Questions about the use of my new MAX6350 Precision Voltage Reference circuit

[t1d]

I built a 5 volt precision voltage standard with a MAX6350. It is a self-contained IC, needing only a few capacitors and, optionally, a 10K trimm pot.

I realized that I might want to look at the circuit’s performance, before installing the trimm pot, to see if the trimm pot is needed. Here are my test results:

DMM                        Reading             Notes
Brymen BM869s                  5.0009  This meters is 2-3 years old. It has never been calibrated. The probes were delta’d out 0.0003v.
ExTech DM110                    5.00      The probes are 0.00v at dead short. This meter is a pocket model, being around 10-15 years old. It has never been calibrated.
Tenma 72-1005                  4.987    The unit is decades old and is likely to have never been calibrated.
CenTech HF1                      5.04      Harbor Freight cheap freebie
CenTech HF2                      5.05      Harbor Freight cheap freebie

I have not accounted for any thermal influence. The air conditioner is set to 75*F/23.89*C. The devices were not warmed for any period, prior to taking the readings.

So, I end up with a question… Which do I assume is more accurate… the high quality Brymen BM869s, or the new MAX6350 Precision Voltage Reference? Whichever one I select as my standard, I will calibrated everything else to that one.

For background, I am just a hobbyist, but I do want things to be as accurate as possible. My main goal is to have everything reading the same, as much as reasonably possible, for consistency’s sake. (I do realize that I will likely never be designing anything to 0.000x accuracy, other than maybe for current draw calculations of battery life.) I have rather nice equipment, for a non-professional, but much of it is decades old and needs to be calibrated. The BM869s has only had light use; it has never been in the field. Additionally, I do not have a friend/source with a proven reference with which to compare.

So, which do I choose as my reference (BM869s, or MAX6350) and why? Or, is there some sort of universally accepted averaging method that I could perform?… Split the difference between the two, etc?

I have an additional question about the trimm pot. Will its center point be at 5K, so that the IC can be adjusted up and down from its beginning point, or will it be at 0R and the adjustment can only add resistance?

Pictures, schematic and data sheet are attached. The yellow arrow on the PCB picture points to the location of the trimm pot, should it be added. The MAX6350 PCB is temporarily attached to the Battery Clip Board with Blu Tack

Thank you for your help. I appreciate it.

[Performa01]

The answer should be simple: The MAX6350 has a specified initial accuracy (i.e. without trim) of 0.02%. The trim is a voltage divider btw, and a tiny portion of its output signal is added to the internal feedback path.

For checking its actual accuracy you need a lab grade bench DMM, so it doesn't make sense to tinker with low resolution / low accuracy handhelds, even if they were recently calibrated, because calibration could only confirm that they meet their (wide) tolerances.

The Brymen BM869s got the required resolution, but not the accuracy, which is essentially the same as the untrimmed MAX6350 and even more importantly it doesn't have the required stability, as the temperature coeffcient alone is some 30ppm/K.

The MAX6350 on the other hand has the potential to be more accurate, because of its low tempco of 1ppm/K and good long term stability of 30ppm/1000h, hence it would make sense to use the trim option for fine adjustment.

According to your measurement, there is a disagreement between the MAX6350 and the BM869s of 0.018% (180ppm). There is no way to know which one is closer to the true value and/or what the true value is. So you cannot adjust the MAX6350 without a proper DMM. The rule of thumb says that you need a DMM that ideally has one order of magnitude lower tolerances (i.e. 0.002% = 20ppm) to calibrate and adjust a 0.02% reference.

It would make sense to ask a cal. lab to adjust your reference for you - or maybe you know someone with a proper calibrated DMM with at least 70ppm accuracy in the 10V range who could help you out.
 

[t1d]

The trim is a voltage divider btw, and a tiny portion of its output signal is added to the internal feedback path.

According to your measurement, there is a disagreement between the MAX6350 and the BM869s of 0.018% (180ppm). There is no way to know which one is closer to the true value and/or what the true value is. So you cannot adjust the MAX6350 without a proper DMM. The rule of thumb says that you need a DMM that ideally has one order of magnitude lower tolerances (i.e. 0.002% = 20ppm) to calibrate and adjust a 0.02% reference.

It would make sense to ask a cal. lab to adjust your reference for you - or maybe you know someone with a proper calibrated DMM with at least 70ppm accuracy in the 10V range who could help you out.

