RFI protection concepts

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
KMN
Posts: 53
Joined: Sat Jan 20, 2018 2:49 pm

Re: RFI protection concepts

Post by KMN »

JR. wrote: Thu Jan 25, 2018 9:46 am Not sure I understand your question, but I would be tempted to put the input caps inside (after) the inductors. Perhaps a single cap outside from + to -. If this is a phono input don't over do the C loading.
Thanks for your tips. And I emphasize an apology for my poor communications skills.

My main question is about the L2 bead. Note it's slightly different location in the two diagrams posted by Gertjan. Is one implementation superior to the other?

In my implementation, the input signal path is first through the beads then to the cap, at the connector/chassis wall. The output path is similar in that the first filter component encountered is the bead, followed by the cap also both at the connector/chassis wall but bead and cap are reversed order with respect to RF energy coming into the box from the i/o cable. This is all per Gertjans diagram.

My understanding from reading this thread is that from an RF standpoint, ie RF getting into the box from the outside on the various io port cables, seeing the bead first then the cap is ideal. But in the case of the output driver, it may not like that configuration, ie having a cap before the bead at it's output and so the bead impedance gets inserted between the output and cap. Hopefully I haven't misunderstood...to keep the output driver happy the bead, cap order gets reversed even though this may not be the best RF rejection.
JR. wrote: Thu Jan 25, 2018 9:46 am the RF caps are generally of little concern to the audio path. If it makes you feel better you can use COG/NPO ceramic caps.
OK that is what I was hoping. Maybe I can run an experiment later to see if I hear anything.
mediatechnology wrote: Thu Jan 25, 2018 11:31 am I wouldn't put capacitors to ground and imbalance the instrumentation amp on the phono input.
You'll just make CMR worse.
I wouldn't over-think this.
Are you having an RF problem and if so what is it?

My instincts tell me that we make a huge mistake when we apply the same RFI/EMI solutions that apply to low impedance mic and line outputs to a balanced phono input.

1) The input already has differential termination capacitance because the cart requires termination.
2) Any common mode capacitances you add is in parallel with the cart's differential termination.
3) Any capacitive imbalance to ground is going to wreck CMR because the source impedance is high and inductive.
4) There's series resistors available to form an RC filter to prevent rectification of CM signals.

I can't argue with putting ferrite beads in series with each input but I would avoid putting CM caps on a balanced phono input.
Just don't do it.
Even better. No caps at all just beads on the input. I'll try it.

Honestly I think I'm just freaking out and being overly analytical about it since this is my first ever solder to audio hardware. I simulate lots but haven't ever listened to anything I've put together. I know I can cause problems in my home audio equipment if I try...key up my RF amplified am signal too close to any of the single ended stuff and it sounds like aliens are coming down. I've also noticed cell phone activity but haven't had that particular type of noise in a while.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: RFI protection concepts

Post by mediatechnology »

Honestly I think I'm just freaking out and being overly analytical
That's why you came here to us. ;)
about it since this is my first ever solder to audio hardware
No way! Your build looks like you're pretty experienced. Better than my stuff.

I edited my post so you might want to go back and have a look-see...

I would shield those input lines from the 5P XLR to the board with real STP cable.
There's more bang-for-the-buck there than anything else....
I've actually found that the best HF CMR performance is obtained when C1/C2 are used for termination, R4/R5 are installed and C3 is linked open.
This is totally counter-intuitive, based on measurement and something I want to research. I've seen this phenomena with both CFB mic preamps and op-amp based INAs.
KMN
Posts: 53
Joined: Sat Jan 20, 2018 2:49 pm

Re: RFI protection concepts

Post by KMN »

Very much obliged for everyone's opinions.

I already upgraded to the STP input lines! I meditate on this forums posts when I have free time. I would note that C1/C2 are indeed some nice, high quality film caps.

I will be taking the advice given here.
mediatechnology wrote: Thu Jan 25, 2018 11:47 am No way! Your build looks like you're pretty experienced. Better than my stuff.
Haha, thanks. Maybe it's an optical illusion. I am soldering on the prototypes I make down at work (renewable energy, RV, inverters, charge controllers, etc.), all the time, when they don't end up working perfectly on the fist try (read always). Nothing production quality, though. Definitely self taught and the results are rarely pretty.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: RFI protection concepts

Post by mediatechnology »

I used a 9 pin D connector for my low-level cart connections.

I found out right away how sensitive to electrostatic fields it was when I installed - out of necessity - a plastic backshell.
As a temporary measure I used aluminum foil to line the inside of the shell and brought the foil out so it grounded to the connector body.
I used it that way for months.
That 1" of exposed cable was a large source of ingress.

Ultimately I bought a nickle-plated backshell.

I can't stress to everyone how important it is to use shielded cables internally.
montemcguire
Posts: 11
Joined: Fri Mar 30, 2018 9:26 pm
Location: Watertown, MA USA

Re: RFI protection concepts

Post by montemcguire »

A great thread! I’d like to add that there are some specialized MLCCs that are designed for specific RF uses, such as base station amplifiers, which have impedance nulls placed near commonly used transmitter frequencies. The point of this is probably to reduce device dissipation in transmitter circuits, but these properties also make them ideal for RFI filters.

For example, a generic 0603 100pF cap like the Murata GRM1885C2A101JA01 will have a fundamental resonance around 650MHz, not a particularly important frequency. However, the Murata GQM1875C2E101JB12 has been sized to have a fundamental null at 840MHz, as well as a secondary null at 1.97GHz.

