DIY RF Condenser Mics

[rogs]

Looking at the new Gerber files --- thanks again --I can't find C14? ...I'm not sure where it can go though ---I think it may need to be a 2.5mm MLCC  rather than a 5mm poly!

I think C7 and C15 will probably need to be mounted on the reverse side now ---- same as the electrolytics on some of the original BM800 circuit boards..

 

[Khron]

I sort of questioned the actual requirement / usefulness of C14, actually. The oscillator section has its own "bypass cap" as it were (C6),  and the JFET's not doing all that much in the HF range(?).

And since, with the added zener regulation, there's less need for "brute force" supply rail smoothing / filtering, won't 6.3mm diameter electrolytics suffice? Surely those could easily fit into those footprints

A 47u/50v on the "input", before R11 and the zener, and 100-220u/16v after that.

 

[rogs]

I sort of questioned the actual requirement / usefulness of C14, actually. The oscillator section has its own "bypass cap" as it were (C6),  and the JFET's not doing all that much in the HF range(?).

And since, with the added zener regulation, there's less need for "brute force" supply rail smoothing / filtering, won't 6.3mm diameter electrolytics suffice? Surely those could easily fit into those footprints

A 47u/50v on the "input", before R11 and the zener, and 100-220u/16v after that.

C14 came from my earlier experiments with my Hi-Z Schoeps....... the low pass filter after the zener - R12 and C15 - gets rid of most of the zener noise, but electrolytics are not always that good at HF decoupling....... So I added C14 to get rid of any HF noise not 'captured' by C15.

As you say, probably overkill....

Yes, I'm sure 6.3mm electolytics would be fine..... I just tend to think in terms of what I already have in stock here! 

 

[ricardo]

C14 came from my earlier experiments with my Hi-Z Schoeps....... the low pass filter after the zener - R12 and C15 - gets rid of most of the zener noise, but electrolytics are not always that good at HF decoupling....... So I added C14 to get rid of any HF noise not 'captured' by C15.

As you say, probably overkill...

You need C14 to stop stuff from the oscillator getting to R10 & the FET drain.

rogs, your blue extra bits in #276 & RF.AMX8Zen is the correct way to do things.  Many 'simpler' arrangements abound but (as you have found) lesser arrangements will fail at times.  It's not overkill

With the 'high' voltage BC560s, D2 & 3 don't have to be Zeners.  They could be 1n4148 or any number of SMD double diodes.  But you must have them or else a momentary cable short will zap the base emitter junctions.  The mike will still work but be forever noisy.

Don't forget to leave enough space for a bigger C13 to flatten response & reduce noise.  I think I said 10n with R4 & 10 4k7 so C13 should be 22n with2k2

If it is still necessary to reduce the sensitivity further, then adding  pad resistors across T2 secondary  - suggested by Ricardo and implemented by Khron in his AMX9 layout - should do the job.
Sadly, they will  not reduce the noise floor by the same degree of attenuation.

If my understanding of the noise sources is right, damping T2 secondary will reduce noise & signal equally for no difference in S/N

Some of you will know I have an obsession with simplification .. but never at the expense of performance or reliability.

The mods we are discussing here all improve reliable performance and we have good evidence they work and are worthwhile.  8)

My musings about feeding the rectifier directly from the bridge come under 'stuff that must be tested' but, IF they work they MAY improve & simplify stuff. 

 

[Khron]

Isn't the oscillator power already sort of "isolated" by R2 and C6? I suppose one could "guild the lily" and replace R2 with an inductor, for extra HF / RF filtering...

You need C14 to stop stuff from the oscillator getting to R10 & the FET drain.

 

[rogs]

With the 'high' voltage BC560s, D2 & 3 don't have to be Zeners.  They could be 1n4148 or any number of SMD double diodes.  But you must have them or else a momentary cable short will zap the base emitter junctions.  The mike will still work but be forever noisy.

