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

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
YMMV, but contrary to popular belief, I'm afraid those 22nF cercaps on the outputs just will not work to suppress UHF interference from cell phones. Not even when mounted directly on the XLR as I will show in the pictures below. ANY wiring from the caps to the chassis, to pin 1 or to whatever ground will be too long to make the caps effective at those frequencies. You'll need a circumferential ground connection from the XLR cable shield to the mic body, like the Neutrik EMC plugs have or an RFI filter like the Takstar CM-63 has.
Did you have the inductors/ferrite beads from the XLR pins/CERAMICs connection to the PCB too?
=====================
The "Calrec CB Radio" was a car ignition system driven by an AC motor with a 1m rod stuck on the spark plug (all Tesla inventions BTW) I could only run it at night cos all the meters in the factory (including the passive multimeters so I'm told) would go blip blip blip. It was an all-wave transmitter.

The BBC sent us a VERY bent Calrec stick mike with the comment that it seemed rather susceptible to RFI. I opened it up and found the PCB earth tracks were (not surprisingly) cracked. As it was Saturday, I could fire up the Calrec CB Radio and played around with Ceramics & earth connections.

I sent it back to them with a note saying we had done a full investigation and quoted an exorbitant price (about half the price of a new mike) for 10min work and 2 ceramics to modify further mikes.

Next thing, a small crate arrives by courier, the entire inventory of BBC Calrec mikes in the North of England. Then an irate phone call asking why we "hadn't done them yet and they had to be back by Fri cos they were the only mikes of any brand that didn't buzz in our new TV studio"

Schoeps mikes have a good reputation for RFI immunity but my Calrecs were better cos I cheated. Alas my cheat isn't applicable to this case.
 
Quick early report:

I've modded both of my XLR cables with the Neutrik NC3FXX-EMC connectors. I can easily use these mics in the studio w/o worrying if someone has a cell phone in their pocket. If I hold my phone right on top of the mics, there's still some interference but almost none in comparison to what they were before. The MP mic with this Neutrik EMC cable still outperforms the DIY in RFI rejection, though RF is also barely there if my cell phone is right next to it.

I ordered a few small SMD ferrite beads thinking I could somehow try that in between the binding post and the FET gate lead. Uh ... they are so small that I think if I even blink they will disappear! So I don't think I'll be trying that at all.

With my Mouser order, I purchased a couple of axial 47uH inductors and I'm wondering if I were to pair those with a couple or 330pF ceramic disc caps (like is done in the U87) if it'd have a similar effect as the Neutrik EMC connector? I'd install those inductors and caps carefully inside the mic body at the XLR pins somehow. It'll be tight and tricky.

1709853449692.png
 
Last edited:
I would say this latest experiment pushed me way over the "hokey" bridge. I implemented the inductor-capacitor shown above being extremely careful that nothing is shorted out. The level of RF with it is not fixed at all (with a non-Neutrik EMC XLR connector). Even if the leads of these components are "too long" I would have expected it to do something in the positive direction. So at least I can put that to rest and go back to my less hokey XLR pin connections.

Pictures included just for amusement:
IMG_4091.jpeg
IMG_4092.jpeg
 
I've modded both of my XLR cables with the Neutrik NC3FXX-EMC connectors. I can easily use these mics in the studio w/o worrying if someone has a cell phone in their pocket. If I hold my phone right on top of the mics, there's still some interference but almost none in comparison to what they were before.
:) Low levels of RFI sometimes give a sense of unease so it's worth doing it right. There's reasons why the Schoeps are so well regarded besides flat response, lo THD bla bla.

The Neutrik connectors deal with RFI on the shield but not on the signal lines.
I purchased a couple of axial 47uH inductors and I'm wondering if I were to pair those with a couple or 330pF ceramic disc caps (like is done in the U87) if it'd have a similar effect as the Neutrik EMC connector? I'd install those inductors and caps carefully inside the mic body at the XLR pins somehow. It'll be tight and tricky.

View attachment 124101
The ceramics need to be on the XLR side of the inductors. I was amused when I found out Dip Ing Wuttke of Schoeps used the same values & arrangements as I did.

Long capacitor leads may be just as bad or worse as they act like aerials.

