Voltage references in series - bad idea ?

[AlexResch]

I have a few multimeters and I discovered each of them shows something else when measuring say a 9V battery. I am thinking of getting a couple of 10V references like REF102CP and placing them in series so that I have some idea how much my multimeters are off. Is this a good idea ? I do not have access to any calibration lab - out of the question. My question is : is there any reference good enough for a 4 and 1/2 digit multimeter ? Will it "stay put" for say, 2 years or so ? After which I could get a new one (s).
I am pretty sure this has been discussed here before and a lot - I apologize for asking an old question again but I am very new to these things and do not know how to ask the right questions or look for the specific answers. I would like to be able to be sure of 0.1V up to around 60V , 0.3V to around 200V and 0.5-1v past that. I would appreciate suggestions on the most suitable IC's. Thank you very much !

[iMo]

Figure 11:
http://www.ti.com/lit/ds/symlink/ref102.pdf

[ogden]

I have a few multimeters and I discovered each of them shows something else when measuring say a 9V battery. I am thinking of getting a couple of 10V references like REF102CP and placing them in series so that I have some idea how much my multimeters are off. Is this a good idea ? I do not have access to any calibration lab - out of the question. My question is : is there any reference good enough for a 4 and 1/2 digit multimeter ? Will it "stay put" for say, 2 years or so ? After which I could get a new one (s).
I am pretty sure this has been discussed here before and a lot - I apologize for asking an old question again but I am very new to these things and do not know how to ask the right questions or look for the specific answers. I would like to be able to be sure of 0.1V up to around 60V , 0.3V to around 200V and 0.5-1v past that. I would appreciate suggestions on the most suitable IC's. Thank you very much !

REF102CP is buffered reference, it is not easy to put such in series unless you power each series reference from individual, isolated supply. In short: bad expensive idea. To check random multimeters you will be fine with just one REF102 that is powered from two 9V batteries. Maybe don't bother and just buy "voltage standard": http://shop.voltagestandard.com/ What you will do when you find that your multimeters are off? Throw them away? Better just buy proper multimeter which you can trust out of the box?

[AlexResch]

Figure 11:
http://www.ti.com/lit/ds/symlink/ref102.pdf

Thank you - looks good and affordable.

[AlexResch]

I have a few multimeters and I discovered each of them shows something else when measuring say a 9V battery. I am thinking of getting a couple of 10V references like REF102CP and placing them in series so that I have some idea how much my multimeters are off. Is this a good idea ? I do not have access to any calibration lab - out of the question. My question is : is there any reference good enough for a 4 and 1/2 digit multimeter ? Will it "stay put" for say, 2 years or so ? After which I could get a new one (s).
I am pretty sure this has been discussed here before and a lot - I apologize for asking an old question again but I am very new to these things and do not know how to ask the right questions or look for the specific answers. I would like to be able to be sure of 0.1V up to around 60V , 0.3V to around 200V and 0.5-1v past that. I would appreciate suggestions on the most suitable IC's. Thank you very much !

1. REF102CP is buffered reference, it is not easy to put such in series unless you power each series reference from individual, isolated supply. In short: bad expensive idea. To check random multimeters you will be fine with just one REF102 that is powered from two 9V batteries.

2. Maybe don't bother and just buy "voltage standard": http://shop.voltagestandard.com/

3. What you will do when you find that your multimeters are off? Throw them away?

4. Better just buy proper multimeter which you can trust out of the box?

1. Sorted.
2. Maybe... But he's out of stock and with shipping a couple of those cost a pretty buck around $350 for 5 or so.
3. Add/subtract. Recalibrate.
4. And what about when the new one is long "out of the box" ?

[Brak]

There's an article here:

http://www.markhennessy.co.uk/dvm_reference/

about stacked LM4040.

Brak

[iMo]

100000 Volt source stacked up by 10000pcs of REF5010s

http://www.ti.com/lit/an/sbaa203/sbaa203.pdf

[guenthert]

2. Maybe... But he's out of stock and with shipping a couple of those cost a pretty buck around $350 for 5 or so.

Uh, you wouldn't want to buy five of them.  Buy one voltage standard and four or five cheap references (e.g. AD584 based ones, but perhaps you can get away with a battery, a resistor and a zener), then trim the cheap ones to match the voltage standard.  They will drift over time and with temperature (much more than the voltage standard), but within, say, 20 minutes and +/-1K or so, should give reasonable good results (check your requirements).

