'Affordable' Reference Standards to Have On-Hand in a Electronics Lab

[fmashockie]

Hello All! This is my first time posting in the metrology forum.  I apologize if this should belong in the 'Beginners' forum.  But anyway, I guess I have a 2 part question:

1)  I found an old Analogic AN3100 DC calibrator on eBay for cheap.  I figured it could be useful as a precision DC source for calibrating my benchtop meters and power analyzer.  But considering it has not been calibrated in a long time, it will likely require calibration to be of any use.  I attached a screengrab from the manual of needed reference materials.  Haven't been able to find much info on them.  Does anyone know of any alternatives I could look for that might be readily available?

2) In general, I've been looking for precision reference standards for voltage (AC and DC) and resistance.  But everything is very expensive! Is there anything out there that is accurate, but relatively affordable that I could look for to have on-hand in my repair lab?  I'd be interested in DIY kits/solutions as well!

Thanks in advance!!

[Nanitamuscen]

Think about what class of accuracy you need a source. There will be further advice from this.
And what budget do you have for it.

[fmashockie]

Think about what class of accuracy you need a source. There will be further advice from this.
And what budget do you have for it.

Thank you for your reply! So I guess for simplicity, I'll start with the Analogic AN3100.  I've attached the full service manual.  Accuracy is +-50uV.  There is a calibration procedure in the manual.  And it has suggested materials for calibration.  But these do not seem to be available anymore.  Can you provide some alternatives that would work for this calibration procedure?  I'm looking to keep budget <$500.

Thanks!

[Nanitamuscen]

I think you can consider Yokogawa calibrators. Both old versions and more modern ones.
You can also visit the website of the respected Ian Scott Johnston. This person produces small batches of very good calibrators.

[bdunham7]

Thank you for your reply! So I guess for simplicity, I'll start with the Analogic AN3100.  I've attached the full service manual.  Accuracy is +-50uV.  There is a calibration procedure in the manual.  And it has suggested materials for calibration.  But these do not seem to be available anymore.  Can you provide some alternatives that would work for this calibration procedure?  I'm looking to keep budget <$500.

Since you've apparently already acquired it, it would be more useful to consider what you can do with it.  You can verify some of the basic functions of a typical 4.5-digit multimeter, which is a good handheld or low-end bench meter.  You can also use the accurate and stable output for other things, but since we have no idea what you are wanting to do, its hard to know what use the device might be to you.

The device and manual are old enough that the calibration procedures and standards are thorougly obsolete.  The Julie Research items listed in the manual might be found in a museum, but not likely on a metrologists bench.  The way this would be done nowadays would be to simply use a good 6.5-digit DMM (or better) with a recent calibration to verify and adjust if needed.  A cal lab would likely just use their 3458A so as not to have any issues with TUR/TAR or guardbanding.  Professional calibration would probably be pretty cost-ineffective.

Metrology is an expensive  and time consuming hobby!  The best thing I can think of for you to do at or near your budget is to round up a decent condition HP/Agilent 34401A. 

[alm]

I'd say there are three important, somewhat independent, concepts here:

• A stable reference
• The ability to transfer to different ranges / functions
• Traceability to an absolute reference (e.g. NIST)

For DC voltage, the common solution to stability is having a few 10V or ~7V voltage references. On the low end of the spectrum you might have the cheap Chinese AD584 references, then higher you might have AD587/AD588/LT1021-based references (e.g. products by Voltagestandard.com and DMMCheckPlus (previously made by Voltagestandard.com)), and basically the sky is the limit after that. In some cases buying a reference gets you some degree of traceability (Voltagestandard.com and DMMCheckPlus.com will ship you a piece of paper with a presumably traceable measurement), but this doesn't have to be linked. You could build a few references yourself, compare them over time (this will tell you relative drift between the references), and then get one of them compared to a reference with a known value. For example look at the USA Cal Club Round 3 topic.

To transfer to different ranges, I would look into a Kelvin Varley Divider (like Fluke 720A, ESI RV722A or the Julie Research VD106) or a Hamon divider (like Fluke 752A) and a null meter. A DMM could make a null meter if the bias current is low enough (you can measure this), but in general an analog dial is easier to use if you're adjusting trimmers until it reads 0. A good explanation and instructions on how to build some of these is Conrad Hoffman's Mini Metrology lab. The trick here is that the voltage dividers (the Fluke 720A or Hamon divider) can be self-adjusted so they provide an accurate ratio without any external validation. Same with the null meter. This makes your DCV reference the only thing that needs external validation.

AC is much more difficult than DC. There are some solutions that transfer an AC value to a DC value, like Thermal Voltage Convertors or the old Fluke 540B. But they themselves need calibration. And to calibrate many multi meters you need a variety of voltages and frequencies, so you need an ACV calibrators (which are generally expensive boat anchors like the Fluke 5200A or Datron 4200A) in addition to the AC-to-DC transfer. So I'd suggest starting off with DCV and forgetting about ACV for the moment.

[fmashockie]

Wow thank you so much alm.  I really like the idea of trying to build some of these myself.  You gave me a few resources to get started with that! 

Thank you to everyone else as well! I have heard of Ian Scott Johnston's precision voltage reference.  They seem pretty affordable, too.

[dmmguy]

Hi, maybe you can begin with these if you only want to test your DMMs

https://www.ebay.ca/itm/155763974742

https://www.ebay.ca/itm/155751553394

[Alex Nikitin]

The Analogic you have now is a potentially very nice voltage source and to re-calibrate it you only need an access to a freshly calibrated 6.5 digits or better voltmeter or, at the very least, a calibrated 10V reference + a sensitive voltmeter (a "null-meter) which could resolve at least 1uV. . The unit is however very old and most likely needs some TLC, possibly new electrolytic caps. The most precious parts of this unit are the resistors set and the switches. If these are performing well and not damaged, you should be able to restore (or even improve) this reference to be a very useful tool.

Cheers

Alex

[tridac]

Analogic kit is well known for quality and references like that are specifically designed to have long term stability and accuracy, even if not calibrated for years. If that's the only reference you have, and providing it works and the outputs make sense across the range, check against the best dvm you have, then it's already probably the most accurate thing you have in your lab, at least until you have something to compare it to. More than adequate to calibrate analog meters and probably dvm up to 4.5 digits. A good system dvm is essential. Older Hp models like the 3456, 6.5 digit, are an affordable option and are so old now, the aging rate is very low. Have a couple in the lab here, neither formally calibrated in years, but track together within a couple of microvolts once warmed up and are within a handful agaist a Fluke dc standard. The 3457 are 6.5 digits via front panel,  7.5 via hpib commands, but from tests, not convinced they are much better than the 3456. Afaics, Fluke make the best voltage standards, while HP make the best voltmeters. There also low cost options, such as online sellers offering precalibrated fixed output references. Quality might vary, but some of them look pretty good and could be usefull for an initial reference to compare against. A handful of microvolts is more than good enough for many. Any better than that and thermal ambient, lead and connector variations start to affect results. If you think you need better than that, then there should be good reason, as it will be an expensive journey. If the work doesn't care if the measurement is miliivolts out, then it's a waste of money to pay for single digit microvolt accuracy, however much we all strive for the best...

[all_repair]

Check this:
https://www.ianjohnston.com/index.php/onlineshop/handheld-precision-digital-voltage-source-2-mini-detail

No relationship with maker, I wish I have one.

