RF Sniffer Probes- what common mode core?

[Conrad Hoffman]

I have several articles on how to build RF sniffer probes. The better ones use a small toroid common mode transformer/filter between the coil and BNC. The coil is typically a 1-turn split shield thing made from rigid coax or similar. What they never seem to tell me is what kind of ferrite to use for the common mode choke, or even how many turns. Warning- I'm not an RF guy, so maybe the answer is obvious, but not to me. I have a good assortment of ferrites from Amidon. What should I use and how many turns? These are typically wide band probes for general EMI work.

[T3sl4co1l]

The basic structure works at HF, up to whatever mode limit it has (i.e., the appearance of non-TEM00 modes, causing frequency response to break up).  Plain (air cored), the bandwidth is usually pretty small, mind -- comparable to the electrical length of the structure, more or less.

The ferrite loading is used to extend the lower limit downward, and perhaps to some extent its losses act to dampen those higher modes, so that while performance at the top end might be poor, it might not be as catastrophic as the whole dips and peaks of an accidentally-resonant structure.

If you only need some to 10s of MHz, a #43 or other high-mu NiZn would be fine.  If you want minimum cutoff, you need maximum mu, typically a MnZn EMI filter / pulse transformer type with mu >= 10k; or amorphous or nanocrystalline with mu > 100k which can bring response all the way down to some kHz.

I don't think there's much value in layering or stacking different types (e.g. flanking a #76 (Fair-Rite designation, MnZn, mu = 10k) with some #43 (NiZn, mu = 800) or even #61 (NiZn, mu = 125)), but that may afford lower/flatter insertion loss, or characteristic impedance, at very high frequencies (as the high-mu types tend to become rather lossy, and perhaps even capacitive (lowering Zo), above their cutoff frequency).

Tim

[thm_w]

Can you link to the article so we can see what you are talking about?

[Conrad Hoffman]

This article doesn't use the common mode choke but gives you the general idea. https://www.edn.com/06-04-98-sniffer-probe-locates-sources-of-em/  There's a good image right here- https://www.eevblog.com/forum/projects/diy-magentic-field-probes/ Also https://emcesd.com/pdf/emc99-w.pdf The better articles are from very old issues of Conformity magazine but are not available online. I did find one that called out the choke and it's equivalent to #75 material with about 7 turns bifilar wound, so I'm good to go! If it's of interest I can certainly post some pictures when I have it done.

[T3sl4co1l]

Those probe types depend on magnetic field from another (non contact, not looped-around) conductor; a core has little effect (e.g. you could put a rod through the solenoid type), or isn't possible anyway (magnetic loop).

Tim

[Conrad Hoffman]

Sorry the linked probes don't show what I'm talking about. The core has nothing to do with the loop itself. It's a common mode filter connected between the loop wires and the BNC. Just went through my box of ferrites and, as usual, I don't have exactly what I want. Need to look at the data sheets for all the materials I do have to see if there's something suitable.

[T3sl4co1l]

Ah. Then most anything will do, preferably with thick cross section so that the inductance/turn^2 is large, so that few turns are needed.

Tim

[Conrad Hoffman]

Testing my probe & seem to be on the right track. I notice that some probes use a 50 ohm resistor in series and others omit it. I can see trying to match the cable impedance, but does it really accomplish anything other than adding the noise of a 50 ohm resistor to the circuit? Cable reflections?

[coppercone2]

I think that EMI sniffer document tells you about the resistor (the one with the tiny coil inside of a brass tube)

[T3sl4co1l]

Which one?

Loss is needed somewhere, to terminate the line.  At least one of source or load is required; both is acceptable, but you must avoid having none at all.  Note the loop/coil by itself is a near total reflection, at whatever phase shift is caused by its reactance.  It doesn't couple strongly back into the EUT, or space; it's a rather poor antenna.

You don't need the source termination resistor (in series with the coil, or parallel or both for that matter, for f > Zo / (2 pi L)), but if you omit it, you must always have a 50 ohm terminated load.  Reading with a 1M scope (without using a tee and terminator, or internal terminator / 50 ohm input mode) will give erroneous results.  Specan should always have input resistance so this shouldn't matter much in that case.  Thus, the source termination is cheap insurance.

Tim

[Conrad Hoffman]

Good explanation, thanks! It clears up my thinking a bit. Can you tell I'm not an RF guy? OTOH, that never stopped me from playing with the stuff.

[JohnG]

I know others have weighed in here, but I'll add my $0.02. The core acts as a CM impedance to prevent E-field (capacitive) pickup. For the case of CM E-field, the shield and conductor more or less act two conductors connected in parallel, so it acts kind of like a single wire. At the instrument end, they don't see the same impedance to ground, so some of the CM current in the cable gets converted to differential mode and is measured. The core raises only the CM impedance and thus reduces CM current in the cable.

As a result, your best bet is a nice, lossy ferrite intended for EMI purposes. Sometimes you can scavenge a good one from an old wall wart. Lossy is good here, because the cable makes a big long loop from the DUT to the instruments, to ground and some set of path(s) back to the DUT. This can ring pretty badly if the path includes capacitance (say, the capacitance of a power supply transformer). Probably not too much of an issue with a field probe, but better safe than sorry.

John

[T3sl4co1l]

Note that the equivalent circuit is an impedance divider:

The source is E-field around the probed trace.  Likely quite a high impedance (an equivalent capacitance or whatever).
This drops into the probe's internal impedance (coil reactance, TL resistance, whichever).  The drop across this element is what's sensed as CM-DM conversion.
Finally, the CM cable impedance (via shield exterior) sinks the CM current to ground or surroundings.

So it's an equivalent of three series impedances, the voltage drop of the middle one being what's measured by the instrument.

We can add ferrite beads to the cable to increase that last impedance, but notice this will only appreciably reduce the current injected by the source, when its impedance is comparable to the source impedance.  So, if that's some kohms say (10kohm is a whopping 0.16pF at 100MHz -- probably typical for a probe right up on a trace), then we need at least as much ferrite bead impedance to reduce the error.  Which is far from practical by ferrite bead alone.

Which is why these things are usually shielded, and balanced to whatever degree they can be (given that the probe's shield/housing will also be ground for convenience).

Honestly, I don't even mind the E-field sensitivity of the stupid little loop I made years ago:



(Notice it's two turns bare unshielded, one side grounded, so it's more directly susceptible than most proper designs.)  I find it's not hard to separate E- and H-oriented components from the waveform, and can use the E-component to follow signals perhaps better than a well shielded one would.  Downside, it's harder to get clean waveforms from, of course.  But it's not like it's calibrated, it's only ever qualitative in use, and for that, just looking at the waveforms, relative amplitude, ringing, spikes, all of that, is perfectly usable.  YMMV.

Tim

[JohnG]

For a lot of stuff you can get away with having the coil unshielded, but I do a lot of work near parts with high dv/dt, and I am trying to estimate currents by integrating the waveform (not very well, but I don't have a lot of options). In this environment, the shielding is pretty important.

The coil shielding doesn't really help with common-mode, though, and that's where the CM cores become helpful.

John