50A 0.1% Bench ammeter?

[smoothVTer]

I have an application which needs to measure current in sweeps which use 10mA steps from 0A - 50A, with an accuracy of 0.1% or better.   Up until now, I've been using 2 34461A's in parallel for up to 20A sweeps with success, but as these current measurements get larger, ballasting/current sharing becomes problematic, not to mention stacking 5 of these meters seems excessive.

I am running an entire bench of instrumentation sequenced using VISA and python, so whatever instrument I end up using needs to speak SCPI and have auto ranging.   I cannot find anything from any of the common vendors like Keysight etc. that satisfies these requirements.

Considerations:
(1) I've already considered using a shunt, however, my application would require 2 or 3 shunts of different values, plus different current sense amps with different gains, plus automated switching between them; so, high current relays.  I do not want to get into the business of designing an ammeter and the serial interfaces required...sounds like a long drawn out custom lab equipment project.
(2) Clamp ammeters are not accurate enough for what we are trying to do, plus, I can't find a clamp ammeter that speaks SCPI.
(3) Autoranging would be required because these sweeps are automated and unattended.  I cannot run a sweep up to 10A, then pause it and change shunts, then unpause it and change shunts again.   
(4) I understand banana cable interfaces are good to about 10A, maybe 15A.    It would be OK to use bus bars with studs to interface these high current paths in my lab bench, that is to say, I would not be limited to standard banana jacks for any of this work.

Does anything like this equipment already exist out there?
If it doesn't, can anyone chime in to why it probably doesn't exist?

Any other helpful thoughts?

Thanks for reading everyone.   

[tszaboo]

It exists. You need deep pockets.
https://www.lem.com/en/product-list/200s

Look around on their website, there might be others that fit your particular need better.

[alm]

Look in this recent thread for some ideas how to measure 50A DC. This device comes to my mind, but it won't do auto ranging, and would require changing /switching connections to change range.

There are current clamps that output a voltage and can be interfaced with any bench DMM, but it probably won't have the required accuracy. DCCTs (see other topic for link) like LEM Ultrastab might be a possible solution since they are very linear.

As for why no such device exists, because using an external shunt allows more flexibility in trading of sensitivity / noise vs dissipation / voltage drop, and allows more heat dissipation. Auto-ranging at these kind of current levels (DC) would require some pretty chunky (solid state) relays. I guess there is not much demand that is not already satisfied with one of the other solutions I mentioned.

What you could do is put different value of shunts in series with a diode across them (that can handle 50A) to limit dissipation. Then measure the voltage across each shunt resistor using different DMMs or a scanner. This should give you a rudimentary form of auto-ranging.

[tszaboo]

Look in this recent thread for some ideas how to measure 50A DC. This device comes to my mind, but it won't do auto ranging, and would require changing /switching connections to change range.

There are current clamps that output a voltage and can be interfaced with any bench DMM, but it probably won't have the required accuracy. DCCTs (see other topic for link) like LEM Ultrastab might be a possible solution since they are very linear.

As for why no such device exists, because using an external shunt allows more flexibility in trading of sensitivity / noise vs dissipation / voltage drop, and allows more heat dissipation. Auto-ranging at these kind of current levels (DC) would require some pretty chunky (solid state) relays. I guess there is not much demand that is not already satisfied with one of the other solutions I mentioned.

What you could do is put different value of shunts in series with a diode across them (that can handle 50A) to limit dissipation. Then measure the voltage across each shunt resistor using different DMMs or a scanner. This should give you a rudimentary form of auto-ranging.

Wow, 7K for half a dozen shunts and an INA. Maybe I should go back to this field. Since I made 0.03% accurate current controllers for 50A how hard can this be?

[alm]

Wow, 7K for half a dozen shunts and an INA. Maybe I should go back to this field. Since I made 0.03% accurate current controllers for 50A how hard can this be?

I agree that it's overpriced, but the question was for an off-the-shelf solution, and this is off-the-shelf, although it won't do auto-ranging.

[jonpaul]

50 A ..100A 100 mV standard meter shunt

6.5 digit system type DVM with HPIB/Lan/RS-232

OR

Instrument diff amp >>24 bit ADC

Shunts here https://simpsonelectric.com/wp-content/uploads/File/datasheets/shunt_datasheet.pdf

1% but can be calibrated to 0.1%

j

[Berni]

Yep flux gate current transformers are the best way to do this

https://xdevs.com/review/ultrastab_review/

I happened to pick up a lot of 3 of these from some decommissioned MRI machine or something on ebay cheep. The older models are made by Danfysik (Now LEM) and come in an orange plastic case, but the internals are pretty much the same.