[Kean]

Check this:
https://www.ianjohnston.com/index.php/onlineshop/handheld-precision-digital-voltage-source-2-mini-detail

No relationship with maker, I wish I have one.

I am very happy with my PDVS2mini.  Highly recommended if you can afford it... and if it is in stock...

[Grandchuck]

I also highly recommend the PDVS2mini.

[DH7DN]

ADRmu as a ~10 V DC voltage reference seems to be rock-solid (as far as I can tell, still testing...)
Build your own, download its cooking recipe here: https://github.com/marcoreps/ADRmu

The BOM price is at $350 - $400 per unit and you need at least 3 of them to gain the voltnut-confidence in stability. Monitoring of voltage references demands good, maintained and calibrated equipment under defined environmental conditions so there go few k$. Maintaining traceability to national standards can cost somewhere from $5 to several hundreds/thousands of dollars (every year). Expensive hobby, wouldn't recommend to dive into the 8.5 digit DC Volts metrology. But then again, it's a hobby and hobbies are meant to burn money for the lulz 

Besides that, a German company called "Knick" has manufactured few capable Direct Volts and Direct Current calibrators which are sold relatively cheap.

[CatalinaWOW]

First question is why you want to calibrate?  Any answer is OK, even "just cuz".  But it has to be YOUR answer and will guide what you do and spend in this direction.

[tggzzz]

First question is why you want to calibrate?  Any answer is OK, even "just cuz".  But it has to be YOUR answer and will guide what you do and spend in this direction.

Precisely. I hope the OP reads that and formulates their answer.

[J-R]

I'm no metrology expert, but I've been dabbling a bit over the last couple years and perhaps was in a similar situation back then when needing to calibrate/adjust some DMMs I repaired.

I'd think having a PDVS2mini would negate the need for the Analogic, at least from a practical perspective.  But there is the cost and stock issue.  It could be 6 months or more before they are back in stock, although now that Ian retired, maybe he'll be faster to get the next batch out!  1,000% recommended, though.

The VoltageStandard.com PentaRef seems like the bare minimum to check the Analogic, but it is also out of stock currently.  1V, 2.5V, 5V, 7.5V & 10V for the five outputs could be one option.  You would want to check the Analogic right after receiving the reference for maximum accuracy.  The Analogic has only one official adjustment, 10V.  The zero and 1V are just for verification to ensure the unit doesn't need repair.

The "hack" to make this possible is to connect the Analogic and the Pentaref in series with a DMM and connect the outputs negative to negative OR positive to positive, then measure the difference with the DMM using the mV range.  A decent handheld or bench meter should be able to handle this.  Something with at least 0.01mV of resolution.  Then you don't need a 6.5 digit DMM just yet.

[fmashockie]

Thank you for everyone's recommendations and input!  I just received the Analogic AN3100 yesterday and it looks great! Appears to work well, too! Though I only have my DMM at home right now to test and it's only a 4 sig fig meter.  I do have a HP3478A in the lab (5.5 digit) that I can use to check it, but it is not calibrated. 

The reason I want a precision DC voltage reference (and other standards for resistance and AC voltage) is that I work as a engineer repairing my company's lab equipment (biotech).  We have a very small budget for the test equipment I need to do this.  So I buy a lot of it used for cheap.  I thought it would be helpful to have calibration standards in case I need to get something like a benchtop meter, power analyzer, etc. back up and running.  We are not under any strict regulations here like GMP/GLP for example, but we will be getting ISO 9001 accredited soon.  My electrical testing equipment won't necessarily fall under the scrutiny of that accreditation (some of my other standards will), but I thought it would be a good idea to have an ability to calibrate them if they ever do. 

[J-R]

If you have a slight attraction to older gear, then the Fluke 515A might also be something to keep an eye out for.  I picked one up a while back and it's a fun gadget to have if you can't handle one of the boat anchor Fluke calibrators!

But based on what you are saying you need for work, I think the proper solution is a calibrated bench DMM.  Trying to dig up all this old equipment in order to get enough references is going to be higher risk, more money and more work than just shipping out a bench DMM every year or so.  Which DMM to get is another discussion.

[trobbins]

A cheaper and more social option may be to say hullo to nearby service/manufacturing centres that may have a 6.5 digit DMM they keep in cal.  Invite yourself over for a coffee and chat, and ask if you can take your 5.5 digit, and perhaps your AN3100, to make some spot tests.  Not a calibration, but may give you the confidence to proceed on, and may allow a future opportunity to do another set of spot tests with acclimatized equipment.

[CatalinaWOW]

Not cheap, but certainly not more expensive than the approach you are chasing if you value your time appropriately is to periodically rent a calibrated instrument.  This is probably easier to properly document for your ISO and other certifications.

[Rax]

Excellent input. I am not so far from where you are, and this community has been extremely helpful and the wealth of knowledge and generosity of sharing is unmatched.

You may want to investigate - possibly from this community - if there are any turbo- hobbyists with some good standards at hand. They may help you assess, and possibly adjust yours.
Also, in the educational side of things, Fluke has an excellent calibration manual available for download. It's based on older standards and equipment, but it will teach you a ton on the fundamentals of practicing this.

[EC8010]

I would second the Fluke "Calibration: Philosophy in Practice" recommendation. I downloaded the first edition after I borrowed the hardback 2nd edition from a chap who had been issued with it on a Fluke calibration course. In truth, the principles are unchanged between editions, but I like paper books and it is very good in either edition. I tried to buy a copy of the 2nd edition but it didn't seem to be for sale, so I emailed Fluke from work asking where I could buy a copy. Their very courteous response was to ask for postal address and they sent me a copy FOC. I quite often refer to the Fluke book.

[huababua]

What do you guys think about the Time Electronics 1017 DC V/I/R Multifunction Calibrator?
https://www.timeelectronics.com/portable-voltage-current-instruments/1017-voltage-current-resistance-calibrator/


I know that this is not a "standard" but a "calibrator" (whats the difference anyway? A standard also needs calibration, does it?)
But looking at the specs it seems to me that this could be a "cheap" standard-option?

[PartialDischarge]

Have a look at the Advantest TR6142, schematic is available on ko4bb

[alm]

What do you guys think about the Time Electronics 1017 DC V/I/R Multifunction Calibrator?
https://www.timeelectronics.com/portable-voltage-current-instruments/1017-voltage-current-resistance-calibrator/


I know that this is not a "standard" but a "calibrator" (whats the difference anyway? A standard also needs calibration, does it?)
But looking at the specs it seems to me that this could be a "cheap" standard-option?

It depends on what you need in terms of accuracy and values. Keep in mind that the instrument has only a single resistance and current range, so if you are near the bottom of the range accuracy and resolution won't be so great. For example resistance accuracy is 0.05% of full scale, so setting it to 10 Ohm would have an uncertainty of 5 Ohms. Stability and resolution will be better than this. So if had another meter, you could adjust it so it read 5 Ohm +/- 0.01 Ohm on that meter.

I'd say a standard in general has a single or a few values, like the AD584 voltage standards that can output 2.5/5/7.5/10V or the DMMCheckPlus. And is focused on stability rather than flexibility. While a calibrator while usually be able to produce a larger range of values, like this example which can range DC voltages, currents and resistances through quite a range. In a cal lab, you would have calibrators doing the daily calibration of all the devices that come in, and use the standards once in a while to verify and/or adjust the calibrators.

