Stable AC current source for "calibration"?

[c4757p]

So, I finally got my Fluke 45 fixed, after multiple broken parts. One of the parts that I had to replace was the EEPROM, which was totally dead, so I'm in the process of recalibrating it. So far I've managed to calibrate everything except the AC A range, because I can't think of anything stable enough to use as a standard. If anybody is curious, I've found these to be quite stable enough for the other modes, and used my HP 3468A to measure them as a transfer standard.

• DC V less than 20: Power Designs 2005 precision power supply
• DC V less than 50: Power Designs TP343A (seriously, this is actually stable enough that even the least significant digit on the 3468A completely froze)
• DC V 50-1000: homemade boost converter
• DC mA/A: homemade constant current source
• Ohms: Resistors mated to heating elements for thermal stability
• AC V less than 7: HP 3325A
• AC V greater than 7: HP 3325A -> buffer -> transformer
• Frequency: HP 3325A locked to a rubidium reference

I have no idea how to get a stable AC current source. For the lower current ranges, I suppose I could feed a constant AC voltage through a resistor, and just measure the current sense resistors in both meters and mathematically compensate for the difference. I don't really like that approach, though, and I like it even less for high current because of the heat generated. Any ideas? (And does the waveform matter? I would assume it should be a sinusoid, but a square wave would be much easier.)

[oscopeface]

I'm not sure if any of that equipment will really get you true calibration? I calibrate those Fluke 45's all the time using a Datron 4808 or Fluke 5500 Calibrator at work.
I don't think a homemade AC current source will do much for you in terms of a real calibration.

[c4757p]

Well, as I said I'm using my 3468A as a transfer standard, and I'm OK with that. I don't need the value to be exactly what I want it to be, I just need it to stay fixed between measuring it on the 3468A and typing the measured value into the 45.

[mzzj]

I would use same strategy  as with AC volts:  HP 3325A -> buffer -> transformer

Just reverse the transformer to get higher amp low voltage output. Add some series resistance to help keeping the current stable.

I don't think it would be that difficult to keep it enough stable.  Besides, You need something like stability only for 1 minute to write down numbers from both meters. 

Another option would be true voltage controlled bipolar current source, but I don't think its really necessary.

[AlfBaz]

This may take a bit of looking into but you could use a transformer. You pass the current through the primary winding (so that is in series with the load) and then run some dc current throught the secondary. by varying the DC current in the sondary you are basically saturating the core thereby creating a "variable" choke

If I remember correctly the more you saturate the core the higher the reactance in the ac winding, so if the AC current increases you divert some of it into the scondary via a rectifier, increasing the reactance. If the AC current starts to drop the dc current in the secondary drops lowering the reactance.

[PA4TIM]

Problem is you often need several magnitudes but also several frequencies. AC calibration is very difficult. The signal often should not have harmonics because if they fall into the range of the RMS converter they give a mistake. For instance my prea measures upto 1 MHz. If rhe AC source has a high harmonic content it will measure that too. But a 10 kHz meter will not see those harmonics. ( we are talking many digit meters, because the " fault " will be not visible on a 4,5 dgit meter or the source must be fery bad.

You can make a very good clean oscillator ( williams has some designs) and compare it to a VDC standard. Or chop a DC Vref voltage and use it as a squarewave. The rms amplitude will be half the DC voltage. If you need a sinewave you can filter it. But then you have a stable AC source of unkown amplitude. But stability is more important as amplitude. Better to have 9.99543V that stays that amplitude as a 10.0000 signal that changes plus min 0.1V over temperature and short time.

For me AC is still te most difficult. I have a Fluke 10V AC standard. All my meters agree very close on DC and resistance but not on AC. Like measuring ripple current. I tested it with 5 rms meters and there was a factor 10 difference. And all gave a much lower ripple as a scope showed.

Besides all this, there is the crestfactor that makes AC measuring hard.

[mzzj]

Problem is you often need several magnitudes but also several frequencies. AC calibration is very difficult. The signal often should not have harmonics because if they fall into the range of the RMS converter they give a mistake. For instance my prea measures upto 1 MHz. If rhe AC source has a high harmonic content it will measure that too. But a 10 kHz meter will not see those harmonics. ( we are talking many digit meters, because the " fault " will be not visible on a 4,5 dgit meter or the source must be fery bad.

You can make a very good clean oscillator ( williams has some designs) and compare it to a VDC standard. Or chop a DC Vref voltage and use it as a squarewave. The rms amplitude will be half the DC voltage. If you need a sinewave you can filter it. But then you have a stable AC source of unkown amplitude. But stability is more important as amplitude. Better to have 9.99543V that stays that amplitude as a 10.0000 signal that changes plus min 0.1V over temperature and short time.

PC  sound card will do reasonably well for this purpose. Decent one has something like -100db noise+harmonics level and wont cost you arm and a leg.
If I remember correctly short-term stability is also more than enough for 4½ digit meter based on measurements I did couple of years ago.
Actually I was wondering is there any USB-based soundcard with a possibility to add your own precision reference..

[c4757p]

Responding while parked in my car, so I will get to a more detailed response in a while, but - actually, the Fluke 45 doea not need multiple AC frequencies for current calibration (all is done at 1 kHz) and only requires two or three datapoints for magnitude. My guess is that because they are using a very good integrated RMS-DC module, much of the calibration can be derived from AC V and DC A.

[c4757p]

Didn't have a suitable transformer for the higher currents and really didn't feel like winding one. Since the manual doesn't seem to specify a current waveform, I just ended up throwing together a bipolar switch, using it to drive a resistive load with a 1 kHz square wave, and waiting for the load temperature to settle. It was reasonably stable (I couldn't get the LSD on the 3468A to settle at lower currents, but that digit doesn't even exist on the Fluke - had to use a different meter for the higher current because the 3468A has a 3A max). TBH I don't really care if it's fully within specs, as long as it isn't completely unreasonable. This calibration will probably be the last time I ever use the AC A mode, and I have a handful of other meters that measure AC A just fine...