CM Chokes sometimes suck

cently hired an EMI guru. He insists on placing CM chokes on every single e nded I/O signal line. These signals can be floating 100K thermistors that a re received by a Pi filter (100K resistor and two 1.0uF caps to ground). Th e miniture CM chokes are 1k at 100MHz (1.6uH).

g thermistor in a motor winding that could be a double ended input with the CM choke and Pi filter or single ended into the Pi filter. I'm stating tha t the added CM choke adds nothing.

If the CM choke is not used then the other lead goes to ground and the sig nal is "single ended". But you knew that already. Cheers, Harry

Mark, you make a good point but this guru will rain on my parade forever and that s$it must stop. He also gets paid a lot more than me and that PMO!!

hired an EMI guru. He insists on placing CM chokes on every single ended I/O signal line. These signals can be floating 100K thermistors that are received by a Pi filter (100K resistor and two 1.0uF caps to ground). The miniture CM chokes are 1k at 100MHz (1.6uH).

thermistor in a motor winding that could be a double ended input with the CM choke and Pi filter or single ended into the Pi filter. I'm stating that the added CM choke adds nothing.

is "single ended".

If I knew it I wouldn't have asked.

hired an EMI guru. He insists on placing CM chokes on every single ended I/O signal line. These signals can be floating 100K thermistors that are received by a Pi filter (100K resistor and two 1.0uF caps to ground). The miniture CM chokes are 1k at 100MHz (1.6uH).

is

There are return lines, at least in my DA42 (you can Google for the type).

an EMI guru. He insists on placing CM chokes on every single ended I/O signal line. These signals can be floating 100K thermistors that are received by a Pi filter (100K resistor and two 1.0uF caps to ground). The miniture CM chokes are

Maybe I'll be shat upon for this, but IMHO, there's something to be said for not bringing out antennas connected solidly to various points on the internal ground plane(s). If there's a Pi filter as you describe on the "ground" line too, then it doesn't matter.

For something like a thermistor, there's no reason I can think of not to use pennies-a-piece tiny ferrite beads. You only need a CM choke when the current is high enough to saturate an inductor.

an EMI guru. He insists on placing CM chokes on every single ended I/O signal line. These signals can be floating 100K thermistors that are received by a Pi filter (100K resistor and two 1.0uF caps to ground). The miniture CM chokes are

If the pcb ground plane is solidly bolted to the box, and especially if the connector shells are bolted to the box and to the PCB ground plane, bringing in a ground through a connector pin won't cause EMI problems.

A pair of beads, one for each end of the thermistor, isn't a bad idea. EMI rectification errors then to be in the 100+ MHz range, and tend to follow PCB resonances, both broken up by a bead. A common-mode choke only attenuates... common mode RF! But grounding one pin, and RC filtering the other, will usually work fine.

A 100K thermistor in a motor winding does get interesting, especially if there's a high-frequency motor driver.

y hired an EMI guru. He insists on placing CM chokes on every single ended I/O signal line. These signals can be floating 100K thermistors that are re ceived by a Pi filter (100K resistor and two 1.0uF caps to ground). The min iture CM chokes are 1k at 100MHz (1.6uH).

By "single ended" do you mean separate signal and ground conductors, or a signal whose ground return is the airframe or something else besides another wire routed next to the signal line? I understand that you aren't referring to a differential line, but "single ended" by itself doesn't say anything about the ground return.

What is the expected offending signal that the EMI guru is trying to filter out?

How's your 1.0 uF capacitor look at, say, 120 MHz? For that matter, how does your 100K resistor look?

-- john

Le Tue, 21 May 2013 12:30:18 -0700, Harry D a écrit:

Hey, EMI wasn't Master's Voice for no reason. And to justify suckers the high expense, EMI experts count power ratios in Nipper as we all know.

Wow, did RCA sell off Nipper as well?

Cheers

Phil Hobbs

y hired an EMI guru. He insists on placing CM chokes on every single ended I/O signal line. These signals can be floating 100K thermistors that are re ceived by a Pi filter (100K resistor and two 1.0uF caps to ground). The min iture CM chokes are 1k at 100MHz (1.6uH).

