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

[jp8]

Even with a hermetically closed metal enclosure, you will probably still notice it will have little or no effect as the RFI will enter the mic through the XLR cable. I think that has been proven already by joulupukki's experiments using the phone as an "EMC sniffer probe". Without an RFI filter in the XLR cable where it enters the Faraday cage, it can be prevented that RFI enters the Faraday cage. That is how cable entries in EMC test rooms ate carried out. Btw, it is not required to connect the Farady cage to earth ground in order to work. Actually, the mic body of a condenser mic is a Faraday cage, unfortunately in most cases containing a wide open door for RF signals we commonly know as XLR connector. But good enough a Faraday cage to keep mains hum outside the mic. Mic bodies are not connected to earth ground either yet they do work as a Farady cage.

Jan

 

[Cqwet Dbdfte]

You want the mike body and the cable shield to be the reference plane, as I stated, "ground" in a general sense is not a useful concept at high RF.
The mike body _could_ be a Faraday cage, if, if ,if, but there are many pitfalls.
If indeed the XLR cable and connection is the culprit, it would be easy to get 40dB rejection from the XLR to the mike circuit.
https://www.digikey.com/en/products/detail/murata-electronics/NFA18SL227V1A45L/4358102
... or similar. "Big" thru hole parts, while easy for DIY, are useless in this context.
A starquad cable with confirmed EMC properties should be tried.
Chances are good this is not a new problem.
https://www.neutrik.com/en/product/nc3mxx-emc

 

[Cqwet Dbdfte]

Even with a hermetically closed metal enclosure, you will probably still notice it will have little or no effect as the RFI will enter the mic through the XLR cable. I think that has been proven already by joulupukki's experiments using the phone as an "EMC sniffer probe". Without an RFI filter in the XLR cable where it enters the Faraday cage, it can be prevented that RFI enters the Faraday cage. That is how cable entries in EMC test rooms ate carried out. Btw, it is not required to connect the Farady cage to earth ground in order to work. Actually, the mic body of a condenser mic is a Faraday cage, unfortunately in most cases containing a wide open door for RF signals we commonly know as XLR connector. But good enough a Faraday cage to keep mains hum outside the mic. Mic bodies are not connected to earth ground either yet they do work as a Farady cage.

Jan

Open up one of those line filters on a Lindgren screen room filters and you will see what I mentioned.
Hermeticity would relate to being gas tight, like glass to metal seals with no requirement for RF attenuation.

 

[rmburrow]

Yes. They have some, but there are multiple values/etc. I’m wondering what specific one(s) would work.

The pack of ferrite beads I have from Amidon contained some large ones for MW frequencies (i.e. AM band), some for HF (2-30 MHz), and smaller ones for VHF and UHF. My guess is you need the UHF beads in that mic. Put a small piece of "spaghetti" over the lead to the input FET in that mic, install the ferrite bead, and resolder the connection. Stabilize the bead so it won't move.

 

[jp8]

Accidentally, I just received a Gefell M295 for repair having a ferrite bead on the Gate pin. It's not visible in the picture, but it is glued to the pin with a small blob of red glue on the JFET body side of the bead. The classic CMC5 Schoeps circuit also has a ferrite bead on its Gate pin.

Interesting off-topic detail: The nose cone seals on the body through a rubber ring. Perhaps this is very common in the more expensive segment of SDCs, but my reference is mainly the budget models.

Jan

 

[joulupukki]

Accidentally, I just received a Gefell M295 for repair having a ferrite bead on the Gate pin. It's not visible in the picture, but it is glued to the pin with a small blob of red glue on the JFET body side of the bead. The classic CMC5 Schoeps circuit also has a ferrite bead on its Gate pin.

Interesting off-topic detail: The nose cone seals on the body through a rubber ring. Perhaps this is very common in the more expensive segment of SDCs, but my reference is mainly the budget models.

Jan

View attachment 123905

Very interesting. I should have ordered a few random ones that would fit in my mic when I placed my recent order with Mouser. I’ll be able to at least try the Neutrik EMC XLR connector this week. If that doesn’t help, these little ferrites are what I’ll be going after next.

 

[jp8]

Ground at 700MHz+ would require a low inductance conductor, think as wide as it is long. Getting a low contact resistant connection to Terra firma use a couple of square yards of copper plate.
Not a viable option, or useful in any practical situation.
Ground at ~700MHz + is not a useful concept, better look at signal differences from a reference plane, in this case the FET, but anything that acts as a semiconductor junction can rectify the signal into the audio range.
Because you suspect one cause another may be overlooked. Confirmation bias is real, so divide and conquer. Exclude other causes, use another mike, pre amp, power supply, another mike cable etc etc.
Order shielded power cords from Amazon and throw out all old standard non shielded types.

