### Author Topic: .  (Read 3599 times)

0 Members and 1 Guest are viewing this topic.

#### AQUAMAN

• Regular Contributor
• Posts: 87
• Country:
##### .
« on: July 21, 2013, 08:55:41 am »
.
« Last Edit: September 04, 2013, 01:37:28 am by AQUAMAN »

#### Rerouter

• Super Contributor
• Posts: 3489
• Country:
• Question Everything... Except This Statement
##### Re: Measuring ESR of MOSFET/IGBT gate (nF/pF) to miliohm precision while IN CIRCUIT
« Reply #1 on: July 21, 2013, 09:33:05 am »
what your trying to do is not all that easy, especially with milliohm accuracy,

how i would attempt measuring the esr would be, trigger the circuit off the gate pin, when the gate passes a threshold voltage,
you measure the voltage across it, you then pump in a decent whack of current and measure the difference, but in both cases you want both the current source and adc disconnected before the mosfet turns off again, thus triggering on the gate,

at 5Khz, you could likely get away with a micro looking for a high or low pin,
same for the gate threshold, a comparitor looking for it exceeding or falling beneath a certain voltage to begin to measure,

you can even chop your current in and out multiple times to get a more accurate measurement,

#### Marco

• Super Contributor
• Posts: 3811
• Country:
##### Re: Measuring ESR of MOSFET/IGBT gate (nF/pF) to miliohm precision while IN CIRCUIT
« Reply #2 on: July 21, 2013, 09:49:45 am »
Is the source at ground? If so couldn't you just take a DSO and average lots of waveforms of the voltages before and after the external gate resistor during turn on with a constant load? This can give you gate voltage and current to high precision.

Then just pick the part near 0 volt (gate capacitance is non linear) and calculate the RLC circuit from the averaged waveform.

#### Marco

• Super Contributor
• Posts: 3811
• Country:
##### Re: Measuring ESR of MOSFET/IGBT gate (nF/pF) to miliohm precision while IN CIRCUIT
« Reply #3 on: July 21, 2013, 02:15:31 pm »
At a couple of microseconds you'll need a hell of a lot faster ADC than in a normal LCR meter. Audio frequency test signals and ADCs won't cut it. Also the way they measure doesn't really work for you, you simply don't have the time to let the circuit get into a steady state response to the test signal before measuring magnitude and phase angle. You'll have to work directly with the partial differential equations to get a fit.
« Last Edit: July 21, 2013, 02:37:24 pm by Marco »

#### Marco

• Super Contributor
• Posts: 3811
• Country:
##### Re: Measuring ESR of MOSFET/IGBT gate (nF/pF) to miliohm precision while IN CIRCUIT
« Reply #4 on: July 22, 2013, 01:42:56 am »
I am aware that I will need to build my own custom made ESR meter to do this. I built a custom made meter to measure the on-resistance of the transistors in the same time constraints to better than mOhm accuracy
That measurement can work with averages for current and voltage, which can be done analogue and thus doesn't need fast sampling.
Quote
I understand now that there are two ways of ESR measurement, constant current and constant voltage?
Hmm, constant current might work I didn't consider that ... if you overpower the inductance fast you can eliminate it from the measurement.

With a step voltage, or any voltage waveform really, the RLC series circuit is going to resonate and you have to determine R/L/C from how it resonates. with a normal LCR meter it's allowed to settle into steady state with a sine input at which point magnitude and phase angle of the current is all you need but that won't work for you because you only have a few cycles to work with at reasonable signal frequencies.

Just some stream of consciousness calculations ... lets say you have 10 nH and 100 nF of input capacitance. If you want to get that input capacitance to say 1 volt in 1 microsecond you have to pump 1e6*100e-9 = 100 mA into it (for large/HV MOSFETs this is going to be a LOT higher).

You'll need to cut off the gate from the driver so you can ram it with a stiff current source, I think you can just ignore the time it takes for the current to reach steady state. So you can sample it with a slow ADC you'll probably have to get tricky with integrators, for instance you could measure the gate voltage both with a differentiator integrator pair (this will get you C*i*t) and an integrator (this will get you i*r*t+0.5*C*i*t^2). The integrators will act both like filters and sample and hold circuits so you can use a slow high resolution ADC.
« Last Edit: July 22, 2013, 04:20:54 am by Marco »

#### fcb

• Super Contributor
• Posts: 1246
• Country:
##### Re: Measuring ESR of MOSFET/IGBT gate (nF/pF) to miliohm precision while IN CIRCUIT
« Reply #5 on: July 22, 2013, 04:50:05 am »
Why do you want to measure the gate resistance?  I thought gates were essentially capacitive? Am I missing something.

Also, if you are feeding a signal into the gate (to measure it's capacitance or whatever), aren't you in danger of triggering the DUT?  Are you trying to measure the DUT while it is in circuit?

#### Marco

• Super Contributor
• Posts: 3811
• Country:
##### Re: Measuring ESR of MOSFET/IGBT gate (nF/pF) to miliohm precision while IN CIRCUIT
« Reply #6 on: July 22, 2013, 06:43:41 am »
These capacitances are very very small, Ciss less than 10nF.
Depends on the switch used, with a TO220 high voltage MOSFET you'll be looking at the uF range ... it will have to be adapted to the switch.
Quote
I can't let the gate voltage get to above 1V I would think.
Just used 1 volt to get a grip of the magnitude of the numbers, you'll presumably use a safety cut off any way which stops the current if the voltage gets too high.
Quote
Do you have any documents on constant current ESR measurement (or any ESR)?
No, but the problem is you're far off the beat part ... I see some which take the difference between peak voltage just before current is turned off and the voltage afterwards, but that measurement seems a bit sensitive to noise in your case. Honestly it's been far too long though since I did any real noise calculations on circuits so my intuitive sense on what might work or might not work could be way off ... you'd have to try.

Smf