An Experimental AC Voltage Calibrator

[trobbins]

mzzj, Room Equalise Wizard (REW) has it in their generator.  The spectrum analyser is used to view the 'output' signal to be nulled and each harmonic's amplitude and phase can be adjusted to null that harmonic down in to the noise floor, with settings for up to 10th harmonic.  The null degrades as frequency and amplitude are changed, as expected.  It's just convenient if you have a need.

[tszaboo]

For generating AC signals for a microcontroller, or something like that, the easiest way is with DDS generators from AD. Select one that has a few MHz range only. You can feed this into an amplifier for gain, and a second one for low impedance output if needed. If you want to measure AC (for feedback for example) similarly, probably the easiest to do is with a RMS to DC converter chip from AD. The LT parts from their portfolio are using delta-sigma modulation to do this conversion, and they are quite accurate once calibrated.

[enut11]

The prototype variable voltage/frequency AC Calibrator is now working to my satisfaction. Provided enough time is given for the setup to settle wrt temperature (>60min) the stability is good enough to allow a valid simultaneous comparison of two or more meters. Maximum frequency is about 50KHz for LV out and 15KHz for HV out.

This design should only be used with pure sine wave input. Square waves can give unpredictable results at low frequencies (core saturation) and high frequencies (attenuation).

I have de-rated the maximum AC outputs to 10vACrms (LV) and 100vACrms (HV). The latter was purely a choice based on the 100v rating of the Jaycar MM1900 5W audio output transformer.

I added a 1Kohm input voltage trim 10-turn pot. This sits at the top of the amplifier 100Kohm input resistor and gives a fine adjustment of about 1% of input signal and allows better than 1mV resolution at 100vAC signal out.

There are still parts of the build that could be improved but that will be part of phase 2 of this project.

[David Hess]

If you want to measure AC (for feedback for example) similarly, probably the easiest to do is with a RMS to DC converter chip from AD. The LT parts from their portfolio are using delta-sigma modulation to do this conversion, and they are quite accurate once calibrated.

If I wanted to measure RMS with as much accuracy as possible, then I would use a high resolution sampling converter and calculate it directly.  Such a sampling converter could be integrated, or if I thought I could get better performance, I might use a discrete high precision sampler in front of a slower but more accurate converter.

The most significant limitation to accuracy would be the passband response of the stages up to and including the sampler.  If the frequency content can be known, then correction could be applied after conversion.

[1audio]

However if you want to be able to compare to the international standards you need to use a thermal converter. They are finicky, fragile and drift a lot but they are still the gold standard for AC.

I was thinking about the limitations of the square wave calibration. You could predict the error from the rolloff I believe and possibly figure out the bandwidth of the meter and potential HF accuracy using several frequencies. Ultimately you would need to compare back to a sine wave reference to be sure. I don't think its a dead end but the bandwidth correction will be essential for accuracy. I'm not sure how something like the HP3458 would react to that, the HP 3458 uses sampling in effect for AC measurement so some strange things could happen at certain frequencies.

The current best efforts for AC at midband (100 Hz to 10 KHz) are around 10 ppm or so.

HP made a sampling AC voltmeter that was good to 1 GHz around 50 years ago. They abandoned it for some reason, like no customers? They also made a Thermal RF power meter with higher accuracy (a version is still in production for $20K +) which does still have customers.

[enut11]

However if you want to be able to compare to the international standards you need to use a thermal converter. They are finicky, fragile and drift a lot but they are still the gold standard for AC.

I was thinking about the limitations of the square wave calibration. You could predict the error from the rolloff I believe and possibly figure out the bandwidth of the meter and potential HF accuracy using several frequencies. Ultimately you would need to compare back to a sine wave reference to be sure. I don't think its a dead end but the bandwidth correction will be essential for accuracy. I'm not sure how something like the HP3458 would react to that, the HP 3458 uses sampling in effect for AC measurement so some strange things could happen at certain frequencies.

The current best efforts for AC at midband (100 Hz to 10 KHz) are around 10 ppm or so.

