DIY RF Condenser Mics

[rogs]

If there was no resonance, I wouldn't be getting hundreds of millivolts RF at the gate.

The amplitude of the RF at the gate can be quite large - even without considering any resonance.
It will depend on the difference in value betwen the capsule capacitance and C4..... In other words, the amount of 'bridge' imbalance.
It is the change in amplitude level at the gate that determines the signal output, not the level of RF present.

 

[shot]

Sadly, removing those cans from an assembled through plate PCB, without causing damage, I've found to be almost impossible.

I must admit that I had been removing cans more than ten times from the same board!
The trick was to use hot air gun! I was just heating it's pins until I was able to pull out the can using minimal force. After that I had to clean pcb thoroughly, then rewind a new core, solder it in, try to calibrate, dissapointment, unsolder all again... repeat... repeat... repeat...
So it's possible!
But very time consuming



Luka

 

[rogs]

I must admit that I had been removing cans more than ten times from the same board!
The trick was to use hot air gun! I was just heating it's pins until I was able to pull out the can using minimal force. After that I had to clean pcb thoroughly, then rewind a new core, solder it in, try to calibrate, dissapointment, unsolder all again... repeat... repeat... repeat...
So it's possible!
But very time consuming



Luka

I have manged to succeed removing cans from th PCBs .. although it's not easy with a through plate boards that has a ground plane!
It usually wrecks the can - or the board -- or both! - and as you say, it's very time consuming....

What I have been trying is to build the oscillator and inductor assembly onto a good old fashioned 'breadboard'... OK, not ideal for 10MHz RF circuits - especially from a 'capacitive leakage' point of view - but it is still possible to calibrate and test different arrangements, by just plugging the cans in and out.
Although the capacitance leakage between the 'breadboard' tracks serve to mess up the calibration somewhat, it's useful to remember that if you only change one component at a time, whatever capacitive leakage is included in the whole assembly remains the same.
Any measured differences must relate to that single changed component.

All a bit crude of course, but quite effective - and certainly better than constantly removing the cans from the PCBs, when simply trying out new ideas!

 

[global-records]

I connected this stock capsule (65 pF, 1 ohm resistance from the ring to the center of the membrane) to the modified rogs circuit, where I replaced T2 and C8 with an inductor with 11 turns, the core turned out to be approximately in the central position (inductance about 2 µH). Also, I bought a 6.5-30 pF smd variable capacitor in parallel to C4=55pF (the one I found in the nearest electronics store). And it works pretty well at 10 Mhz. And it feels like it works a little better with the cap and screen removed. If the capsule is of an even smaller capacity, then the quality factor should be higher, but I did not check. In general, my plans are to reduce the capsule gap to increase sensitivity. I wonder if the diaphragm is glued to the ring?
voice test

 

[rogs]

I connected this stock capsule (65 pF, 1 ohm resistance from the ring to the center of the membrane) to the modified rogs circuit, where I replaced T2 and C8 with an inductor with 11 turns, the core turned out to be approximately in the central position (inductance about 2 µH). Also, I bought a 6.5-30 pF smd variable capacitor in parallel to C4=55pF (the one I found in the nearest electronics store). And it works pretty well at 10 Mhz. And it feels like it works a little better with the cap and screen removed. If the capsule is of an even smaller capacity, then the quality factor should be higher, but I did not check. In general, my plans are to reduce the capsule gap to increase sensitivity. I wonder if the diaphragm is glued to the ring?
voice test

Sounds pretty good! ....... Can you post a schematic of your current circuit ? ... I'm still trying to work out the details of exactly how you're connecting the FET gate to the bridge....

 

[global-records]

 

[rogs]

View attachment 140928

Looks interesting .... What is your reason in replacing T2 with a simple inductor?... If you wind it onto a similar former, I can't see the benefit of not creating a second transformer.... it's only a few more turns onto the former.

Although the Spectrum coils are not perfect for the task, I would think that trying out different winding ratios for a second transformer might result in a higher 'Q' - with potentially more noise free signal gain than from a single inductor ? ..... I may be wrong of course!

 

[Gerard]

Fascinating to read of others who have found the project interesting.... Thanks for the link.
That forum seems to include some experiments with actual hardware, and not just with 'sims'
Interesting to read of the continuing experiments with various alternative transformer options..
This is one area where 'sim' software can be difficult (too many unknown variables!).
I was glad to read that there has been some more successful builds of variations of the project...

Interesting comment in https://forum.vegalab.ru/showthread.php?t=95404&p=3243405&viewfull=1#post3243405 and https://forum.vegalab.ru/showthread.php?t=95404&p=3243445&viewfull=1#post3243445. He suggests that some capsule diaphragms have a high resistance coating but others have a low resistance coating, and that, while a high resistance coating is immaterial in a conventional hi-Z mic circuit, it may be detrimental in an RF circuit. Is this another variable we should consider? How do we even measure it, without risking damaging the diaphragm?

