Affordable Useful AC & DC references

[Majorassburn]

I saw that this explosively growing market had a gap of available, affordable (very low cost) but still useful "ballpark" references to do some function testing of their low cost devices.

Frankly, I don't see that market need. I already have a "ballpark reference", namely the digital multimeter I bought. If I were to spend time and money to obtain another reference on top of that, I would want something a bit more definitive.

"I have always wondered whether I can trust my meter. Nevermore! Now I wonder whether I can trust my cheap homebrew voltage reference."

Funny. But, there are 30+ eBay buyers who would disagree with you. Check my feedback.

[EC8010]

My point was that oscilloscopes are very poor for revealing distortion and that by the time you can see distortion on an oscilloscope (such as flat topped mains), it can be pretty bad, and that some distortions (flat topped mains) cause significantly different measurements between different methods of measurement. Sadly, just switching between a sine wave and a square wave does not reveal all the measurement errors. I found a Rigol oscilloscope that measured cycle RMS acceptably when switched between square and sine, but when I varied the mark/space ratio it got it wrong. Clearly, its internal calculations were wrong, yet it's not hard to calculate cycle RMS perfectly for any waveform provided that your sampled measurements are correct. When I attempted to use it to measure noise it was hopeless - a factor of three out if I remember correctly. To sum up, you need lowish distortion (<0.1%) to be certain that the errors of your generator are insignificant compared to the meter to be tested, and no oscilloscope is good enough to verify that. (Despite their puff, most oscilloscopes are roughly 8 Effective Number Of Bits, and that's because it's good enough.)

By the way, your oscilloscope trace doubled the errors achievable because you only exercised half of its ADC's range, throwing away a bit of resolution. And even as shown, it's not a smooth sine wave - it has visible steps.

There are no excuses in test and measurement. If you want to calibrate even a 3 1/2 digit DMM, then you need something ten times better, so you can't verify your tester with an oscilloscope. You either need your tester to be better by design or verified by something better. A recording quality soundcard is a very good start.

[Majorassburn]

My point was that oscilloscopes are very poor for revealing distortion and that by the time you can see distortion on an oscilloscope (such as flat topped mains), it can be pretty bad, and that some distortions (flat topped mains) cause significantly different measurements between different methods of measurement. <snip>

By the way, your oscilloscope trace doubled the errors achievable because you only exercised half of its ADC's range, throwing away a bit of resolution. And even as shown, it's not a smooth sine wave - it has visible steps.

There are no excuses in test and measurement. If you want to calibrate even a 3 1/2 digit DMM, then you need something ten times better, so you can't verify your tester with an oscilloscope. You either need your tester to be better by design or verified by something better. A recording quality soundcard is a very good start.

Here's some better resolution for you. See any flat tops or bottoms or weirdly shaped curves other than what looks like perfectly shaped sine waves? The "steps" you see are from the digital oscilloscope's low resolution screen recreating & displaying the sine wave. BTW, I didn't quote distortion specs in my ads based upon viewing an oscilloscope screen. I agree, that scopes don't agree nicely with DMM measurements. I just view them as guidelines/approximations.

[shapirus]

Here's some better resolution for you. See any flat tops or bottoms or weirdly shaped curves other than what looks like perfectly shaped sine waves?

Even if you don't see them, it doesn't mean that the wave isn't distorted and there are no harmonics. It looks good, but we still can't say how actually good it is without a proper spectrum analysis.

[ArdWar]

Here an easy test, can you tell whether the bottom waveform is a pure sine wave or not?

[tggzzz]

My point was that oscilloscopes are very poor for revealing distortion and that by the time you can see distortion on an oscilloscope (such as flat topped mains), it can be pretty bad, and that some distortions (flat topped mains) cause significantly different measurements between different methods of measurement. <snip>

By the way, your oscilloscope trace doubled the errors achievable because you only exercised half of its ADC's range, throwing away a bit of resolution. And even as shown, it's not a smooth sine wave - it has visible steps.

There are no excuses in test and measurement. If you want to calibrate even a 3 1/2 digit DMM, then you need something ten times better, so you can't verify your tester with an oscilloscope. You either need your tester to be better by design or verified by something better. A recording quality soundcard is a very good start.

