DIY RF Condenser Mics

[rogs]

I don't know anything about RF stuff, but regarding FM solution would it be too simple to use an oscillator like used on the common FM bugs (is it Hartley oscillator though in FM bugs only single coil is used) with the Foster-Seeley FM demodulator/discriminator so that the oscillator coil is actually the L1 in the demodulator:

http://aaronscher.com/Circuit_a_Day/week_by_week/August_2016_FM_Foster_Seeley_detector/FM_Foster_Seeley_Detector.html
There is also ltspice emulation asc file for the circuit:
http://aaronscher.com/Circuit_a_Day/week_by_week/August_2016_FM_Foster_Seeley_detector/documents/Foster-Seeley-Simulation.asc

L1 and L2 are "lightly coupled" which means the coils are in close proximity but are not really a single transformer (if I understood correctly).
The condenser capsule could replace C1, which is the parallel capacitor like in this FM bug:

http://www.next.gr/uploads/542-b12fce8580.gif

Also in some FM bugs the capacitor on base is replaced with a varicap diode which is then feed the audio signal to modulate the frequency, so I guess it's another spot for the capsule connection.

http://danyk.cz/stenice.gif
http://educypedia.karadimov.info/library/fm2w.gif

Wonder how Rode did it on NTG3...

It would be interesting to see how a practical build of something like his works out --To create a high quality FM mic, you need to use an  oscillator with really low phase noise....
There's a lot of expertise among the members here, so maybe someone might like to give it a try?....

I've never found schematics for any of the Rode RF mics online -- I've always assumed they've taken a similar approach to Sennheiser, but maybe not?.....

 

[rogs]

That sounds wonderfully promising! Oh, where are those PCBs and capsules? China Post, please hurry!

China post is usually pretty quick here in mainland UK  - apart from during Chinese New Year ! 

A further thought on the value of R6.  There are likely to  be  some minor variations between the phantom power supplies provided by different pre-amps / recorders.
I've found using 1K or higher seems to keep the current drain below 8mA (4mA per leg)...which is OK with all the phantom powers I've tried
Going as low as 47R will  let you grab a couple of extra dB in sensitivity, but it might be pushing some phantom supplies to their limit?.......

 

[Gerard]

I've never found schematics for any of the Rode RF mics online -- I've always assumed they've taken a similar approach to Sennheiser, but maybe not?.....

In my research about RF condenser microphones, before I started this topic here in GroupDIY, I found references to three or four different RF strategies for condenser microphones:
1. Use the mic capsule to modulate the RF oscillator directly. I think no-one uses this method.
2. Use the mic capsule to modulate an RF bridge, with and AM demodulator (Baxandall article in Wireless World 1963, and Rogs design here)
3. Use the mic capsule to modulate an FM demodulator. (Sennheiser use this method)
4. Use the mic capsule to modulate a phase discriminator. (Uwe Beis' design)

Three and four above may be the same; I don't have enough theory or electronics design knowledge to know if they are the same, or even variants of the same principle.

AKG also made an RF mic in the 1960s; I have found little about it other than mention of its existance. Ruud mentioned an "RF" mic designed by the Netherlands broadcaster (NOS?), also in the 1960s; apparently quite a lot of them were made

 

[rogs]

AKG also made an RF mic in the 1960s; I have found little about it other than mention of its existance........

That may have been the one that was mentioned in posts #2 and#3 of this thread......
this one maybe?...  https://ia800503.us.archive.org/3/items/bbc-rd-reports-1966-32/1966_32.pdf

 

[RuudNL]

The article I referred to, may have been published in "Omroep Technische Mededelingen nr. 1, 1961"
The Dutch radio and television relied on the NRU (Netherlands Radio Union) for the development and design of their technical installations.
As far as I know, some of the NRU microphones were designed by Professor Geluk, that is why they were often referred to as "Geluk microphones".  (Freely translated: "microphones that will bring you luck"...)
I have searched to find this article in "Omroep Technische Mededelingen", but thus far I haven't found it.
At the moment a friend of mine, with good connections in the broadcasting world, is trying to obtain a copy of the article.

