DIY RF Condenser Mics

[Khron]

Unless i'm horribly mistaken, i believe the test-capsule used is a single-sided K67 clone, which may well explain some(?) of the things on the higher end of the spectrum... Possibly.

[Edit]

Well, then again, if need be, one could consider Henry Spragens' method of toning down the top end in a Schoeps circuit (with a high-shelf attenuation) Not quite sure yet where i could shoehorn that in (3 resistors and a cap), on the existing board size, though ;D

Actually, maaaaaaybe just. But that'll involve four vertical resistors, though

 

[rogs]

....rogs, how are the mikes mounted for the Pink Noise measurements?  Is this your 30cm measurement?

30mm  measurement... The mics were packed into a box stuffed with towels,  with the Sennheiser transducer placed within 30mm of each capsule..
The box was covered in blankets , and stuffed under a duvet in a quiet bedroom.
I do believe it may have started raining outside slightly during the latter (10KHz and pink noise) part of the experiment ...That may account for the  increase in noise level in the 10KH file.

It really was just intended  as a  simple comparative noise floor test ....The AMX8 capsule was a cheap Chinese edge terminated C12 type .. so not very linear.
I'm not sure how linear the Sennheiser transducer is either?.... It was from an old pair of HD450s

And of course packing everything into a tiny space  doesn't allow for variations  that may be caused by the different cardioid status formed by the different mounting arrangements (the C12 in the BM800 body is constructed as an end address unit)

Many thanks for taking the trouble to analyse the results... it certainly gives an independent confirmation that  my own results were  not total nonsense!

I'm not sure about adding EQ to the mic itself?.....  I've never been convinced that a simple passive HF roll off can hope to compensate accurately for the complex frequency response of these various cheap capsules...
I've always preferred to either add EQ in post for recorded files -- or add active EQ into the mic channel chain - at line level - for 'live' use.
I've also found that adding a passive low pass filter before the output emitter followers tends to add noise!..

As usual, lots more to investigate and experiment with..... thanks again for your advice and comments.....

 

[rogs]

Unless i'm horribly mistaken, i believe the test-capsule used is a single-sided K67 clone, which may well explain some(?) of the things on the higher end of the spectrum... Possibly.

[Edit]

Well, then again, if need be, one could consider Henry Spragens' method of toning down the top end in a Schoeps circuit (with a high-shelf attenuation) Not quite sure yet where i could shoehorn that in (3 resistors and a cap), on the existing board size, though ;D

Actually, maaaaaaybe just. But that'll involve four vertical resistors, though

No the capsule was a Chinese edge terminated C12 style 'clone' .....not quite as peaky as a K67, but still pretty 'HF heavy'...

I've not had much luck with Henry's passive EQ in the mic itself. .... The addition of a simple passive low pass will certainly 'tame' the HF -- but not necessarily correct the complex frequency response of the capsule very accurately ?
I've found it also adds noise - especially if the series resistors  are more than about 1k.

I prefer to add EQ in post for recordings, or add active EQ -  at line level - into the mic channel for live situations (as I mentioned in my previous post to Ricardo)

On a different note ......the PCBs are here! ... and looking good.  ;D

I may have to file out the mounting holes a bit ... although I may get away with it if I select a suitable BM800 frame (the mounting hole spacings do vary slightly)
I  dug up a few of the original BM800 PCBs (no idea why I kept them ---just a bit of a hoarder I'm afraid    )  and they seem to solve the different spacing problem by placing the mounting holes  on 30mm centres, and making the hole diameter 3mm (it may even be 1/4" - I'm not sure?) ...
Anyway, it's only a minor issue...

I shall put one together as soon as I've finished my other chores (once I've started, I shall probably stay in my study until bedtime!!)

As I say, looking good so far..... again, many thanks for your efforts....

 

[Khron]

I'm not sure how linear the Sennheiser transducer is either?.... It was from an old pair of HD450s

Well, since both mics were tested using the same signal source, the analysis is relative (as opposed to absolute), so the linearity, or lack thereof, of the signal source is (or should be) pretty much irrelevant

I've also found that adding a passive low pass filter before the output emitter followers tends to add noise!..

For what it's worth... http://www.audioimprov.com/AudioImprov/Mics/Entries/2014/2/10_EQ_Pt.4_-_Noise.html

On a different note ......the PCBs are here! ... and looking good.  ;D

I may have to file out the mounting holes a bit ... although I may get away with it if I select a suitable BM800 frame (the mounting hole spacings do vary slightly)
I  dug up a few of the original BM800 PCBs (no idea why I kept them ---just a bit of a hoarder I'm afraid    )  and they seem to solve the different spacing problem by placing the mounting holes  on 30mm centres, and making the hole diameter 3mm (it may even be 1/4" - I'm not sure?) ...
Anyway, it's only a minor issue...