Thank you, for your reply. You have provided good information and thinking. Let's see if I have the "take away" understanding correctly:

- For the MAX6350, I just plain have to find an appropriate proven reference and that I should not install the trimm pot, until the MAX6350 base voltage output is known.
- As the pot R/D Circuit can only add voltage, I should not install it, until the MAX6350 base voltage is known.
- The MAX6350 really won't be an appropriate reference for the BM869s, because it is not an order of magnitude greater accuracy.

I did not make it clear that I am mostly interested in having the capability to calibrate the BM869s and the Tenma 72-1005. (The Tenma 72-1005 is a multi-device unit, having a PSU, DMM, Frequency Generator and Frequency Counter.) The other meter readings were only intended for reference.

I think I only have a few possibilities for finding the needed better reference source...
- A friend of mine has connections to a local school and Makers group.
- Advertising for help... EEVBlog (hint, hint), CraigsList (Ooo, CraigsList,) Etc.
- Pay for a calibration service... Not really a good option, due my budget. I spent the budget to make the MAX6350; hah!

I greatly appreciate your support and efforts in replying.

[Performa01]

Thank you, for your reply. You have provided good information and thinking. Let's see if I have the "take away" understanding correctly:

- For the MAX6350, I just plain have to find an appropriate proven reference and that I should not install the trimm pot, until the MAX6350 base voltage output is known.

Not quite. You should not install the trimmer unless you're able to measure the output of the MAX6350 with sufficient accuracy. Without the trimmer, you can rely on the specification that it will be within its initial accuracy of +/-0.02%. After installing the trimmer, you can't rely on that anymore. You could adjust it to give the same reading on your BM869S as without the trimmer, but then you have to make sure that both devices have acommodated to the ambient temperature and warmed up for a sufficient amout of time (>30 minutes) and that the ambient temperature did not change by more than 1K between the two measurements.

- As the pot R/D Circuit can only add voltage, I should not install it, until the MAX6350 base voltage is known.

No, it adds a tiny amout of the adjusted portion of the output voltage to the internal feedback network. If the trimmer outputs a lower voltage than the internal fixed (and laser trimmed) voltage divider, it will lower the feedback voltage, hence increase the output voltage - and vice versa. So you can of course lower the output voltage as well.

- The MAX6350 really won't be an appropriate reference for the BM869s, because it is not an order of magnitude greater accuracy.

It can easily be trimmed to a much better accuracy; +/-25ppm should easily be achievable if a calibrated Fluke 8842A, 8846A, Keitley 2001 or similar is available for monitoring the MAX6350 output. Then it will stay within +/-30ppm for a temperature window of +/-5K, but in theory you have to expect an additional 30ppm error for every 1000 hours (41 days). But this is a worst case figure, in practice the long term drift should be lower (and it might change sign quite often) and it will decrease a lot as the reference ages. After a few years, there will most likely be barely any drift anymore. High quality references are artificially aged (at an accelerated rate) for this reason (putting them through temperature cycles).

So or so a a regular calibration with adjustment should be done, and by making a chart from the calibration protocols you can see the trend and decide what the true long term stability is. Once it is good enough, you can increase the calibration intervals accordingly. 

I think I only have a few possibilities for finding the needed better reference source...
- A friend of mine has connections to a local school and Makers group.
- Advertising for help... EEVBlog (hint, hint), CraigsList (Ooo, CraigsList,) Etc.
- Pay for a calibration service... Not really a good option, due my budget. I spent the budget to make the MAX6350; hah!

Yes, maybe a fellow EEVBlog member (preferably a volts nut ) happens to live near you and could do you a favour.
I for one would be happy to assist, but unfortunately there is a big pond between the two of us...

[Old Printer]

Voltage standard.com aka Malone Electronics has a calibrated 5 volt reference for $28. He will recal it for $5 for the first two years. He is well respected here and provides the most cost effective way for a junior volt nut to get started. I believe your BM869 is software calibrated and without proper equipment and knowledge you will likely do far more harm than good. Malone has other products that run up to $90 if you want to get in deeper. The key to his products is its calibrated just prior to shipment on some serious equipment that is also regularly calibrated, something you cannot match at home.

[CDaniel]

If you don't have calibrating equipment that pot shouldn't be populated and you rely on the initial accuracy of the reference . Because obviously you could make it worse with the pot , you don't know in which position should be turned  .

[t1d]

Thank you, for your reply. You have provided good information and thinking. Let's see if I have the "take away" understanding correctly:

- For the MAX6350, I just plain have to find an appropriate proven reference and that I should not install the trimm pot, until the MAX6350 base voltage output is known.