A hi-Z circuit relies on a constant charge on the diaphragm, so (effectively) no current flow. Therefore, a high resistance diaphragm is not an issue. But for an RF-polarised diaphragm, the charge changes at 7-10 MHz. Could it be that some of the cheaper capsules have diaphragms that are not truely metallised, but have a high resistance conductive coaring? These would be unsuitable for use in an RF condenser mic. Maxim87 refers, in https://forum.vegalab.ru/showthread.php?t=95404&p=3260713&viewfull=1#post3260713, the comment above yours, rogs, to trying to silver his own diaphragms (before being shot down by Expresso with comments about sulphide formation on reactive silver).

 

[rogs]

I found many of the points in that Russian thread interesting.... Capsule capacitance values, and membrane tensions particularly so (I'm assuming the tranalation of the phrase 'high resistance coating' refers probably to the effect that deposition has on membrane flexibility?).
The subject of potential non-linearity as the bridge approaches balance made in this post also makes fascinating reading.
Comments about transformer construction - and imbalances - also introduces further variables that could make for some interesting experiments.

I think the difficulty of being able to model some of the components accurately for use in 'sim' software makes physical experiments more important for this type of project.
The variable results that one can get from (apparently) identical prototype builds certainly indicate to me that there are a number of factors in this approach to RF mics that need further investigation....
I still find it a fascinating project - It was great to read that others had too!

 

[ricardo]

He suggests that some capsule diaphragms have a high resistance coating but others have a low resistance coating, and that, while a high resistance coating is immaterial in a conventional hi-Z mic circuit, it may be detrimental in an RF circuit ... Therefore, a high resistance diaphragm is not an issue.

That isn't the case. Resistance in the coating directly affects the the electronic noise in a HiZ mike.

Zephyr.doc shows even 2k4 in the signal path can degrade the S/N of a good LDC mike by about 1dB

 

[rogs]

Reading through that Vegalab thread , and looking again at the notes kingkorg made during his experiments with the Sennheiser push/pull capsule has made me look - once again- at the inductor options for this project.

I've been experimenting with different inductor values, to better match the 'bridge' capacitor values (that is the series value of the capsule and C4).
Calculating the inductor values to be closer to the resonant frequency of the 'bridge' helps to keep up the high 'Q' of the inductor assembly....
And that can make quite difference!

I've tried out a range of capsules and inductor values, and have been able to create several prototypes where the S/N ratio is always better than 80dB... a couple are more like 85dB ... Which is not too bad!

I've updated the project notes (see HERE) to include a table of suggested inductor values, and coil winding details to construct them.

Trying to get the 5.3uH Spectrum coils to be a 'one size fits all' was - in retrospect - probably not a very good idea!

 

[rogs]

...... Trying to get the 5.3uH Spectrum coils to be a 'one size fits all' was - in retrospect - probably not a very good idea!

Further to that comment, I've finally started to get some results that are more consistent and predictable.
Taking a step back to consider some of the points we have discussed over the course of this thread, I think one or two more precise conclusions need to be attempted...
'The bridge has to be slightly unbalanced, and must never cross the balance point, even with the loudest sound inputs'.
Trouble is with that statement is that it doensn't give much of clue as to what degree of imbalance 'slightly' means...

Referring back to Baxandall's original 1963 article he makes the observation that a SPL of 74dB (1dyne/cm2) will change the capsule capacitance 'in the region of 0.001pF' (page 594 column 2).
Extending that observation, we can calculate that an SPL of - say - 140dB would result on a capacitance change of around 2pF.....
An SPL of 148dB might be around 5pF
So, if the value of the bridge capacitance C4 is calculated to be no closer than 5pF to the value of the capsule, and no more than 10pF different then we can at least attempt to put some sensible figures to 'slightly' !

To confuse the matter further, we have so far not specified whether that capacitor should be larger - or smaller - than the capsule value.
For best results, it should be smaller. That will improve the sensitivity a little.
Earlier experimental results suggested that the RF level should be as low as possible, to keep the noise figure low.
That conclusion is probably wrong. It is the oscillator amplitude noise that seems to be the dominant cause of system noise, not the RF level.
Improving the 'Q' of the inductor assembly also improves the S/N ratio by helping to minimise oscillator noise.

The new inductor values and winding notes I've post in the v.6 project notes can help to optimise both the inductor and capacitor selections, for various capsule capacitance values.
Applying these more precise caclulations has allowed more repeatable samples of the circuit to be contructed .... and with some pretty low noise figures. (<10dBA) in most cases. The worst sample I've built since the introduction of these more exact values has a S/N ratio of 80dB.

The choice and construction of the inductors, and a more careful selection of capacitor values can make quite a difference with this project.

'5u3H as one size fits all' for the inductors was - as I mentioned earlier - not actually a very good idea! ...