Current Measurement from uA to 100A

[luky315]

I'm looking for a solution able to measure from low currents (standby /leakage) up to 50A (preferably 100A) for use in  lab equipment (development of power electronics), so low cost is not really a target. But I have problems finding a solution which is able to switch between current ranges without interruptions. Is there even a theoretical solution for this problem? The accuracy of the measurement should be "as good as possible", but the more important point is that there are no artefacts from switching between ranges and a voltage drop <1V even at 100A.

[ch_scr]

Either switch the shunts overlapping (make-before-brake) or have a single shunt and switch the ranges of the input amplifier.
The latter is limited by shunt power dissipation on the upper end and thermal voltages on the PCB on the lower end.

[coppice]

Switching shunts is always going to have issues. If you are on the microamp level shunt, and the loads suddenly goes to 100A, you will both get a horrible voltage sag, and very little time to switch shunts before something overheats. Maybe multiple flux gate sensors could be the answer. I don't know how well a high sensitivity flux gate would tolerate 100A, while a low sensitivity one measures that high current properly. Maybe that would defeat this approach.

[mk_]

Switching shunts is always going to have issues. If you are on the microamp level shunt, and the loads suddenly goes to 100A, you will both get a horrible voltage sag, and very little time to switch shunts before something overheats.

Protection can be done with an high speed comparator (reaction time 1-5 ns)  driven high current FET (300A-FETs with 1 or 2milliohms Rds_on are aviable) shorting the shunt when overcurrent-detection gets activated (can be done below 100ns delay from overcurrent until shunt shorted).... and restart measuring with the highest current-Shunt. Nothing for AC but for DC-measurements it`s fine.

Protection with shorting the shunt avoids a lot of problems specialy at high currents.
Our device, runnig up to some 100As, floats against all other Voltages; reactiontime throu some isolators from overvoltage on shunt to gate ~180ns, most time (70ns) is consumed from the isolated gatedriver.

hth

[Alex Nikitin]

My 2 cents.

If your application can tolerate a non-linear voltage drop, two shunts for two ranges (say, 1 mOhm for 100mA-100A and 1 Ohm for 100nA-100mA) in series, with a pair of clamping high current Schottky diodes over 1 Ohm. You may have a problem with the forward leakage of Schottky diodes though * .

With a single 10mOhm shunt capable of 100W dissipation you'll get 1V max drop at 100A and the noise level below 100uA. You would need to watch out for thermal voltages in this version.

A 1mOhm shunt + a sensitive DC current clamp, capable of sensing microamps and at the same time of surviving a 100A overload is another possible option.

Cheers

Alex

P.S. - * it is possible to make an accurate MOSFET "soft" (i.e. no hard switching) clamp at, say, from 100mV to 150mV over 1 Ohm shunt, that would allow for more accurate measurements at the low end of the range.

[Phil1977]

I´m afraid the best out-of-the shelf solution would be a really powerful sourcemeter:

https://www.calplus.de/keithley-2651a.html

SourceMeter, 100nV...40V, 100fA...5A, 2kW (50A Pulse Mode), Vierquadranten Betrieb,

For such a large current range it´s an advantage to build a power source, not only a meter. A source can internally compensate for shunt losses and transparently switch the shunts - or whatever kind of magic Keithley is using there.

With a single shunt it´s really difficult. 1V voltage drop at 100A means 10mohm resistance. That corresponds to 10nV per uA - not absolutely impossible to measure but very complex, expensive and prone to error. If you switch shunts you also open the door for many complications.

Any kind of magnetic measurement will also be incredibly sensitive to distortions. Even with many windings the resulting magnetic field of 1uA is simply too small.

[luky315]

A big SMU would be interesting, but even the €€€ model from Keithley is not powerful enough - we need 50A (sometimes even nearly 100A) continuously - so I think that some solution with different measurement paths is needed.

[Phil1977]

Maybe it´s a stupid idea, maybe not:

Use two power supplies in parallel. Let´s assume you want to simulate a 120V-battery:

- Set power supply 1 to 121V and 100mA limit, measure the current with a 10 Ohm-Shunt and uV-meter
- Set power supply 2 to 119V and 100A limit, measure the current with a 1mOhm-Shunt and uV-meter.

Probably you need some backflow protection for power supply 2. Because 100A are quite much for a Si-diode and Schottky-diodes have too much leakage current, you could try a MOSFET-switch that opens as soon as the output voltage is below 120V.

As long as the consumed current is lower than 0.1A the PS with the higher output voltage delivers all the current. As soon as the current rises, the lower voltage PS adds current. You can log the current of both sources and add both values for the report.

Of course, it´s not a perfect and not a beautiful solution because the voltage is not really stable. But maybe it´s a solution anyhow.