With the caps on the PCB side, the RFI goes through the inductors and then through the ceramics to earth.
Depending on their construction, the inductors may be transmitting aerials for the RFI

The CERAMICs also need to be AS SMALL AS POSSIBLE with SHORT LEADS. Use cheapo X7R ceramics 10n - 22n cos they are usually the smallest.

Have you dressed & twisted the transformer wires?

Don't assume a big name German is always good. Some Neumanns are among the worst mikes for RFI
 
Last edited:
:) Low levels of RFI sometimes give a sense of unease so it's worth doing it right. There's reasons why the Schoeps are so well regarded besides flat response, lo THD bla bla.

The Neutrik connectors deal with RFI on the shield but not on the signal lines.

The ceramics need to be on the XLR side of the inductors. I was amused when I found out Dip Ing Wuttke of Schoeps used the same values & arrangements as I did.

Long capacitor leads may be just as bad or worse as they act like aerials.

With the caps on the PCB side, the RFI goes through the inductors and then through the ceramics to earth.
Depending on their construction, the inductors may be transmitting aerials for the RFI

The CERAMICs also need to be AS SMALL AS POSSIBLE with SHORT LEADS. Use cheapo X7R ceramics 10n - 22n cos they are usually the smallest.

Have you dressed & twisted the transformer wires?

Don't assume a big name German is always good. Some Neumanns are among the worst mikes for RFI
Lots of info here. Just trying to do one change at a time, checking the results each time. I have tested twisted transformer wires. It did barely improve things. I’m sure there must be some optimal lead dress on those. I don’t know what it’d be. Once I cut them shorter there isn’t much wiggle room.

I think my next attempt of tests is gonna be to track down ferrite beads that I could slide over the gate input of the FET. …and/or try out a J305.
 
I would say this latest experiment pushed me way over the "hokey" bridge. I implemented the inductor-capacitor shown above being extremely careful that nothing is shorted out. The level of RF with it is not fixed at all (with a non-Neutrik EMC XLR connector). Even if the leads of these components are "too long" I would have expected it to do something in the positive direction. So at least I can put that to rest and go back to my less hokey XLR pin connections.

Pictures included just for amusement:
View attachment 124102
View attachment 124103
 
At high UHF frequencies the parasitic properties will dominate, making leaded parts useless. If you cannot get the self resonant frequency of a certain part, (or measure it) itsa a waste of time.
Murata, Johanson has that info easily available.
For bypass caps, SMT 0402, maybe 0604, size will work, 68pF has SFR of around 800MHz, lower values less, traces should be short, or none, with poured ground planes with a lot of vias. The layout is as important as the part values.
Not a typical DIY job, but possible. Until these requirements are satisfied turn off the cellphone.
 
At high UHF frequencies the parasitic properties will dominate, making leaded parts useless. If you cannot get the self resonant frequency of a certain part, (or measure it) itsa a waste of time.
Murata, Johanson has that info easily available.
For bypass caps, SMT 0402, maybe 0604, size will work, 68pF has SFR of around 800MHz, lower values less, traces should be short, or none, with poured ground planes with a lot of vias. The layout is as important as the part values.
Not a typical DIY job, but possible. Until these requirements are satisfied turn off the cellphone.
I'll just use the cables with the Neutrik EMC connectors. Shouldn't have to worry anymore about odd cell noises with those. I think a better board layout would work, but since I've already got four of these style of mics now, that's as much as I need.
 
Hopefully those filtered connectors work with your RFI challenge, if not those PCB's can be left in a drawer.
 
Did you have the inductors/ferrite beads from the XLR pins/CERAMICs connection to the PCB too?
=====================
The "Calrec CB Radio" was a car ignition system driven by an AC motor with a 1m rod stuck on the spark plug (all Tesla inventions BTW) I could only run it at night cos all the meters in the factory (including the passive multimeters so I'm told) would go blip blip blip. It was an all-wave transmitter.

The BBC sent us a VERY bent Calrec stick mike with the comment that it seemed rather susceptible to RFI. I opened it up and found the PCB earth tracks were (not surprisingly) cracked. As it was Saturday, I could fire up the Calrec CB Radio and played around with Ceramics & earth connections.

I sent it back to them with a note saying we had done a full investigation and quoted an exorbitant price (about half the price of a new mike) for 10min work and 2 ceramics to modify further mikes.