The mike should be a Faraday cage, with a continuous metallic contact around any susceptible point. Any opening should be a "waveguide beyond cutoff". Any metallic weave or screen should be cut from a flat sheet and expanded, then bonded to the mike cavity for continuous peripheral contact.
The mike body should be copper plated, then nickel plated, maybe tin plated for solderability.
Any insulation should not in any way stop the contact.
Add a 2k - 10k resistor in the gate lead, wirewound is OK, max 1mm from FET.
The leads from the XLR to the PCB could be (wirewound) resistors, 50-ish Ohms should be tried, followed by caps max ~100pF to case, with a low inductance connection, like BeCu fingerstock soldered to the PCB then circumferentially connected to the conductive mike tube's inside wall.
The easiest solution is the best, but without a real pinpoint understanding of cause, too many options exist.
RF does not practically behave in any way like 60Hz so methods for this can be put aside.
A test chamber for the mike could be a 6 foot long copper tube, capped at the end, and steel wool crammed around the cable.
Even a test with the
XLR cable alone, then terminated, then on the mike with the signal leads disconnected, then disconnect parts on the PCB until the noise stops.

I see where you come from, and I agree with many of your statements from technical point of view, but I think at least some of your proposals might prove to be overkill or may even introduce other issues. There are plenty of reasonably RF-immune mics around without some of your, imho rather extreme RFI countermeasures included. Let me review your statements and allow me to give my point of view. Admittedly, I'm not an analog RF wizzard and don't want to pretend I'm on the same level wrt RF knowledge and experience as you are, but I do have a background in electronics and 40+ years of hardware design. I apologize in advance if you consider my comments to be Too Long To Read, but some things are not easily explained in a few words (at least, not by me).

"Confirmation bias is real, so divide and conquer. Exclude other causes, use another mike, pre amp, power supply, another mike cable etc etc.": Agree with that. When reading all the experiments OP already and described in this thread and another one, he is definitely open minded and not afraid to experiment.

"The mike should be a Faraday cage, with a continuous metallic contact around any susceptible point. Any opening should be a "waveguide beyond cutoff". Any metallic weave or screen should be cut from a flat sheet and expanded, then bonded to the mike cavity for continuous peripheral contact.": Agree with the Faraday cage statements. As I already said in another post, the weak point is the XLR for its opening and for the almost inevitable inductance between XLR cable shield (for what it's worth as a shield at cell phone frequencies), via XLR pin 1 to the mic body. Whether your metallic weave proposals are practically feasible and attractive...? Perhaps you have a clear view of how to implement this in a practical way while at the same time not being too visually disturbing. A round PCB with ground plane connected to pin 1 and the mic body by as many points as possible, like shown in one of the earlier posts, seems more practical to me, to be honest. But for a DIY already difficult enough to build, especially the connections to the mic body.

"Add a 2k - 10k resistor in the gate lead, wirewound is OK, max 1mm from FET.": I would not do that, at least not above 2k as it would add noise. A ferrite bead having the same RF resistance at the frequencies of interest will not add noise in the audio band.

"The leads from the XLR to the PCB could be (wirewound) resistors, 50-ish Ohms should be tried, followed by caps max ~100pF to case, with a low inductance connection, like BeCu fingerstock soldered to the PCB then circumferentially connected to the conductive mike tube's inside wall.": From technical point of view, I agree. I actually made similar statements here in this and other threads: large caps on the output won't help against UHF cell phone RFI. It's the inductance to the shield/Faraday cage/mic housing which should be kept as low as possible. The question is only how to do that in a practical way. The BeCu fingers you mention are effective and commonly used in the RF arena, but I have yet to find those fingers small enough to be of practical use in an SDC and do not get stuck when inserting the PCBA with XLR into the tube. There are some XLRs having a proper ground contact to the mic tube, but I cannot find them as a commodity part. Takstar CM-63 has one. Where to get those...?