 

[rogs]

A further thought on the value of R6.  There are likely to  be  some minor variations between the phantom power supplies provided by different pre-amps / recorders.
I've found using 1K or higher seems to keep the current drain below 8mA (4mA per leg)...which is OK with all the phantom powers I've tried
Going as low as 47R will  let you grab a couple of extra dB in sensitivity, but it might be pushing some phantom supplies to their limit?.......

OK further to that - and as a result of further attempts at optimising oscillator current,  I've decided that the best compromise for the present circuitry  means several changes of component values. 
(Ruud's prototype PCBs are still valid - you just don't fit all the  part now!)

So --
• R1 is now 220k (was 100k)
• R2 is now a link (was 100R)
• R4 is not fitted (was 10k)
• R5 is now100R (was 470R)
• R6 is now 1K (was 100R)
• C1 now 220pF (was 100pF)

EDIT: I have removed the instruction to leave out R3 (1M) -- there can be some oscillator instability at certain settings if it's not fitte....

With those values, the phantom power drain is around 3.5mA per leg - so the 6K8 resistors now dissipate around 85mW each

The sensitivity now allows the noise floor to be >-65dB below 0dB FS, using a preamp gain of 20dB

 

[rogs]

I've now built a number of prototypes - using Ruud's prototype PCB -  with different capsules and configurations.

What is becoming clear is that my attempts at reducing oscillator current as much as possible had side effects with oscillator stability and noise levels.

I've now built 3 prototypes of this latest version of the circuit: http://www.jp137.com/lts/RF.AMX5.pdf  which has proven to be the best so far.
The phantom power current drain is around 7mA - so 3.5mA per 'leg' - (only about 80mW dissipation  in the phantom power 6k8 resistors)

The noise floor is now approaching -70dB , which I don't think is too bad for such a simple circuit.

(Short audio speech sample here: http://www.jp137.com/las/RF.AMX5.wav )

My latest prototype is an RF end addressed cardioid mic...... a 34mm cardioid suitable for outdoor use maybe?....

 

[Khron]

I assume that noise floor was measured with a fixed capacitor replacing the capsule?

 

[rogs]

I assume that noise floor was measured with a fixed capacitor replacing the capsule?

No  - test was made  using the parameters I mentioned in post# 125  --

The actual test was made by playing a 1KHz tone directly into the mic capsule - via a Sennheiser headphone transducer - with the level set to record around -0.5dB below full scale into my Tascam DR100 Mk3, with the mic pramp gain set at 20dB.

The test tone was stopped and transducer was then removed , with the recorder continuing to record the ambient room noise (late at night).

The resultant level was around -70dB below that full scale tone recording. No other levels were changed.

There is a copy of the recording here:  http://www.jp137.com/las/FS.1k.wav
(I have removed the noises made by the removal of the transducer)

If you open the recording in your DAW and increase the gain of the 'silence'  by 40 or 50dB, you can hear the clock ticking in my room  (about 3 metres away) -- and my breathing!

I thought the noise floor was quite good for such a simple circuit?....

 

[Khron]

The reason i asked was because i remembered Henry Spragens having compared the electrical noise floor of some Schoeps-ish mic circuits with and without the capsule attached, and the noise of "just" the air hitting the diaphragm swamped the circuit noise by quite a few orders of magnitude.

Just trying to say the SNR of this might potentially be even higher / better than what you've measured

 

[rogs]

The reason i asked was because i remembered Henry Spragens having compared the electrical noise floor of some Schoeps-ish mic circuits with and without the capsule attached, and the noise of "just" the air hitting the diaphragm swamped the circuit noise by quite a few orders of magnitude.