Both need calibration and you could use a standard for calibration (if it happens to have the value you need), and you could use a calibrator as standard. It's just that calibrators are often larger and heavier so less practical to ship around than a small voltage or resistance standard. It's also more affordable in terms of space and money to have multiple voltage standards for comparison rather than multiple calibrators.

[Majorassburn]

Hi. I just joined. I make precision dc references, ac sine wave voltage references, precision resistance references & small linear power supplies especially for DIY calibration needs. If you want to correspond, my email is:  majorassburn@gmail.com and I sell on eBay as SQWARREL. Here's a link or two to some of my listed items:  https://www.ebay.com/itm/285497640527 and https://www.ebay.com/itm/285489340860  and  https://www.ebay.com/itm/285491468993
Please accept my apologies for my amateurish products. I am making them for hobbyists and do-it-yourselfers who want to verify their under-$100 digital multimeters (which I own several ), although I also own a few Fluke 87V's with cal certs.
I have always been fascinated by the fact that no matter how much you are willing to spend on superlative cal equipment, none ever meet your highest expectations! 
I love this forum because it is a VAST source of useful information!

[J-R]

I bought an Advantest R6144 a while back and despite having to beat all the covers back into shape it's quite nice to have around.

I still say having a pile of calibrated references/standards alone is a dead end from a practical perspective.  Then there is the cost of having all the references/standards calibrated.  So the best solution by far is to have one or two of your best DMMs shipped out for calibration, then use those along with non-calibrated sources to check the lesser equipment.

Even a common exception, such as having equipment that only needs a specific calibrated reference becomes questionable once you consider how limited it is along with the calibration cost-effectiveness.

HOWEVER, in the context of a hobby, then of course those things do not matter since we are having fun and doing what we want...


With regard to this recent post of two eBay "references", I unfortunately have many concerns with them:
- general issue with having to buy many references as mentioned above
- these specific references are not really precise enough to be considered appropriate for calibration purposes
- stability is not mentioned
- multiple problems with the general construction of the references, such as no enclosures or board stand-offs, awkward connection points and use of universal proto-boards rather than a manufactured PCB
- apparently the maker only has a calibrated Fluke 87V which is not really good enough for providing calibration data
- future calibration services are not mentioned

[alm]

I still say having a pile of calibrated references/standards alone is a dead end from a practical perspective.  Then there is the cost of having all the references/standards calibrated.  So the best solution by far is to have one or two of your best DMMs shipped out for calibration, then use those along with non-calibrated sources to check the lesser equipment.

I don't disagree but would like to add the following side notes. Having discrete standards like a 10 VDC standard is most useful if you have a way to transfer this 10 V DC to other values like 100 V, 1 V, 100 mV etc. This is how cal labs used to do it (look at Fluke 7105 which used just a 1.018V Weston Cell to calibrate any DCV value from 10 mV to 1000V. This is still how standards labs do it since their primary standards like Josephson Voltage Standards can't generate all voltages they need. So you would only calibrate a single 10 VDC standard. You might want to have multiple so you can estimate how much they're drifting, but you can have one calibrated and compare the others to that one. Same with resistance. Like I wrote, AC is more problematic. So your pile of standards might boil down to 4x 10 VDC and 4x 10 kOhm.

Obviously this is a lot more work than sending a DMM for calibration. It depends on if you have more time or money to spend. Buying / making a standard and tools like a voltage divider can be cheaper, and the voltage standard can be calibrated using a cal club or a budy with a calibrated DMM. Having your DMM calibrated by a buddy however depends on having another calibrated device (like another DMM) and signals for all the ranges (e. g. 100 mV-1000V DC, 100 Ohm - 10 MOhm, etc). So that for most people requires sending out the meter every year or so, incurring a recurring cost. But obviously if you're a professional needing uncertainties down to say 20 ppm, this is the way to go.

[Majorassburn]

Hi & Thanks for the valuable feedback. I agree with all your comments about my eBay listed AC & Resistance calibrator modules.

I don't make excuses when constructively criticized but I do like to clarify as to the limitations that you have accurately pointed out.

"With regard to this recent post of two eBay "references", I unfortunately have many concerns with them:
- general issue with having to buy many references as mentioned above
- these specific references are not really precise enough to be considered appropriate for calibration purposes
- stability is not mentioned
- multiple problems with the general construction of the references, such as no enclosures or board stand-offs, awkward connection points and use of universal proto-boards rather than a manufactured PCB
- apparently the maker only has a calibrated Fluke 87V which is not really good enough for providing calibration data
- future calibration services are not mentioned"

As I said in my initial post, my eBay devices are designed for the under-$100 DMM owner who wants to verify the readings of his/her Chinese cheapie DMM.  Such DMM's are usually manufactured to be self-calibrating and non-adjustable so, verification is the only owner option for this ubiquitous category of cheap DMM's.

All of my eBay listings are intentionally vague on specifications so that I don't mislead buyers into thinking that they're going to receive a laboratory-quality or NIST-traceable device. These are really just hobbyist-grade but very useful and VERY accurate. 

Here's an example using the AC Sine Wave module: After 15-minute warmup, in a 78 degree, non-drafty environment, attached to the matched Power Supply module, the AC Sine Wave output will stabilize at 6.000VACrms, +/-2mV and hold that voltage for several hours. I think that's pretty good for a $39 Sine Wave generator with a 6Vrms, low-impedance output, huh?

And, it certainly will indicate whether your cheapie DMM is working properly at low AC ranges where most such DMM's self-calibrate. My instructions clearly state that a "good reading" on a 0-6VAC range doesn't necessarily equate to a valid reading at hgher and dangerous AC ranges.  But, the AC Calibrator does provide a useful, stable, non-fluctuating Sine Wave that is so necessary to be able to reasonably accurately READ a DMM's AC display, right?

To keep selling costs at a minimum, I offer these modules simply as modules to the DIY'er who are familiar with what they can do with modules and not as finished products which sophisticated calibration freaks like myself would be attracted to!    Therefore, I sell hand-wired perf boards, no standoffs and clunky header pin connection points. For the module DIY'er, these are not deal-breakers.

And, yes, I love and use my calibrated 87V's. They certainly speed up breadboarding custom designs and module revisions. My $1,000 calibrated Keithley 2110 5.5 Digit Multimeter with Dual Display just keeps watch over my Fluke Fleet. ( like that: Fluke-Fleet!  )  I do use the K to calibrate all my amateurish products before they're listed, though.

Finally, there is no mention of re-cal of any of my modules because of their intended use. But, if anyone who buys one wants to re-cal, I would do it for free as long as they pay shipping back & forth. Same for repairs unless they let the magic smoke out of the entire module. 

So, none of my products are of lab-quality nor, in the class of a DMMCheck or other high-precision test or reference equipment. That market is well-served and too pricey for old, fixed-income geezers like me. 

Again, Thank you for your valuable feedback. You have given me the opportunity to clarify and I appreciate that.

[Majorassburn]

One other thought: Actually, a question:

It is my observation that most newer-manufactured "quality" (cough-cough) Chinese, under-$100 DMM's use a single-DMM-chip with true-rms conversion done internally.

I have noticed that their t-rms readings are reasonably accurate on sub-1KHz sine wave and triangle wave inputs but are considerably less accurate with same-frequency square wave inputs.

My suspicion is that their t-rms converter circuitry is optimized for (low crest factor) sine waves to make good marketing specs but that they rapidly fall short when presented with distorted or higher crest factor inputs.

