DIY RF Condenser Mics

Help Support GroupDIY:

rogs

Well-known member
Joined
Jun 23, 2006
Messages
350
Location
UK
Couple of further 'tweaks' -- ( I've updated the project notes to version 4.2 )

• Now that R2 is no longer a link, C6 only decouples the oscillator supply. So I've now fitted a 100nF cap into the 'vacant' D1 location, to provide additional HF decoupling across C15. Probably overkill ? .... but it's only a few pence!

• I was wondering if C5 and C10 fitted as 220nF extended the LF response too far? Depends on what impedance you're working into of course, but with the BC557B emitter followers having a typical Hfe of around 280, I thought it might be better to try 100nF.

Not exactly straightforward to measure the frequency response of an RF mic circuit without the capsule connected, so I referred back to the wisdom of Peter Baxandall, and knocked up a little interface based on his ideas around fig.6 on page 593 (PDF sheet 6) of his 1963 paper. (Copy HERE ).

I've attached a schematic of my testing interface below.

Connecting to my Sound Devices USB 1.5 preamp, I find that with C5 and C10 fitted as 100nF the LF -3dB point is around 20Hz.

Initial test suggest a pretty linear frequency response (within 0.5dB) up to around 18KHz, with a -2dB drop in level at 20KHz.
The initial THD+n figure at 1KHz reads as 0.045% -- although as the source reads 0.0285%, it may be slightly better. So not too bad.

Experiments (as always!) continue :) ...
 

Attachments

  • AMX10 test interface.jpg
    AMX10 test interface.jpg
    72 KB · Views: 45
Last edited:

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,277
Location
Marcelland
• I was wondering if C5 and C10 fitted as 220nF extended the LF response too far? Depends on what impedance you're working into of course, but with the BC557B emitter followers having a typical Hfe of around 280, I thought it might be better to try 100nF.
I wouldn't worry so much about LF response. With 100n, the +3dB point is about 12Hz and response at 20Hz is -1.3dB.
However I would be much more concerned about the impedance unbalance at LF.
See attached table.
It is considered good practice to maintain impedance unbalance under 20 ohms, for adequate CMRR.
An unbalance of 20 ohms limits the max CMRR to 50dB with an otherwise perfect mic input.
See how 220n barely scratches it, when 100n fails.
That's why I chose 470n in the latest builds of my Micbooster.
 

Attachments

  • Scoeps Z unbalance and LF response.jpg
    Scoeps Z unbalance and LF response.jpg
    21.5 KB · Views: 34

rogs

Well-known member
Joined
Jun 23, 2006
Messages
350
Location
UK
Once again, I had followed Schoeps lead ... he has 100nF for both capacitors -- (although he does go to the trouble of recommending matching emitter follower transistors! )
I think I should probably return the value of C5 to 220nF, and leave C10 as 100nF.
As the PCB has had ongoing project mods, it is not easy to fit C10 with the larger size 220nF capacitor very neatly!

According to your table, C5 as 220nF and C10 as 100nF should still keep the Z value differences at less than 20....
 
Last edited:

Khron

Well-known member
Joined
Apr 8, 2010
Messages
1,983
Location
Finland
To be fair, in balanced connections, impedance is the whole point; whether the signal happens to be symmetrical / present on the "cold" line as well is a nice-to-have, but not a necessity.
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,277
Location
Marcelland
The table in my last post makes no reference to the respective levels of the outputs, only to their impedances, and shows that, although the midrange impedances are quite well matched, at low frequencies (where it matters the most) the deviation can be a problem.
 
Last edited:

shot

Well-known member
Joined
Nov 28, 2008
Messages
643
Location
Zagreb, Croatia
Back to the capsule question!
Are you using the same capsule in every version and simply changing the parts on the pcb, or you always assemble new mic with new capsule? Have you ran into problems with different capsules?
I'm asking because I'm struggling with three mics that I'm attempting to build. They sound like sh*t and after going back and forth with components, even rebuilding one mic from the scratch back to original v1.0 circuit it's always the same - noise noise noise!

I've built my first RF back when this journey started with aliexpress 67-ish capsule (rogs recommended) and v1.0 circuit. That mic is stellar! It is loud and it has insanely low amount of noise. Basically preamp noise dominates and mic's noise feels as non-existent. And this mic's output it like 12-15db louder than any other RF mic I've built!

Almost year later I ordered two more capsules and I've put them in a drawer for some time. Few months later I ordered two more. Always the same aliexpress capsule. And I've built three mics with those newly bought capsules - they all have too much noise. At some point I finally got the tester tool that can measure small capacitance so I've tested three of those capsules and they have 33pf, 44pf and 48pf of capacitance. Really low! I didn't measure the first capsule I bought that is in the good sounding mic. But I'll do that when I have time.