Here's some better resolution for you. See any flat tops or bottoms or weirdly shaped curves other than what looks like perfectly shaped sine waves?

That's as convincing as putting a voltmeter on a battery and asking whether the displayed voltage is wierd.

I agree, that scopes don't agree nicely with DMM measurements. I just view them as guidelines/approximations.

If the instruments are in cal, they should "agree". That's "agree" within the limits of their specification, of course.

Anybody selling calibration tools realluy ought to know that instinctively. It shouldn't need to be said.

[Majorassburn]

My point was that oscilloscopes are very poor for revealing distortion and that by the time you can see distortion on an oscilloscope (such as flat topped mains), it can be pretty bad, and that some distortions (flat topped mains) cause significantly different measurements between different methods of measurement. <snip>

By the way, your oscilloscope trace doubled the errors achievable because you only exercised half of its ADC's range, throwing away a bit of resolution. And even as shown, it's not a smooth sine wave - it has visible steps.

There are no excuses in test and measurement. If you want to calibrate even a 3 1/2 digit DMM, then you need something ten times better, so you can't verify your tester with an oscilloscope. You either need your tester to be better by design or verified by something better. A recording quality soundcard is a very good start.

Here's some better resolution for you. See any flat tops or bottoms or weirdly shaped curves other than what looks like perfectly shaped sine waves?

That's as convincing as putting a voltmeter on a battery and asking whether the displayed voltage is wierd.

I agree, that scopes don't agree nicely with DMM measurements. I just view them as guidelines/approximations.

If the instruments are in cal, they should "agree". That's "agree" within the limits of their specification, of course.

Anybody selling calibration tools realluy ought to know that instinctively. It shouldn't need to be said.

OK, Guys. Like most threads, the helpful responses tend to occur in the first page(s) or so. After that, the latecomers, nay-sayers and perpetual-post-pessimists and nit-pickers seem to weigh in and dredge up all kinds of minutia and picayune reasons that everything discussed prior to their late arrival is either worthless, misstated, or wrought with technical shortcomings. Enough.
If I listened to that kind of unhelpful crap, I'd never produce a product.

I'm appreciative and grateful for the input from those who have reached out with constructive criticism and helpful suggestions, so.....

Here's my preliminary spec for the AC Reference Module:
Adjustable Sine Wave Output 0-6VACrms @ 100Hz, into 10MegOhm Non-capacitive Load, Less Than 1% THD, 6 Month Accuracy 0.5%, 76 Degrees F, 60% Humidity, Free Periodic Re-Cals.

Here's my preliminary spec for the 10VDC Reference Module:
10VDC, Trimmed to 10.0000 +/- 2LSD, into 10MegOhm Non-capacitive Load, 6 Month Accuracy 0.1%, 76 Degrees F, 60% Humidity, Free Periodic Re-Cals, All Additional Specs Per LT1236-10 Data Sheet.

How's that for a start? Suggestions? Questions? Comments? 

Remember, these are $20 or so devices aimed at non-professional folks who own low-priced test equipment and not the DMMCheck crowd.

Thanks in advance.

[bdunham7]

My point was that oscilloscopes are very poor for revealing distortion and that by the time you can see distortion on an oscilloscope (such as flat topped mains), it can be pretty bad, and that some distortions (flat topped mains) cause significantly different measurements between different methods of measurement.

In practical terms, discussing actual devices that might reasonably be used with the sort of reference mentioned here,  I think that distortions not visible on an oscilloscope trace are unlikely to cause significant errors when comparing meters that use a TRMS conversion system.  This is assuming a fundamental of 100Hz as shown.  Things would change if one of the meters was a reasonably accurate average-responding system.  Unfortunately I don't have time or bench space to spare at the moment, but a simple experiment with an AWG could show this quite easily.

[bdunham7]

If I listened to that kind of unhelpful crap, I'd never produce a product.

Here's my preliminary spec for the AC Reference Module:
Adjustable Sine Wave Output 0-6VACrms @ 100Hz, into 10MegOhm Non-capacitive Load, Less Than 1% THD, 6 Month Accuracy 0.5%, 76 Degrees F, 60% Humidity, Free Periodic Re-Cals.