Duly noted, i've now spread out the mounting holes to 29.85mm (to stick to some sort of grid), and made them 3mm.

I also took the liberty to make some room in the middle, in preparation for the high-shelf. At the end of the day, if one doesn't wish to employ it, one can just solder in wire jumpers where the base resistors would be

PS: 1/4" is 6.35mm; i'm guessing that was supposed to rather be 1/8" (3.175mm) ;D


PS(2): Re: Chinese CK12's - i don't know about "peaky"; i ran some sweeps on mine, and if anything, the response is more like "tapered", with a relatively "straight" (or constant / consistent) rise towards the top end.
https://www.dropbox.com/s/blkhgb8eq0z07y9/Screenshot%202019-01-11%2020.06.30.png?dl=0
(Three separate mics, capsules from the same batch, all relative to my calibrated measurement mic.)

 

[Gerard]

Wow! What a lot of new messages since Ricardo's posting this morning. Some of them may supersede my thoughts below.

This is my analysis of the files that rogs posted in #196.

What a fascinating preliminary analysis. And it opens the way for a discussion about how we want to trim the frequency response of Rogs' circuit design. I, for one, wait with bated breath Ricardo's fuller analysis. I also note that this preliminary analysis is the result of measurements on a single example with a C12 capsule. Those of us who are making early examples of the mic should measure ours to an agreed common method and upload the results for Ricardo (or someone with equal knowledge and capability) to analyse, to come up with a typical set of measurements.

Here are a few random thoughts:

• How do we intend to use our mics? For voice or acoustic instruments or for electronic instruments?
• What sort of response would we like from the microphone? These measurements show a significantly brighter top end than the Rode NT1. Do we want something with a reasonably flat response? Or perhaps something like our favourite dream mics, whether it be a Neumann U47 or U67 or an AKG C141 of your preferred model - or whatever!
• How extended do we want the HF and LF regions to extend? As far as the capsule can measure? Or less? The unfiltered RF circuit can probably measure capsule movement down to subsonic - great for recording that 64' organ or ambient noise due to an underground train passing half a mile away
• Can all the preferences be accommodated by a simple set of filters that could be built into the mic as required (I can't imagine that we should try to accommodate all of the preferences, with switching between them (Khron's later post confirms how tight space is); maybe just a single preference at each end of the frequency range, selected from a table at build time)
• What would be an acceptably good measuring procedure, one that we could all reasonably be able to follow?

Maybe others have thoughts to add too?
;D

 

[Gerard]

Well, then again, if need be, one could consider Henry Spragens' method of toning down the top end in a Schoeps circuit (with a high-shelf attenuation) Not quite sure yet where i could shoehorn that in (3 resistors and a cap), on the existing board size, though ;D

That is what I was thinking - both Henry's method and the space limitation for fitting in more circuitry. If necessary, we could consider a longer donor body, with room for a longer PCB; either the longer version of the one of the type Rogs mentioned for his first (stripboard) build, or a BM900 (looks longer; I haven't seen one in the metal).

 

[RuudNL]

I've not had much luck with Henry's passive EQ in the mic itself. .... The addition of a simple passive low pass will certainly 'tame' the HF -- but not necessarily correct the complex frequency response of the capsule very accurately ?
I've found it also adds noise - especially if the series resistors  are more than about 1k.

I can confirm this. Years ago I tested this and the addition of 10 K resistors between the FET and the PNP transistors gave an audible increase in self noise. On his website Henry states that the increase in noise level is so low that the noise caused by the air molecules on the capsule is higher, but that is not my experience.
The good thing is that with 820 pF it approaches the needed frequency compensation for a K67 capsule.
(An American company is still selling this mod as an "update" for certain Chinese microphones.)

 

[Khron]

Well, as i mentioned, i can still squeeze in the two series resistors with relative ease, but arguably for the sake of flexibility, the trimmer / resistor, and/or the capacitor may be easier to attach on the bottom side of the board.

That is what I was thinking - both Henry's method and the space limitation for fitting in more circuitry. If necessary, we could consider a longer donor body, with room for a longer PCB; either the longer version of the one of the type Rogs mentioned for his first (stripboard) build, or a BM900 (looks longer; I haven't seen one in the metal).

Would going with lower values for the series resistors (say, 1k or so) have any detrimental effects? I guess the cap value might need to go up in that case, but that's not the end of the world anyway, is it?