I have just ordered 4 different PCBs with the KM84 circuit, KM84 + CMOS oscillator for capsule bias voltage and two other circuits I don't want to share (yet). They were designed for the Takstar CM-60 and 63. All SMT designs, except for some parts like the elcaps. Will do some experiments with my RF jammer to see what helps and what doesn't. One of the designs has below RF filter with X2Y cap, Common Mode filter and ferrite bead in the ground path. Inspired by the Takstar CM-63 RF filter, which proved to be very effective. But as said, that may be attributed to their good, low inductance pin 1 to mic body connection, rather than the CM filter. Anyway, whether it is effective or not, such a filter cannot be applied to OP's PCBAs.



Jan

 

[ricardo]

There's loadsa good stuff suggested by people here but practically none of them emphasize that EMI/RFI protection is CRITICALLY DEPENDENT on the EXACT PHYSICAL ARRANGEMENT.

So I shall repeat my request ...

I don't understand this pic. Is this the back of the XLR pins on the mike? Can you label p1 p2 p3 & the 'screw terminal tab'? What is the Grey conical bit at the top of the pic?

The circuit you show has no EMI/RFI protection at all. I'll post some stuff for this but first we need to confirm the EXACT way the cable shield gets to

• the body of the mike
• and the circuit.

Also may I clarify that when you put the MP PCB in your DIY housing, there is no RFI from your phone?

If so, that's good news as it means your DIY housing can achieve good RFI immunity.

 

[joulupukki]

There's loadsa good stuff suggested by people here but practically none of them emphasize that EMI/RFI protection is CRITICALLY DEPENDENT on the EXACT PHYSICAL ARRANGEMENT.

So I shall repeat my request ...

Also may I clarify that when you put the MP PCB in your DIY housing, there is no RFI from your phone?

If so, that's good news as it means your DIY housing can achieve good RFI immunity.

You’ve got it backwards. If I put the DIY PCB into the MP body, the cell phone RFI stays exactly the same. That leads me to believe that the problem is not at all with the mic body housing.

I would try what you suggested (MP circuit board into the DIY Takstar body) but the nose cone of the MP build is just barely too large and will not fit.

 

[joulupukki]

Could it be that the 2N3819 FET I’ve got in my builds is more sensitive than the J305 FET that is part of the MP build? My DIY build edges out the MP build by a few decibels in terms of volume that comes out of the mics at the same preamp level. Perhaps this is also a contributing factor to the difference?

 

[Matador]

There's loadsa good stuff suggested by people here but practically none of them emphasize that EMI/RFI protection is CRITICALLY DEPENDENT on the EXACT PHYSICAL ARRANGEMENT.

I appreciate your expertise, as I'm sure you've forgotten more about microphones that I can ever hope to learn in a lifetime.

I'm finding it difficult to grok how a ferrite bead on the gate would help, given the gate is already being driven by an essentially infinite source impedance. Did you ever do anything like that at Calrec, andd was it effective?

 

[ricardo]

I'm finding it difficult to grok how a ferrite bead on the gate would help, given the gate is already being driven by an essentially infinite source impedance. Did you ever do anything like that at Calrec, and was it effective?

We never had to do this on our mikes as far as I can remember but Dip Ing Wuttke did it for some of his Schoeps mikes; I think those that had 'extension' capsules screwed onto the front so there was another long cable entering the mike body.

RFI happens cos the RF is demodulated by the most sensitive part of the circuit, the FET. You don't have to get rid of the RF completely; just reduce it enough so it doesn't get demodulated.

The ferrite bead 'inductor' forms a lossy voltage divider with 'Cin' of the FET.

 

[ricardo]

View attachment 123927

Thanks for this. Everything is clear now.

You’ve got it backwards. If I put the DIY PCB into the MP body, the cell phone RFI stays exactly the same. That leads me to believe that the problem is not at all with the mic body housing.

What I'm looking for is the distance the 'Ground Tab' (actually the cable shield) travels before it 'merges' into the Faraday Cage of the mike body. Millimeters matter here. One cm may cause problems.

The fancy Neutrik cable/plug has a circumferential capacitor so that the shield is 'connected' to the plug body (and hopefully the chassis or mike body) at RF OUTSIDE the Faraday Cage; which is the best way. Any distance inside the Faraday Cage may be an antenna.

 

[jp8]

I'm finding it difficult to grok how a ferrite bead on the gate would help, given the gate is already being driven by an essentially infinite source impedance.