You should be able to put the "unhelpful crap" into perspective with mathematical arguments or experimental demonstrations.  For example, how much difference between TRMS meters might you see with a signal that has 1% THD?  (probably not much) Also, which harmonics are present in your signal?  (It matters, possibly a lot) 

As for your AC specs, I would observe that 10M "non-capacitive" is probably a grossly insufficent spec for an AC source since many meters have much lower impedances, like 1M + 150pF.  You'd need to know your output source impedance (it's probably much, much lower than 1M) and figure out what errors you might have with various loads.  You might also want to explain how you're substantiating the 1% THD spec and the 6-month accuracy spec.  If the answer is "it's just my best guess, but hey they're cheap", well I suppose that'll have to do.

[Majorassburn]

My point was that oscilloscopes are very poor for revealing distortion and that by the time you can see distortion on an oscilloscope (such as flat topped mains), it can be pretty bad, and that some distortions (flat topped mains) cause significantly different measurements between different methods of measurement.

In practical terms, discussing actual devices that might reasonably be used with the sort of reference mentioned here,  I think that distortions not visible on an oscilloscope trace are unlikely to cause significant errors when comparing meters that use a TRMS conversion system.  This is assuming a fundamental of 100Hz as shown.  Things would change if one of the meters was a reasonably accurate average-responding system.  Unfortunately I don't have time or bench space to spare at the moment, but a simple experiment with an AWG could show this quite easily.

Thank You for putting this distortion discussion in such practical terms. You nailed it.

[Majorassburn]

If I listened to that kind of unhelpful crap, I'd never produce a product.

Here's my preliminary spec for the AC Reference Module:
Adjustable Sine Wave Output 0-6VACrms @ 100Hz, into 10MegOhm Non-capacitive Load, Less Than 1% THD, 6 Month Accuracy 0.5%, 76 Degrees F, 60% Humidity, Free Periodic Re-Cals.

You should be able to put the "unhelpful crap" into perspective with mathematical arguments or experimental demonstrations.  For example, how much difference between TRMS meters might you see with a signal that has 1% THD?  (probably not much) Also, which harmonics are present in your signal?  (It matters, possibly a lot) 

As for your AC specs, I would observe that 10M "non-capacitive" is probably a grossly insufficent spec for an AC source since many meters have much lower impedances, like 1M + 150pF.  You'd need to know your output source impedance (it's probably much, much lower than 1M) and figure out what errors you might have with various loads.  You might also want to explain how you're substantiating the 1% THD spec and the 6-month accuracy spec.  If the answer is "it's just my best guess, but hey they're cheap", well I suppose that'll have to do.

Excellent points.
1) Thank You. My source impedance is an Op-Amp 20mA short-circuit protected output, 10uF capacitor-coupled to a 10KOhm, 20-turn Cermet pot to Ground, center wiper tap. So, plenty of juice to play with at 1megOhm. I will change all of my specs to "into a 1MegOhm + 150pF Load" because it is also inclusive of most handheld and bench DMM's as well as scopes.
2)  Thank You Again. The 6-month accuracy spec is by actual testing units for almost 2 years. In fact, the tested units are actually well within a 0.25% spec. since power-cycled and constantly powered during that period. Of course, they vary with ambient temp so it is hard to be more exact and that's why I opened it up to 0.5%.

As for the 1% THD, that's my experimenter's best guess, having studied many app notes and circuitry specs.. I have a hunch that it might be much lower than 1%. But for what these references are designed to do, 1% is plenty adequate. I'll let someone with a $10,000 Spectrum Analyzer prove me wrong. 

Your last comment about the harmonics has me confused. With substantial output low-pass filtering built in to my modules, I didn't think harmonic content would matter much based upon the suppression provided by the LP filtering.    Comments?

[CatalinaWOW]

I actually agree there is a need for the type of product you are trying to provide.

Here is a relatively off the wall suggestion.  There has been much discussion of the cheap Chinese references.  Several people have found them both stable (I know that requires definition, but remember the thread we are on) and with outputs matching the attached sticker.  I am one of those.  But there is clearly wide variation in the product as indicated by a large number of other reports.

