ACV stuff for HP3456A calibration

[tatel]

Looking to calibrate or at least check ACV ranges on a 3456A.

Yeah, I know the manual says to use a F5200+F5205. Well, none of these are even in sight. Moreover, Ohms and DCV have priority to me. And the series of 19 videos about the F5200 from defpom is scary. So I thought it wouldn't happen anytime soon. But...

...I happened to stumble upon a Gertsch RT-7 ratio transformer on ebay and won the bid.

So far I tested it from mains using a variac. It works as expected. Now looking for a more stable AC source. Since Gertsch docs say V=0.35*freq, 350 V max, I guess some variable voltage, variable frequency AC source would come handy. If I'm getting it right, Gertsch RT-7 best accuracy is at 400 Hz, and I guess it would do 1 kHz with no problem. I don't know what would be the max frequency this Gertsch could deliver while maintaining accuracy.

3456A manual asks for up to 250 kHz. I would be more than happy with about 300Vrms at 1 kHz. More frequency would be better of course, but I could easily live without it. I would like to have lower frequencies available however.

So far I have seen some different ways. I would rather follow the one that would allow me to came as close as possible to adjust the ACV ranges. I guess it would also be the most expensive one and that it will require external calibration. In that case I just will allow plenty of time for the project, but would aim to complete it anyway.

Being an ignoramus, I would like to hear some general opinions/ideas about all the approaches I have found. Perhaps it would be wise to go the cheap way while learning and hope some F5200 will come into reach sometime in the future.

First, a couple threads from enut11 and the links he provided, about signal generator+ audio amplifier + reversed mains transformer (I have a Rigol DG800):

https://www.eevblog.com/forum/metrology/an-experimental-ac-voltage-calibrator/msg3840836/#msg3840836

https://www.eevblog.com/forum/projects/diy-variable-ac-voltage-and-frequency-source-for-calibration/msg5552145/#msg5552145

Second, almost the same thing seems to be available from Feelelec. It's said it can work with other AWGs, not just FeelTech:

https://aliexpress.com/item/1005004153177445.html

Third, the DIY ultra low distortion oscillator mentioned by branadic, it seems to have also basic AGC somehow:

https://www.eevblog.com/forum/metrology/ac-voltage-standard/msg3602024/#msg3602024

And last, the SWR200 way, which, if I'm getting it right, will output a sinewave from an integrated zener reference, with AGC and variable frequency using two external capacitors. It would need a DAC and anyway the same amplifier + reversed mains transformer thing I guess. Available from Mouser:

https://eu.mouser.com/ProductDetail/Apex-Microtechnology/SWR200C?qs=TiOZkKH1s2QO3EEJBtAi3w%3D%3D

So what do you guys think?

[J-R]

Volt-nut in-training here... Are you trying to calibrate/adjust the HP3456A directly off these outputs or in combination with another calibrated DMM?

My quick opinion is that I don't see how any of the methods listed would provide the desired absolute accuracy alone, but the short-term stability is probably acceptable when used in a transfer standard context and another calibrated DMM.

I have an RT-5 but the problem is that you still need an initial high-precision source and a way to double-check everything.  Many times I got an output that wasn't expected, perhaps sometimes due to noise, RF interference, or even just poor accuracy when changing frequencies.

Ultimately I did go the route of a function gen with a transformer, which can get you up a few DMM ranges alone depending, and then for higher voltages/frequencies add an audio amp in between.  Then I was able to calibrate/adjust unknown DMMs against known calibrated DMMs no problem.  This worked well enough for me since I had multiple calibrated 6.5 digit DMMs and was repairing other 3.5-5.5 digit DMMs.  But I think for a 6.5 digit DMM you'd want to just ship it out for a proper calibration.

[tatel]

Volt-nut in-training here... Are you trying to calibrate/adjust the HP3456A directly off these outputs or in combination with another calibrated DMM?

You might be right, I'm afraid.... Hopefully I will remain as a mild case. My goal is to be where you are telling me. To have a calibrated 6.5 digit max, to calibrate/adjust lesser, 3.5/4.5 devices. Not interested in say, maintaining the standard volt.

IIRC it was also you who told me it would be easier/cheaper to get a calibrated DMM. I can say now you were right, I'll probably expend more money than a brand new 34465 would cost. But where the fun is in that approach?