I had just copied the 6v2 protection diodes as fitted in the original Schoeps CMC5 schematic ... I see your point of  them simply acting as diodes, should either pins 2 or 3 be shorted directly to ground - but I had always assumed that using a zener was to protect against  excess voltage presented across the transistor, in the event of the 47uF cap being fully discharged.
If it was then presented with a 'live' phantom power 'plug in'  it might present excess voltage across the CE junctions before the capacitor was charged?......
A bit remote I would have thought? .... but  as Schoeps had fitted them - and at a cost of only at 3p per diode - I decided 'hang the expense'! 

If my understanding of the noise sources is right, damping T2 secondary will reduce noise & signal equally for no difference in S/N

Yes of course..... I had assumed that the dominant noise source might be the resistor noise  from the source and drain resistors, which is 'post' attenuator... but a quick practical test confirms your assumption that the noise floor is indeed reduced as well.
(I do need to confirm the actual values for the -10 and -20 dB pads ... the provisional values of 3k3 and 1k  on both the AMX9 and AMX10 schematics may not be correct...

My musings about feeding the rectifier directly from the bridge come under 'stuff that must be tested' but, IF they work they MAY improve & simplify stuff. 

I looked again at Baxandall's paper and found this comment on page 594:

"....Ordinary conversational speech at a foot or two corresponds to about 1 dyne/cm' alternating pressure, and this causes, with a typical modern electrostatic microphone element, a capacitance change in the region of 0.001 pF....."

I decided that's not a lot! -  and changes that small would need all the help they could get....... 
Hence my decision - at this stage - to stick with the 'noise free' gain  from the step up voltage - and 'Q' - provided by T2 ....

Isn't the oscillator power already sort of "isolated" by R2 and C6? I suppose one could "guild the lily" and replace R2 with an inductor, for extra HF / RF filtering...

...
C6 was fitted to decouple the collector of the oscillator transistor, which helps with the linearity of the oscillator waveform ...R2 was selected to allow the optimum oscillator level - at max 'Q' setting - to be determined.

The idea of C14 was to add extra decoupling  to keep any HF zener noise out, should the electrolytic not be good enough at high frequencies. 
In fact it would also act - with R2 - as a low pass filter to help keep any remaining oscillator noise away from the drain resistor as well.

In the past, I have had problems with HF noise and electrolytics, - but that was some years ago, and modern electrolytics are probably much better with HF..... old habits die hard! 

 

[ricardo]

I had just copied the 6v2 protection diodes as fitted in the original Schoeps CMC5 schematic ... I see your point of  them simply acting as diodes, should either pins 2 or 3 be shorted directly to ground - but I had always assumed that using a zener was to protect against  excess voltage presented across the transistor, in the event of the 47uF cap being fully discharged.
If it was then presented with a 'live' phantom power 'plug in'  it might present excess voltage across the CE junctions before the capacitor was charged?......
A bit remote I would have thought? .... but  as Schoeps had fitted them - and at a cost of only at 3p per diode - I decided 'hang the expense'! 

The reason Dip. Ing. Wuttke used Zeners is cos he has quite low voltage PNPs as the outputs.  There's no problem with using Zeners though your 'high voltage' BC560s can do with plain diodes.

Yes of course..... I had assumed that the dominant noise source might be the resistor noise  from the source and drain resistors, which is 'post' attenuator... but a quick practical test confirms your assumption that the noise floor is indeed reduced as well.

This was what prompted my original reply.  But your extra 16dB sensitivity compared to Rode NT1 makes all that moot.

"....Ordinary conversational speech at a foot or two corresponds to about 1 dyne/cm' alternating pressure, and this causes, with a typical modern electrostatic microphone element, a capacitance change in the region of 0.001 pF....."

I decided that's not a lot! -  and changes that small would need all the help they could get....... 
Hence my decision - at this stage - to stick with the 'noise free' gain  from the step up voltage - and 'Q' - provided by T2 ....

Listening to the noise on your samples, I can barely hear those 'clonks' and something starting up with the signal magnified greatly in Audacity.