Some of the negative comments against these references are actually not well founded.  A used LM399 is not necessarily a pejorative, but an indication of a well aged part.

It seems there might be a business opportunity in providing a screening service for these devices.  With your existing equipment you could screen them to hobby levels of precision.  Whether this would be profitable would depend on two factors.  First, the yield of "good" devices.  And second, the marketability of the screened devices given the reputation these things have developed.  The latter would be positively affected if you provided a traceability trail for your results.   This doesn't necessarily involve a formal calibration, but would require you to analyze the tests you have performed to "confidence test" your equipment and provide a justifiable set of error bars on your results.  This analysis would be beneficial to you regardless of whether you take on this screening activity.

[bdunham7]

I have a hunch that it might be much lower than 1%. But for what these references are designed to do, 1% is plenty adequate. I'll let someone with a $10,000 Spectrum Analyzer prove me wrong. 

Your last comment about the harmonics has me confused. With substantial output low-pass filtering built in to my modules, I didn't think harmonic content would matter much based upon the suppression provided by the LP filtering.    Comments?

It wouldn't be that difficult to measure your THD to the levels you need with a PC soundcard input or a decent oscilloscope with FFT.  No SA needed and in fact most won't even work for this purpose.  I wouldn't take a victory lap yet because I would think that a good portion of your customers might actually have average-responding meters where even the lower harmonics matter.

The harmonic content matters because some TRMS meters have very low BW--1kHz or even less in some cases--and if a substantial portion of your THD is out-of-BW harmonics, then the resulting error will be larger.  Put another way, I'd expect different TRMS converters to respond to distortion that is within their specified BW in much the same way, but if the distortion is out-of-BW for one of them then they would differ.

[Kleinstein]

I don't see the distortion that critical. The accuracy of the cheap meters is not that high and they should still respond reasonable with a slightly distorted sine. The BW is usually up to at least 1 kHz - just the specs may not be for a higher frequency. The upper limit for the specs is different from a -3 dB bandwidth limit.  Part of the limitation can be a slew rate limit and thus not an issue for moderate low harmonic content, but a big issue with the square wave.

It would still make sense to check the distortion to have some relible specs, e.g. with a sound card. If one wants to use the scope, that in combination with a notch filter to suppress the base frequency. This is the old style method.

Driving the output directly from an OP-amp can be an issue. The current is sufficient, but op-amps don't like driving capacitive loads. E.g.  150 pF (1.5  m of cable or a DMM input)  can be enough to make an TL07x wired as a buffer oscillate. Even those OP-amps that are specified to drive quite some capacitance are not not really happy about capacitive loading.

[Majorassburn]

I don't see the distortion that critical. 
<SNIPPED>
Driving the output directly from an OP-amp can be an issue. The current is sufficient, but op-amps don't like driving capacitive loads. E.g.  150 pF (1.5  m of cable or a DMM input)  can be enough to make an TL07x wired as a buffer oscillate. Even those OP-amps that are specified to drive quite some capacitance are not not really happy about capacitive loading.

Thank you for your input.  Attached is a simplified schematic showing the output driving a capacitor. No oscillations have been observed in prototypes yet. Comments/suggestions?
Also attached is a TI Note regarding Op Amp compensation for driving capacitive loads. Such a method doesn't seem necessary at my low frequency unless I'm missing something?

[miro123]

I am a little bit confused.
1. From one side , the TI application note clearly defined two components - Cf and Riso. But I don't see it them back in your circuit.

2. Another issue with your circuit is output potmeter creates two problem - your output impedance extremely high for any calibration instruments.
What is your attenuation driving nominal specified capacitive load ? And what about specified ppms?
3. What is the transfer function vs temperature oof LPF - 5,1K /1u. How this applies to AC accuracy tests
4. What about the polarity of capacitors.
5. What about stability of  initially set of potentiometer ratio?

Can you provide the specified accuracy spec within 6months? - if you take into account my remarks 2, 3, 4 and 5

[Kleinstein]

In the last circuit the ouput is not directly driven by the OP-amp. Except for 100% of the amplitude the pot at the output will isolate the external load from the OP-amp. However the resistance from the pot will also add to the output impedance and the resistance is thus depending on the amplitude setting.  This is not good at all. In AC mode some DMMs are one 1 M input resistance and there upt to 5 K of output resistance would be a detectable error of up to 0.5%.