I don't get the beads, I am not trying to buy NY city. The thermistor comes into the PCB in a shielded twisted pair. One lead goe s directly to ground, the other thru a Pi filter (220nF, 100K, 220nf) to a Hi Z ADC input. The caps are X7R. The signal BW < 1.0Hz. The noise bandwi dth is >10 MHz but the Pi filter has a BW of

hired an EMI guru. He insists on placing CM chokes on every single ended I/O signal line. These signals can be floating 100K thermistors that are received by a Pi filter (100K resistor and two 1.0uF caps to ground). The miniture CM chokes are 1k at 100MHz (1.6uH).

directly to ground, the other thru a Pi filter (220nF, 100K, 220nf) to a Hi Z ADC input. The caps are X7R. The signal BW < 1.0Hz. The noise bandwidth is >10 MHz but the Pi filter has a BW of Regards, Harry

In the ungrounded lead. As I said, you can probably get away without it.

This is a thermocouple input, and it's heavily lowpass filtered by RCs, and then by software filtering. The tc cable is a shielded pair. Unfortunately, the tc is inside an NMR probe, along with high field levels of RF in the 50-500 MHz range. The RF did (always will) get into the signal conditioning opamps and get rectified to small offsets, in a system where 1 microvolt matters. I added this bead

and it helped by something like 30 dB, and especially broke up narrowband resonances.

I got this business by doing this better than their previous vendor. An RF generator could shut down his controller from clear across the room. I've sold over 3000 controllers so far, not bad for a "stinkin bead."

Le Wed, 22 May 2013 09:55:25 -0400, Phil Hobbs a écrit:

I only knew Nipper as La Voix De Son Maitre (His Master's Voice), EMI, JVC and Deutsche Grammophon, which already is an all mess.

The actual situation seems to be even worse...

All (ahem) the story is here

and there

hired an EMI guru. He insists on placing CM chokes on every single ended I/O signal line. These signals can be floating 100K thermistors that are received by a Pi filter (100K resistor and two 1.0uF caps to ground). The miniture CM chokes are 1k at 100MHz (1.6uH).

directly to ground, the other thru a Pi filter (220nF, 100K, 220nf) to a Hi Z ADC input. The caps are X7R. The signal BW < 1.0Hz. The noise bandwidth is >10 MHz but the Pi filter has a BW of > Regards, Harry

then

is

range.

this

In such unwanted rectification scenarios there is also another option: Swap the opamp against a CMOS type if one is available that has otherwise acceptable specs. Those do not have BE junctions behind the input pins that could rectify.

In one case (across the Bay from you guys) that measure alone killed the noise dead, as John Wayne would have put it. It was GSM cell phones getting in there, something where regular beads aren't very effective anymore unless you use the special Murata thingamagics, and those need super-direct ground contact plus shield in order to work.

ntly hired an EMI guru. He insists on placing CM chokes on every single end ed I/O signal line. These signals can be floating 100K thermistors that are received by a Pi filter (100K resistor and two 1.0uF caps to ground). The miniture CM chokes are 1k at 100MHz (1.6uH).

goes directly to ground, the other thru a Pi filter (220nF,  100K,  220 nf) to a Hi Z ADC input. The caps are X7R. The signal BW < 1.0Hz. The noise bandwidth is >10 MHz but the Pi filter has a BW of > Regards, Harry

nd then

e tc is

range.

d this

sold

"Good bead" :^) Say if you passed one of the TC leads through it in the other direction would that make it a one turn CM choke? (Just trying to expand my transformer understanding.)

George H.

.highlandtechnology.com  jlarkin at highlandtechnology dot com

Yes, very effective indeed.

John

They do tend to substitute their own noise for the external noise..

ntly hired an EMI guru. He insists on placing CM chokes on every single end ed I/O signal line. These signals can be floating 100K thermistors that are received by a Pi filter (100K resistor and two 1.0uF caps to ground). The miniture CM chokes are 1k at 100MHz (1.6uH).

goes directly to ground, the other thru a Pi filter (220nF, 100K, 220nf) to a Hi Z ADC input. The caps are X7R. The signal BW < 1.0Hz. The noise ban dwidth is >10 MHz but the Pi filter has a BW of

nd then

e tc is

range.

d this

PG

sold

John, your TC signal came into a diff-amp that rectifies at >1.0MHz. My sin gle ended signal, after filtering, comes into a ADC with a bandwidth

hired an EMI guru. He insists on placing CM chokes on every single ended I/O signal line. These signals can be floating 100K thermistors that are received by a Pi filter (100K resistor and two 1.0uF caps to ground). The miniture CM chokes are 1k at 100MHz (1.6uH).

directly to ground, the other thru a Pi filter (220nF, 100K, 220nf) to a Hi Z ADC input. The caps are X7R. The signal BW < 1.0Hz. The noise bandwidth is >10 MHz but the Pi filter has a BW of >> Regards, Harry

then

is

range.

this

sold

Or else a couple of my nice Russian feedthroughs. ;)

Cheers

Phil Hobbs