HP made a sampling AC voltmeter that was good to 1 GHz around 50 years ago. They abandoned it for some reason, like no customers? They also made a Thermal RF power meter with higher accuracy (a version is still in production for $20K +) which does still have customers.

Hi @1audio. Your are correct of course with your comments about the thermal converter method being the gold standard. However, it is a finicky method and requires systematic best practice to work well. I wanted a 'quick and dirty' method for comparative measurements. With my setup, assuming you have a good reference meter, you can compare several instruments simultaneously under arguably identical conditions.
enut11

[tszaboo]

If you want to measure AC (for feedback for example) similarly, probably the easiest to do is with a RMS to DC converter chip from AD. The LT parts from their portfolio are using delta-sigma modulation to do this conversion, and they are quite accurate once calibrated.

If I wanted to measure RMS with as much accuracy as possible, then I would use a high resolution sampling converter and calculate it directly.  Such a sampling converter could be integrated, or if I thought I could get better performance, I might use a discrete high precision sampler in front of a slower but more accurate converter.

The most significant limitation to accuracy would be the passband response of the stages up to and including the sampler.  If the frequency content can be known, then correction could be applied after conversion.

I think you underestimate the complexity of such systems.
Let's take the LTC1968. It will work to 500khz without too much trouble, costs 4 USD and it needs one capacitor (film preferably) and a slow 12-16 bit ADC to digitize it.
If you want to make a similar system with sampling, you have to make a DC ad AC accurate antialiasing filter, Sample it with an ADC at least 1MSPS, but realistically, much higher, I would say 5MSPS. Then you have to make sure that your sampling frequency is not the multiple of the incoming signal, because then you sample it the same phase. Or you undersample it, and use an ADC with high input bandwidth. It will be a SAR ADC in either case, others will not have the sample rate.
Nevertheless, driving these is not trivial, needs a high bandwidth opamp, in the order of hundreds of MHz otherwise the input of the ADC will miss codes and all kinds of nastiness.
Then we arrive at the microcontroller.
When you all verified the system, wrote the code, and finished it up... Sounds like a lot of work doesn't it? Could be more accurate? Yes, sure, but be ready to those debugging sessions, when "I wish I would have a 16 bit scope to see what my ADC driver is doing, and why do I have INL errors in my output code".

[1audio]

And you would still need one of those infernal thermal voltage converters to validate your efforts. The whole thing is a rabbit (rat) hole. Just sorting out my vintage calibrator has been a challenging exercise in who do you trust? Every instrument deviates as the frequency goes up and not in the same ways.

[David Hess]

HP made a sampling AC voltmeter that was good to 1 GHz around 50 years ago. They abandoned it for some reason, like no customers? They also made a Thermal RF power meter with higher accuracy (a version is still in production for $20K +) which does still have customers.

Racal-Dana had one also.  I have a couple of Tektronix 7S11 samplers so could put something with similar or better performance together in about 10 minutes with what I have on hand.  Tektronix even published an application note showing how to do it.

The advantage of sampling measurement over linear measurement (for lack of a better term) is that the frequency response of the sampler can be calculated from the sampling gate width which can be measured with an unleveled RF source.  So if you know the frequency components of the measured signal, which will often be the case with an RF sampling voltmeter measurement, a simple lookup will correct for the passband response.  Racal-Dana included the correction curve for their sampling RMS voltmeter in their manual.

If I wanted to measure RMS with as much accuracy as possible, then I would use a high resolution sampling converter and calculate it directly.  Such a sampling converter could be integrated, or if I thought I could get better performance, I might use a discrete high precision sampler in front of a slower but more accurate converter.

The most significant limitation to accuracy would be the passband response of the stages up to and including the sampler.  If the frequency content can be known, then correction could be applied after conversion.

I think you underestimate the complexity of such systems.
Let's take the LTC1968. It will work to 500khz without too much trouble, costs 4 USD and it needs one capacitor (film preferably) and a slow 12-16 bit ADC to digitize it.

If you want to make a similar system with sampling, you have to make a DC ad AC accurate antialiasing filter,

Nope, one of the chief advantages of the sampling RMS measurement is that no antialiasing is required at all.

