Checking Multimeter Calibration at Home

One of the most important tools in an e-biker's toolbox is a multimeter (aka Digital Volt Meter or DVM). It is sometimes needed to be used to measure voltages with a high degree of accuracy, for example looking at the output of a charger to tell if it is set up to the correct voltage or not. Unfortunately, unless you spend a lot of money, the chances are you will have a low cost Chinese made device with an accuracy at best of around +/- 1% and sometimes this can be as bad as +/- 2%. From experience, I've found that a lot of DVMs are not too bad on the 20 V scale, but on the 200 V scale things can be quite poor.

 

1% doesn't sound like much, so what's the big deal? Well +/- 1% of 42V can be a difference of as much as 0.42 V higher or lower than the indicated voltage. So if your meter gives you a reading of 42 V from your charger, the actual value might be as low as 41.58 V or as high as 42.42 V. As most battery BMS modules are set to balance at 4.18 V/cell and above, then at the bottom end, 41.5 V, the pack would never balance. At the top end of the error range it would be over charging the cells and potentially causing damage. Reduced cycle life is likely. If the error is as much as 2%, the situation can be far worse. Some meters I've had have been out by more than a volt when trying to measure 42 V.

 

I am lucky in that I have access to meters with a high level of accuracy, so I can check things and make sure my chargers are set up properly and I can check the accuracy of my cheaper meters. I am also interested in charging to 4.1 V/cell to increase pack cycle life. This requires voltage monitoring with a reasonably accurate DVM, otherwise you are lost.

 

So how can the average person with a little technical knowledge check that their meter is giving them a sensible reading?

 

There are IC chips you can buy that are used as voltage references for different kinds of circuits. You can get them in different voltages and also levels of accuracy. Typically, the most accurate versions seem to be designed to work at 5V. The idea is to get one of these ICs, wire it up to a power supply and connect it to your DVM to see how accurate it is.

 

I chose the Analogue Devices IC - MAX6126B50+. It has an accuracy of at least 0.02%:

 

https://www.analog.com/media/en/technical-documentation/data-sheets/max6126.pdf

 

I got a couple from RS:

 

https://uk.rs-online.com/web/p/voltage-references/1914296?gb=s

 

It's a fairly simple circuit, all it needs are a few capacitors for stability and filtering:

 

 

For super accurate measurements, you can even set up a 4 wire measurement to compensate for lead resistance in your DVM connecting wires:

 

But I thought I'd just start with the basic circuit, with GND / GNDS and OUTF /OUTS connected directly together as in the top circuit

 

The ICs arrived and then I opened the package. I knew they would be small, normally used in surface mount PCBs, but I was not expecting this small!

 

 

I mounted the chip on a piece of acrylic using super glue. I soldered in the caps on the "top" side of the board, and used the clipped off legs of the caps to connect the IC leads to the Vero board. This was fine for the GND / GNDS and OUTF /OUTS connections as it was soldering two sets of pins together. The really tricky part was soldering to Vin and NR. I pulled a piece of multi strand wire apart, so I could source wire thin enough and soldered these single strands to the right points:

 

 

It was not easy and as you can see from the solder, it took a few attempts - plus I got Vin and NR the wrong way round the first time...always check your work!

 

The top side:

 

 

I used 1 uF caps for stability and a 10 uF cap on the Vin side.

 

I connected it to a bench top PSU running at 7.5 V , but you could use anything that gives 5.2 V up to about 12 V.

 

Then the moment of truth, connecting it to a meter on the 200 V range:

 

 

What a relief, 5 V followed by 3 zeros. On the 5 V range, I got 4.9994 V. So that could be the limit of the calibration on my old Fluke, or it could be the limit of the accuracy of the voltage reference.

 

So what about a cheap multimeter?

 

On the 20 V range:

 

 

And on the 200 V range:

 

 

 

So 0.1 V out on the 200 V range, compared to an accurate DVM. 5 V is at the bottom end of the 200 V range. Ideally it would be nice to have a voltage reference at a higher voltage, as an error of 0.1 V may not be a fixed systematic error across the whole 200 V range. However, I have measured various sources with this meter and the Fluke and it does seem to be consistently 0.1 V out. So this DVM is not too bad and I know the error to expect and I can compensate for that. Other meters I've checked have been far, far worse.

 

If I did this again, I might look for a 10 V reference chip with a similar level of accuracy. I would certainly try to find one with a standard dual in line package to make soldering easier!