Although I understand that a so-called perfect square wave has a crest factor of 1, it appears that these t-rms circuits don't work well unless they see more easily calculated and converted sine waves.

Has anyone else noticed this "feature" in these newer import DMM's? Thanks.

[bdunham7]

I have noticed that their t-rms readings are reasonably accurate on sub-1KHz sine wave and triangle wave inputs but are considerably less accurate with same-frequency square wave inputs.

My suspicion is that their t-rms converter circuitry is optimized for (low crest factor) sine waves to make good marketing specs but that they rapidly fall short when presented with distorted or higher crest factor inputs.

Although I understand that a so-called perfect square wave has a crest factor of 1, it appears that these t-rms circuits don't work well unless they see more easily calculated and converted sine waves.

Has anyone else noticed this "feature" in these newer import DMM's? Thanks.

It isn't just ultra-cheap or low end products, it is anything with a low-bandwidth (typically on-chip) TRMS converter.  These are primarily designed to be used with mains-derived AC voltages of 50/60Hz.  It is a BW issue, not crest factor, although these same meters will struggle with high CF signals as well simply because anything with a high CF will typically have harmonics well above the fundamental.

A quick test on a Fluke 116 shows me that 5.000VAC (sine) @ 1kHz gives me a reading of exactly 5.000, but a square wave of the same V_RMS and 1kHz gives me 4.880 with about 10 digits of flicker.  Reducing the fundamental frequency improves the square wave performance considerably.  At 60Hz, the difference is only 2-3 counts of flicker with the square wave which indicates that for mains systems, the TRMS function is accurate and useful to the extent you need TRMS at all.

[Mickle T.]

The only reason why Chinese DMM front-end chips have a True-RMS bandwidth limitation of about 1 kHz is that they do not use analog conversion, but mathematical processing of samples from a relatively slow sigma-delta ADC by a relatively fast microcontroller. At least all SDIC Microelectronics solutions have this architecture.

[Kleinstein]

There are DMM chip sets that include digital RMS with a relatively slow SD ADC. This comes with the rather limited BW, but it also comes with positive sides:
The most obvious is the fast reaction: remove the probes from an AC signal and one gets 1 in between reading and than a solid zero. Analog RMS may need a few seconds.
Another point is good linearity also down to low values, like 1% or even less, where analog RMS often does not work well, though the limited resolution still hits.
A final point is the performance with lower frequencies and non sine waveform, like 30 Hz - because of compromise with settling the performance of analog RMS often suffers below some 40 Hz.

A square wave is tricky for many meters. Besides the higher frequency content, there can also be slow rate issues, e.g. at an electronic rectifier. So while it is easy to generate a square wave of known amplitude it is not a great test signal for a DMM. I would add a reasonable know (e.g. calculate with Spice) low pass filter to reduce the highest frequencies (e.g. > 1 kHz) and the slew rate. As there is anyway not that much power in the higher harmoncis the accuracy of the filter is not that critical.

[J-R]

alm, I think we are somewhat agreeing.  I do have some 10V references (among others) and they are great to have for the hobby.  But still there are a mountain of calibration points on a typical DMM, such as the previously mentioned AC but also quite a few for resistance, capacitance, and frequency.  I have multiple ways to generate these inputs, but mostly only by verifying with another calibrated DMM.  I've repaired about a dozen bench meters and they frequently have completely different calibration points.  So simply adding one more DMM to your bench could require another pile of references (even considering the tools you mentioned).

Awesome14/Kaysert, is that you?

[alm]

alm, I think we are somewhat agreeing.  I do have some 10V references (among others) and they are great to have for the hobby.  But still there are a mountain of calibration points on a typical DMM, such as the previously mentioned AC but also quite a few for resistance, capacitance, and frequency.  I have multiple ways to generate these inputs, but mostly only by verifying with another calibrated DMM.  I've repaired about a dozen bench meters and they frequently have completely different calibration points.  So simply adding one more DMM to your bench could require another pile of references (even considering the tools you mentioned).

No, the important point that using the techniques I described, you can generate DC voltages from 100 mV to 1000V using just a stable adjustable voltage source and a 10 VDC standard. The 10 VDC standard is the only thing that needs calibration. Not a "pile of standards". You can do something similar with resistors and AC voltage, and by extension DC and AC current. Frequency is also easy to divide. I think there are similar tricks for capacitance using ratio transformers, but I've never looked into that. There are commercial devices that use this exact principle to adjust themselves from a few artifact standards (Fluke 57xx). Before this people in cal labs used to do it by hand (the Fluke 7105B system I mentioned above).

If you want to calibrate one DMM using another DMM of reference, you still need those adjustable, stable voltages, resistances, currents etc from 100 mV to 1000V. The difference is that you don't need the 10 VDC reference or any ratio devices, and that it saves time in exchange for paying more money.

[mawyatt]

A square wave is tricky for many meters. Besides the higher frequency content, there can also be slow rate issues, e.g. at an electronic rectifier. So while it is easy to generate a square wave of known amplitude it is not a great test signal for a DMM. I would add a reasonable know (e.g. calculate with Spice) low pass filter to reduce the highest frequencies (e.g. > 1 kHz) and the slew rate. As there is anyway not that much power in the higher harmoncis the accuracy of the filter is not that critical.

While we agree the squarewave may be tricky for some meters, and is the simplest of all waveforms, outside DC, to create accurately, very accurately actually. However from the user perspective and actual use, this is where we respectfully disagree.

First off, the user can create an extremely accurate waveform with some simple CMOS logic and an accurate DC reference, without the need for anything other than an known accurate DC meter to measure the DC reference voltage, say 5.000V for example. By design the relative accuracy of the waveform can be inferred to a high degree of precision and confidence, and tightly referenced to the accurate DC reference as the result is simply Vref/2 for the squarewave amplitude. The waveform has a unique property that no other waveform posses other than DC, the Average and RMS are the exact same, precisely Vref/2!!

Secondly, filtering the waveform to slow the slew rate and reduce the harmonics is exactly what one SHOULD NOT do. Since this very effort relaxes the DC reference voltage and the waveform Average and RMS precise 1/2 relationship to said reference.

Since the squarewave is comprised of only odd harmonics that fall of at a rate of 1/n for voltage, power rate 1/(n^2), where n is the harmonic, by filtering the waveform one is throwing away energy in these odd harmonics so the meter can't attempt to record such and thus producing an associated reading error by ignoring or attenuating these harmonics. One must remember that "RMS" is the overall "Heating" effect of all aspects of the waveform, DC, harmonics and such. By utilizing a low frequency waveform, not using any low pass filtering, this places many of the higher frequency harmonics within the meters' bandwidth and thus are recorded, certainly better than just throwing away or purposely attenuating those harmonics.

We've utilized this technique with VDD reference 5.000V for a buffered CMOS Flip-Flop (at low frequency to assure squarewave symmetry), with excellent results awhile ago, and still utilize such to compare various higher end DMMs (3 KS34465A, DMM6500, HP34401A, AG34401A, DMM3065X).

This simply works, and works quite well indeed 

https://www.eevblog.com/forum/testgear/ac-rms-dmm-tests/

The sine wave is certainly a better test and calibration waveform and why it's used by cal labs. However, creating a pure sinewave at precision levels acceptable for DMM comparison/use is difficult and expensive and why most users don't have direct access to this level of expensive equipment.