I can confirm this. Years ago I tested this and the addition of 10 K resistors between the FET and the PNP transistors gave an audioble increase in self noise. On his website Henry states that the increase in noise level is so low that the noise caused by the air molecules on the capsule is higher, but that is not my experience.
The good thing is that with 820 pF it approaches the needed frequency compensation for a K67 capsule.
(An American company is still selling this mod as an "update" for certain Chinese microphones.)

 

[RuudNL]

Besides the increase in self noise, my main objection is that low-pass filtering adds frequency dependent phase shift.
Also it influences the transient response and adds distortion. (I did measurements.)
It would be possible to reduce the value of the resistors in series with the coupling capacitors (and increase the value of the filtering capacitor), but that would mean a heavier load on the FET outputs.  Because of the higher impedance on the drain, this would also unbalance the AC levels on the output.

 

[abbey road d enfer]

I can confirm this. Years ago I tested this and the addition of 10 K resistors between the FET and the PNP transistors gave an audible increase in self noise.

This is absolutely predictible. Increasing the impedance seen by the PNP's by a factor of 5+ makes their noise current perceptible. In addition, it increases the circuit's output impedance, which is detrimental to the CMRR.
There are better options that have been discussed here and on the micbuilders list, particularly adding caps across the drain and/or source resistors or across the FET.

On his website Henry states that the increase in noise level is so low that the noise caused by the air molecules on the capsule is higher, but that is not my experience.

The white nature of the noise current makes it perceptible against the pink-ish molecular agitation (brownian) noise. Of course it also depends very much on the capsule's size and tuning.

 

[abbey road d enfer]

Besides the increase in self noise, my main objection is that low-pass filtering adds frequency dependent phase shift.

Any kind of analog EQ results in phase-shift. It is often desirable. Because the problem EQ endeavours solving has already created its own phase-shift. As long as the mic is a minimum-phase transducer (it is for the most part of its response), its phase response is a conjugate of its frequency response. When properly EQ'd, the resulting phase response is linear.

Also it influences the transient response

Same comment.

and adds distortion. (I did measurements.)

It shouldn't; if it does, it's because it's been poorly implemented. It is true that the little space available in mics often does not leave enough room for well-designed EQ.

It would be possible to reduce the value of the resistors in series with the coupling capacitors (and increase the value of the filtering capacitor), but that would mean a heavier load on the FET outputs.

Loading the drain has no negative consequences, since its a current source.

Because of the higher impedance on the drain, this would also unbalance the AC levels on the output.

Which is a non-issue. It has consequences on the CMRR, but that would be in a favorable way.

 

[Khron]

At the end of the day, the whole thing about balanced connections is noise rejection, isn't it? So the output impedance is arguably (or should be?) more important than the absolute symmetry between signals.

Otherwise, mics like the C414 TL's or Oktava MK012's, or other asymmetrical-output- wouldn't work (as well as they do)

 

[rogs]

My first sample of Khron's original PCB (pre 'pad' version) is now assembled, using the component values for RF.AMX8 ... see here: http://www.jp137.com/lis/RF.AMX8.jpg

Works fine - exactly as expected!  ;D ... Everything fits... although  - as I mentioned earlier - you may need to open up the mounting holes a tiny bit with some BM800 frames. (Now corrected on latest issue)

I decide to fit C4 and the capsule connections on the track side of the PCB for convenience.

As Khron has already mentioned, the 7.5mm spacing for the 1/4w resistors is a bit tight ... but they do fit (see attached photo).
You could use MF12 1/8w resistors -- the highest power consumption is for  R2 at around 85mW, so no real problems with power ratings.

 

[Khron]

Not that i'm patting myself on the back, but that's "pwetty"! ;D

And since it works, i'm quite glad to hear the "tweaked" layout and extra ground plane on the top, didn't mess things up too-too much

It might be interesting to look at the output signal with an "RF" spectrum analyzer, i kinda-sorta wonder if there's any difference between the two iterations (ie. if any RF from the oscillator section "bleeds" or doesn't, into the output).
But that's just a "random" curiosity on my part

Would it help if i moved the C4 footprint to be parallel with C3 ("SOT"), between the cans?


PS: If one (really) wanted to, even on this pre-pad version, one could tack on an extra resistor off the JFET's gate, maybe "inside" a dab of RTV / silicone, and run the other end to a pad switch :

My first sample of Khron's original PCB (pre 'pad' version) is now assembled, using the component values for RF.AMX8 ... see here: http://www.jp137.com/lis/RF.AMX8.jpg

Works fine - exactly as expected!  ;D ... Everything fits... although  - as I mentioned earlier - you may need to open up the mounting holes a tiny bit with some BM800 frames. (Now corrected on latest issue)

I decide to fit C4 and the capsule connections on the track side of the PCB for convenience.