 

Using this simple circuit, you can get an idea as to the accuracy of your meter, without spending a lot of money having it professionally tested.

Edited August 30, 20232 yr by WheezyRider

 

These are a couple of handheld multimeters. The cheaper one's accuracy varies which is bad news at values away from the voltage reference. Then again the pricier meter's less accurate.

I set the charger 0.5V lower to err on the safe side and I'll be treating myself to a better meter.

 

Incidentally my Ruideng regulator does a lot better than the claimed ±(0.3%+3 digits) so that's become a makeshift voltage reference.

https://vi.aliexpress.com/item/1005001488070904.html

Nice project.

 

That device comes in an uSOP package, looks like pin spacing is 0.65mm.

 

You can make a DIP adapter from a small square bit of stripboard, stick the IC upside down on a square of stripboard and run fine wires to 0.1" pin headers on the edges, or you can buy adapters such as these;

 

Thanks for your very interesting post [mention=26000]WheezyRider[/mention] - I've squirrelled away the details of that useful but mega fiddly looking tiny soldering challenge. My DVM probably contains an adjustable potentiometer held in position with hot glue... I'm now wondering how self-calibration to reference could be built into a cheap multimeter - being able to press a button marked "Calibrate" would be convenient (especially if it worked). It'd be even more convenient, if the DVM calibrated itself after being switched on...

Edited August 30, 20232 yr by guerney

So, you manage to get your voltmeter to measure accurately. How are you going to deal with the ripple from your charger? Does your voltmeter measure the peak value, average value or RMS value? How big is the ripple from your charger -actually, not what it says in the spec.?

So, you manage to get your voltmeter to measure accurately. How are you going to deal with the ripple from your charger? Does your voltmeter measure the peak value, average value or RMS value? How big is the ripple from your charger -actually, not what it says in the spec.?

 

From the chargers I've looked at, ripple hasn't been such a problem, drift has been more noticeable, especially in conditions where there is no load, but only around 50 mV or so. If ripple is a concern you could always put a capacitor of a reasonable size across the output.

[ATTACH type=full" alt="53644]53644[/ATTACH]

 

These are a couple of handheld multimeters. The cheaper one's accuracy varies which is bad news at values away from the voltage reference. Then again the pricier meter's less accurate.

I set the charger 0.5V lower to err on the safe side and I'll be treating myself to a better meter.

 

Incidentally my Ruideng regulator does a lot better than the claimed ±(0.3%+3 digits) so that's become a makeshift voltage reference.

https://vi.aliexpress.com/item/1005001488070904.html

 

Interesting to see how the accuracy varies over a range of just 10V. How did you set the charger 0.5 V lower?

 

With the voltage 0.5 V lower, it's likely that you will not get to 4.18V/cell needed for balancing to occur. Are you worried about that?

From the chargers I've looked at, ripple hasn't been such a problem, drift has been more noticeable, especially in conditions where there is no load, but only around 50 mV or so. If ripple is a concern you could always put a capacitor of a reasonable size across the output.

All chargers have ripple. It's just a quiestion of how much. How much ripple did you measure on your own charger?

So 0.1 V out on the 200 V range, compared to an accurate DVM. 5 V is at the bottom end of the 200 V range. Ideally it would be nice to have a voltage reference at a higher voltage, as an error of 0.1 V may not be a fixed systematic error across the whole 200 V range. However, I have measured various sources with this meter and the Fluke and it does seem to be consistently 0.1 V out. So this DVM is not too bad and I know the error to expect and I can compensate for that. Other meters I've checked have been far, far worse.

 

If I did this again, I might look for a 10 V reference chip with a similar level of accuracy. I would certainly try to find one with a standard dual in line package to make soldering easier!

 

Using this simple circuit, you can get an idea as to the accuracy of your meter, without spending a lot of money having it professionally tested.

I think all DVMs have built in voltage reference. Multitesters are limited usually by the number of digits of their LCD. There are plenty of cheap DVMs with 10,000 counts which can measure 41.xxV within +/-0.01V.

I think all DVMs have built in voltage reference. Multitesters are limited usually by the number of digits of their LCD. There are plenty of cheap DVMs with 10,000 counts which can measure 41.xxV within +/-0.01V.

Why do they go so far off then when the battery runs down?