Here's where the simple CMOS squarewave technique can offer users a alternative at a very modest cost, even attractive for hobbiest

Anyway, hope this clarifies some of the subtle parameters and effects of the simple squarewave for consideration as an RMS reference waveform. We can offer more details, but don't want to detract this thread unless folks are interested.

Best,   

[tggzzz]

Cheap sanity check for generating audio sine waves with a known amplitude...

• Take a 10MHz DDS function generator, quality of your choosing
• Generate a 0.05Hz sine wave
• Attach a DVM which captures min and max readings, set it to DC volts on the fastest sampling rate. I use an Agilent 34410A
• The min and max readings correspond to the peaks and troughs, so calculating Vpp and Vrms is easy
• Increase the DDS output frequency and assume that the Vpp is unchanged. That shouldn't be a problem if the frequency is a few orders of magnitude below its max frequency

[Kleinstein]

I fully agree that the square wave can be created rather accurately. The problem is however that there is a good chance that a DMM may have problems with exactly that square wave shape, especially the high slew rate part. Even an other wise well working AC RMS part may fail under this condition and show a reading that is off a little. This makes the square wave with sharp transitions not suitable for any electronic RMS circuit - it would be only OK for an unbuffered thermal converter.

Filtering will effect the RMS value, but not that much if the cross over is significantly higher (e.g. > 10 x the frequency). A simple filter (e.g. 2nd order passive RC) can be calculated reasonable well. It takes a little math to get the correction factor of some 0.995. There is some uncertainty in that factor, but one can at least estimate it reasonably well, e.g. from the part tolerances and assumed parasitics.
There is nothing bad in having a calculated factor, even if it is not a simple one like 1/2 that can be done in the head.

A know filter is way better than BW limit of the DUT with a usually not well defined BW limit. The BW specs of the meters are usually minimal values and for testing a meter one should rely as little as possible on the quality of the meter.  A nasty point here is that with the usual AD637 and similar analog RMS chips the BW limit is known to be amplitude dependent - that larger the amplitude the faster. Ideally there is a lower BW limit in front, but when aiming for high BW this may be skiped.
Another point are possible slew rate limits and zero crossing delays in the active rectifier. So it is not just the frequnecies, but also possibly additional (nonlinear) effects from the sharp edges.
This alone makes a square wave a poor test signal for amplitude accuracy. It would be more like an extra test - good to pass, but a meter could still be useful if it does not work well with a square wave.

For checking the frequency response of the DMM the square wave is anyway not suitable as a mix of frequencies - this is more a thing to use a DDS type sine generator.

For the lower frequency end (e.g. 10-50 Hz)  it may be interesting to check how the meter reacts to both a more sqare wave and sine wave. However this is beyond the normal tests done, though it is known that the analog RMS chips have some problems at low frequencies that depends on the waveform. AFAIK there are no esteblished test methods for this.

[IanJ]

I'd think having a PDVS2mini would negate the need for the Analogic, at least from a practical perspective.  But there is the cost and stock issue.  It could be 6 months or more before they are back in stock, although now that Ian retired, maybe he'll be faster to get the next batch out!  1,000% recommended, though.

Down to costs rather than time. Some components are still through the roof or not available yet.

Ian

[mawyatt]

I fully agree that the square wave can be created rather accurately. The problem is however that there is a good chance that a DMM may have problems with exactly that square wave shape, especially the high slew rate part. Even an other wise well working AC RMS part may fail under this condition and show a reading that is off a little. This makes the square wave with sharp transitions not suitable for any electronic RMS circuit - it would be only OK for an unbuffered thermal converter.

Filtering will effect the RMS value, but not that much if the cross over is significantly higher (e.g. > 10 x the frequency). A simple filter (e.g. 2nd order passive RC) can be calculated reasonable well. It takes a little math to get the correction factor of some 0.995. There is some uncertainty in that factor, but one can at least estimate it reasonably well, e.g. from the part tolerances and assumed parasitics.
There is nothing bad in having a calculated factor, even if it is not a simple one like 1/2 that can be done in the head.

A know filter is way better than BW limit of the DUT with a usually not well defined BW limit. The BW specs of the meters are usually minimal values and for testing a meter one should rely as little as possible on the quality of the meter.  A nasty point here is that with the usual AD637 and similar analog RMS chips the BW limit is known to be amplitude dependent - that larger the amplitude the faster. Ideally there is a lower BW limit in front, but when aiming for high BW this may be skiped.
Another point are possible slew rate limits and zero crossing delays in the active rectifier. So it is not just the frequnecies, but also possibly additional (nonlinear) effects from the sharp edges.
This alone makes a square wave a poor test signal for amplitude accuracy. It would be more like an extra test - good to pass, but a meter could still be useful if it does not work well with a square wave.

For checking the frequency response of the DMM the square wave is anyway not suitable as a mix of frequencies - this is more a thing to use a DDS type sine generator.

For the lower frequency end (e.g. 10-50 Hz)  it may be interesting to check how the meter reacts to both a more sqare wave and sine wave. However this is beyond the normal tests done, though it is known that the analog RMS chips have some problems at low frequencies that depends on the waveform. AFAIK there are no esteblished test methods for this.

Just for fun, we turned on a few of our DMM meters and gathered our handhelds, an almost 3 decade old Fluke 87, a Keysight UL233A and a Uni-T UT210E for a quick-n-dirty test.

Using our DIY DMM Tester mentioned above set at 250Hz, and without letting the instruments even warm up, or even taking the time to let the plug-in stabilize, just simple Plug-N-Play as quick as possible!!!

DMM                       Vref in volts                        Vac(rms) in volts
Fluke 87                  5.00                                   2.506
KS UL233A              5.000                                 2.509
Uni-T UT210E          4.99                                   2.45
SDM3065X              5.00079                             2.49789
AG34401A              5.00078                              2.49997
DMM6500               5.00090                              2.49995
KS34465A (new)     5.00088                              2.50009
KS34465A (1.5yr)   5.00087                              2.49988
KS34465A (3yr)      5.00088                              2.50019

So we must have some "Magic Squarewaves" then 

Would be interesting to see how well this works with an HP3458A. Maybe Dr. Frank is listening and might consider this with his nice HP3458A, he certainly knows this instrument, the various AC modes, and how to properly apply such!!

Anyway, as always, YMMV 

Best,

[bdunham7]

Here's where the simple CMOS squarewave technique can offer users a alternative at a very modest cost, even attractive for hobbiest

If the hobbyist has meter with capacitor-blocked front-ends for the AC ranges and BW orders of magnitude above the fundamental of your square wave, then your DC-derived square wave is good enough for a "not broken, hasn't drifted" verification test and may even be quite precise.  However, if the hobbyist has any variety of budget DMM that doesn't work well with square waves, especially one-sided ones, your technique becomes less attractive.  The person that started this conversation claims to sell a product that puts out a clean sine wave of 6.000V +/- 2mV.  For a 6600-count DMM, that seems like a better choice for verification.

Would be interesting to see how well this works with an HP3458A.

You would expect any good TRMS, AC-blocked and high BW meter to accurately indicate the AC component of your square wave.  The whole reason you have them is to be able to read the TRMS value of complex signals.  However, if you don't have all three of those characteristics--TRMS, AC-block, high BW--the sine wave will be accurately read but your square wave will not.  You can see your UNI-T is starting to struggle at 250Hz already.

[mawyatt]

Yeah, was kinda surprised how well the old Fluke 87 behaved, and the relative cheap KS UL233A (was part of KS promotion that was "given" away with the purchase of a KS34465A back when we got our 1st 34465A (Tan case), well before the 34465A became seriously backordered). Even the ultra cheap Uni-T didn't do so bad!!