Extra noise free gain would only be an advantage to drown the noise of a less-than-stellar preamp.  Your Sound Devices is about as quiet as its theoretically possible.

Even the Rode's sensitivity makes it quiet with lesser preamps so yours is well above what is required for these.

I'm investigating Infinite Impedance Detectors.  Most seem to suggest larger Drain resistors eg 10k.

Was there some reason to go from 4k7 to your present 2k2?

I'm just trying to understand how all this works.

 

[rogs]

The reason Dip. Ing. Wuttke used Zeners is cos he has quite low voltage PNPs as the outputs.

According to the data sheets, the CE breakdown voltage for Wuttke's BC416 and for the BC560s are the same .. at - 45V ....
So  I'm probably just following Wuttke's lead and being ultra safe! 
 

Extra noise free gain would only be an advantage to drown the noise of a less-than-stellar preamp.  Your Sound Devices is about as quiet as its theoretically possible.

My Sound Devices unit is a (now rather old) USBPre 1.5 ... and yes, the preamps are quite quiet...
Not quite as quiet as my Tascam DR100Mk3 though...Even my cheap Behringer UMC404HD 'Midas' preamps give quite a good account  of themselves , noise wise.
So I'm a bit spoilt for choice with quiet preamps ...

I'm investigating Infinite Impedance Detectors.  Most seem to suggest larger Drain resistors eg 10k.

I came across the infinite impedance concept almost by accident, and thought it might be worth a try.  I was immediately impressed by the results over a simple diode rectifier...and that seems to be a common thought among those who have used infinite impedance detectors.
Usually for higher quality AM detection in radio, from what I've read..
I would guess that the drain resistor is selected to suit the  FET and the supply voltage?......

Was there some reason to go from 4k7 to your present 2k2?

I changed to 2k2 resistors after I had added the voltage regulation components, as they seemed to suit the FET requirements better, with a supply voltage now around 9V. 
That has certainly been the case with my Hi-Z Schoeps style circuit..... and the smaller value resistors seem to generate less noise - although I'm guessing that's marginal?

I'm just trying to understand how all this works.

Me too!    ..... I think there is more to be done on calculating optimum FET biasing.
It looks like there is going to be a quite narrow range of ideal  FET gate voltage input from T2 ?...Too low and the noise level increases  as the FET doesn't seem to turn off completely in the +ve half cycles? ...Too high and the 'on' voltage swing seems to become non linear....
Problem is, it's the calibration of T1 - not T2 -  that  seems to determine the optimum setting for noise and distortion - which is slightly confusing to a simple hobbyist like me?...
Investigations continue!

 

[Khron]

I could be wrong, but calculating things will only get you so far with JFETs, what with them being not exactly tight-tolerance (or tightly-binned) devices. Hence the 1M trimmer in the Schoeps circuit, or the "hand-selected" source resistors in KM8x's and U87's.

But of course, having a starting point, or some values to aim for, would obviously help

Me too!    ..... I think there is more to be done on calculating optimum FET biasing.
It looks like there is going to be a quite narrow range of ideal  FET gate voltage input from T2 ?...Too low and the noise level increases  as the FET doesn't seem to turn off completely in the +ve half cycles? ...Too high and the 'on' voltage swing seems to become non linear....
Problem is, it's the calibration of T1 - not T2 -  that  seems to determine the optimum setting for noise and distortion - which is slightly confusing to a simple hobbyist like me?...
Investigations continue!

 

[rogs]

I could be wrong, but calculating things will only get you so far with JFETs, what with them being not exactly tight-tolerance (or tightly-binned) devices. Hence the 1M trimmer in the Schoeps circuit, or the "hand-selected" source resistors in KM8x's and U87's.

But of course, having a starting point, or some values to aim for, would obviously help

I'm sure you're right.... 'hands on' and you're  checking what actually works!...