Not quite. You should not install the trimmer unless you're able to measure the output of the MAX6350 with sufficient accuracy. Without the trimmer, you can rely on the specification that it will be within its initial accuracy of +/-0.02%. After installing the trimmer, you can't rely on that anymore. You could adjust it to give the same reading on your BM869S as without the trimmer, but then you have to make sure that both devices have accommodated to the ambient temperature and warmed up for a sufficient amout of time (>30 minutes) and that the ambient temperature did not change by more than 1K between the two measurements.

Sorry, I worded my reply poorly... I was thinking that I would have to be there with the technician, and if the MAX6350 was out, initially, I would have to add the pot and the tech could then finish the calibration... Thanks for making that clear for everyone.

- As the pot R/D Circuit can only add voltage, I should not install it, until the MAX6350 base voltage is known.

No, it adds a tiny amout of the adjusted portion of the output voltage to the internal feedback network. If the trimmer outputs a lower voltage than the internal fixed (and laser trimmed) voltage divider, it will lower the feedback voltage, hence increase the output voltage - and vice versa. So you can of course lower the output voltage as well.

Okay, this is what I meant by asking if the 10K pot would be centered at 5K, so you could add, or subtract, with it.

- The MAX6350 really won't be an appropriate reference for the BM869s, because it is not an order of magnitude greater accuracy.

It can easily be trimmed to a much better accuracy; +/-25ppm should easily be achievable if a calibrated Fluke 8842A, 8846A, Keitley 2001 or similar is available for monitoring the MAX6350 output. Then it will stay within +/-30ppm for a temperature window of +/-5K, but in theory you have to expect an additional 30ppm error for every 1000 hours (41 days). But this is a worst case figure, in practice the long term drift should be lower (and it might change sign quite often) and it will decrease a lot as the reference ages. After a few years, there will most likely be barely any drift anymore. High quality references are artificially aged (at an accelerated rate) for this reason (putting them through temperature cycles).

A 1000 hours of run time, or shelf time, too?

So or so a a regular calibration with adjustment should be done, and by making a chart from the calibration protocols you can see the trend and decide what the true long term stability is. Once it is good enough, you can increase the calibration intervals accordingly.

Good info.

I think I only have a few possibilities for finding the needed better reference source...
- A friend of mine has connections to a local school and Makers group.
- Advertising for help... EEVBlog (hint, hint), CraigsList (Ooo, CraigsList,) Etc.
- Pay for a calibration service... Not really a good option, due my budget. I spent the budget to make the MAX6350; hah!

Yes, maybe a fellow EEVBlog member (preferably a volts nut ) happens to live near you and could do you a favour.
I for one would be happy to assist, but unfortunately there is a big pond between the two of us...

It is kind of you to offer, even if it is not practical; thank you. I think I may start a separate thread to ask for volunteers. I guess this would be the best forum category... Thank you for your wonderful reply.

[t1d]

Voltage standard.com aka Malone Electronics has a calibrated 5 volt reference for $28. He will recal it for $5 for the first two years. He is well respected here and provides the most cost effective way for a junior volt nut to get started. I believe your BM869 is software calibrated and without proper equipment and knowledge you will likely do far more harm than good. Malone has other products that run up to $90 if you want to get in deeper. The key to his products is its calibrated just prior to shipment on some serious equipment that is also regularly calibrated, something you cannot match at home.

Thanks for the lead. I will check them out.

[t1d]

If you don't have calibrating equipment that pot shouldn't be populated and you rely on the initial accuracy of the reference . Because obviously you could make it worse with the pot , you don't know in which position should be turned  .

Agreed. That's why I posted this thread, initially. It is a new area of learning for me and I knew you guys could set me straight. Thanks for everyone's generosity.

[t1d]

Voltage standard.com aka Malone Electronics has a calibrated 5 volt reference for $28. He will recal it for $5 for the first two years. He is well respected here and provides the most cost effective way for a junior volt nut to get started. I believe your BM869 is software calibrated and without proper equipment and knowledge you will likely do far more harm than good. Malone has other products that run up to $90 if you want to get in deeper. The key to his products is its calibrated just prior to shipment on some serious equipment that is also regularly calibrated, something you cannot match at home.

Turns out that I had come across this company, when doing my original research... But, being a DIY guy, I wanted to try it, myself. I might still go this way; I will just have to think on it.

What I really would rather do is save my pennies and by a top-shelf Bench DMM. The Siglent3065x 6-1/2 Dual Display Meter has my attention. I bought my oscope from Saelig. They are good people...
http://www.saelig.com/siglent-dmm/sdm3065x.htm

Yes the BM869s is software adjusted. I found the instructions on how to do it, but your caution has much merit. It has calibration points at 5, 50 and 500 volts.