But what I was suggesting is either precision sampling followed by slow precision RMS conversion, or the direct computational approach where a high frequency high resolution sampling ADC is used for low frequency RMS conversion.  I was told about 20 years ago that some multimeters had started using the later method in lieu of analog computation but have not been able to verify that.

The Racal-Dana and HP RF voltmeters mentioned above use the former method, and I am suggesting extending this idea to low frequency measurement.

Sample it with an ADC at least 1MSPS, but realistically, much higher, I would say 5MSPS. Then you have to make sure that your sampling frequency is not the multiple of the incoming signal, because then you sample it the same phase. Or you undersample it, and use an ADC with high input bandwidth. It will be a SAR ADC in either case, others will not have the sample rate.

The bandwidth is determined by the sampling gate width, which produces a non-linear passband response that can be easily calculated.

I suggested a simple system relying on a sampling ADC with high input bandwidth used at much lower frequencies, but an alternative is to implement a precision external sampler with a slower ADC.  I messed around with the later many many years ago but not with this application in mind and I mostly learned what does not work for precision sampling.

Nevertheless, driving these is not trivial, needs a high bandwidth opamp, in the order of hundreds of MHz otherwise the input of the ADC will miss codes and all kinds of nastiness.

The chief advantage of the sampling method is that within the input signal range, no processing is required before sampling, removing those error contributions, which is the major reason to use it.  I can design and fabricate  a sampler or sampling system, but constructing a high accuracy thermal RMS converter would be much more difficult, at least for me.

Then we arrive at the microcontroller.
When you all verified the system, wrote the code, and finished it up... Sounds like a lot of work doesn't it? Could be more accurate? Yes, sure, but be ready to those debugging sessions, when "I wish I would have a 16 bit scope to see what my ADC driver is doing, and why do I have INL errors in my output code".

Calibration requires an unleveled RF source to measure the sampling gate width.  If you are desperate, processing after sampling can be done with any low frequency RMS measuring tool.  I could set something up which makes RMS measurements to GHz frequencies in about 10 minutes with what I have on hand, but it would not be real suitable for the low frequency high resolution measurements we are discussing, although it would work.

[tszaboo]

HP made a sampling AC voltmeter that was good to 1 GHz around 50 years ago. They abandoned it for some reason, like no customers? They also made a Thermal RF power meter with higher accuracy (a version is still in production for $20K +) which does still have customers.

Racal-Dana had one also.  I have a couple of Tektronix 7S11 samplers so could put something with similar or better performance together in about 10 minutes with what I have on hand.  Tektronix even published an application note showing how to do it.

The advantage of sampling measurement over linear measurement (for lack of a better term) is that the frequency response of the sampler can be calculated from the sampling gate width which can be measured with an unleveled RF source.  So if you know the frequency components of the measured signal, which will often be the case with an RF sampling voltmeter measurement, a simple lookup will correct for the passband response.  Racal-Dana included the correction curve for their sampling RMS voltmeter in their manual.

If I wanted to measure RMS with as much accuracy as possible, then I would use a high resolution sampling converter and calculate it directly.  Such a sampling converter could be integrated, or if I thought I could get better performance, I might use a discrete high precision sampler in front of a slower but more accurate converter.

The most significant limitation to accuracy would be the passband response of the stages up to and including the sampler.  If the frequency content can be known, then correction could be applied after conversion.

I think you underestimate the complexity of such systems.
Let's take the LTC1968. It will work to 500khz without too much trouble, costs 4 USD and it needs one capacitor (film preferably) and a slow 12-16 bit ADC to digitize it.

If you want to make a similar system with sampling, you have to make a DC ad AC accurate antialiasing filter,

Nope, one of the chief advantages of the sampling RMS measurement is that no antialiasing is required at all.

But what I was suggesting is either precision sampling followed by slow precision RMS conversion, or the direct computational approach where a high frequency high resolution sampling ADC is used for low frequency RMS conversion.  I was told about 20 years ago that some multimeters had started using the later method in lieu of analog computation but have not been able to verify that.

