AC Voltage Standard

[Kennylxix]

So I'm fairly new to metrology but I find it very fascinating (I just sent my first bench dmm off to calibration. . I've found tons of information on methods to create very precise and cool standards for pretty much every measurement under the sun except one. AC RMS.  I've tried and failed to find a single source of a DIY reference or standard (The closest I've come is a dmmcheck but it's pretty limited on amplitude).  I also failed trying to find an IC that does any meaningful inversion.

Am I missing something or is AC voltage just not really a concern for anyone in the diy community?  I want to get to a point to where can calibrate a lower resolution DMM using only home made equipment and this is the one thing I just can't find anything on

Thanks for any info!

[capt bullshot]

Don't know a actual project, but from my knowledge a thermal converter would be your way. Start checking some of Jim Willams' application notes, LT built once a nicely integrated solution. But there's no secret stuff inside a thermal RMS converter, so you should be able to build one from scratch (more mechanical challenging than electronically).

[Kennylxix]

I've seen thermal converters and they are really cool but what i really want is an actual source.  Something that outputs a precise AC RMS voltage .  Even better would be a programmable one.  Oscillators pretty much already do this but they are designed for precise frequency not amplitude.  I know it's possible because multi thousand dollar calibrators have the capability.

I'm learning that pretty much anything that can be bought for thousands can be made for hundreds. I just don't know how it's done.

[CalMachine]

I'm in the works of a PWM version of a 10V reference.  Once I get that design finished and working as well as I would like it to, I plan on trying an DC->RMS comparator circuit to give out a DC/AC Transfer value.  I'm shooting for the stars, here!   

I contacted Guildline in regards to their 7410... but they told me it is no longer available and they have nothing even remotely similar to offer.  I wonder who paid them to stop producing that unit?   
http://www.weiku.com/products-image/13742223/Model-7410-AC-Voltage-Reference.html

Guildline took down all information regarding it, so it was a little more difficult finding something about it.

[Kennylxix]

Hmm so maybe a precision sine wave generstor that is programmable for amplitude and frequency going into a step up transformer with something like an lt1966 RMS to DC converter monitoring the generator output before the transformer.  Then an nice ADC on the RMS to DC converter with an Arduino to both monitor the ADC and control the generator.  It would needs some serious calibrating but once it was it should be pretty stable.  The lt1966 has a .02% linearity.

(Please forgive anything stupid I just said, I'm new to electronics;)

[Kleinstein]

Classical AC sources with precision amplitude use a thermal converter to do feedback on the amplitude. This is limited in amplitude though and might need stable dividers - precision transformers (e.g. with double core) can be pretty good in this. A lot of work, but in principle DIY possible.

For the low frequency range (e.g. 50 / 60 Hz) there is a chance to go sampling a sine and have a precision DAC that can also output DC. One can reasonably check the speed by sending out steps. As the usual DAC chips are reasonable fast, this might not be so hard, if you don't need super high accuracy. The individual DAC steps (or a reasonable part of it) could be measured in DC.

The complementary way would be a fast ADC to measure AC voltage by sampling and do RMS conversion in software. So more or less do what HP called "true volt" or the 3458 does for AC measurements, only with an integrated ADC. The problem here could be the speed and compensation of the input stage - so make sure amplitude is frequency independent. Directly at the output of a generator one might not need amplification at all - just some protection. However the ADC might have internal delays / frequency dependence. To a limited extend one could check it with a step response or a square wave signal. So even if you want only 50/60 Hz - have a MHz BW amplifier and adjust like a scope probe to have it straight in the low frequency range.

[Kennylxix]

Hmm yeah I understood about half of that:). I would definitely want it pretty accurate as it would be used to check calibration of dmms. I definitely want to look into this more. I bet if someone made one and could produce it even somewhat reasonably priced there would be a market for it.

[CalMachine]

Hmm yeah I understood about half of that:). I would definitely want it pretty accurate as it would be used to check calibration of dmms. I definitely want to look into this more. I bet if someone made one and could produce it even somewhat reasonably priced there would be a market for it.

I'm in the same boat with you, lol.  I'm very new to the design side of things and reading what a lot of people post on hear skims right on my hairline.  I get the general theory behind most things, but then implementing, properly, is a whole different story.  I love visiting these forums though, as I learn SOO much and it really helps me with my job!   

[pelule]

Have a look to folowing thread. Offers some ideas...
https://www.eevblog.com/forum/metrology/diy-precision-ac-rms-to-dc-transfer-standard/?all
I personally would put my money ito the DAC based solution.

[Kennylxix]

Generating the sine wave with a DAC seems like I really good idea.  I would imagine that a you wouldn't even have to measure the voltage coming out after the initial setup because the DAC should output the same thing (within the limit of the reference) every time.  Is that correct?