I'm actually not sure if my noise problems are down to the capsule, but I'm out of ideas and I'm more and more towards blaming the low capacitance of the capsule for it. It's output level is probably too low so SN ratio gets reduced. Btw, no problems tuning T1/T2 on them.
I wouldn't be surprised if aliexpress ships whatever they have in stock so no consistency at all.
And I don't have at hand any other spare capsule right now to test

:(

Luka
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,277
Location
Marcelland
so I've tested three of those capsules and they have 33pf, 44pf and 48pf of capacitance. Really low!
Even if the good capsule measured ar 60-80pF, the one with 33pF would sound only about 6-8dB below, and have no more noise.
You should probably look elsewhere. Can You check the RF level at T1's primary, against teh one that workd good?
 

rogs

Well-known member
Joined
Jun 23, 2006
Messages
350
Location
UK
I have built around 10 of these mics, and all have had acceptable noise floors. The lowest noise and best sensitivity came from a version using a 3 micron edge terminated capsule, which measured at around 90pF.
That same capsule would also collapse if you applied more than around 50V DC from a 'conventional' circuit. So I have concluded it probably has a lower tension membrane - which is ideal for an RF mic!
It also allows the system to be run at 8MHz, which is closer to the 7MHz specified on the inductor spec sheet.... So the Q is probably a bit higher.

I have tried several 'K.67' type Aliexpress capsules and have found them to be quite sensitive. Those have all had a capacitive value of around 65pF.
I have also tried a couple of Chinese'K.47' style mics - again with values around 60/65pF - but these were a lot less sensitive (by around 10dB).

If you use a capsule with a significantly lower capacitive value, I've found the bridge imbalance is often too extreme to allow a 'useful' calibration point. With the other bridge capacitor being significantly larger than the capsule, the resonant inductor tuning point generates a higher level of RF carrier - and with a smaller proportion of capsule capacitive value 'variation'.
This results in a lower modulation amplitude, with a higher RF carrier level, and this will tend to generate more noise.
I've also found that a significantly unbalanced bridge seems to allow several 'peaks and troughs' in the inductor tuning range - with no obvious dominant peak setting.
With a higher value capsule - and a more balanced bridge set up - there tends to be very much a dominant 'peak' setting -- which is often quite dramatically obvious. A much improved signal to noise ratio as a result of course...

Ideally, using a capsule with a capacitive value of - say - 40pF you would probably get better results from an inductor value of 10 or 12 uH. Unfortunately I've not been able to locate any such device off the shelf -- although I imagine some experienced hobbyists might consider 'winding their own'?....
Using a 5.3uH inductor with such a low capacitive value will require a lower value bridge 'balancing' capacitor. That will require a much higher oscillator frequency -- which will probably result in a much reduced system 'Q' -- and higher noise!

The more recent project revisions have concentrated on reducing the system current. One major advantage of RF condenser mics is that the low impedance allows for outdoor use, and low current drain is a useful feature for battery powered recorders.
The lower oscillator voltage does reduce the system signal to noise ratio a little -- but not dramatically.

The project notes still include a link to the original circuit set up -- see HERE .

As I say, I've had my best results from a low tension 3 micron 90pF capsule. And the bridge being almost balanced seems to give the best results. (The self biased infinite impedance detector sits automatically at 'Vp' so it doesn't need much to turn it on and off!)
 
Last edited:

shot

Well-known member
Joined
Nov 28, 2008
Messages
643
Location
Zagreb, Croatia
Even if the good capsule measured ar 60-80pF, the one with 33pF would sound only about 6-8dB below, and have no more noise.
You should probably look elsewhere. Can You check the RF level at T1's primary, against teh one that workd good?
Yeah, I'll do that!
I just need to remember to bring back the good one from the studio back on my home bench
 

shot

Well-known member
Joined
Nov 28, 2008
Messages
643
Location
Zagreb, Croatia
I have built around 10 of these mics, and all have had acceptable noise floors. The lowest noise and best sensitivity came from a version using a 3 micron edge terminated capsule, which measured at around 90pF.
That same capsule would also collapse if you applied more than around 50V DC from a 'conventional' circuit. So I have concluded it probably has a lower tension membrane - which is ideal for an RF mic!
It also allows the system to be run at 8MHz, which is closer to the 7MHz specified on the inductor spec sheet.... So the Q is probably a bit higher.

I have tried several 'K.67' type Aliexpress capsules and have found them to be quite sensitive. Those have all had a capacitive value of around 65pF.
I have also tried a couple of Chinese'K.47' style mics - again with values around 60/65pF - but these were a lot less sensitive (by around 10dB).

