Measuring Rds-on - bandwith limitation from clamping circuit

[eTobey]

Hi,

im reading about measuring Rds-on of Mosfets. There are not many resources i can find on that. I only found one topic in this forum: here. There is a paper (APT0407), that shows some clamping circuits.

There is one thing, i dont understand in the the figure attached.
They say, that there is only a reliable measurment of about 20kHz possible due to the RC time. It seems to me, that this is wrong. My understanding is, that the delay by the RC time is only until the brown curve hits the blue curve (how i expect point B to be), which happens about 3..5us. But doesnt these 3.5us mean that it would be suitable for 280kHz?

Another question i have:
Why ist that final value not taken right after the reaction time? For me the time where they take the final value would only make sense, if that curve goes like A.
Also: Why does the brown curve slowly raise after the reaction time?

Have a great day.
Toby

[PartialDischarge]

Hi,

Not sure where the brown curve comes from, but it goes up to show how the Rds gets higher after current is switched, so Vds goes up.

The blue A curve is the one with the low bandwidth limitation, and this bw gets worse as the Vds becomes higher.
There are really not good circuits out there to measure low Vds (or Rds), either the precision is bad (large offset) or the response is too slow or the response has ringing. So for IGBTs and low frequency switchers it's ok but for devices that are ON during a few us it is quite difficult to obtain a precise measure of Vds.

I should know as I'm working in one device to do this.

[jonpaul]

Very old and we'll studied problem.

See IR, Silicon x, etc app NOTES ON FET Rds measurement from 1970s..1990s.

This is a bad circuit, wrong measurement technique.

Question the source.

Such a measurement Must consider Zener and probe capacity, this ckt is limited by very high 100 k R series and scope overload recovery time.


j

[inse]

I have recently performed similar measurements on IGBTs.
The setup was similar to the one you showed.
Except the bus voltage was only 14V and the inductance was 1mH.
No signal clamping had been used.
I had good results with a 12 bit HD scope using the lowest possible range that did not show overdrive distortion.
The IGBT was turned on for some ms and current and Vce has been recorded.
I measured up to 15A.
It’s another topic if you have to do that measurement at high bus voltage, good luck on that.

[PartialDischarge]

Very old and we'll studied problem.

Absolutely not true.
The problem is solved for steady state turn-ons of semiconductors (which is how manufacturers measure their rdson by the way, big deal) and low frequency switchings (10's of us turn on time) . All current circuits available are slow and unsuitable for real time study of turn-ons of 100ns for example.
Also offsets are huge and need to be corrected, and ringing is common during the switching.

There is actually a lot to be studied here including new circuit concepts or the information obtainable from real time observation of the Vsat (bond wire lift off in IGBTs for example)

[jonpaul]

au contraire

Rds on pulsed, dynamic was researched and measured long ago

just one....
https://eepower.com/technical-articles/dynamic-on-resistance-measurement-techniques-for-gan-power-transistors/

j

[PartialDischarge]

au contraire

Rds on pulsed, dynamic was researched and measured long ago

just one....
https://eepower.com/technical-articles/dynamic-on-resistance-measurement-techniques-for-gan-power-transistors/

j

Now you quote an article from 2021, I thought you said this was and "old" problem and measured "long ago". And that measurement method some would consider slow, as they mention "The response time of the clamp circuit in our double pulse test system proved to be less than 100 ns, which is fast enough for most applications".
Also that method has other problems.

A couple of years ago I was already obtaining better results, faster response and smaller negative dips.

[jonpaul]

quoted most recent article, many from decade's ago.
j

[eTobey]

Question the source.

Such a measurement Must consider Zener and probe capacity, this ckt is limited by very high 100 k R series and scope overload recovery time.

There are questions, that i would like to be answered. As i understand from that paper, those things have been taken into account.

If you cant give me some good arguments or answers, i question you ;-)

[eTobey]

I have recently performed similar measurements on IGBTs.
....
I had good results with a 12 bit HD scope using the lowest possible range that did not show overdrive distortion.

Can you answer me one of my questions? I wonder, what pitfalls there are, when implementing rds-on measurement in a BLDC controller.

[inse]

The question is why you want to measure the Rdson, maybe I haven’t figured out your requirement.
I had to do it to match the suppliers datasheet with our application requirements to have a common data base to compare different components.
If you want to determine the losses in your circuit, then separate between conduction losses and switching losses.
Usually the switching losses are the main contributor.
What is exactly your question?
Do you have to deal with some specific physical effects?

[eTobey]

The question is why you want to measure the Rdson, maybe I haven’t figured out your requirement.
I had to do it to match the suppliers datasheet with our application requirements to have a common data base to compare different components.
If you want to determine the losses in your circuit, then separate between conduction losses and switching losses.
Usually the switching losses are the main contributor.
What is exactly your question?
Do you have to deal with some specific physical effects?

Current sensing in a BLDC Controller.

My questions are in my first post...

[inse]

Sorry I got this wrong as you were asking for Rdson, not current sense.
I cannot contribute to this problem.
The setup with oscilloscope has mislead me as well.