Agree, the 6V +-2mv clean sinewave is a better waveform for this application, IF this is verifiable, stable and repeatable performance!! With all due respect, don't think that what Mr Majorassburn has divulged would stand up well against higher end DMMs in that respect, however he stated that's not the intended audience!

We do know the squarewave concept is extremely stable, doesn't depend on any precision components, nor calibration except reading the DC reference value, then simply divide by 2. Of course if one wants to dive into the intricate details of trapezoidal waveforms which include finite non-equal Rise Fall times and asymmetry, then a "fudge" factor can be applied. A deeper dive which includes exponential Rise and Fall characteristics gets even more interesting!!

However, all this aside, think we've shown it works pretty well, without the need complex setups, stabilization, "fudge" factors and such, at least with the various types of DMMs we own.

Would be very interesting to see what other folks find wrt this squarewave concept.

Best

[mawyatt]

Would be interesting to see how well this works with an HP3458A.

You would expect any good TRMS, AC-blocked and high BW meter to accurately indicate the AC component of your square wave.  The whole reason you have them is to be able to read the TRMS value of complex signals.  However, if you don't have all three of those characteristics--TRMS, AC-block, high BW--the sine wave will be accurately read but your square wave will not.  You can see your UNI-T is starting to struggle at 250Hz already.

Exactly, this would likely revel the intricate details of the waveform rather that the reverse where the waveform revels the limitations of the meter!! Wish we had a 3458A, but have no justifiable use for such an exceptional instrument, can't even conjure up an "excuse" 

Maybe in our next life we'll get one....more likely we'll end up with a very hot broken Simpson or Triplett to measure burning hot coal

Best,

[alm]

Would be very interesting to see what other folks find wrt this squarewave concept.

If only there would have been another topic where the same solution and problems with it were already discussed by the same people.

[Majorassburn]

Agree, the 6V +-2mv clean sinewave is a better waveform for this application, IF this is verifiable, stable and repeatable performance!! With all due respect, don't think that what Mr Majorassburn has divulged would stand up well against higher end DMMs in that respect, however he stated that's not the intended audience!

Hi, again. I love these types of discussions because they remind me that there is no simple answer to any particular need when it comes to electronics! Everything is a compromise of some sort. That's why I enjoy making my amateurish calibrators. Basically, they accomplish a simple purpose, albeit not very well, but at very low cost.

As for my AC Pure Sine Wave generator modules, I have tried 555 & 4047-based square-wave-to-sine-wave generators using phase shift & integrator schemes. Both work well as crude "Yup, your DMM seems to work OK" verifiers. The perfect square wave is one trick to master and the conversion technique to sine wave output has its own problems to address.

One of my challenges   is to end up with minimal-count, low-cost componentry that achieves reasonably repeatable, clean & stable output sine waves. Of course, predictable, repeated, accurate amplitude hinges on  the power supply because the amplitude varies with power supply levels. Refining & decoupling that requires additional circuitry that raises costs to where there is no market other than the occasional curiosity-seeker.

So, I concentrated on a generation technique where I use an op-amp square wave generator to feed the first integrator whose triangle wave output becomes a sine wave out of the second integrator & then buffer/amplify that for the 6V+ sine wave. Lots of distortion but t-rms DMM's don't seem to choke on that!

Also, selecting 50-60Hz generator frequencies results in too much output fluctuation (+/- 5mV) caused by environmental influences & shielding is not an affordable option. So, I use 100Hz as a compromise that results in a nice, non-fluctuating output.

However, the predictable & useful 6VACrms amplitude wholly depends on a controlled power source & the most affordable option is a 12V regulator IC on board. That works good enough for my intended audience because, netting out all the positve and negative tempco's of all the various components used, my module produces a +/- 2mV amplitude predicatbility after warmup in a controlled environment on the 32nd day of the month when the moon is full! 

I would encourage one of you professionals to try one & put it to the test & report the results for all of our benefit. I lack the required test equipment to do any such thing but would love to see whether my modules are any good so I can raise my selling price if they are! 

Finally, the reason I started making the AC sine wave DIY modules is that, short of owning a rather expensive function generator, there doesn;t seem to be a quick & dirty way to get a high enough amplitude, low frequency sine wave to reasonably & accurately get a stable, non-fluctuating reading from a DMM.

Now, go easy on me  ....remember, I'm just an old geezer AMATEUR! 

[bdunham7]

triangle wave output becomes a sine wave out of the second integrator

Well, not quite but perhaps close enough...

[Majorassburn]

triangle wave output becomes a sine wave out of the second integrator

Well, not quite but perhaps close enough...

Thanks. Actual output:

[Gyro]

A small suggestion for your ebay listings, I've no problem with your modest accuracy claims, but you could do with loosing the "Top Quality Double-Sided Fibreglass FR4 PC Board" label in the images. We can all see that it is ordinary FR4 proto board, nothing wrong with that form of construction for a modest spec device, especially when done neatly. It probably puts some of us in mind of a certain controversial 10V reference by an ex-member though. He made extravagant / magical claims (2ppm) for a bodged together matrix board,copper foil, and dodgily soldered unit that he still sells on ebay to this day for >$1200 (do a forum search for 'D-105' if you're interested in the sorry tale).

Just a matter of not calling a spade an earth inverting horticultural implement. You don't need it.

[Majorassburn]

to GYRO: "A small suggestion for your ebay listings, I've no problem with your modest accuracy claims, but you could do with loosing the "Top Quality Double-Sided Fibreglass FR4 PC Board" label in the images. We can all see that it is ordinary FR4 proto board, nothing wrong with that form of construction for a modest spec device, especially when done neatly."

DONE!  Thank You for the suggestion. 

[mawyatt]

Would be very interesting to see what other folks find wrt this squarewave concept.

If only there would have been another topic where the same solution and problems with it were already discussed by the same people.

This would be in reference to someone with an HP3458A!!

Best,

[mawyatt]

..... It probably puts some of us in mind of a certain controversial 10V reference by an ex-member though. He made extravagant / magical claims (2ppm) for a bodged together matrix board,copper foil, and dodgily soldered unit that he still sells on ebay to this day for >$1200 (do a forum search for 'D-105' if you're interested in the sorry tale).

Just a matter of not calling a spade an earth inverting horticultural implement. You don't need it.

Remember that "Devine Intervention" type reference, with the magical thermal equilibrium protoboard construction, "Heavenly Derived" thermal compensation diode and "Majestic Thermal Pipe". Reading those posts was entertainment for the evening, of course along with ones favorite brew

Best,

[deepfryed]

Thanks to you fellas I wasted the evening reading that thread but I did have a good chuckle. He's still selling the divine device for over $1000 now 

[J-R]

While it's true that Kelvin, Kelvin-Varley and Hamon dividers, various bridges and a null meter could be used to accomplish many DMM calibration steps, and this method is frequently used in older calibration procedures, the problem remains that you need to acquire or construct all that equipment which is not trivial or cheap.  I also consider it to be more prone to errors/mistakes, as you're not using another calibrated DMM to verify the outputs before setting a calibration point.

So a solid solution still is to just buy a DMM that you can afford to have calibrated regularly.