[ogden]

2. Maybe... But he's out of stock and with shipping a couple of those cost a pretty buck around $350 for 5 or so.

Then consider this one instead: https://www.eevblog.com/forum/beginners/t17704/

[edit] More about cheap DVM checkers: http://www.markhennessy.co.uk/ad584_references/

4. And what about when the new one is long "out of the box" ?

Anyway you shall not calibrate it using self-made voltage "standard" that is never calibrated.

[Kleinstein]

Of the shelf voltage reference chips usually have a limited absolute precision. They could still be used to stack a few to get a higher voltage. However one would need to measure there individual voltage. An amplifier with a chain of good quality resistors might still be an alternative.

There are some relatively cheap (e.g. $20 range) voltage reference boards with measured voltage. With some luck and looking for the right ones you get them individually measured with a well calibrated 6 digit DMM. There are quite some threads about these reference.

[AlexResch]

Of the shelf voltage reference chips usually have a limited absolute precision. They could still be used to stack a few to get a higher voltage. However one would need to measure there individual voltage. An amplifier with a chain of good quality resistors might still be an alternative.

There are some relatively cheap (e.g. $20 range) voltage reference boards with measured voltage. With some luck and looking for the right ones you get them individually measured with a well calibrated 6 digit DMM. There are quite some threads about these reference.

Datasheet for REF102 shows 2.5mV precision in 10V. That's at least 8 times better than what I need ( see my first post ). Five of them will be better than that, proportionally and from 50V ( given my modes needs ) I can build up with resistance dividers up to 1000V. I see absolutely no need for accuracy exceeding say 2.5% given the kind of simple stuff I am busy doing. I have absolutely ZERO access to calibration labs or better equipment. I have no "well calibrated 6 digit DMM". I could buy one but then a year or two down the line I am back to square one and I'll end up buying the meter to calibrate the meter, to calibrate the meter...

[mzzj]

Multiple REF102 in series would give you pretty good "reference", they are stable enough for your purposes for next 10 years or more and you can compare individual values with each other if you suspect that one of them might be bad.

Not calibration lab grade solution but good enough for many hobby uses.

[drussell]

Figure 11 in the linked REF102 datasheet linked above shows how to do stacked REF102s...

[MadTux]

Even highly accurate calibrators use stacked voltage references, e.g. Fluke 5440B/AF uses 2 stacked LT-FLU references to get a very stable reference voltage around 15V.

Thereby I think it's a very good way to get stable voltage references at higher voltages. In terms of absolute error, it's probably better than averaging several references in parallel (less noise).

[Kleinstein]

To get lower noise (relative to voltage), there is little difference between using 2 reference in series and 2 reference in parallel with averaging resistors. The difference is in a different voltage level and maybe that the amplifier noise is less important at a higher voltage.


There is no need to have a 6 digit meter yourself - there are small ready made reference boards available that come with measurement data from an good DMM in China. There are lots of offers in the $5-20 range for something like this:
https://www.ebay.de/itm/AD584-4-Channel-2-5V-5V-7-5V-10V-High-Precision-Voltage-Reference-Module/221991435819?hash=item33afb87e2b:g:9EIAAOSwKtlWktEP
(just a random pik from German eaby). Chances are the Chinese also deliver to Africa.

The main problem with these is to find a reliable source, so that the calibration of the reference is true and not fake. While it's easy to get cheap parts and even fakes, it rather difficult to find reliable data in china. It is not an official calibration, but just a small company using a hopefully well calibrated meter.  They are reasonable cheap (often less than a new reference chip without the board and data) so that one could even afford 2 from different sources and do a cross check.

[AlexResch]

To get lower noise (relative to voltage), there is little difference between using 2 reference in series and 2 reference in parallel with averaging resistors. The difference is in a different voltage level and maybe that the amplifier noise is less important at a higher voltage.


There is no need to have a 6 digit meter yourself - there are small ready made reference boards available that come with measurement data from an good DMM in China. There are lots of offers in the $5-20 range for something like this:
https://www.ebay.de/itm/AD584-4-Channel-2-5V-5V-7-5V-10V-High-Precision-Voltage-Reference-Module/221991435819?hash=item33afb87e2b:g:9EIAAOSwKtlWktEP
(just a random pik from German eaby). Chances are the Chinese also deliver to Africa.