Hah! I just realized my equipment is way smarter than I am. I am very fortunate to have that kind of a problem, being a retired guy on a tight budget...

Thanks guys!

[Performa01]

Okay, this is what I meant by asking if the 10K pot would be centered at 5K, so you could add, or subtract, with it.

It needs not be the center position where the influence of the trimmer is zero. It depends on the actual internal reference voltage, which might be just around 1.2 volts. This is then amplified by an OpAmp with an internal feedback network that is laser trimmed to yield the intended output. You have noticed that there are different versions of that chip, with different output voltages. The internal reference voltage will always be the same, just the subsequent amplification is different.

It might be that there is an internal scaling network that actually puts the neutral adjustment point (with no influence on the output voltage) near the center of the trimming range, but you certainly shouldn't rely on that and even if it's done that way, it would only be a coarse approximation.

A 1000 hours of run time, or shelf time, too?

It would be for run time, but:

1. A good reference should usually be running all the time. You don't want to wait for hours to stabilize after switching it on and even then its output voltage will usually always be slightly different than before. Of course this isn't really an issue as long as we're talking about tolerances as high as 30ppm, but it becomes significant once we're entering single digit ppm territory.

2. There is some aging in the shelf as well. It should be much slower, but certainly not zero.

[t1d]

More good information, Performa01; thank you.

[MosherIV]

I think I only have a few possibilities for finding the needed better reference source...
- A friend of mine has connections to a local school and Makers group.
- Advertising for help... EEVBlog (hint, hint), CraigsList (Ooo, CraigsList,) Etc.
- Pay for a calibration service... Not really a good option, due my budget. I spent the budget to make the MAX6350; hah!

I doubt schools have their dmm calibrated, not the ones regularly used by the kids. Maybe one used by the science lab technicians but maybe not. Well that is my experience in UK.
Makers groups are like you, ametures with little resource but better at begging or clubing together.

Yes, you are almost in the right place. In the metrology section, one if the threads is about a club in US to post a traveling 10V reference that some have accurately set up.
Ask if someone there can set up your 5V reference for you.

Some calibration services can be as low as $50. You may be thinking that calibration means adjustment, NO!
It means a check against known standards and a report on how far your device is off. If the device is within specification they leave it alone and just give you a report. If it is out of specification, they will do as you request - adjust or leave it.
The buget cal requires an additional fee to do the adjustment.

Your Brymen is fairly unlikely to need adjustment. Even if it does, pay to have it done.

Edit: I do the same thing at home for fun but I must admit that I am not as serious about it as others here. I used AD584 which has 3 Voltages - 2.5V, 5V and 10V. I also got 3 of those Chinese 6.9V low noise reference chips but have not got round to building it yet. I just like to see how much my 15 odd dmms differ from each other 

[t1d]

I doubt schools have their dmm calibrated, not the ones regularly used by the kids. Maybe one used by the science lab technicians but maybe not. Well that is my experience in UK.

Ooo, good point...

Yes, you are almost in the right place. In the metrology section, one if the threads is about a club in US to post a traveling 10V reference that some have accurately set up.
Ask if someone there can set up your 5V reference for you.

I will check there.
Thank you for the help.

[t1d]

This looked like the calibration group thread.
https://www.eevblog.com/forum/metrology/usa-cal-club-round-2/
I have sent cellularmitosis a pm asking how to participate. I look forward to hearing from him. Thank you for the lead.

[Old Printer]

As the metrology guys say, first you need a calibrator, then you need a calibrator for the calibrator. A slippery slope indeed. 😀

[RandallMcRee]

This looked like the calibration group thread.
https://www.eevblog.com/forum/metrology/usa-cal-club-round-2/
I have sent cellularmitosis a pm asking how to participate. I look forward to hearing from him. Thank you for the lead.

Member of the US cal club, here. Cellularmitosis passed the mantle off to Vindoline--best to message him!

It will be like a year before you see that travelling reference, since we're following a star pattern and you will be at the last, of course. I, and, I am sure, many other members would be happy to measure your new max6350 against their gear. I can tell you the difference between your reference and a Malone 5v reference, for example.

It would be best if you have let your reference burn-in about 1000 hours before sending it off to me or anybody else, then it will be most stable. Let me know when and if you want a measurement. You pay for shipping both ways, of course.

Edit--pic of dmmcheckplus measurement. Its calibration lists it at 5.0000 volts, so in spec still.