These two nulls are close to the GSM band frequencies, as well as the DECT cordless phone frequencies. Because of this, when the part is used as a shunt filter capacitor, connected directly from an input terminal to chassis, these nulls can provide attenuation that is often significantly greater than what the bulk capacitance of the part could provide, and can also provide meaningful attenuation above the cap’s fundamental self resonant frequency.

By using these specially designed capacitors, an RFI filter can thus have higher attenuation at important frequencies without the gotcha of trying to increase attenuation by increasing the bulk capacitance, which usually makes the part electrically larger and shifts its self-resonant frequency lower, placing the interfering frequencies deeper into the inductive portion of the cap’s impedance curve.

I’ve picked the Murata parts for examples here because I use SPICE extensively, do not have a VNA, and they publish detailed models for a number of their offerings. Other vendors may also produce specialized engineered resonance RF capacitors like this, but I haven’t been able to find them yet - please let me know if you’ve found them!

You can get the Murata models from their SimSurfing web site. It allows you to look at some detailed graphs of impedance and such, as well as download some pretty detailed SPICE models. Here’s the link: http://ds.murata.co.jp/simsurfing/mlcc.html?lcid=en-us

The schematics in this thread show a 220pF input or output filter cap, but Murata does not have models for (or seem to actually sell) a 220pF cap designed for these types of RF uses. However, they do specify a few 100pF parts that, because of their engineered resonances, may perform better than a generic 220pF MLCC. The GQM1875C2E101JB12 seems to be the ideal part for this use, but the GQM2195C2E101JB12 and GQM22M5C2H101JB01 also have detailed SPICE models, and show a slightly different tuning of resonant frequencies, which seem a bit low for cell phone filtering.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: RFI protection concepts

Post by mediatechnology »

montemcguire wrote: Sun Apr 01, 2018 2:37 am For example, a generic 0603 100pF cap like the Murata GRM1885C2A101JA01 will have a fundamental resonance around 650MHz, not a particularly important frequency. However, the Murata GQM1875C2E101JB12 has been sized to have a fundamental null at 840MHz, as well as a secondary null at 1.97GHz.

These two nulls are close to the GSM band frequencies, as well as the DECT cordless phone frequencies. Because of this, when the part is used as a shunt filter capacitor, connected directly from an input terminal to chassis, these nulls can provide attenuation that is often significantly greater than what the bulk capacitance of the part could provide, and can also provide meaningful attenuation above the cap’s fundamental self resonant frequency.

By using these specially designed capacitors, an RFI filter can thus have higher attenuation at important frequencies without the gotcha of trying to increase attenuation by increasing the bulk capacitance, which usually makes the part electrically larger and shifts its self-resonant frequency lower, placing the interfering frequencies deeper into the inductive portion of the cap’s impedance curve.
Thank you for visiting and joining us montemcguire! Thanks also for posting this info.

DECT 6.0 interference is a real problem for the LME49710 MEE49720: viewtopic.php?f=6&t=939
montemcguire
Posts: 11
Joined: Fri Mar 30, 2018 9:26 pm
Location: Watertown, MA USA

Re: RFI protection concepts

Post by montemcguire »

Honored to join the group, and a hearty welcome to all here!

Since I'm using the LM4562 heavily for my current project, that thread about DECT rectification certainly caught my eye. I've had good luck avoiding RFI problems using the LM4562, but maybe it's been through pretty meaty 4 layer PCBs that leave most of the copper on the board, arranged as low impedance ground pours and power planes, the use of these special caps, and maybe also some good 'shielding luck'. I need to get a DECT rig just to test this - any pointer on the model that you used that causes mishaps? I'm working on a high end monitor controller, and the last thing anyone needs to hear is RFI and wonder whether it's "just in the monitors" --> fail. My cellphone won't tickle this problem, so maybe I'm just lucky - PM me with that cordless model!

Regardless, this is one of the few cases where SPICE component models can show you detailed RF performance that is normally not modeled. Most SPICE models don't address GHz range behavior, and blending real RF tools and RF specified components (S parameters and such) with LF oriented SPICE tools is not that simple either. I guess the answer is to get a VNA at some point, but it sure is a lot easier (and quicker and cheaper) to do a simulation and/or a lot of math than to spin up a nice PCB and test it, even with a nice network analyzer. In the end, a prototype is always needed, but I need to have that done only after all of the simulation or calculation details have been proven. Nobody can afford the cycle time of a board spin, fabrication, build, and testing, especially compared to math and simulations.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: RFI protection concepts

Post by mediatechnology »

Welcome aboard!

It's a Panasonic KX-TG9331T base station.

My cell phone has to be right on top of it and active before I see anything so I'm not surprised you can't tickle it.
Dittos for a WiFi access point. It's DECT thing.

For the DIP LME49720 I found that if it were completely enclosed in a metal box they were OK.
When in an open environment with a metal plate and ground plane underneath I found that the ingress seems to be into the actual package.
A small piece of grounded copper foil stuck to the top of the DIP seemed to quiet it down a lot.
It doesn't seem to be conducted via the device pins but radiated into the top of the die.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: RFI protection concepts

Post by JR. »

Sounds like maybe they need to bring back metal can packages.

JR
Cancel the "cancel culture", do not support mob hatred.
Post Reply