These are very stable and linear from miliamps to 100s of amps. The ratio of input/output is defined by the number of turns, so it is very stable, then to get there a nulling method is used, meaning there are no linearity distortions as the magnetic field inside is always 0.

The one problem these have is zero offset because it will pick up earths magnetic field and running current trough them while not powered (so no nulling action) can magnetize the core, resulting in an offset visible on the output. So you do need to null them out with 0A flowing when you first power it up, but afterwards they are rock solid.

[smoothVTer]

Very interesting idea, since I already have a 20CH 34972A.   Would be rather quick and cheap too.   Thanks!

[smoothVTer]

Thanks everybody for all these great ideas.  I have some research to do now.  My sincere gratitude to ya'll for sending me on the right path. 

[PwrElectronics]

I am going to ask test engineering what they used in a functional tester for a product I designed about 6-7yrs ago.  The unit used 100u-ohm shunts and amp to measure current but it was not accurate enough as-is so the functional tester ran I think 200A and 400A through it and cal fudge factors were programmed into the firmware.  I don't remember what they used in the tester to measure that current anymore.  It would have interfaced to the tester computer somehow.

When I was doing hand cal on protos, I used a 0.25% meter shunt and a 5-6 digit bench DMM as I recall.  But, at least in my case, I didn't need to measure over such a wide range.  We actually ignored the 0A error or offset as the product would never be sourcing low currents so accuracy at the bottom end was not a concern.

[Berni]

The problem with shunts at such high currents is that the P=R*I^2 gets really out of control, so you need really low resistances to not be dissipating an unreasonably high power as heat in the shunt resistor.

Even precision resistors tend to drift a bit with temperature, so that impacts your accuracy. At the same time the output voltage is tiny, making it difficult to measure properly. Then since it gets hot it also causes thermocouple voltages at the connections to the shunt, at the low shunt resistances you can easily get a thermocuple voltages equivalent to a few amps flowing trough the shunt. Okay but the whole shunt gets hot equally... well not really, the cables connecting it could be removing heat at different rates on each side, the contact resistance of the massive cable lugs might not be exactly the same on each side...etc.

This is why flux gate current transformers are so good. The wire passing trough the center can be made as thick as you need, allowing for potentially 1000s of Amps without getting too hot. If it gets hot it is no problem since the number of turns on a core is not temperature dependent. The winding ratio can be chosen so that you get a nice big signal in the 100s of mA on the output, making it easy to measure. It also does not suffer from thermocuple effects (since no part of this uses voltage, it is all currents) and so it means a transformer made for 1000A will still measure 1mA reliably.

[PwrElectronics]

OK, looks like they used a LEM IT 1000-S/SP1.  That is a 1000A range unit and around $4k USD.  Overkill for the OP but smaller ones are available.  This one used a +/-15V supply and a precision load resistor.  The DAC unit in the tester measured the voltage on that resistor.

[alm]

While I agree that current shunts don't scale well with current because the voltage scales linearly while dissipation scales quadratically, I think 50A should still be manageable: 1 mOhm would dissipate 2.5W and drop 50 mV.

[Berni]

Yeah brand new these LEM current transformers are going to be insanely expensive.

But they are really easy to find for under 50 bucks a piece on ebay:
https://www.ebay.com/itm/334966674098?hash=item4dfd91d2b2:g:RiEAAOSwZvJkyVaf&amdata=enc%3AAQAIAAAAwBnESRXNfqPrwU4GSWNzyOvyc4VtTMLwEddKiDhdbG2TFvuwQlFfQFSbG%2Fgg8c1Jysha9j7VlKcAQiIWnffE9YmfHyeO7wtKRC1%2FrDirIoiDWFJW6gHKF0sMiYibbsMK5%2BbGLEVstPa16crxTG%2BhXEzHR3Y7aL0UTCynLoCusyV1p3pWWDoO%2BcED6fj1GbLvwZ32bxj1Vobs6UjBgvMBye9PWHxK0gvz7x5p7ZhGiFfcjpvHWUHDI5%2FePQufQJPnIw%3D%3D%7Ctkp%3ABk9SR56999fgYg

[alm]

What load will your DUT tolerate? Shunts will be fairly resistive at lower frequencies, while a DCCT will be quite inductive. See this article about the LEM/Danfysik Ultrastab by TiN where his current source is getting unstable above 16A due to the inductive load.