The thing I'm currently looking at is how the adjustment of T1  is the one that appears to be performing almost identically to the adjustment of the 1M trimmer in the Schoeps circuit.... 
On a spectrum analyser, you can actually see the change in the  2nd harmonic distortion  - down to a very similar level to the Schoeps -  as you adjust the core of T1.
I would have expected it to be T2 (which is the one that changes the amplitude of the signal presented to the gate) that made the difference?...
EDIT: Bit of a red herring ... sorry!...... Yes, tuning T1 can appear to change the  level of distortion, but only because  changes the level of the oscillator output, and thus the amplitude of the T2 output as well.
So tuning either core can affect the sensitivity and the linearity as the FET bias is changed. 
There does appear to be only one 'sweet spot' -  although there are other (smaller) peaks as you tune for the optimum setting against a tone reference.

As I say, investigations continue... 

 

[Gerard]

My order of RF.AMX8 PCBs arrived yesterday, some time after Khron uploaded the RF.AMX10 PCB layout last week.  :  My previous PCB order, for the original diode detector layout, also arrived after Rogs had advanced the design to the 'Infinite Impedance Detector' circuit. It seems delivery of my PCBs are always a couple of generations behind the state of the design. 

All my present choral concerts will be finished early next week, so maybe I will have some evenings free and can get back to making some of the latest generation of this design.

Quick question: Which capsule seems to work best for choral music and which for solo voice? Has anyone tried them in real life? I notice Rogs started with the RK-67 type, but now seems to mention the C12. Is that becayse they are what is to hand, or is it as a result of trying them in a real recording environment and, if so, what were you recording?

 

[RuudNL]

Off topic, but great to see how this project has grown since the start!

 

[rogs]

My order of RF.AMX8 PCBs arrived yesterday, some time after Khron uploaded the RF.AMX10 PCB layout last week.  :  My previous PCB order, for the original diode detector layout, also arrived after Rogs had advanced the design to the 'Infinite Impedance Detector' circuit. It seems delivery of my PCBs are always a couple of generations behind the state of the design. 

I know what you mean about PCB deliveries!.... I received my second batch of Ruud's original prototype PCB the same day as Khron's posted his first BM800  version......

Only to be expected I suppose, when you have such useful contributions to a thread like this....... and I repeat, I don't think we should be anywhere near this far on without such great  input from others -- especially  Ruud and Khron !

I don't have any AMX10 PCBs yet, but I have managed to add the voltage regulator mods to the AMX8 PCB on a small stripboard - and without cutting any tracks on the PCB. (Photo attached).
It's only really important when used with poor phantom power supplies ... I can jot down a few notes if that would help?...

Quick question: Which capsule seems to work best for choral music and which for solo voice? Has anyone tried them in real life? I notice Rogs started with the RK-67 type, but now seems to mention the C12. Is that becayse they are what is to hand, or is it as a result of trying them in a real recording environment and, if so, what were you recording?

I've tried out K47, K67 and C12 style capsules, simply because that's what I have here.... They are only Chinese versions, sadly...

The C12 is the most sensitive and thus has the lowest noise floor. The K67 is almost as sensitive, but requires a different value for bridge balance capacitor... ( around 68pF for K67 and  68p in parallel with 47pF for the C12.

The K47 was lot less sensitive -- and surprisingly had the largest capacitive value, at around 100pF.

I've only done voices tests here so far, and I prefer the C12 personally... The K67 is too bright for my liking. 
I haven't done any  internal EQ testing -- I tend to prefer adding EQ externally -  so the AMX10 FET load values of 2k2 and 4n7 should give an HF -3dB cutoff frequency at around  16KHz.  Other values may suit different versions better?....

 

[rogs]

Off topic, but great to see how this project has grown since the start!

Not quite sure how far it would have got without the huge 'nudge' your prototype PCB gave it ! .... thanks again ...

 

[Gerard]

I have managed to add the voltage regulator mods to the AMX8 PCB on a small stripboard - and without cutting any tracks on the PCB. (Photo attached).
It's only really important when used with poor phantom power supplies ... I can jot down a few notes if that would help?...