Back to references, DMMCheckPlus.com & VoltageStandard.com surely are the lowest-end worth considering.  Both of these companies have good reputations and have many public reviews.  It defeats the entire concept of a reference/standard to have something that is surrounded by unknowns.  Buying these eBay references is worse than not having them because it lulls people into a false sense of security such that they might adjust their equipment, and it wastes their money in the process.

I see our eBay-er has added the "NEW PRECISION MULTIMETER CALIBRATION LAB" listing for approximately $110.
For under $175 delivered, you can get a new Brymen BM235 from Welectron with an ISO calibration, which is going to cover all of the DMM's functions and ranges, not just a small sub-set.  Under $275 for the BM789 which is a 50,000 count DMM.

[alm]

While it's true that Kelvin, Kelvin-Varley and Hamon dividers, various bridges and a null meter could be used to accomplish many DMM calibration steps, and this method is frequently used in older calibration procedures, the problem remains that you need to acquire or construct all that equipment which is not trivial or cheap.  I also consider it to be more prone to errors/mistakes, as you're not using another calibrated DMM to verify the outputs before setting a calibration point.

I linked to the mini metrology lab tutorial which builds these devices quite cheaply. Sure, it won't give you standards lab level performance, but neither will give any DMM. That you trust a DMM that has been calibrated in the past more than standards and bridge-circuits that can be verified before every important measurement is your opinion. I trust my standards over anything else in my lab. For me DMMs are just ratio devices, like the old dividers, to measure the ratio between an unknown value and a reference. We can agree to disagree there.

I see our eBay-er has added the "NEW PRECISION MULTIMETER CALIBRATION LAB" listing for approximately $110.
For under $175 delivered, you can get a new Brymen BM235 from Welectron with an ISO calibration, which is going to cover all of the DMM's functions and ranges, not just a small sub-set.  Under $275 for the BM789 which is a 50,000 count DMM.

Okay, so now you have a DMM that was calibrated on all ranges and functions, and you have another DMM that you want to verify on all ranges and functions. What now? You still need sources for all ranges and functions. I don't see how buying yet another DMM solves that.

[bdunham7]

You still need sources for all ranges and functions. I don't see how buying yet another DMM solves that.

Actual standards worthy of the name are one thing, but at this level for sources you can get an FY6x00 AWG and a few select non-precision test resistors and perhaps a few PP capacitors. 

[alm]

Actual standards worthy of the name are one thing, but at this level for sources you can get an FY6x00 AWG and a few select non-precision test resistors and perhaps a few PP capacitors.

I have done that. It's... interesting with bench meters (which is what the topic started with). But the argument was that a DMM gives you all the ranges and functions, as opposed to a cheap 'calibrator' that will just test a few functions and ranges. But in my experience getting the sources to compare multimeters is actually the hard/expensive/bulky part. Especially high voltage DCV, ACV beyond what a function generator can produce, and AC current.

[bdunham7]

But in my experience getting the sources to compare multimeters is actually the hard/expensive/bulky part. Especially high voltage DCV, ACV beyond what a function generator can produce, and AC current.

If you're doing informal verification, you can live with what you have.  If you want to do an actual calibration with adustment on even a simple handheld, then you need a lot of stuff.  There are piles of such stuff in my office (working) and in my garage (awaiting repair). The better handhelds are actually more difficult in some ways than bench meters because they need exact nominal sources while typical decent bench meters only need stable and accurately known sources.

But yes, actually coming up with 750VAC @ 50kHz is a challenge.  The solution is in my garage and weighs a few hundred pounds.

[Majorassburn]

So a solid solution still is to just buy a DMM that you can afford to have calibrated regularly.

Back to references, DMMCheckPlus.com & VoltageStandard.com surely are the lowest-end worth considering.  Both of these companies have good reputations and have many public reviews.  It defeats the entire concept of a reference/standard to have something that is surrounded by unknowns.  Buying these eBay references is worse than not having them because it lulls people into a false sense of security such that they might adjust their equipment, and it wastes their money in the process.

I see our eBay-er has added the "NEW PRECISION MULTIMETER CALIBRATION LAB" listing for approximately $110. For under $175 delivered, you can get a new Brymen BM235 from Welectron with an ISO calibration, which is going to cover all of the DMM's functions and ranges, not just a small sub-set.  Under $275 for the BM789 which is a 50,000 count DMM.

Hi. Thanks for referring to my amateur calibrators on eBay. It gives me the opportunity to clarify why I think they offer some value to a certain segment of the market that professional, calibrated, expensive standards and DMM's just can't serve.

!.  Some might interpret your comments, "Buying these eBay references is worse than not having them because it lulls people into a false sense of security" as unfounded because:  a) The comment assumes that any such buyer is ignorant as to the realities of any calibrator's limitations and, therefore, b)  Such a buyer should not, under any circumstances, be playing with anything that runs on batteries, is plugged into the wall or, is powered by alternate energy sources of any kind! 

2.  Your suggestion that buying a 50,000-count Brymen for a hundred or so more is an alternative but really impractical because you still lack a stable source for the accurate comparison of various DMM's, without regard to the actual calibrated accuracy of that source.

I think what you may have missed here, probably because I failed to emphasize it earlier, is the point that possessing, for example, an AC Sine Wave Calibrator or a 10 Volt Reference or a Precision Resistor module gives one the ability to use IT as the source for comparison of the various DMM's attached to it so that you can note the varied readings of your attached DMM's.

As an example, while possessing a NIST-traceable, calibrated, multi-thousand-dollar 10V reference, your attached Brymen and Agilent and Keysight, etc. 10-12-digit DMM's are ALL going to disagree as you move toward LSD's. So, which one do you end up loving and which one finds a new home? 

My eBay amateur 10V Reference, 0.05% basic accuracy, trimmed to 0.0001% accurate, Multi-Output reference module will allow you to do the same thing but, for only 50 bucks! And, you get TWO other useful and usable reference outputs for the 2-6V ranges thrown in! Whatta Deal! 

Granted, you cannot trust my eBay references to do much more than get you in the ballpark. BUT, after you buy them, you can still afford to make your mortgage payment! 

I think the overall message in this discussion is that, short of possessing a slew of expensive calibrating standards AND incredibly expensive calibrating equipment AND the experience and ability to re-calibrate various DMM products that allow such re-calibration, none of us can really calibrate ANYTHING to perfection. And, I also acknowledge the fact that most lower-cost meters cannot be "calibrated" at all, in the traditional sense. No knobs or dials or 'twisties' inside. They can only be verified, right?

I do believe that there is some meaningful value to a lower-cost DMM owner in being able to access a $30 sine wave source that can put out a non-fluctuating, usable 100Hz, 0-6VACrms, Lo-Z voltage source to at least verify that when you stick the probes into a 120VAC wall socket and it bounces around and finally reads 124.2VAC, that you're reasonably sure the damned thing is reading "close enough for guvmint work". Same for the other "calibrators" I offer.

Two other suggestions:  1) Having access to low-cost but properly-presented references like the ones I offer is better than having nothing at all. And, it's much cheaper than buying a calibrated multi-thousand-count DMM to compare that $30 do-all, Shanghai-Special to.

2) I agree that the DMM-checkers you referred to certainly are excellent quality and trustworthy but, they, too are limited in capability as initial verifiers of functionality. And, they may also mislead the lesser informed DMM owner into a false sense because 'DDS-ing' a square wave and grinding off its inconvenient edges to APPROXIMATE a sine wave has its own inherent faults, as well.