[Kennylxix]

So the AD5504 can output up to 60v which is perfect.  It has a built in reference as well.  I tried to look up how you go about actually telling one what to do and came up empty though.

[Kennylxix]

The biggest issue my ignorant viewpoint can see is translating the desired RMS value to a p-p value for the DAC.  This could be somewhat ignored by using a RMS to DC converter to show what the actual output is and something like a pot to control p-p.  That would be a kind of "you'll be happy with what we give you" kind of situation rather than being able to program exact wanted values but as long as the RMS sensing is accurate outputting somewhat arbitrary values isn't really an issue for dmm calibration.

This is fun.  Once again I apologize if what I'm saying doesn't actually make sense.

[pelule]

As a group of setvalues build  the output sinus (lets assume 10 setpoints for a half cycle of the sinus wave for easy calculation), the full sinus wave requires 20 values, thus an output frequency of 50 Hz requires 20*50 = 1000 values per second.
So settling time of the DAC defines the maximum output frequency. Thus I would use a DAC with a parallel input.
In the old days we used a ROM containing the set values and clocked the adress with different clock speed to control the ouput frequency. Nowadays I would use a SRAM (or FRAM) to have the output signal form programmable (getting an abritary signal generator)
Design critical are:
- the switching speed of the DAC / defines the max possible frequency
- the switching failure of the DAC generates dissortion
- the drifit (tempco, longterm stability) of the DAC the and reference and the impact of load all influence the stability of the output amplitide.

[Conrad Hoffman]

I've got a Fluke 540B thermal transfer standard but even those had a calibration constant. I think Thaler made an AC chip that's no longer available. Here's some info from a N&V article- http://www.nutsvolts.com/magazine/article/the-ac-volt

[David Hess]

The way I have done it in the past is to generate a precise 50% duty cycle square wave with accurate peak voltages but this only works if a square wave is suitable.

What I have considered is the sampling method which can return both the peak and calculated RMS values over a wide frequency range while the sampler itself can be calibrated with a clean square wave.  The problem then becomes how to build a high accuracy sampler.

[Kennylxix]

I found a paper where they created the waveform with a DAC which they then used a calibrated 8.5digit dmm to measure each step of a single cycle then calculated the RMS value from those results.  This gave them a calibrated value for each use of the standard which eliminated issues with long term stability.  I think this could be done with an ADC instead of a dmm.  I'm pretty sure a 24bit ADC would have enough resolution to get an accurate measurement and would be a lot quicker than the 40mins it took them with a dmm.

[Kennylxix]

As long as the DAC and reference were accurate enough I don't think the DAC would be needed at all but it would give an extra layer of confidence comparing what the ADC thinks its outputting and what the DAC calculates from the output.

[timb]

As long as the DAC and reference were accurate enough I don't think the DAC would be needed at all but it would give an extra layer of confidence comparing what the ADC thinks its outputting and what the DAC calculates from the output.

You really can't easily get a DAC with the required accuracy, INL, DNL etc. to do calibration work without some sort of digital feedback.

Besides, even a 18-bit precision DAC can have quite a large temperature variation even at DC bandwidths, which is why modern calibrators still use ADC feedback.

There's Jim Williams app note where he uses two precision 16-bit DACs (summed with a precision op-amp) and a 24-bit ADC for feedback. There's even a circuit in the appendix for creating a class-AB amplifier out of discrete transistors capable of a +-120V PP output.

I'd start there.

[Moon Winx]

It's kind of funny that after decades of advancements in electronics, we still are left with a thermal converter as the most accurate way to measure an AC signal. NIST has developed a rather mature AC Josephson Standard or JAWS (Josephson Arbitrary Waveform Standard) that is fundamentally accurate and produces waveforms up to 1 V at 100 kHz or so. They don't use it as the official AC standard as of now but that is the goal. It is a quantum-accurate DAC and works by pulsing Josephson junctions with bitstreams.

[amspire]

I wanted to do a project to learn FPGA's and the thing I picked was a PWM sinewave generator with the first harmonic of significant amplitude at something like the 54th harmonic.

Currently I am just working to a 4ns pulse accuracy, but I should get to 1ns or lower accuracy once I sort out a few tricks with the FPGA - I am still a VHDL beginner. The thing about this approach is that as long as you can minimize switching issues, the AC accuracy comes down to the DC accuracy, plus errors that are calculable. I only need something like 40 PWM pulses a cycle, so at 1KHz, there is only a 40kHz pulse rate. Using CMOS switches with a 5ns rise time connected to an accurate DC reference, the errors can be very low. It is only the differences between the rising and falling edges that matter, and they can be observed and compensated for.  When you only have to make a low pass filter to eliminate the 54th harmonic and above, you can make filters with very accurate gain at the output frequency. You are not dependent with the stability of resistors in a DAC at all.

