Is there a way to reduce cell phone interference in DIY KM-84?

[TLRT]

Very easy, you can add a LCR network inside the mic body or inside a dedicated XLR connector

 

[Cqwet Dbdfte]

You may need an RF spectrum analyser to find the offending signal, but there is so much stuff in the spectrum, could even be public service bands. Decoupling at the output is meaningless if the RFI already got rectified into the signal, nF bypassing won't do squat. Bypassing the input signal could kill it. You may need a self resonant cap at the offending freq, like 68pF 0402 NP0, which has a dip around 800MHz. The cap needs to find a good ground to be effective, hard to do on that PCB.
Keeping RF from entering the resonant cavity should be at top of list.

 

[TLRT]

You may need an RF spectrum analyser to find the offending signal, but there is so much stuff in the spectrum, could even be public service bands. Decoupling at the output is meaningless if the RFI already got rectified into the signal, nF bypassing won't do squat. Bypassing the input signal could kill it. You may need a self resonant cap at the offending freq, like 68pF 0402 NP0, which has a dip around 800MHz. The cap needs to find a good ground to be effective, hard to do on that PCB.
Keeping RF from entering the resonant cavity should be at top of list.

U87's LCR HPF filter cuts above 1.2MHz...isn't that just easier?

 

[Matt Nolan]

Off-the-wall suggestion.

I see you have a tant electrolytic in the source resistor bypass position, whereas the non-DIY ones have alu electrolytics. Is it possible that you have a non-polarised tant here or have it accidentally wired in reverse polarity? This could make it more susceptible to RF pickup which is then demodulated by the FET. If I have my understanding of tant construction right, that is...

 

[MicUlli]

22pF to 68pF SMD cap (smaller than 1206) directly at the XLR inlet should help a lot.
Additionally it may help to add a ferrite bead at the capsule connection wire.
My 1 cent

 

[joulupukki]

Off-the-wall suggestion.

I see you have a tant electrolytic in the source resistor bypass position, whereas the non-DIY ones have alu electrolytics. Is it possible that you have a non-polarised tant here or have it accidentally wired in reverse polarity? This could make it more susceptible to RF pickup which is then demodulated by the FET. If I have my understanding of tant construction right, that is...

They are polarized. I've got them in the proper orientation.

 

[Delta Sigma]

Read this quickly, so I may have missed it, but did you break the XLR Pin 1 to PCB connection?

Pin 1 to mic body, PCB pin 1 pad to mic body as @ccaudle suggested.

 

[joulupukki]

Read this quickly, so I may have missed it, but did you break the XLR Pin 1 to PCB connection?

Pin 1 to mic body, PCB pin 1 pad to mic body ass @ccaudle suggested.

Yes. That's how I've currently got it wired. I'm not getting any weird ground or 60-cycle hum at all. I think it's parasitic coupling (or something) somewhere in the circuit that's picking up the cell phone transmissions. And it's in the audible range of frequencies so I'm not convinced a filter is going to remove it. My best guess is that it's something to do with the board layout and/or choice in components. Maybe the transformer leads or even the fact that the copper wire trace going from C4 to the transformer primary pad runs along the outside of the board. I don't know (obviously).

 

[ccaudle]

And it's in the audible range of frequencies so I'm not convinced a filter is going to remove it.

Noise pickup from cell phones and other RF sources does not actually start out in the audible range, it starts somewhere in the MHz range, and non-linear effects in the transistor junctions create low frequency signals based on the modulation of the signal. Same effect used to intentionally demodulate AM radio signals in old radio designs, just happens accidentally.
So you filter the high frequency signals as early as possible, and then there is no signal to demodulate down into the audio range (or at least the signal is reduced in amplitude enough that it is no longer audible, even if technically remnants remain that you could see with test equipment).

 

[Ike Zimbel]

We’ve got 5G networks here, but my house, especially in the basement where my studio is, barely hangs on to a data connection via LTE.

Which means a cell phone in your house is using the MOST power it can output to reach the network.

 

[Cqwet Dbdfte]

U87's LCR HPF filter cuts above 1.2MHz...isn't that just easier?

IF the RFi is at the FET gate, i.e. RF rectification, the signal is already damaged, in the audio band, included in the output signal where filters would have no effect.
But if the RFI crawls up the XLR connection a filter there would pf course help.
Attenuating the RF getting to the FET would help. Making mike body and a screen a Faraday cage would be a primary objective.