As Khron has already mentioned, the 7.5mm spacing for the 1/4w resistors is a bit tight ... but they do fit (see attached photo).
You could use MF12 1/8w resistors -- the highest power consumption is for  R2 at around 85mW, so no real problems with power ratings.

 

[rogs]

Not that i'm patting myself on the back, but that's "pwetty"! ;D

And since it works, i'm quite glad to hear the "tweaked" layout and extra ground plane on the top, didn't mess things up too-too much

It might be interesting to look at the output signal with an "RF" spectrum analyzer, i kinda-sorta wonder if there's any difference between the two iterations (ie. if any RF from the oscillator section "bleeds" or doesn't, into the output).
But that's just a "random" curiosity on my part

Would it help if i moved the C4 footprint to be parallel with C3 ("SOT"), between the cans?


PS: If one (really) wanted to, even on this pre-pad version, one could tack on an extra resistor off the JFET's gate, maybe "inside" a dab of RTV / silicone, and run the other end to a pad switch :

You should pat yourself on the back -- it's  a pretty elegant PCB!    ... especially as it's ' issue 1' , as it were.

I don't 'feel'  there are any significant RF issues... it tunes well, and there is very little noise.

One little quirk  I have discovered ... the K67 capsule in my 'end address' version  need to be connected one way round. 
The connection to the bridge balance point - the low voltage 'end' -  needs to go to the edge termination, with the higher voltage to the centre terminal...It's function of the proximity of the edge connection to the head basket, which can cause some handling noise.
It's still a bit strange getting used to being able to operate the mic with the head basket removed - if necessary-  without incurring horrendous hum!.....still coming to terms with a low impedance LDC condenser capsule of course.

(You also no longer need a double mesh on the head basket either)...

I think C4 is OK where it is... it's easy to fit, and to change it if necessary ...I think in between the cans might be less accessible....

Do you have the Gerber files for your 'pads' version  (which I think will be  RF.AMX9 ?)...
I think the layout looks really useful, and it should be possible to add either pad by just fitting a jumper link to the pad header....
In the case of a BM800 it's no big deal to simply unscrew the case to gain access to the link......a lot easier than fitting a switch somewhere on that tiny casing!

I know you're continuing to develop ideas with the EQ, but  I would like a copy of the non EQ  version - but with pads - if it's possible? ...Only if it's easy to do though! 

Thanks again for your efforts.
I think we now may have a version that  might just be ready for others to try and build?.... 

 

[Khron]

Aw, shucks! : You'd better quit with all the flattery, before it goes to my head ;D

No, i hadn't yet exported the Gerbers, but as before, that's just a few clicks away Looks good on JLC's viewer.

https://www.dropbox.com/s/sc1y3m0l4uvwxs8/RFmic_gerber.zip?dl=0

Re: pads - that sounds like a great application for those computer-mainboard-like shorting-jumpers. If no attenuation's needed, just place it on the middle pin, at 90deg from either pad "option" (ie. "in line" with R11).
https://sc02.alicdn.com/kf/HTB1ia26KVXXXXXwXFXXq6xXFXXXq/227803161/HTB1ia26KVXXXXXwXFXXq6xXFXXXq.jpg

You should pat yourself on the back -- it's  a pretty elegant PCB!    ... especially as it's ' issue 1' , as it were.

I don't 'feel'  there are any significant RF issues... it tunes well, and there is very little noise.

One little quirk  I have discovered ... the K67 capsule in my 'end address' version  need to be connected one way round. 
The connection to the bridge balance point - the low voltage 'end' -  needs to go to the edge termination, with the higher voltage to the centre terminal...It's function of the proximity of the edge connection to the head basket, which can cause some handling noise.
It's still a bit strange getting used to being able to operate the mic with the head basket removed - if necessary-  without incurring horrendous hum!.....still coming to terms with a low impedance LDC condenser capsule of course.

(You also no longer need a double mesh on the head basket either)...

I think C4 is OK where it is... it's easy to fit, and to change it if necessary ...I think in between the cans might be less accessible....

Do you have the Gerber files for your 'pads' version  (which I think will be  RF.AMX9 ?)...
I think the layout looks really useful, and it should be possible to add either pad by just fitting link to the pad header....
In the case of a BM800 it's no big deal to simply unscrew the case to gain access to the link......a lot easier than fitting a switch somewhere on that tiny casing!