Next thing, a small crate arrives by courier, the entire inventory of BBC Calrec mikes in the North of England. Then an irate phone call asking why we "hadn't done them yet and they had to be back by Fri cos they were the only mikes of any brand that didn't buzz in our new TV studio"

Schoeps mikes have a good reputation for RFI immunity but my Calrecs were better cos I cheated. Alas my cheat isn't applicable to this case.
No beads included in the ground path. My sole purpose of the test was to see if having the ceramics on the PCB or on the XLR would make a difference as that is what many people think they do. Adding a bead would help, I guess. That's why you see it included in the RF filter schematic I shared in one of my previous posts. Btw, this ground bead is also in the Neutrik EMC plugs to prevent RF CM currents flowing through your PCB. So it's not only relying on the circumferential connection of the cable shield to the XLR connector housing.

If you look at the Schoeps CMC5 schematic, you'll immediately see that they took EMC measures seriously: a bead in the JFET gate (Dr5 = Drossel 5 = Inductor 5). And it has the LC filter on the XLR output. But those parts are wired components not suitable to suppress UHF frequencies as cqwet-dbdfte already pointed out. So whether it will hold up against cell phone RFI...?

Jan
 
No beads included in the ground path. My sole purpose of the test was to see if having the ceramics on the PCB or on the XLR would make a difference as that is what many people think they do. Adding a bead would help, I guess. That's why you see it included in the RF filter schematic I shared in one of my previous posts. Btw, this ground bead is also in the Neutrik EMC plugs to prevent RF CM currents flowing through your PCB. So it's not only relying on the circumferential connection of the cable shield to the XLR connector housing.

If you look at the Schoeps CMC5 schematic, you'll immediately see that they took EMC measures seriously: a bead in the JFET gate (Dr5 = Drossel 5 = Inductor 5). And it has the LC filter on the XLR output. But those parts are wired components not suitable to suppress UHF frequencies as cqwet-dbdfte already pointed out. So whether it will hold up against cell phone RFI...?

Jan
Refresh my memory Jan … in your prototype PCB where you’ve made room for the Takstar CM-60 rolloff switch, you incorporated the LC filter on the XLR output, right? If that ends up being effective, I’ll be curious to know what values you used.

If I were to order a bead that’d slide over the JFET gate in my build, which one should I use? I’m still a bit confused how the ferrite values and impedance works. On the impedance graphs should I be looking for one where the values are high for the frequencies I want to attenuate?
 
Yes, a higher impedance bead for the relevant frequencies is better, forms an LC attenuator with the gate capacitance. These could be used universally on any audio FET gate.
Those snap on ferrite clumps you see sometimes may only gave 50 ohm Z at 100MHz, likely some dudes bandaid to get thru a required EMC test.
 
https://www.mouser.com/ProductDetail/TDK/MPZ1608D101B?qs=chjFIDm9dbrhujE6gAH3rg==

Rather than bottom fishing for a low probability solution start at the top with the part most likely to work. The current is extremely low so core saturation won't be an issue.
To install a FB in trace, take an xacto knife, scrape off insulation a litte longer than the part, make a cut in the trace, tin the surface, add the part.
If no trace, but a post, an 0603 part could be used, one side on the post, the other to the FET, more finnicky but doable. A new PCB would be easier.
Forget those 1950's parts.
 
Last edited:
https://www.mouser.com/ProductDetail/TDK/MPZ1608D101B?qs=chjFIDm9dbrhujE6gAH3rg==

Rather than bottom fishing for a low probability solution start at the top with the part most likely to work. The current is extremely low so core saturation won't be an issue.
To install a FB in trace, take an xacto knife, scrape off insulation a litte longer than the part, make a cut in the trace, tin the surface, add the part.
Forget those 1950's parts.
What trace are you thinking, specifically. I have a couple of those around that value, but they are so small that if I even think about touching them, they disappear. ;)
 
Working with small components is hard without lenses and tweezers, thin long soldering tip, good light, steady hands, or have it made in China by GoGo, JLC etc. .
Not sure about that PCB layout, if the gate lead is in a trace it can dissected.
The power traces may be easier. From other mike designs its evident that the connections from the XLR is getting serious filtering, both from conducted RFI and also radiated.
If indeed the damage is done from the XLR end, the gate can be left alone and efforts focused on signal, power and ground filtering, where additionally there is no penalty for capacitive bypassing to "ground".
 

Latest posts

Back
Top