The Racal-Dana and HP RF voltmeters mentioned above use the former method, and I am suggesting extending this idea to low frequency measurement.

Sample it with an ADC at least 1MSPS, but realistically, much higher, I would say 5MSPS. Then you have to make sure that your sampling frequency is not the multiple of the incoming signal, because then you sample it the same phase. Or you undersample it, and use an ADC with high input bandwidth. It will be a SAR ADC in either case, others will not have the sample rate.

The bandwidth is determined by the sampling gate width, which produces a non-linear passband response that can be easily calculated.

I suggested a simple system relying on a sampling ADC with high input bandwidth used at much lower frequencies, but an alternative is to implement a precision external sampler with a slower ADC.  I messed around with the later many many years ago but not with this application in mind and I mostly learned what does not work for precision sampling.

Nevertheless, driving these is not trivial, needs a high bandwidth opamp, in the order of hundreds of MHz otherwise the input of the ADC will miss codes and all kinds of nastiness.

The chief advantage of the sampling method is that within the input signal range, no processing is required before sampling, removing those error contributions, which is the major reason to use it.  I can design and fabricate  a sampler or sampling system, but constructing a high accuracy thermal RMS converter would be much more difficult, at least for me.

Then we arrive at the microcontroller.
When you all verified the system, wrote the code, and finished it up... Sounds like a lot of work doesn't it? Could be more accurate? Yes, sure, but be ready to those debugging sessions, when "I wish I would have a 16 bit scope to see what my ADC driver is doing, and why do I have INL errors in my output code".

Calibration requires an unleveled RF source to measure the sampling gate width.  If you are desperate, processing after sampling can be done with any low frequency RMS measuring tool.  I could set something up which makes RMS measurements to GHz frequencies in about 10 minutes with what I have on hand, but it would not be real suitable for the low frequency high resolution measurements we are discussing, although it would work.

OK, an RF sampling ADC. I'm not familiar with those, so actually don't know how well they would work, so I take your word for it.
I can tell you that this wouldn't work with a SAR ADC.
Those LTC196x chips actually are not thermal based, they are using some sigma-delta conversion, and they are super simple to use. But I guess, each to their own.

[David Hess]

OK, an RF sampling ADC. I'm not familiar with those, so actually don't know how well they would work, so I take your word for it.

I can tell you that this wouldn't work with a SAR ADC.

It would work with a *sampling* SAR ADC (not all SAR ADCs are sampling ADCs), and it is what I would try first, but performance will be limited by the relatively low input bandwidth.  Some modern parts are very fast now.

I have been told that some modern multimeters do exactly this with a fast high resolution sampling SAR ADC operating at a relatively low sample rate and RMS conversion done on the digital side, but I have not confirmed it directly.

Those LTC196x chips actually are not thermal based, they are using some sigma-delta conversion, and they are super simple to use. But I guess, each to their own.

I suspect they use switched capacitor computation techniques.

[Kleinstein]

I have been told that some modern multimeters do exactly this with a fast high resolution sampling SAR ADC operating at a relatively low sample rate and RMS conversion done on the digital side, but I have not confirmed it directly.

The relatively popular DMM chip set HY3131 (e.g. used in one of Daves DMMs) have the option to use digital RMS with a relatively fast internal SD ADC.
The BW is still a bit limited in this case, and an external RMS chip may still be used. Both ways have there advantages.
The new DMM6500 is supposed to use digital RMS with an extra ADC chip (should be SAR type). Chances are they added the fast 2 nd ADC for the digital RMS and than also used it for the ditizing mode.

The  LTC196x would use switched capacitors like other SD ADC chips and many modern SAR ADC chips. So the driving needs can be similar.

For the RMS measurement a defined bandwidth can be an advantage. Higher BW is not allways better as this also means more noise. Especially with a calibrator the signal is relatively well behaved with a known frequency and low crest factor. The higher harmonics should not contribute too much. So things are a bit easier than with a normal DMM input. If filtering is needed some of this could be before the output and no extra anti aliasing needed at the ADC.
For the RMS calculation alising can be good and include frequencies higher than Fs without much extra effort. So the limit is not from the sampling rate, but the sampling part of the ADC and the input amplifier.

