsules working in fig-8 and omni.
 
Last edited:
Interesting point. With my double sided C12 capsules having 62 pF each side, C4 would need to be 120 pF. Presumably I should take the opportunity to use an 8 MHz oscillator frequency.

What about C8? Should I use 68 pF for the 8 MHz frequency?
Yes - I think 8MHz would be best for those larger values -- and, as you suggest, use 68pF for C8 .
 
Back
Top