I know you're continuing to develop ideas with the EQ, but  I would like a copy of the non EQ  version - but with pads - if it's possible? ...Only if it's easy to do though! 

Thanks again for your efforts.
I think we now may have a version that  might just be ready for others to try and build?.... 

 

[rogs]

Aw, shucks! : You'd better quit with all the flattery, before it goes to my head ;D

No, i hadn't yet exported the Gerbers, but as before, that's just a few clicks away Looks good on JLC's viewer.

https://www.dropbox.com/s/sc1y3m0l4uvwxs8/RFmic_gerber.zip?dl=0

Re: pads - that sounds like a great application for those computer-mainboard-like shorting-jumpers. If no attenuation's needed, just place it on the middle pin, at 90deg from either pad "option" (ie. "in line" with R11).
https://sc02.alicdn.com/kf/HTB1ia26KVXXXXXwXFXXq6xXFXXXq/227803161/HTB1ia26KVXXXXXwXFXXq6xXFXXXq.jpg

Many thanks for the Gerber files .. much appreciated..

It wasn't flattery --I genuinely think that's  a very elegant piece of work!  ...... 
A double sided through plated PCB  - with a ground plane - and prototypes available for less than £10?.....amazing !
It's all come along way from hand drawing and Letraset tape masters!! 

I think I shall order a few MF12 resistors,  and try them out on one of my other prototype PCBs

And I think I need to try and get my head round quite where this new Infinite Impedance version fits with regard to the bridge balancing....I don't think the bridge does necessarily need to be unbalanced any more? ...

The grounded gate and the 4k7 drain and source resistors will allow the FET to self bias........and are  ideally chosen to suit the IDSS of the FET and establish a reasonable 'Q' point.
The 1nF across these 4k7 resistors determines the HF cut off frequency (Fc - currently around 30KHz) . But  the actual 'measured DC' across the resistors will vary considerably, depending on the level of 10MHz 'unbalanced' voltage. ..
Now how this 'rectified ' DC - added to the original bias voltage - affects the FET biasing, I'm not sure? 
The actual demodulated audio signal is only of the change of amplitude across the R/C time constant, and is not dependent on a minimum DC level, as the original diode  rectifier was..

On top of that, the FET is also acting as phase splitter to provide a balanced audio output... It's busy  little device in this circuit! 

 

[Khron]

The cap across the drain resistor does indeed "perform" a high-cut, but (and i could be horribly wrong on this) the one across the source resistor will actually cause a high-frequency boost.

Could that potentially be the (or a) source (no pun intended) of the rising HF noise in Ricardo's analysis?  ??? Just speculating / spitballing here...

[Edit] Well, kinda-sorta - see attachment (plot is of the drain voltage).

 

[Khron]

Adding a drain capacitor does indeed end up in a "net" high-cut, albeit very slight. Disconnecting the source cap accentuates the high-cut (note carefully, the scales on the left edge of the graphs).

PS: "Even" the drain-cap-only version is "only" 1.4dB down @ 20kHz (relative to the flat region, at -0.5dB or so), and 0.4dB down @ 10kHz.

PS(2): Admittedly, i have no clue how the input RF and the demodulation plays into that, but...

 

[ricardo]

This is absolutely predictible. Increasing the impedance seen by the PNP's by a factor of 5+ makes their noise current perceptible. In addition, it increases the circuit's output impedance, which is detrimental to the CMRR.
There are better options that have been discussed here and on the micbuilders list, particularly adding caps across the drain and/or source resistors or across the FET.
The white nature of the noise current makes it perceptible against the pink-ish molecular agitation (brownian) noise. Of course it also depends very much on the capsule's size and tuning.

In https://groups.yahoo.com/neo/groups/micbuilders/files/Ricardo/ChinaMod%2BU87hybrid/ I analyse adding bits across the drain resistor in detail.  You have to join.

Since then Abbey has provided me with an LTspice model of U87 which is more accurate than what I used with the linear circuit analysis DOS programme I wrote circa late 80's.  This beach bum needs to get off his arse and update ChinaMod+U87hybrid.doc but the findings there are all still valid.  In fact my mod is ideally suited to EQ for AMX8 with the capsule rogs used, where a HF shelf is appropriate.  Only a single cap is required.

I second everything Abbey & Khron says about Henry's EVIL resistors including stuff about Minimum Phase which was of great interest to me in 1980.  I would add that my EQ mod REDUCES NOISE while his increases noise.  If you look carefully at Henry's measurements, you'll see that he actually does have increased noise.