Please; that would help me. My main recorder is a Mk. 1 TASCAM DR-680, which has noisy phantom power (I've added extra smoothing on the P48 supplies for inputs 1-4, but I can't do the same for inputs 5-6; I use inputs 1-4 for my Ambisonic mic).

I look forward to seeing any final updates or comments. But mainly I look forward to trying out these mics and asking some professional recording friends for their comments. I'll post their comments and my own.

 

[rogs]

Please; that would help me. My main recorder is a Mk. 1 TASCAM DR-680, which has noisy phantom power (I've added extra smoothing on the P48 supplies for inputs 1-4, but I can't do the same for inputs 5-6; I use inputs 1-4 for my Ambisonic mic).

Some notes on the actual experiment here:

www.jp137.com/lts/RF.AMX8voltreg.pdf

Certainly helps keep out the rubbish from the  noisy phantom power I have on a preamp here ( part of a cheap Behringer mixer)

 

[Gerard]

Thank you, Rogs. Much appreciated.

When things quieten down, I'll make up some of the RF.AMX8 boards with the add-on power supply stripboard and see how I get on. Gotta make up a power supply for a Sennheiser MKH 110 too - that's the one with a controlled frequency response down to 1 Hz (the MKH 110/1 allegedly went down to 0.1 Hz! That's more like a high-speed barometer). However, I'm taking part in two choral concerts tomorrow and I've got to finish and present the accounts and treasurer's report for one of my choral societies next week; so I probably won't get anything done electronics-wise before the bank holiday weekend.

 

[rogs]

Thank you, Rogs. Much appreciated.

When things quieten down, I'll make up some of the RF.AMX8 boards with the add-on power supply stripboard and see how I get on. Gotta make up a power supply for a Sennheiser MKH 110 too - that's the one with a controlled frequency response down to 1 Hz (the MKH 110/1 allegedly went down to 0.1 Hz! That's more like a high-speed barometer).

I look forward to seeing how you get on in due course .....

Re: the extreme Sennheiser LF response ...
You may have noted that Baxandall's paper  includes the observation (on page 594 in the section  'Linearity Measurements') that "this microphone system has a response extending down to zero frequency" ... So providing the coupling caps are large enough  I can't see why you couldn't get a response that extended down to 1Hz  from our mic?...
The precise  capacitor values would need to be based on the actual input impedance of the emitter followers - which depends in large part on the Hfe of the devices.. so some calcs to be done there...

And I'm not sure quite what you might use it for?...recording elephant infra-sound maybe? 
The lowest organ 32' C pedal has a fundamental of 16Hz , and that would push most  recorders to their limit....
But below that?... pretty specialised stuff I would think! 

 

[Gerard]

Yes, it's an instrumentation mic and reputedly may have had a military application back in the '60s. A pair were sold recently on eBay by a recording hire company in Ireland (actually the later MKH 110/2, but otherwise I believe the same spec as my older MKH 110). Mine came from a theatre that was disposing of older equipment. I got it for recording big organs; though I agree it is rather overkill. It was not very expensive.

I want to get that "trouser-flapping" effect of the low pedal stops. If the mic works properly, next stop is to check the  recorder LF response. And then to hope the recordings are not overwhelmed by environmental infrasonic noise :

Yes, I am aware of Baxandall's comment re LF response; that was one of the reasons I started this project. It will be interesting to see how our mic compares with the MKH 110. 

 

[Gerard]

Yes, it's an instrumentation mic and reputedly may have had a military application back in the '60s.

Just checked. The Sennheiser Micro-Revue 70-71 (http://lcweb2.loc.gov/master/mbrs/recording_preservation/manuals/Sennheiser%20Micro-Revue%2070-71.pdf, page 35) refers to "military and research purposes". The frequency response curve looks suspiciously flat to 10 kHz, as if it comes out of the art department rather than the development or production department