The more normal way for AC outputs is to have a 50 ohm output impedance from a resistor after the amplifier. 50 ohms would be OK with a 1 M input impedance meter, except for the high end and if constant one can still correct for this.
The need to isolate the amplifier from capacitive load does not depend on the signal frequency, but the choice (faster one tend to be more sensitive) of the amplifier.

The oscillator is a square wave followed by a relatively simple low pass filter. With LM358 OP-amps the amplitude may not be very stable, as the ouput swing of the OP-amp is not very stable (e.g. it can depend on the temperature). Chances are also that the -12 V part may not be that stable. The LM358 also tends to produce quite some cross over distortion - often already visible at 100 Hz.

[Alex Nikitin]

As much as I like to do things in a purely old fashioned analogue way, an accurate, reasonably stable and inexpensive AC source is best to be made by digital synthesis with a microcontroller and a DAC. There are microcontroller chips made with internal DACs, band-gap references and even look-up tables for sinewave generation. All you need to add is an output buffer/filter and you'll get a reference AC source with better than 0.1% accuracy and stabiltiy.

Cheers

Alex

[Majorassburn]

As much as I like to do things in a purely old fashioned analogue way, an accurate, reasonably stable and inexpensive AC source is best to be made by digital synthesis with a microcontroller and a DAC. There are microcontroller chips made with internal DACs, band-gap references and even look-up tables for sinewave generation. All you need to add is an output buffer/filter and you'll get a reference AC source with better than 0.1% accuracy and stabiltiy.

Cheers

Alex

Can you suggest a few part numbers for us to look into? Thanks.

[Majorassburn]

I love this particular Metrology forum because I'm in the company of many highly talented electrical engineers and advanced hobbyists. Therefore, I learn.

However, I'd like to re-post here what I started out saying when I opened this thread:

"My intent is to make and sell very low-priced but USEFUL references for those of us who are not yet ready to resort to costly lab calibration. We may want to be able to perform basic, "ballpark" function checks of our DMM's to see if they're even working correctly or reasonably holding calibration.  And, we can't justify spending $150 or more for higher quality devices like the DMMCheck+ and others.

AC Voltage Reference:  Here's a device that generates a relatively pure, NON-FLUCTUATING, adjustable sine wave from 0 to 6VACrms @ 100Hz (fixed frequency) for those who do not have a quality signal generator or alternative. (100Hz was chosen as a DMM-friendly "sweet spot" for avoiding EMI-influence in the working environment.)"

I'm mainly a marketing guy. I realize that engineers and dedicated electronic theoreticians may not appreciate that but here's my philosophy regarding my original intent in this thread:

!)  A low priced electronic product's salability decreases proportionately to the degree of electronic quality and design sophistication built into it.

What I mean by that is, if the goddamned thing works well enough to satisfy the target market that it's aimed at, build it and sell it!  Continued attempts at perfecting it will lead to fewer and fewer sales! 

2)  You've all heard the old phrase "Build a better mousetrap and the world will beat a path to your door."

Well, I'm not building mousetraps and I don't want the world at my door, either! There are enough bill collectors and tax men there already! 

I'm trying to offer products that are affordable and useful and, by virtue of those two objectives, are necessarily filled with shortcomings of many kinds. My stuff will never appeal to knowledgeable, experienced and talented electrical engineers because those professionals exist on a much higher plane.

But, for the unsophisticated hobbyist, student, homeowner, DIYer and "electronically curious", my stuff is attractive and provides solid value to them. 

Just look at Chinese electronics today. They are, in some cases, vastly superior to American made products (the dwindling few) and are usually much less expensive to acquire. Are they perfect? Hell, no! BUT, that doesn't stop people from buying them in mass quantities, does it. 

Compare that to Keysight, Fluke, Tektronix, etc. and other truly high-quality brands and you will see that these superior brands and products are much LESS salable to the masses because they are truly outstanding in design, capability, accuracy, durability, etc., and MORE EXPENSIVE, right?