I don't have any 6.5 calibrated DMM at this moment.

My quick opinion is that I don't see how any of the methods listed would provide the desired absolute accuracy alone, but the short-term stability is probably acceptable when used in a transfer standard context and another calibrated DMM.

Then I'll take the cheap way for now I think... I was wondering about SWR200, but if it can't be good enough, it is what it is.

I'll get that 3456A externally calibrated eventually, but I want to learn more and to get more fun, checking it myself before doing that. I'm gonna collect the toys anyway.

Thank you very much

[J-R]

Yeah, collecting and tinkering is a fun part of the hobby.  You could go down the rabbit hole of the AC-DC thermal converter?

It will still be interesting to hear what the true experts say on this query, probably will take some time for everyone to weigh in.

Another option: Youtuber atkelar just finished a repair on a Fluke 5100B calibrator and he mentioned his plan was to rent the equipment necessary to calibrate/adjust it.

[tatel]

Yeah, collecting and tinkering is a fun part of the hobby.  You could go down the rabbit hole of the AC-DC thermal converter?

At this moment I would say no...

It will still be interesting to hear what the true experts say on this query, probably will take some time for everyone to weigh in.

I was expecting SWR200 could perhaps be useful, since it's a expensive, hermetic IC which has quite a few useful things in it. At the very least, I was expecting it could give accurate, stable amplitude, and perhaps frequency also. Otherwise I can't see how that price (€320, up to 500 for the selected M variant) could be justified. It makes AD5791 look cheap. But, I guess they could be milking the cow a little bit too much?

Another option: Youtuber atkelar just finished a repair on a Fluke 5100B calibrator and he mentioned his plan was to rent the equipment necessary to calibrate/adjust it.

Well, I got a €200 quote for having that 3456A calibrated. If, after getting the toys and tinkering at will, I become reasonably confident it will pass that calibration, then it will go to the cal lab. To, say, get confirmation I did it right. So far it doesn't look as being really out of whack, but I still have no confidence it's in 90 day specs. Voltage reference and some resistors will do that trip before.

I guess, to calibrate a 5100B is another entirely different can of worms.

[EC8010]

It would help to know what your application is. For audio frequencies, there are audio test sets. They typically can produce up to about 20VRMS sine wave down to their noise floor at frequencies from 10Hz to 300kHz (300kHz was needed to allow audio test sets to test amplifiers passing time code replayed by a video tape machine running at x50 normal speed). These days, audio test sets are effectively posh sound cards. In fact, there are various software packages that will allow you to literally use a soundcard as a generator and measurement system. Sound cards in a computer tend to pick up interference, so the next step up is to buy an outboard ADC/DAC intended for the music recording industry. They can work very well, but lack the convenience (and cost) of a dedicated audio test set. Beware that mains typically has 3% distortion, so the accuracy of your transformer ratio bridge will be destroyed by using mains as a source, even before you consider possible saturation of the inductors at that low a frequency. I've never used a transformer ratio bridge myself, but they are the ultimate way of making accurate alternating measurements. It really all depends on what you're trying to do. Do you have a genuine application, or are you simply trying to use your new toy? If the latter, you might find "Coaxial Electrical Circuits for Interference-Free Measurements" to be useful. Written by three national standards metrologists, it assumes rather a lot of prior knowledge, but it goes into transformer bridges and their design and use in considerable detail.

[tatel]

"Looking to calibrate or at least check ACV ranges on a 3456A."

As I understand it, one would have some sine signal at some frequency and output voltage, and the ratio transformer would allow to divide that output AC voltage like a KV divider would do with DC. With small output impedance. That way one could check, and ideally adjust, at least some AC ranges. Am I missing something?

I used variac and mains just to check ratio transformer is working good. Brymen 869s shows each digit changing accordingly to the changes in ratio transformer, so I think it's OK. Now I'm looking for a good enough AC source.

I do have a posh (internal) sound card lying around. You made me aware I could perhaps use it to measure/check an AC source I could use/build, so thank you very much. It was going under the radar, I'll have a look at it.

SWR200 comes from Thaler but it's now owned by Cirrus. I know this article from Nuts&Volts is known by most of you. https://www.nutsvolts.com/magazine/article/the-ac-volt

I wonder if something like that could be good enough to adjust ACV ranges. After the article, it seems the author thinks so. If it's no good enough, then I will probably try a Wien oscillator with AGC.