"Infinite" for DC and audio frequencies, that is. But we're talking UHF frequencies hrte, where literlaly every mm of wire or copper track should be regarded as an inductor and every stray pF counts. Looking at a mic schematic, you may only see a JFET, a bunch of resistors, caps and other parts which are actually drawn there and are physically there on the PCBA. An RF guy will see a multitude of parasitic inductors, capacitors, transmission lines and what have you which are at least as important as those parts drawn on the schematic

Those ferrite beads constitute a mainly resistive element at RF frequencies and hence will absorb RF energy, rather than reflect them like inductors do. And they will dampen parasitic LC circuits, e.g. consisting of Gate capacitance and inductance in the whole loop of capsule - JFET capacitance - PCBA ground - XLR pin 1 - mic tube back to the capsule. Maybe the RF current loop is actually shorter due to a large capacitance between ground plane (if it exists) and mic body, but you get the idea.

Jan

 

[Delta Sigma]

"Infinite" for DC and audio frequencies, that is. But we're talking UHF frequencies hrte, where literlaly every mm of wire or copper track should be regarded as an inductor and every stray pF counts. Looking at a mic schematic, you may only see a JFET, a bunch of resistors, caps and other parts which are actually drawn there and are physically there on the PCBA. An RF guy will see a multitude of parasitic inductors, capacitors, transmission lines and what have you which are at least as important as those parts drawn on the schematic

Those ferrite beads constitute a mainly resistive element at RF frequencies and hence will absorb RF energy, rather than reflect them like inductors do. And they will dampen parasitic LC circuits, e.g. consisting of Gate capacitance and inductance in the whole loop of capsule - JFET capacitance - PCBA ground - XLR pin 1 - mic tube back to the capsule. Maybe the RF current loop is actually shorter due to a large capacitance between ground plane (if it exists) and mic body, but you get the idea.

Jan

I remember working on old UHF trunks that used PCB traces as inductors. Just a little swirl.

 

[jp8]

Could it be that the 2N3819 FET I’ve got in my builds is more sensitive than the J305 FET that is part of the MP build? My DIY build edges out the MP build by a few decibels in terms of volume that comes out of the mics at the same preamp level. Perhaps this is also a contributing factor to the difference?

Could be, could be not. In this RF arena, every single detail could be the root cause. The PCB layouts are different, different capacitors, different resistors, different transformer, different anything. Every detail counts in the RF domain, even where you least expect it. That's what is making it difficult to solve. It can take EMC Gurus days to solve EMC issues, which in hindsight look so simple (talking from own experience here when we hired in an EMC expert at work).

But why not just try and swap the J305 for a 2N3819 JFET? You never know...

Jan

 

[Cqwet Dbdfte]

The ferrite bead 'inductor' forms a voltage divider with 'Cin' of the FET.

The ferrite bead increases the inductance of the lead, and forms a second order lowpass filter with the gate capacitance (+Miller effect) You could look at it as an impedance divider. Ferrite is low Q and lossy, a poorly chosen bead may be 50 ohms, at the wrong freq where a resistor would be broadband, if you can tolerate the thermal noise, a 1k resistor should not add much, -124dBv over 20kHz,
and very easy to try.

 

[Cqwet Dbdfte]

I see where you come from, and I agree with many of your statements from technical point of view, but I think at least some of your proposals might prove to be overkill or may even introduce other issues. There are plenty of reasonably RF-immune mics around without some of your, imho rather extreme RFI countermeasures included. Let me review your statements and allow me to give my point of view. Admittedly, I'm not an analog RF wizzard and don't want to pretend I'm on the same level wrt RF knowledge and experience as you are, but I do have a background in electronics and 40+ years of hardware design. I apologize in advance if you consider my comments to be Too Long To Read, but some things are not easily explained in a few words (at least, not by me).

"Confirmation bias is real, so divide and conquer. Exclude other causes, use another mike, pre amp, power supply, another mike cable etc etc.": Agree with that. When reading all the experiments OP already and described in this thread and another one, he is definitely open minded and not afraid to experiment.

"The mike should be a Faraday cage, with a continuous metallic contact around any susceptible point. Any opening should be a "waveguide beyond cutoff". Any metallic weave or screen should be cut from a flat sheet and expanded, then bonded to the mike cavity for continuous peripheral contact.": Agree with the Faraday cage statements. As I already said in another post, the weak point is the XLR for its opening and for the almost inevitable inductance between XLR cable shield (for what it's worth as a shield at cell phone frequencies), via XLR pin 1 to the mic body. Whether your metallic weave proposals are practically feasible and attractive...? Perhaps you have a clear view of how to implement this in a practical way while at the same time not being too visually disturbing. A round PCB with ground plane connected to pin 1 and the mic body by as many points as possible, like shown in one of the earlier posts, seems more practical to me, to be honest. But for a DIY already difficult enough to build, especially the connections to the mic body.