I do remember reading Henry's  comments on this....... Certainly the ambient noise 'swamps' the actual noise floor, and if I hadn't made the recording without an 80Hz  HPF the 'room rumble' would have made the result quite a lot worse!

I think there is an element of 'Schoeps noise' as well - although it's not too significant. 

As with the best of the DIY Schoeps circuits, I think we're entering the realms of the law of diminishing returns here. .....

Yes, the circuit is noisier than my reference mic (a new 'black' type Rode NT1)  but not by a lot -- and it is already ambient noise that dominates the floor....

As a comparison, the noise floor is about 20dB better than an unmodified BM800,  measured with the same setup ......

And as I mentioned earlier, I don't think it's too bad for such a simple circuit? ....

 

[mhelin]

How do you compare the sound in comparison with a regularly DC biased one? At least the electrostatic force caused by DC bias is avoided so the output should be a lot less non-linear. Some (https://repository.tudelft.nl/islandora/object/uuid:6e23606d-c5a6-4635-8f8d-ea3b2b630763/datastream/OBJ/download) claim this is the main reason for the nonlinearities - at least in context of MEMS capsules but they are not so much different.

 

[Khron]

At least in theory, i would imagine the resonance of the diaphragm is shifted down in frequency (as opposed to when it's constantly pulled by the bias voltage). Whether that's audible (and/or to what extent) or not, is another story...

 

[abbey road d enfer]

I don't know anything about RF stuff, but regarding FM solution would it be too simple to use an oscillator like used on the common FM bugs Simulation.asc

This solution is discussed in the Sennheiser document, and the limiting factor here is phase noise of a free-wheeling oscillator. Getting very stable amplitude is much easier than stable phase (or frequency). That's why they decided to switch to variable demod of a fixed-amplitude signal.

 

[homero.leal]

Just trying to say the SNR of this might potentially be even higher / better than what you've measured

Yes... I agree with Khron! It will be interesting to do a quick test with a small C0G ceramic cap replacing the capsule, just to check for noise floor sans ambience noise.

And also, what do you think about THD? Sergei Steshenko noted  on a personal message, that a possible advantage of RF mics could be that they doesn't need a JFET as impedance converter. If JFET gate to drain capacitance causes THD to increase and this circuit doesn't need a JFET, shouldn't THD be better? Just wondering.

Regards!

HL

 

[Gerard]

PCBs have arrived, but not the RK67 capsules! However, some C12 capsules intended for another project also have arrived; they might get sacrificed for the RF mic project.

Now to find some time when I'm not at work and not singing. Hopefully at the weekend after rehearsals (three long rehearsal workshops this w'end, for three different groups!).

 

[abbey road d enfer]

At least in theory, i would imagine the resonance of the diaphragm is shifted down in frequency (as opposed to when it's constantly pulled by the bias voltage). Whether that's audible (and/or to what extent) or not, is another story...

Remember that, although different (AC vs. DC), AC bias does not eliminate diaphragm pull. It is generally lower because AC bias does not need to be of the same amplitude as DC bias, but it's still producing electrostatic pull.

 

[ricardo]

I'd love to have another go at Baxandall's RF mike but as a beach bum, I don't really have the facilities or inclination.

But a couple of points ...

1. Use the mic capsule to modulate the RF oscillator directly. I think no-one uses this method.
2. Use the mic capsule to modulate an RF bridge, with and AM demodulator (Baxandall article in Wireless World 1963, and Rogs design here)
3. Use the mic capsule to modulate an FM demodulator. (Sennheiser use this method)
4. Use the mic capsule to modulate a phase discriminator. (Uwe Beis' design)

Beis is a nearly exact copy of one of the Sennheiser versions.

As far as I know, some of the NRU microphones were designed by Professor Geluk, that is why they were often referred to as "Geluk microphones".  (Freely translated: "microphones that will bring you luck"...)

Baxandall references several Dutch papers which are almost certainly related to this.