[EC8010]

Some of the software for making a soundcard behave as an audio test set is a bit clunky, but you only need to check what's going on, rather than use it as a regular instrument, so that shouldn't bother you.

Beware Wien bridge oscillators. I know they were very popular and only use one op-amp but they're not actually all that good because the Wien network is inherently low Q (3, if I remember correctly). What that means is that the amplitude control circuitry has to be very subtle to avoid causing distortion. But it also means that it transfers low frequency noise from that circuitry into your signal as modulation of amplitude, so the distortion might be low but amplitude is not terribly stable. A better analogue solution is a state variable oscillator (using three op-amps) because they can be made to have a sufficiently high Q (>300) that they can tolerate Zener diodes as amplitude limiters yet still have low distortion (I've measured 0.003%).

[tatel]

Some of the software for making a soundcard behave as an audio test set is a bit clunky, but you only need to check what's going on, rather than use it as a regular instrument, so that shouldn't bother you.

We'll see what's available on linux

Beware Wien bridge oscillators. I know they were very popular and only use one op-amp but they're not actually all that good because the Wien network is inherently low Q (3, if I remember correctly). What that means is that the amplitude control circuitry has to be very subtle to avoid causing distortion. But it also means that it transfers low frequency noise from that circuitry into your signal as modulation of amplitude, so the distortion might be low but amplitude is not terribly stable. A better analogue solution is a state variable oscillator (using three op-amps) because they can be made to have a sufficiently high Q (>300) that they can tolerate Zener diodes as amplitude limiters yet still have low distortion (I've measured 0.003%).

Thank you for the heads-up. Anyway the first step will quite probably be to use my rigol DG800 as signal source. So audio amp and mains transformer in reverse would be the first bits to put together. After that I will look elsewhere, assuming it will not be enough to check my ACV ranges accurately enough.

[EC8010]

I blew up an amplifier driving a mains transformer in reverse feeding a turntable motor. It worked for a few seconds then expired. I suspect the load made the amplifier oscillate at HF. Use an amplifier you won't cry over, and if you have a means of measuring impedance as a function of frequency, a Zobel network across the output of the amplifier where it feeds the transformer might be a good idea.

I'd be interested to hear how you get on.

[tatel]

I blew up an amplifier driving a mains transformer in reverse feeding a turntable motor. It worked for a few seconds then expired. I suspect the load made the amplifier oscillate at HF. Use an amplifier you won't cry over, and if you have a means of measuring impedance as a function of frequency, a Zobel network across the output of the amplifier where it feeds the transformer might be a good idea.

I'd be interested to hear how you get on.

I'm collecting quite a few different artifacts to check my 3456A. RT-7 is just the last one, and AC has the lowest priority to me. I have still to work on Ohms and DCV things. Not that much spare time. So I won't do it tomorrow. But I will gladly post about this matter when I finally get to it. Thank you for your attention.

[enut11]

Some of the software for making a soundcard behave as an audio test set is a bit clunky, but you only need to check what's going on, rather than use it as a regular instrument, so that shouldn't bother you.

We'll see what's available on linux

Beware Wien bridge oscillators. I know they were very popular and only use one op-amp but they're not actually all that good because the Wien network is inherently low Q (3, if I remember correctly). What that means is that the amplitude control circuitry has to be very subtle to avoid causing distortion. But it also means that it transfers low frequency noise from that circuitry into your signal as modulation of amplitude, so the distortion might be low but amplitude is not terribly stable. A better analogue solution is a state variable oscillator (using three op-amps) because they can be made to have a sufficiently high Q (>300) that they can tolerate Zener diodes as amplitude limiters yet still have low distortion (I've measured 0.003%).

Thank you for the heads-up. Anyway the first step will quite probably be to use my rigol DG800 as signal source. So audio amp and mains transformer in reverse would be the first bits to put together. After that I will look elsewhere, assuming it will not be enough to check my ACV ranges accurately enough.

Hi @tatel. I don't recommend a mains transformer for this task, they are designed for 50/60Hz and have poor frequency response. Also, their secondary windings do not suit an audio amp drive.
Have a look at the 'audio line transformer' that I used, in reverse, in my mini ACv Calibrator, M1120. This goes up to 20KHz or more.
enut11

[tatel]

Understood. Thank you very much.