[t1d]

As the metrology guys say, first you need a calibrator, then you need a calibrator for the calibrator. A slippery slope indeed. 😀

Lol... This is my exact situation...

[t1d]

I have contacted vindoline and he has graciously added me to the roster. Woot!

[t1d]

This looked like the calibration group thread.
https://www.eevblog.com/forum/metrology/usa-cal-club-round-2/
I have sent cellularmitosis a pm asking how to participate. I look forward to hearing from him. Thank you for the lead.

Member of the US cal club, here. Cellularmitosis passed the mantle off to Vindoline--best to message him!

It will be like a year before you see that travelling reference, since we're following a star pattern and you will be at the last, of course. I, and, I am sure, many other members would be happy to measure your new max6350 against their gear. I can tell you the difference between your reference and a Malone 5v reference, for example.

It would be best if you have let your reference burn-in about 1000 hours before sending it off to me or anybody else, then it will be most stable. Let me know when and if you want a measurement. You pay for shipping both ways, of course.

Edit--pic of dmmcheckplus measurement. Its calibration lists it at 5.0000 volts, so in spec still.

Sorry, Randall, I made a lengthy reply and it appears that it has disappeared? And, I really don't know what all I said. Here's some of what I recall...

Firstly, thank you so much for your offer to help. That is very gracious of you. I am still strategizing, but I will contact you, if I need the calibration you offered.

Secondly, as just a hobbyist, I am not sure of the viability of burning anything for 1000 hours... But, I am thinking on using a power supply to burn and switching back to batteries to test. I do wonder if the noisy power supply might damage the MAX6350, somehow.

Thirdly, as a present strategy, I think I will use my BM869s DMM as the common reference. It is just a few years old and has had extremely light use and good care. That will do two things... Equate all the gear to a reasonable common value. And, give me an opportunity to learn how to calibrate my gear, prior to the arrival of the group's reference.

Thanks to everyone for your excellent help.

[MosherIV]

Secondly, as just a hobbyist, I am not sure of the viability of burning anything for 1000 hours... But, I am thinking on using a power supply to burn and switching back to batteries to test. I do wonder if the noisy power supply might damage the MAX6350, somehow

1000 hours is only 45 days.
Ie keep the max6345 powered continuously for 45 days or more.
I built a simple linear with lm7812 regulator for my AD584.

If you have not built main psus before or do not have the confidence to work with mains transformers, use an old wall wart with tranformer to supply your reference. You will know if it is transforer based, they are heavy compared to modern wall warts.

[t1d]

Secondly, as just a hobbyist, I am not sure of the viability of burning anything for 1000 hours... But, I am thinking on using a power supply to burn and switching back to batteries to test. I do wonder if the noisy power supply might damage the MAX6350, somehow

1000 hours is only 45 days.
Ie keep the max6345 powered continuously for 45 days or more.
I built a simple linear with lm7812 regulator for my AD584.

If you have not built main psus before or do not have the confidence to work with mains transformers, use an old wall wart with tranformer to supply your reference. You will know if it is transforer based, they are heavy compared to modern wall warts.

Thanks, Mosher. I have no problem building one up from a transformer... I probably already have several that will do. I just did/do not know if the remaining ripple is okay for the burn-in time. And, what do I use to emulate the multimeter, so that the IC is doing some work? Should I check my meter manual and see what its burden resistance is and use that value?

[MosherIV]

I just did/do not know if the remaining ripple is okay for the burn-in time. And, what do I use to emulate the multimeter, so that the IC is doing some work?

So, check what the input volt range is for your reference. Use a transformer around 5V above that, then use a voltage regulator, like a 7812. The voltage regulator should reduce the ripple.

Do not need a load on the reference. In fact too much load and the reference will not regulate.
I actually buffered mine with op-amps. Both to improve the output drive of the reference and to give some output protection.

The point of burn in is to just run the reference non stop to get over the initial drift uncertainty.
Just keep it powered in 24/7 for 46 days.
I have mine powered continously now for over 2 years.

[t1d]

I still have the complete project that I made in KiCad. I didn't think to offer to share the files, when I first originated this thread. But, I would be glad to do so, for anyone wanting to make one for their own personal (non-commercial) use. Just send me a private message on this forum with your direct email address. I need that, because you can't send attachments with messages through the forum.

I ordered my PCBs from OSH Park, because they charge by the square inch and the board is small/cheap. Cheap enough to not bother building this project on perf board. Way easier to build, than perf, too. I might even have an extra board to freely share (mailing to the USA/48 only.) (I want bother to look for it, until someone asks for it.)