 

[TLRT]

IF the RFi is at the FET gate, i.e. RF rectification, the signal is already damaged, in the audio band, included in the output signal where filters would have no effect.
But if the RFI crawls up the XLR connection a filter there would pf course help.
Attenuating the RF getting to the FET would help. Making mike body and a screen a Faraday cage would be a primary objective.

Km84's were sensitive to these high frequency noises,see u87A schematics with LCR filter after the input. That can be adapted inside a xlr female connector, it helped a lot, I had the same situation. Of course mic needs proper grounding.

 

[Matador]

I have heard this exact same sound MANY times when debugging tube guitar amps with my cell phone near. Interestingly enough, you can hear different versions of it depending on the type of data connection the phone is using (e.g. if it is tunneling the cell connection over WiFi, etc).

For a guitar amp, I've corrected it in most cases by bandwidth-limiting the first stage: adding a cap to ground right on the first stage grid helps some (on the order of 100pF) , and adding a cap in parallel with the plate load resistor helps quite a bit more (68pF over a 100K works quite well). Both in tandem can reduce it to the point where it's inaudible.

I'm wondering if bypassing the drain resistor might help? Standard KM84 drain resistor is 47K, so maybe a 150pF would band limit above 20kHz.

 

[Jakethesnake3000]

I had this problem recently with a DIY LDC mic using a KM84 circuit. Like with your mic it seemed to be extra sensitive around the transformer, but that was a red herring. I managed to significantly reduce it by cleaning/reducing the length of the high impedance lead going to the gate...

 

[joulupukki]

I had this problem recently with a DIY LDC mic using a KM84 circuit. Like with your mic it seemed to be extra sensitive around the transformer, but that was a red herring. I managed to significantly reduce it by cleaning/reducing the length of the high impedance lead going to the gate...

That’s really interesting! As far as I can tell I’ve got it super clean (cleaned everything up there with 99% Isopropyl Alcohol). Would there be a good way of making the gate lead shorter in this build given that the PCB has the JFET so far away from the isolated post?

If I mounted the JFET so the gate lead is really short but the source and drain wires were longer, would that help at all or just move the problem to a different spot?

 

[Khron]

If I mounted the JFET so the gate lead is really short but the source and drain wires were longer, would that help at all or just move the problem to a different spot?

That's at least worth a try - both outputs are FAR lower-impedance nodes than the gate connection.

 

[nsalis]

I didn't follow the discussion in detail; just one question: Ike Zimbel recommended using the Neutrik EMC connector. Have you tried that? It is the easiest way to make a product with a Pin1 problem functional.

https://www.neutrik.com/en/product/nc3fxx-emc-b

Nick Salis

 

[Cqwet Dbdfte]

Of course mic needs proper grounding.

Grounding at 800+MHz and 20KHz is very different.
What constitutes "ground" at 20KHz could be a very good antenna at 800MHz. A wifi antenna can be size of your pinky nail.
A series inductance (air coil, ferrite,..) in the gate lead, and bypass the drain with
a SMT 0402 68pF or less, to the FET source should be tried .
An inductance as small as 27nH in the drain is a good load for an RF amplifier.
Keeping RF from getting into the mike would be my first effort. No plastic insulation on the screen, good metal to metal contact to the mike body etc.
The XLR should be easier to handle with ferrites and SMT caps.
The whole inside of the mike body may be treated with RF absorber material.

 

[TLRT]

Grounding at 800+MHz and 20KHz is very different.
What constitutes "ground" at 20KHz could be a very good antenna at 800MHz. A wifi antenna can be size of your pinky nail.
A series inductance (air coil, ferrite,..) in the gate lead, and bypass the drain with
a SMT 0402 68pF or less, to the FET source should be tried .
An inductance as small as 27nH in the drain is a good load for an RF amplifier.
Keeping RF from getting into the mike would be my first effort. No plastic insulation on the screen, good metal to metal contact to the mike body etc.
The XLR should be easier to handle with ferrites and SMT caps.
The whole inside of the mike body may be treated with RF absorber material.

This RF sensitivity was a known problem for older neumann mics,
I suggested to check the u87 model which has a RLC filter at the input and less sensitive to mobile devices. It is an easy mod to adapt this filter.

 

[Delta Sigma]

Does the MP body tube have a better connection to the bottom bell than the DIY?

Or do they both have poor bonding and that's why the mics need to have a short gate lead?