[enut11]

All interesting stufff guys but what I would like at the moment is feedback on the analog building block solution that I am working on.
My approach has been to use better components. Is there any practical way to improve the design?
How can the 3 major components, sig gen, power amp and output transformer, all currently independent, be coupled into an effective AC Voltage calibrator?
enut11

[mawyatt]

Not sure exactly how the DMM6500 performs RMS, but it agrees very well with the KS34465A which also uses a digital computational method. Maybe the Keysight and Keithley methods are similar. We found the old 34401A using the analog method also agree well with the DMM6500 which is our latest DMM only a few weeks old. Superb instrument BTW!!

I guess one could assume that both methods should give acceptable results if the Crest Factor and Waveform Fundamental Frequency are low enough. That was the basis for using the squarewave as a comparison waveform, since it's relatively easy to get a precise known peak value.

Best,

[2N3055]

Not sure exactly how the DMM6500 performs RMS, but it agrees very well with the KS34465A which also uses a digital computational method. Maybe the Keysight and Keithley methods are similar. We found the old 34401A using the analog method also agree well with the DMM6500 which is our latest DMM only a few weeks old. Superb instrument BTW!!

I guess one could assume that both methods should give acceptable results if the Crest Factor and Waveform Fundamental Frequency are low enough. That was the basis for using the squarewave as a comparison waveform, since it's relatively easy to get a precise known peak value.

Best,

There is a Keithley engineer dropping by on the forum occasionally. Real nice guy, too..
RMS method was mentioned and he said DMM6500 uses sampling + RMS calculation. No exact details of implementation, but sampling it is. There is no RMS converter chip inside...

[David Hess]

I have been told that some modern multimeters do exactly this with a fast high resolution sampling SAR ADC operating at a relatively low sample rate and RMS conversion done on the digital side, but I have not confirmed it directly.

The relatively popular DMM chip set HY3131 (e.g. used in one of Daves DMMs) have the option to use digital RMS with a relatively fast internal SD ADC.
The BW is still a bit limited in this case, and an external RMS chip may still be used. Both ways have there advantages.

That might explain what I was told and why I could not confirm it.  In a discussion about the Fluke lawsuit, it was reported to me that the Tektronix DMM916 did exactly what I described, but inspection of the printed circuit board shows an Analog Devices AD737, and the converter chip datasheet reveals no capability for digital RMS processing.  Of course the report may have been referring to something external to both.

The  LTC196x would use switched capacitors like other SD ADC chips and many modern SAR ADC chips. So the driving needs can be similar.

And like Linear Technologies long experience with switched capacitor filters.

For the RMS measurement a defined bandwidth can be an advantage. Higher BW is not allways better as this also means more noise. Especially with a calibrator the signal is relatively well behaved with a known frequency and low crest factor. The higher harmonics should not contribute too much. So things are a bit easier than with a normal DMM input. If filtering is needed some of this could be before the output and no extra anti aliasing needed at the ADC.

For the RMS calculation alising can be good and include frequencies higher than Fs without much extra effort. So the limit is not from the sampling rate, but the sampling part of the ADC and the input amplifier.

The tradeoff is that the sampling bandwidth, and its noise contribution, is fixed by the sampling gate width and no filtering before can reduce that, and no filtering after it can remove aliased noise.  So input bandwidth to some arbitrary cutoff can be extended, but only at the expense of increased noise unless the sampling rate is also increased to prevent aliasing.

I seem to recall that about 50 kHz was the sweet for sampling rate in old designs to allow reasonable operating life of the needed avalanche pulse generator.  A precision sampler would work differently of course.

[enut11]

Perhaps start with an 8k resistive load that reflects 8 ohm back to the amplifier (eg. for the M1120, the 1.25W tap presents 8kohm for an 8 ohm speaker load, with a 31.6:1 turns ratio).  You could sweep through the frequency range to see what voltage response you get, but at low voltage (as an 8k load would consume some power for higher voltages).  You may still get a reasonably flat response for higher resistance loads.