I think we all agree that lesser-quality DMM's may handle DDS-ed and AC square waves, no matter how many ppm's accurate they may be, differently and produce somewhat varied 'readings' as to the accuracy of the attached DMM's AC range really are. Just hook up a gaggle of $30 DMM's to a DMMCheck-generated, bi-polar, square wave output and you'll get a wildly different reading from each one. 

So, in summary, I invite your further suggestions and constructive criticisms of my efforts so that I can continually improve my amateur low-end offerings. Thanks in advance.

[J-R]

Of course the odds are extremely high that an amateur will be ignorant of your calibrators' limitations.  Also, your listings are filled with confident exposition which will surely tickle the ears of potential newbie shoppers...

Short-term stability is far easier to obtain than accuracy, so having a calibrated DMM in general is still more usable than having a reference standard.  To compare the calibrated DMM to another you only need to check the readings at the same moment, which is extremely trivial and can be done with equipment found on most electronics benches.  And you can check many ranges, not just the ones your calibrators cover.

The VoltageStandard.com 0.001% 10V reference is only $130, not thousands.  The individual TC graph is particularly a great value-added feature.

There is no issue with DMMs and the references disagreeing, because at minimum you have published accuracy specs and temperature variations to consider.  I can check the calibration data for my calibrated DMMs and references to see the reasons for any displayed discrepancies.

I've repaired and calibrated/adjusted about a dozen DMMs in the last few years and it's really not that hard if you have another, better calibrated DMM to use.

Ultimately it's just illogical for someone to buy $110 worth of calibrators to check only a few ranges/functions on a $30 DMM.  If you want to know if your cheapie DMM is working correctly, it's trivial to do this with things you can find around your house.  If you want accuracy, then you would not be putting all your trust in a $30 DMM, but instead you would go the route I presented which is buy one or more calibrated DMMs, such as from HPAK/Fluke/Brymen/etc..  And it's just not that expensive to go this way.


If you are serious about selling references, then I would suggest at minimum:
- obtain a quality 6.5 digit DMM and have it calibrated yearly
- use proper PCBs ($1-$2 per board from any of the popular PCB makers?)
- at least add stand-offs or feet to the bottom of the PCB
- perform long-term tests/characterizations of your calibrators, burn-in, drift etc.
- provide a calibration printout with each unit sold (values + environmental data)
- offer a recalibration service
- ship some samples to various Youtubers for review

[alm]

Short-term stability is far easier to obtain than accuracy, so having a calibrated DMM in general is still more usable than having a reference standard.  To compare the calibrated DMM to another you only need to check the readings at the same moment, which is extremely trivial and can be done with equipment found on most electronics benches.  And you can check many ranges, not just the ones your calibrators cover.

For 3.5-4.5 digit DMMs yes, but this topic started off with bench meters (which I interpret as 4.5-6.5 digits) and 50 ppm uncertainty. At that level finding a sufficiently accurate DMM or stable source get difficult. I've tried comparing 6.5 digit DMMs against bench supplies, function generators and metal film resistors, and trying to find correlation between the signals.

What equipment do you find on the average bench to generate > 90 Vdc, > 100 mA AC or > 20 Vrms AC, particularly something like 100 V 50 kHz or 1000V 100 Hz? I really struggled with this, and have done things like adjusting a 300 Vac range at 7 Vrms 1 kHz because that was all my function generator could produce. Eventually I got a calibrator that can do it, but that's not exactly common lab equipment.

[Majorassburn]

If you are serious about selling references, then I would suggest at minimum:
- obtain a quality 6.5 digit DMM and have it calibrated yearly
I already have a calibrated DMM that adequately verifies for this category of calibrators.
- use proper PCBs ($1-$2 per board from any of the popular PCB makers?)
Can't do that and still keep price in line with market demographic.
- at least add stand-offs or feet to the bottom of the PCB
Will do that for new listings. Thanks.
- perform long-term tests/characterizations of your calibrators, burn-in, drift etc.
Have done so but will start to publish that.  Long term will be in hours, not months, etc.
- provide a calibration printout with each unit sold (values + environmental data)
Already do that in packaging.
- offer a recalibration service
Will do that, Thanks.
- ship some samples to various Youtubers for review
I'd rather you evaluated my stuff because you're really serious about calibration. Wanna try?

Thanks for your suggestions. It'd be fun and a real eye-opener if somebody with your enthusiasm, experience and equipment took a look at one of my modules and reported their findings. I'd certainly like to know just how good or bad they are under your critical eye. But, remember that these are for the $30-$100 DMM buyer and selling price minus fees are an overriding consideration, I can't have you demanding 5ppm performance over 5 years with a 5,000 hour burn-in and still offer them for 20 bucks or so, right? 

[bdunham7]

Thanks for your suggestions. It'd be fun and a real eye-opener if somebody with your enthusiasm, experience and equipment took a look at one of my modules and reported their findings. I'd certainly like to know just how good or bad they are under your critical eye. But, remember that these are for the $30-$100 DMM buyer and selling price minus fees are an overriding consideration, I can't have you demanding 5ppm performance over 5 years with a 5,000 hour burn-in and still offer them for 20 bucks or so, right? 

It would actually save you time and money if you just laid out a board in KiCad and had someone produce it for you vs using perf board. 

If you really want a 'review', send me one of your AC modules or whatever you like.  My equipment is all just laying in a heap right now for a few weeks, but I can easily set your module up and log it for a week or so to see what its drift and tempco are like.   I'll send it back when done.  PM me if you want my address.

[alligatorblues]

If you are serious about selling references, then I would suggest at minimum:
- obtain a quality 6.5 digit DMM and have it calibrated yearly
I already have a calibrated DMM that adequately verifies for this category of calibrators.
- use proper PCBs ($1-$2 per board from any of the popular PCB makers?)
Can't do that and still keep price in line with market demographic.
- at least add stand-offs or feet to the bottom of the PCB
Will do that for new listings. Thanks.
- perform long-term tests/characterizations of your calibrators, burn-in, drift etc.
Have done so but will start to publish that.  Long term will be in hours, not months, etc.
- provide a calibration printout with each unit sold (values + environmental data)
Already do that in packaging.
- offer a recalibration service
Will do that, Thanks.
- ship some samples to various Youtubers for review
I'd rather you evaluated my stuff because you're really serious about calibration. Wanna try?

Thanks for your suggestions. It'd be fun and a real eye-opener if somebody with your enthusiasm, experience and equipment took a look at one of my modules and reported their findings. I'd certainly like to know just how good or bad they are under your critical eye. But, remember that these are for the $30-$100 DMM buyer and selling price minus fees are an overriding consideration, I can't have you demanding 5ppm performance over 5 years with a 5,000 hour burn-in and still offer them for 20 bucks or so, right? 

My lab has 4 732Bs, 2 SR104s, a 720A,  a calibrated 3458A, a 34420A, 10 6.5-digit meters, along with many lower precision standards, fixed dividers, a 1.0 ohm standard to 10ppm. I can get repeatable voltage 10VDC measurements to 0.01ppm. But that is quite a project.

So, I'd be willing to give your standard a run through the lab. I'd be interested in just what is available in that price range.

[Conrad Hoffman]

Pretty crummy lab if you don't have a Josephson junction voltage reference.   
FWIW, way back when I did the Mini-Metrology articles, I spent over a year going down to the lab every morning and intercomparing several standards with an ovenized standard cell array. After a year I had reasonable confidence in the performance. This stuff doesn't happen overnight.