The main problem with these is to find a reliable source, so that the calibration of the reference is true and not fake. While it's easy to get cheap parts and even fakes, it rather difficult to find reliable data in china. It is not an official calibration, but just a small company using a hopefully well calibrated meter.  They are reasonable cheap (often less than a new reference chip without the board and data) so that one could even afford 2 from different sources and do a cross check.

And pray, how on Earth would I find (sic!) a "reliable source" given I have no equipment to test them and absolutely ZERO access to a calibration lab ? Go to China ?

I'm trying to work out a reasonable solution appropriate for my modest needs. The one I found is EIGHT TIMES better than that and does fine to 100kV . Why would I buy some uncertain Chinese device instead of trusting a reputable manufacturer ?

Solutions must be commensurate with the problem : I need to be sure on a tenth of a Volt up to 50V. Being sure on a microvolt on 50V is completely unnecessary in my case. And probably in most any other cases....

[Kleinstein]

There were a few threads bout those Chinese references: most of them seem to be reasonable. The best chance to get a good one would be to check something like this forum for other peoples results (those who can check) and buy from such a source. Also getting 2 from separate sources and a cross check might be good enough.  If the requirements are nor that higher - e.g. more like a crude check on analog and maybe 3 digit meters, an of the self voltage would be Ok, but even those boards with fake reading would meet that low limit, as the refs used are also usually reasonably good.

To really do a calibration on a 4.5 digit meter the reference should be considerably better than the meters specs. So more like  0.01% level - few chips are available at this level.

Another practical way at this very low level would be to buy a new DMM every 1-2 years and use that calibration as a reference point. At something like 2 V or 20 V even the cheap meters are usually relatively accurate, often better than a similar priced reference chip.

[AlexResch]

1. There were a few threads bout those Chinese references: most of them seem to be reasonable. The best chance to get a good one would be to check something like this forum for other peoples results (those who can check) and buy from such a source. Also getting 2 from separate sources and a cross check might be good enough.  If the requirements are nor that higher - e.g. more like a crude check on analog and maybe 3 digit meters, an of the self voltage would be Ok, but even those boards with fake reading would meet that low limit, as the refs used are also usually reasonably good.

2. To really do a calibration on a 4.5 digit meter the reference should be considerably better than the meters specs. So more like  0.01% level - few chips are available at this level.

3. Another practical way at this very low level would be to buy a new DMM every 1-2 years and use that calibration as a reference point. At something like 2 V or 20 V even the cheap meters are usually relatively accurate, often better than a similar priced reference chip.

1. What can one say....
2. Well, what's really needed is ONE digit more. But then, if you take the trouble and read my first post you'll see I don't need that much. Also, 0.01% of 10V is 10/100/100 = 0.001V . REF102 is 2.5 times worse than that and 20 times better than what I need. And I can at least halve that by having them in series.
3. You really think a cheap Chinese meter in storage gawd knows for how long will better the 2.5mV in 10V from a reputable manufacturer

[Terry01]

One of the things you could do is find another EEVBlog member who lives near you, even the same country will do so long as they have the test equipment you need. Send them whatever you want measured and a SAE so they can send you your stuff back with the results once measured.
I did that for another member to save him having to buy equipment for just 1 use. I'm sure most of the guys here wouldn't mind helping you if they can.

[AlexResch]

One of the things you could do is find another EEVBlog member who lives near you, even the same country will do so long as they have the test equipment you need. Send them whatever you want measured and a SAE so they can send you your stuff back with the results once measured.
I did that for another member to save him having to but equipment for just 1 use. I'm sure most of the guys here wouldn't mind helping you if they can.

 

I'll definitely do that when I'll need to. THIS time, the REF102 is WAY better than what I actually need. I have absolutely no need to measure better than a tenth of a Volt in 50. Even that is exaggerated.

[iMo]

@AlexR: This guy is a well equipped metrologist, your neighbor btw, I would guess..
https://www.youtube.com/user/mjlorton

[ManateeMafia]

mjlorton has moved to the USA.

[Terry01]

Yes he has. I saw that on a video somewhere too.
I'm sure there must be other members near you though???