MK221 challenges ...

[gentlevoice1]

Hi all,

Greetings from a very rainy day here in Denmark ...

As it is I have been making a measurement microphone for the MK221 (Microtech Gefell) and BK 4133 (B&K) ½" capsules to be used for headphone measurements. "Everything" seems to be working fine except that for some reason the output from the microphone (i.e. the opamps) almost vanishes after maybe 20-30 seconds. I have a guess why this could be happening but before changing the actual PCB I was wondering if one of you might have a different suggestion ...

The design is based on Scott Wurcer's microphone articles in Linear Audio some years ago (phase splitter with opamp output) and just FYI I have attached the schematic, the PCB layout, and a couple of pictures of the finished microphone design. As can be seen from the pictures all of the high impedance circuitry is encapsulated in an aluminum box - thus the overall noise level is quite low.

However, as I mentioned above, the issue is that the microphone output (~1.5 V from the LM4562 opamp when using a Fluke multimeter to measure the output and when shouting lowly into the microphone) disappears after 20-30 seconds. My own guess is that this is related to possible leakage currents from the connection points between the high impedance parts and the PCB (pcb assembly shown on pictures 2 & 3). These parts are inside the aluminum box - yet the connection wires are isolated from the PCB with relatively thin (50um thickness) virgin PTFE film. And I just wonder if this is sufficient - or I need the PTFE to be thicker?

Or - and this would be worse - maybe I have made some kind of error in the basic connection of the parts around the microphone capsule (left top side of schematic in picture 5)? The resistors isolating the microphone capsule are 5Gohm values and the 1.2nF capacitor isolating the 19pF microphone capsule from the JFET input (2SK222/LSK189) is polystyrene, 630 VDC rating.

I would much appreciate if one of you may know what could be amiss here

Cheers & have a fine day,

Jesper

 

[gyraf]

Have you verified that your +200V polarisazion keeps up during the fail?

 

[gentlevoice1]

Hi ... Good question. It does keep up, however, there is a slight drop in the voltage over time, say 5 volts per minute. I reckon it is mainly the SMD capacitors I use which are X7R types (and have a quite high leakage current - to be replaced, eventually). Do you think this will matter? I can try to keep the polarization charger "on" - then the voltage stabilizes around 204 VDC.

Cheers, Jesper

 

[Khron]

Hi ... Good question. It does keep up, however, there is a slight drop in the voltage over time, say 5 volts per minute. I reckon it is mainly the SMD capacitors I use which are X7R types (and have a quite high leakage current - to be replaced, eventually). Do you think this will matter? I can try to keep the polarization charger "on" - then the voltage stabilizes around 204 VDC.

Cheers, Jesper

Are you implying you're actually turning off the 200V bias supply at some point, for whatever reason? And you're surprised the signal disappears after a while?

 

[gentlevoice1]

Hi Khron ... Yes, I am turning it off and the reason is to have a low noise floor. The coil I use is ever so slightly noisy and this noise is transferred to the microphone. Normally, the capacitors have no problem keeping the voltage stable - but likely due to the X7R capacitors it currently drops a little.

Anyway, if I hear both of you correctly you think the issue is the drop of the polarization voltage? Should be relatively easy to test, if that is the cause ...

Cheers, Jesper

 

[Khron]

Well... Without putting much too fine a point on it... the phrase "duuuh" comes to mind

True, in theory, the capsule and all the other capacitors should indeed not leak any current through, but... You'll have to admit, your fault description sounds strikingly similar to "i'm turning the power off, and my device stops working soon after that".

Does the same thing happen if you keep the bias voltage generator on?

What frequency is that 555-based boost-converter supposed to be switching at? Just thinking you could achieve the same with some more recent-technology device, perhaps a bespoke boost-controller, able to switch at some several-hundred-kHz (in order to get the switching noise well outside the audio range).

 

[gentlevoice1]

@Khron: Well, there's a bit more to it than just that "ooups, it doesn't work".

- As I mentioned I prefer to have a low noise floor as this allows for both low noise and low distortion. To this end I measured distortion yesterday - on a table without the aluminum box so likely a good bit higher noise level - and without fine adjustment reached -112 dB THD for a 0.65Vpp (balanced) signal. Essentially only 2H. Fully acceptable to me, yet this depends on a low noise floor.

- The capacitors I normally use for non-listening circuitry (Rubycon ZL) IME have a vanishingly low leakage current. They typically can be charged and then essentially will almost keep their voltage for days. The higher voltage capacitors I use here (X7R and Nichicon PM) have shown to have a much higher leakage current - something that was a bit unexpected (this is not really apparent from the PM data sheet).

- I have used some time trying to find an SMPS IC that fits my needs, however, besides the LT3751 I haven't really found any. In my experience the challenge typically is that either the circuitry itself may become a bit too complex or costly, and second, that the IC may/might need custom transformers, and third, that these SMPS circuitries typically radiate EMR, as well as are noisy on their output, even if switching at a higher frequency. Which may require either additional shielding or filtering. But if you know of an SMPS IC that is relatively simple to implement I would appreciate hearing about it ...

- I tried to simulate the frequency response of the microphone capsule-near circuitry this morning and it does indeed look like slow polarization voltage changes feed through more or less directly to the output. So, given that the PM/X7R capacitors are indeed more leaky than expected I will have to do something about this. Would be great if this proves to be the issue. Thanks to you both for pointing this out ...

Cheers, Jesper

P.S.: In relation to a high voltage generator a feasible solution could be a non-coil/non-transformer solution - with lower radiation - however, as far as I can see these are not made for voltages higher than 70-80 volts ...

 

[gentlevoice1]

@gyraf: Just want to say thank you again for your tip on the polarization voltage. I have now replaced the X7R and electrolytics with a film capacitor (polyester) and everything now appears to work as it should. So great tip - I hadn't realized that this could cause what it did

Also, the polarization voltage circuitry goes from 0 VDC to 201 VDC in just about 2 seconds and can then be switched off for lower noise. Just as I hoped for.

Also, this small circuitry is capable of quite an impressive performance. This morning I adjusted it again and with a 1.7 Vpp input signal it is capable of appr. -105 dB THD. At 2.9 Vpp it does around -100 dB - and I reckon it can be adjusted to do even better if time allowed.

I would make some changes if the circuitry were to be used for actual recording but for measurement it is now beyond my expectations. Kudos to Scott Wurcer (& Schoeps I think it was who first thought of the phase splitter?) ...

Thanks again & best wishes

Jesper