Practical limits of z0 probing

[law]

Hey all,

I'm probing the high and low side FETs of a half bridge on a low voltage 3 phase inverter.
I'm looking at the gates and the drop across the D-S of the FETs themselves.
I currently use a Micsig DP1502 differential probe for the high side gate and for high side D-S and I'm just using some Rigol 10:1 passive probes for the low side. (probing setup picture included)
My oscilloscope has 50R inputs.
I'm interested in experimenting with Z0 probing, as I have not used it before.
For probing the low side, I did some quick numbers to see what outputs to expect. Gate voltage is 12V and the Vds across the low side FET will be around 30V. High side gate will then sit at around 42V.
For the lower ratio probes (10:1) I see around 200mW across the tip resistor. Obviously the power dissipation goes down as I go up to 20:1 (1K+50R) and 100:1 (5K+50R).
For the 30V across the D-S of the low side FET, 10:1 sees about 1.5W! I would have to go higher in attenuation.
Single ended probing of the high side with Z0 seems like I would have to push the attenuation up even more. For a maximum 48V, a 200:1 (10K+50R) dissipates about 230mW. This seems OK?
Is there a practical attenuation limit? I see Keysight, etc have a 100:1 probe. They have a maximum voltage of 21V.
Any suggestions on this?

[law]

I probably should have said - I'm mainly interested in looking at the rise and fall of the FET switching and catching any ringing. 500Mhz - 600Mhz bandwidth is probably ok.

[electron_plumber]

The "z0 probe" as you called it will provide less capacitive loading than the (~10pF) 10:1 passive probes, but more resistive loading. For a given series resistor, those resistive and capacitive impedance asymptotes will intersect at some frequency, above which the "z0" probe will load the circuit under test less than the capacitance of the 10:1 probe. E.g., 10pF versus 1kOhm

You're trying to capture high frequency LC resonance structures (e.g. in your the gate drive loops), so loading the gate at all is going to change the q-factor and modify the response.

Plug it into ltspice and see for yourself .

[jbb]

Probing the low side gates should be possible, but probing high side gate drivers is really hard. This is because the high side gate is floating on top of the switch node voltage, which experiences high dV/dt during switching.  Unfortunately, most differential probes are not good for the task because they have limited Common Mode Rejection Ratio at high frequencies.

If cost is no object, the Cleverscope CS548 is said to be an excellent device; it builds the isolation into the scope channels and is intended for measuring high side gate drivers.

[mtwieg]

With the fancy off-the-shelf transmission line probes, you'll have trouble finding something that will handle 30V continuously. They are generally optimized for bandwidth (like >1GHz), not for power handling.

Years ago I had a P6158 20:1 probe which I used for pulsed RF power amplifiers. Max rated voltage was 22Vrms, but I would apply >100V rms for a few milliseconds at a time with a low duty cycle, and the probe didn't seem to mind at all.

It might be worthwhile to DIY your own transmission line probe, especially if you only need ~500MHz of bandwidth. That way you can experiment with using higher power attenuator resistors, and easily repair it when you burn it out. If you go this route you may want to get a good pulse generator for testing the bandwidth and flatness of your probe. Here's a popular option which isn't very expensive

Another caution: be careful probing the gate voltage of the upper FET with a transmission line probe if you're using a bootstrapping gate driver. The DC current pulled from it may pull down the bootstrap capacitor voltage very fast, leading to big problems...

[coppercone2]

Hmm I wonder what the failure mode on that probe will be after its been overloaded too many times.

Descriptions of VHF megaoverloads are kind of scary, it seems to vaporize stuff.

I wonder if a megaohm meter would pick up any damage to dielectric from overloading. I guess the voltage is not that high at 100v, but it might be prudent to attempt monitoring

could it spot polarize the dielectric?

I think it would blow a hole some where when it finally decides to fail, possibly in a thinner area or area of minor impedance mismatch? I.e. slightly different braid distribution, surface conductivity, etc. Some where along the transmission line you have a higher voltage.

I think dielectric failure has something to do with polarization, it seems to be what happens to dielectrics under high voltage. Thermal effects are weird to think about for very fast signals, but I think it might get hotter somewhere too.

Someone might be able to leave a EFT machine running continuously on a cable

Would the foam start to densify in some spots? Or maybe become brittle? Mechanical properties of polarized materials might be interesting to read about. I think it would become brittle and then electrostriction would wear it more and more (cracking) until it broke through. I am thinking electrostriction on a soft dielectric might be similar to a dog tugging on a rope chew toy.  Fans of basketball shorts might notice how the nylon rope  (built in belt) begins to behave after alot of wear and tear, the nylon seems to grow together making for very tough knots. Given that alot of coaxial cables have foam dielectric, I expect they might not fair well to overload. The sponges, particularly the ones backed with scotch brite (which is kind of like the stiff coaxial cable wall), even seem to spall for some reason, like chunks fall off the bottom with alot of use.

and failure modes of foam slippers.

Also makes me wonder about ultrasound testing of coaxial cables.