If you use a capsule with a significantly lower capacitive value, I've found the bridge imbalance is often too extreme to allow a 'useful' calibration point. With the other bridge capacitor being significantly larger than the capsule, the resonant inductor tuning point generates a higher level of RF carrier - and with a smaller proportion of capsule capacitive value 'variation'.
This results in a lower modulation amplitude, with a higher RF carrier level, and this will tend to generate more noise.
I've also found that a significantly unbalanced bridge seems to allow several 'peaks and troughs' in the inductor tuning range - with no obvious dominant peak setting.
With a higher value capsule - and a more balanced bridge set up - there tends to be very much a dominant 'peak' setting -- which is often quite dramatically obvious. A much improved signal to noise ratio as a result of course...

Ideally, using a capsule with a capacitive value of - say - 40pF you would probably get better results from an inductor value of 10 or 12 uH. Unfortunately I've not been able to locate any such device off the shelf -- although I imagine some experienced hobbyists might consider 'winding their own'?....
Using a 5.3uH inductor with such a low capacitive value will require a lower value bridge 'balancing' capacitor. That will require a much higher oscillator frequency -- which will probably result in a much reduced system 'Q' -- and higher noise!

The more recent project revisions have concentrated on reducing the system current. One major advantage of RF condenser mics is that the low impedance allows for outdoor use, and low current drain is a useful feature for battery powered recorders.
The lower oscillator voltage does reduce the system signal to noise ratio a little -- but not dramatically.

The project notes still include a link to the original circuit set up -- see HERE .

As I say, I've had my best results from a low tension 3 micron 90pF capsule. And the bridge being almost balanced seems to give the best results. (The self biased infinite impedance detector sits automatically at 'Vp' so it doesn't need much to turn it on and off!)
I was always able to find a tuning point. Never had a situation that any of the capsules won't even tune. But noise is mostly constant, in some positions it falls off. But at that position sine wave response also falls of significantly, so in practice SNR at that point is even worse.

I've tried with different values for C4 and C8 and they just change the tuning point. And they change the timbre of the capsule. At some values mic gets darker (usually too dark) and at some values it starts to loose low end. It can actually benefit to pick an exact value based on a particular capsule to get a tone you like! And after most of capacitor swaps I was still able to retune the inductors. But that damn noise was in most cases the same. Maybe better or worse by few db's but still very noticeable.

I need to bring back the first mic I've built back to the bench and do some measuring and compare it to those newly built ones.
 

rogs

Well-known member
Joined
Jun 23, 2006
Messages
350
Location
UK
I was always able to find a tuning point........
I can confirm that I too have always been able to find a tuning point.....sometimes several, with a particularly 'unbalanced' bridge.
It's whether it is the primary tuning point - and with a high enough 'Q' - that seems to make the difference!

The 5.3uH Spectrum coils have a specified inductance range of 3 to 7.5uH. How far the 'Q' drops towards the edge of those limits is not specified?

The selection of the appropriate tuning capacitor for T2 has always seemed fairly straightforward..... 68pF for 8 MHz and 47pF for 10MHz should allow the inductor to be tuned at pretty much the middle of its range. (5.8uH for 8MHz and 5.4uH for 10MHz ).
The JFET doesn't really present any additional loading (unlike to the earlier diode rectifier did!)

The maths surrounding the tuning of T1 are way out of my league . Abbey did attempt a simulation of the inductor assembly earlier in this thread, but - as he said - there are so many unknown variables it's only possible to make an approximation.

The simple capacitive load across T1 secondary is C4 and the capsule in series. So, with both the capsule and C4 as 68pF, we would get a load of 34pF which is just within the specified range of the inductor at 7.45uH.
In fact T2 primary is fitted across the bridge -- and that makes things a lot more complex. Certainly, with a 65pF capsule and C4 as 68pF I have always found it possible to find the optimum tuning point of T1 - with the inductor core somewhere near the centre of its range.

I have only briefly tried out capsules with smaller values, and found the tuning much less precise (although there were some 'peaks' ). The system 'Q' certainly seemed a lot lower.
I decided there would probably be a need for alternative inductors to get useful improvements but - as I mentioned earlier - I haven't been able to locate any as yet.

I suspect there maybe be some major component changes required to get good results from capsules with lower capacitive values. Either higher oscillator frequencies or higher inductor values -- or maybe both?

There has been a lot of expert input on this thread, and it maybe that someone has already tried alternatives, and not yet posted their results?
I think that's why a forum like this is so useful for the experimental nature of this kind of project.....
 
Last edited:

Latest posts

Top