3)  If you want to sell a lot of stuff, you need to have just about the lowest selling price of any product in that category or very few people are going to buy it! That's an indisputable fact of consumer behavior in today's world, like it or not. Just look at stuff on eBay for proof!

And, the more you try to educate and inform the masses that your product is technically better by design, the more likely they will tune you out quickly and buy something that is presented more simply and is cheaper!  Today's low end buyers do not want to be confused by FACTS! 

Okay, I got that off my chest, so....

I really appreciate everybody's input to my various threads and the help that provides me as I try to balance my offering a low-cost alternative to the other calibration-checkers out there.

If anyone has more suggestions or input, please continue to weigh in with an eye towards what I reiterated above.   
Thanks again.
Major

[tggzzz]

...
I'm mainly a marketing guy. I realize that engineers and dedicated electronic theoreticians may not appreciate that but here's my philosophy regarding my original intent in this thread:

Several people have, belatedly, begun to smell that. It would have helped everybody if you had made that clear earlier.

What I mean by that is, if the goddamned thing works well enough to satisfy the target market that it's aimed at, build it and sell it! 

A key part of marketing is to give sufficient information to determine what a product won't do.

It would have helped if you did that explicitly, rather than leaving it to people to ask questions and make inferences.

We've all seen claims/patents for better mouse traps and apple corer/peelers. Almost all aren't any such thing.

[artag]

not to mention professional appearance (especially if you pay a little extra for black solder mask lol), which may improve sales.

I *despise* black solder mask. I want to see where the traces go, even if I'm not repairing it, or have the plots available. I'd regard it as 'wannabe-cool' which is most definitely not the same as 'professional'.

[shapirus]

I *despise* black solder mask. I want to see where the traces go, even if I'm not repairing it, or have the plots available. I'd regard it as 'wannabe-cool' which is most definitely not the same as 'professional'.

the remark regarding the black solder mask was obviously a (half) joke.

[Majorassburn]

...
I'm mainly a marketing guy. I realize that engineers and dedicated electronic theoreticians may not appreciate that but here's my philosophy regarding my original intent in this thread:

Several people have, belatedly, begun to smell that. It would have helped everybody if you had made that clear earlier.

What I mean by that is, if the goddamned thing works well enough to satisfy the target market that it's aimed at, build it and sell it! 

A key part of marketing is to give sufficient information to determine what a product won't do.

It would have helped if you did that explicitly, rather than leaving it to people to ask questions and make inferences.

We've all seen claims/patents for better mouse traps and apple corer/peelers. Almost all aren't any such thing.

Ya know, I've never been a fan of your arrogant, self-serving, hyper-critical posts in my own, and several other threads. Guys like you NEVER have anything good to say unless you're touting your own imagined success so, your remarks and criticisms and admitted paranoid suspicions as to where I'm coming from are all wasted hot air on me.

I've noted previously how your baselessly superior attitude tends to embarrass and chase away those participants who you deem to be less learned than you and, therefore, not worthy of your respect to at least be courteous toward their requests for help in some of your replies.

Do me a favor and put me on your ignore list and I'll do the same for you. That way, other participants won't have to witness your desperate attempts to throw a wet blanket on any more of my posts.

EDIT: deleted an asinine reference I made to his homeland. Sorry. Temper got the best of me for a moment.

[mawyatt]

Here's quick drawn schematic of a simple Wein-Bridge 100Hz (Wien or Wein) Oscillator using a simple gain limiting for amplitude stability, followed by 2nd Order Active LPF to reduce harmonic content. Bridge Oscillator should produce a respectable sine-wave output if not forced into excessive limiting, so LPF may not be required.

BOM for just electronic components should be between $0.34 and $1.56, so should suffice as a Cheap Sine-Wave AC source.

Anyway, we haven't built this, someone give it a try and would expect this to outperform the schematic shown earlier. If anyone builds this, please report back results.

In the meantime we are considering a very stable and accurate Sine-wave source (another completely different design) that has somewhat performance guaranteed by design rather than direct calibration (performance verified by indirect means). If interested (serious) contact us, we may move forward with this.

Edit 100Hz not 100KHz
 
Best,