"Add a 2k - 10k resistor in the gate lead, wirewound is OK, max 1mm from FET.": I would not do that, at least not above 2k as it would add noise. A ferrite bead having the same RF resistance at the frequencies of interest will not add noise in the audio band.

"The leads from the XLR to the PCB could be (wirewound) resistors, 50-ish Ohms should be tried, followed by caps max ~100pF to case, with a low inductance connection, like BeCu fingerstock soldered to the PCB then circumferentially connected to the conductive mike tube's inside wall.": From technical point of view, I agree. I actually made similar statements here in this and other threads: large caps on the output won't help against UHF cell phone RFI. It's the inductance to the shield/Faraday cage/mic housing which should be kept as low as possible. The question is only how to do that in a practical way. The BeCu fingers you mention are effective and commonly used in the RF arena, but I have yet to find those fingers small enough to be of practical use in an SDC and do not get stuck when inserting the PCBA with XLR into the tube. There are some XLRs having a proper ground contact to the mic tube, but I cannot find them as a commodity part. Takstar CM-63 has one. Where to get those...?

I have just ordered 4 different PCBs with the KM84 circuit, KM84 + CMOS oscillator for capsule bias voltage and two other circuits I don't want to share (yet). They were designed for the Takstar CM-60 and 63. All SMT designs, except for some parts like the elcaps. Will do some experiments with my RF jammer to see what helps and what doesn't. One of the designs has below RF filter with X2Y cap, Common Mode filter and ferrite bead in the ground path. Inspired by the Takstar CM-63 RF filter, which proved to be very effective. But as said, that may be attributed to their good, low inductance pin 1 to mic body connection, rather than the CM filter. Anyway, whether it is effective or not, such a filter cannot be applied to OP's PCBAs.

View attachment 123915

Jan

As we do not have the full story on everything in this picture, we can only guess. I just threw most basic stuff in there.
Fan grills on some spectrum analyzers use the expanded sheet metal weave, likely bonded to the case. Any "floating" metal can act as resonator/antenna/coupling link. I don't know how much that mesh in the mike is conductive and metal-to-metal bonded to the cavity.
A proper RF measurement would not hard to do with the right equipment.
I'm not convinced that the RF enters thru the cable. If so it should be an easy fix .
Anyway, that PCB is not laid out with RFI in mind.
I design even my non-RF PCB's like RF was everywhere, as this is now more the rule. Copy somebody's 1970's layout and expect to be disappointed.

A round PCB covering the XLR inside the tube, SMT parts can go on both sides, and couple power and signal thru vias. Springs on the perimeter should touch the tube.

Somebody mentioned a switching device for the bias voltage, maybe there is a better idea.

Chasing RFI can be so much fun.

ttps://www.copper-mesh.com/coppermesh/micro-expanded-copper-mesh.html

 

[Cqwet Dbdfte]

I feel bad pushing this recipe on you. I have not had this problem.
Engineers in the orchestra world have told me that km84’s could expose a musician who brought a cell phone onto the stage against union rules. I haven’t run into this issue with one of my builds, but I haven’t explicitly tested for it either.
Thanks for the learning experience. I appreciate that you are trying 1 variable at a time. We’ll figure it out.

I had a 105 open the other day that made an interesting connection to the shell of the mic. View attachment 123568

https://scotttroyer.com/2018/10/fix-neumann-kms-105/

"...Its transformerless output minimizes electromagnetic interference, such as hum...."
Neumann
Looks like a HF (coreless) common mode choke on the connector side.
Strange. Pretty extreme and fragile. Poured ground planes top and bottom. SMD ferrites at input?
The 14-pin chip maybe a DC/DC switched cap multiplier?
Some RF guy's work.
Can't find a schematic.
JFET input parts are potted, or "glob topped". Hiding the "secret sauce"?
Some black epoxy and plastics may contain carbon, and be very lossy for UHF.
Case connection in capsule area very positive.

 

[joulupukki]

I'll have to try tomorrow. Maybe I can tie it onto the ground rod that's grounding my ham shack.

Does wrapping the entire thing out to the cable in tin foil count? If so, a full tinfoil wrap doesn't seem to lessen the RFI at all even if the tinfoil has a solid ground connection into the ground terminal into my wall outlet.