The plan is to also have PWM voltage dividers in the FPGA, that I use as the DC reference for the AC PWM, so I can have a super stable and accurate AC voltage that has an accurately programmable amplitude.

The point of this approach is the AC amplitude is exactly calculable from the DC reference voltage and all the error sources are measurable and can if needed be compensated for.

[chris_11]

I am a bit on a loss why you are interested in a super precise AC calibration of an DMM. 100% - a few ppm of the time it is used to measure mains voltages or 50/60Hz AC, derived with a transformer from the mains.
Mains has several % of harmonics due to all sorts of rectifier and triac style circuits. Look it up with a scope (over a transformer) if interested.
To make a calibration source for that, any stable 50/60 Hz sine generator will do. Simple DAC type with a decent reference or a Wien bridge with decent amplitude control.
If you are interested in the AC meter bandwidth hook it to an signal generator and dial the frequency through. Will be short term stable enough to give you the 3dB point and other surprises.
The RF guys measure in dB. 0.1 dB is in the range of 1% amplitude error.
If you are in the electrical meter calibration business that is a different deal, but for the troubleshooting DMMs anything better than 1% is overkill.

[amspire]

When I started electronics, most people were using 5% accurate analog meters, and they managed pretty well.

Most devices do not need precision of any sort.

But so what? What if, for example, someone wants to measure the stability of an Analog Devices RMS converter, just because they want to find out how stable the reading of a device they made is. The reason might be to work out how many digits they need on the display for their device. To do so, they need something more accurate. If the reference is 10 times better then the AD chip, then you get accurate numbers. If you try and use a 0.1% accurate meter to test an AD chip and see a 0.1% drift, you have absolutely no idea what was drifting.

There is no one explanation that answers your question. 100 different engineers can have 100 different reasons for needing an AC reference.

"Because it is fun" is a really excellent reason.

[Kleinstein]

When using PWM to make an accurate DAC, the logic part made in a FPGA is only the small part to start with. The more tricky part is getting the switches and filter part. TiN had some information an the Fluke 57xx calibrator that uses PWM for setting DC values. It take quite a lot of effort to do that and this is not in the logic part - that is easy today and possible with most µCs.

Today ADCs tend to be more accurate than DACs and thus the more promising way is using an ADC to measure a clean sine. The signal could still be with samples from an DAC and a filter, but amplitude reading would be from the ADC. One trouble with DACs is that they tend to also give higher frequency noise that normally can be filtered, but that filter would also add uncertainty. It is easier with the ADC: here, if the signal is pure, there is no need for an extra anti-aliasing filter. It is also possible to directly read the output - thus no more critical amplifier/buffer behind the DAC.

[chris_11]

I assume that if you pimp the reference a bit in a standard audio USB Stick you get a decent good low harmonic source. They usually have between 16 and 24 bits. The high resolution A/D comes free too. And that with two channels in each direction. The isolation between the channels is limited, because for stereo audio not of much concern. Run both channels in parallel. Would be interesting how good the amplitude accuracy can become, if you improve the reference.

At least I found a gain drift spec from those guys of 20ppm/K

http://www.es.co.th/Schemetic/PDF/AK4395VFP-E2.PDF

That's a starter. This is a typical audio D/A used in the better sound cards.

[amspire]

I have no idea of the design choices of the Fluje 5700A, but all you need for an accurate DC PWM integrator is an resistor and a capacitor followed by a very high impedance unity gain zero offset opamp. You do not need an active integrator circuit. The active integrator introduces a whole swamp of extra possible sources of error. The amplifier/filter for an AC PWM output and a DAC are not much different. It is all manageable of you are not running anywhere near the cutoff point of the filter.

There are parts available today that were not available in the 70s or 80's - whenever the 5700A was designed.

Switches can be as simple as a cmos logic chip. The resistance of the cmos outputs do not matter - only the difference between the high and low mosfet switches. It all can be measured and the errors calculated.

It is easy think that the PWM is somehow too simple, but when you build and test them, they are brilliant.

There are many high end references that rely on DAC and ADC components. I am sure the companies using those parts have spent a huge R&D budget on verifying the performance of those parts, and they probably have a tight QC process to ensure the parts used in manufacture meet then needs.

However, the PWM is simple and does not rely on anything hidden. As I said, all the key parameters are visible and testable. 0.01% accuracy is achievable with very little effort. To get down towards 1ppm, there are always many subtle effects that probably take more effort.

Some of the things I have now are FPGA's which can give me a 1ns pulse accuracy as opposed to microprocessor PWM circuits that may only give a 100ns resolution. I also have many great choices for auto-zero offset, low input current opamps. The PWM frequency is irrelevent - only the short term PWM clock jitter matters.