Thanks @trobbins. I tried an 8.2K 2W resistor across the transformer high voltage output and it had a big affect in flattening the high frequency response. Good suggestion
enut11

[Kleinstein]

The signal genrator is relatively straight forward these days.  DDS genration should give a pretty stable and clean enough sine wave. Using a wien bridge or similar can get lower distortion but much more tricky to stabilize the frequency and amplitude.
The power amplifier part is the easy part. At up to some 20 kHz audio should be good enough. The tricky part could be however if the transformer is to be driven in closed loop and not open loop.
I an afraid that a more normal transformer may not be very linear. It depends quite a bit on the type and amplitude to use. A simpel resistive load should not make that much difference. So ideally one would need some feedback directly from the output. One may get away with direct linear FB to have the transformer in the ampliers loop. The alternative may be feedback via a way to measure the amplitude. This could be true RMS, but also simple rectification, as the waveform is known. With a measurement  fron the output, the stability of the generator, amplifer and transformer would no longer be important.

In the extremes there would be the option to have a special precision transformer with seprate drive, correct and sense. This can make up a very stable and accurate gain, but at high costs / effort.

[enut11]

I have carried out some spot frequency checks on the Jaycar MM1900 5W audio transformer (HV Out). Response is flat up to ~30KHz and useful to around 50KHz. The higher frequencies improved dramatically after loading the transformer output with an 8.2Kohm 2W resistor.

Remember, the AC Voltage Calibrator is just a stable AC voltage/frequency source relying on a known reference meter to measure the actual value and transfer this to another instrument.
enut11

[enut11]

Frequency response of the LM1875 power amp was also checked and found to be good to around 75KHz.

I had to change the recommended filter on the LM1875 amp output (1ohm+0.22uF) to prevent the resistor from overheating at high frequencies.  I used a 15ohm 1W resistor in series with a 0.022uF cap.

Ti also recommend the following for high frequency stability:
"A method commonly employed to protect amplifiers from low impedances at high frequencies is to couple to the load through a 10Ω resistor in parallel with a 5 μH inductor."

[Kleinstein]

The RC daming element at the amplifier output is there for a reason. Changing this comes with the risk of oscillation or ringing. So one should at least check with the scope and square wave signal if the ringing gets excessive.

The series RL element can help has it isolates possible capacitive load. It comes with the downside of possibly a slight faster drop in the amplitude for higher frequencies. So one may have to use a smaller inductance. In audio ampliers the inductor is often just some 5 turns around the resistor so more like 0.2 µH.

[enut11]

Thanks @Kleinstein.  For the amplifier only (LVout), I tried both a 1KHz and 10KHz square wave at full output voltage (30vPP) and both were clean. Is this a good enough test for HF stability?

[Kleinstein]

A square wave from the input side is a reasonable test, at least for the lower frequency part. It however does not really test possible ringing mich higher than the amplifiers BW. It can be enough for the test, but it may miss the upper frequency end, which is often the point where ringing happens. The case with no loading to the amplifier is also the easy case and should be OK. It may get tricky with some added capacitive load, even of just some 1 nF (e.g. 10 m of coax). Ringing gets sometimes stronger with smaller amplitude, so that the amplifier is not at it's slew rate limit. The frequency does not really matter, it is about the edges.

[bdunham7]

It however does not really test possible ringing mich higher than the amplifiers BW. It can be enough for the test, but it may miss the upper frequency end, which is often the point where ringing happens.

Just to reinforce that point, when a 'typical' Class AB audio amplifier starts oscillating do to a malfunction, design issue or replacement part difference, the frequency of the oscillation is often several MHz.  So depending on the exact amplifier involved and what sort of input response it has, I suppose ringing might be up there as well.

[trobbins]

enut11, just to clarify can you confirm your scope has sufficiently high bandwidth for the square wave test, and the squarewave input has sufficiently low rise time, and what the load (or loads) were, and whether you had added any load RC or RL networks?