With those values, the phantom power drain is around 3.5mA per leg - so the 6K8 resistors now dissipate around 85mW each

The maximum power you can draw from P48V is when you take a total of 7mA.  The voltage would have dropped to half so this is about right.

i remembered Henry Spragens having compared the electrical noise floor of some Schoeps-ish mic circuits with and without the capsule attached, and the noise of "just" the air hitting the diaphragm swamped the circuit noise by quite a few orders of magnitude.

In https://groups.yahoo.com/neo/groups/micbuilders/files/Mic%20Measurements/ Zephyr.pdf I show detailed noise measurements of several HiZ mikes on pages 10 & 11.  You have to join.  These are 'constant relative bandwidth' measurements eg like a 1/3 8ve spectrum analyser

You see the red  '1G noise' dropping at 3dB/8ve and the white 'resistor noise' rising at 3dB/8ve.  The 'acoustic resistance' noise is white like resistor noise.

I can't remember doing a spectral plot of Baxandall's circuit circa 1980 but it is likely that these RF circuits don't have '1G' noise.  But '1G' noise isn't all bad.  It sounds very much like ambient noise so is less objectionable than white 'resistor / acoustic noise'.

My 1980's Calrec 2050 designs have quite high '1G' noise cos they have effectively 500M or less (Good 1G resistors were rare in dem days) but you only notice this when you compare them with something better ... like the Sennheisers or Baxandall.

If JFET gate to drain capacitance causes THD to increase and this circuit doesn't need a JFET, shouldn't THD be better?

There's a Sennheiser paper that goes into detail on THD sources but the main reason for their better THD at high spl is their 'symmetrical' push pull capsule.  Dem capacitive pads used by all & sundry aren't really that audible.

Remember that, although different (AC vs. DC), AC bias does not eliminate diaphragm pull. It is generally lower because AC bias does not need to be of the same amplitude as DC bias, but it's still producing electrostatic pull.

In most (all?) RF mikes, the AC bias is push-pull so has 'no nett effect' on diaphragm tension.

I'd really like to encourage more work on Baxandall's circuit.  I can't seem to find the  transformer core data I collected when I was in touch with Beiss.  The cores that Beiss used and those I used circa 1980 are all Unobtainium for decades.

Today my starting point would be to use the
- smallest RM cores of the correct frequency,
- 1 or 2 turns for the smallest windings (the base windings for the oscillator & the switching transistors)
- 'Bifiliar' secondary winding to give about the right frequency
- emitter winding to give 25V rms on the secondary

I'm nervous about this as this is going back 30+ yrs and I don't like posting stuff I haven't actually tried. 

I think its also possible to do a Baxandall on rogs' circuit at 10MHz with the Spectrum Comms cores but my doodlings are all too complicated at present .. with at least 3 transformers.

I like rogs' Schoeps type output.

BTW, if you use BC560 or other 'high' voltage PNPs (BC560 at Vceo 45V is marginal but I would risk it), D2 & 3 don't have to be Zeners.  1n4148 is fine but DON'T LEAVE THEM OUT.

 

[abbey road d enfer]

In most (all?) RF mikes, the AC bias is push-pull so has 'no nett effect' on diaphragm tension.

Agreed. However, the original comment was made in view of the actual experiments, that use conventional single backplate capsules.
A dual backplate capsule could certainly be made to operate with opposite DC biases, resulting in zero diaphragm pull (at rest).

 

[ricardo]

In most (all?) RF mikes, the AC bias is push-pull so has 'no nett effect' on diaphragm tension.

Duu..uh!  Abbey is right.

The diaphragm 'pull' on an RF mike is a 'rectified' version of the RF voltage.  So a 20V rms RF bias (28Vp) would have a nett 'pull' equivalent to about 14V DC.  But the 'bias effect' on a Baxandall or rogs mike, is likely about 56V ... making many dodgy assumptions about circuits  8)