I think all DVMs have built in voltage reference. Multitesters are limited usually by the number of digits of their LCD. There are plenty of cheap DVMs with 10,000 counts which can measure 41.xxV within +/-0.01V.

 

Not in my experience. Lots of multimeters I've tested from a range of manufacturers are not accurate enough for our purposes over the 200 V range, regardless of the number of digits. +/-1% is a typically quoted accuracy level, which can be 0.42 V at 42 V. Some are only +/-2% and some I've tested have been more than a volt out at 42 V. Accuracy costs money to implement. Just one of these reference ICs with 0.02% accuracy costs more than some cheap multimeters. These will have cheaper reference ICs, maybe just a 431 chip, it costs pennies, but with an accuracy of only about 1%.

 

A lot of these multimeters were not intended for high accuracy work. A lot will just be used to see if a voltage is there or not, or used to test continuity.

 

At best, I have been able to get reasonable cheap meters with a consistent error of about 0.1 V. But even with that, is it under or over reporting? 42 V may be reported as 41.9 V, so balancing ok - under the 42 V threshold, or it might report 42.1 V and causing some unnecessary concern if you don't know the amount of error. Or, if the charger is outputting 42.1 V and the meter reports it as 42 V you might think all is fine, when actually you are at risk of over charging. I think you need an accuracy of better than 50 mV at 42 V to have peace of mind.

https://www.ebay.co.uk/itm/126010252383

The cells are very underutilised and checked for balance from time to time. They're charged with a constant current power supply, its ripple can be 300mV but has a narrow spiky waveform so there's little energy in it.

 

By avoiding conditions where imbalance has much effect the exact charging voltage is less critical from a safety POV and more a trade-off between range and battery life, my compromise is a heavier battery.

https://www.ebay.co.uk/itm/126010252383

The cells are very underutilised and checked for balance from time to time. They're charged with a constant current power supply, its ripple can be 300mV but has a narrow spiky waveform so there's little energy in it.

 

By avoiding conditions where imbalance has much effect the exact charging voltage is less critical from a safety POV and more a trade-off between range and battery life, my compromise is a heavier battery.

[ATTACH type=full" alt="53650]53650[/ATTACH]

There's a big difference in ripple between a £1500 house battery charger/inverter and a £10 ebike charger.

Ripple's a good point, if the battery can't discharge through the charger presumably it goes on charging to the peak voltage including ripple, albeit increasingly slowly. This might be a concern if float charging, which isn't the usual way to charge lithium batteries... although if left overnight?

 

Out with fag packet: a charger with 0.5V pk-pk ripple would have a half-wave RMS around 0.18V I think, overcharging at 0.6A. Hmm, spreadsheet... overcharging typically averages 180mA (!), although that's with sinusoidal ripple, so in the worst case enough to overcharge to 42.25V overnight. Real world figures would be good, are you able to give it a try?

Ripple's a good point

 

So how do you decide if battery B can be effectivly charged by charger C ?

 

I dont recall seeing the tolerable 'ripple' spec on a charger being quoted very often or indeed how much 'ripple' a particular battery can safely accept under charge.

 

And of course if 'ripple' is that important, you would need to have a calibrated scope handy to measure it.

So how do you decide if battery B can be effectivly charged by charger C ?

 

I dont recall seeing the tolerable 'ripple' spec on a charger being quoted very often or indeed how much 'ripple' a particular battery can safely accept under charge.

 

And of course if 'ripple' is that important, you would need to have a calibrated scope handy to measure it.

 

I haven't looked for a while. When I get the chance I'll dig out the scope and look into it. From memory, when I last looked it was quite small. Chargers have inbuilt filtering to remove most ripple. Won't get a chance this week though.

> So how do you decide if battery B can be effectivly charged by charger C ?

 

As others have said the right target voltage matters a lot e.g. 10S/13S/14S x LiIon/LiFe/etc, and the current too to some extent, which is what led to this suggestion:

https://www.pedelecs.co.uk/forum/threads/conversion-kits-to-blame-in-40-of-ebike-fires-says-london-fire-brigade.45909/post-690258

 

If the voltage and current match OK (and the maker hasn't tied the battery to its charger with comms or edicts), then ripple, voltmeter error, thermal drift and ageing are all small compared to the voltage range between full and empty. They're given significance by building a top-balancing BMS into a battery, obliging us to charge it into the zone of dwindling return and eat away at its safety margin.

 

Were it not for that and with good labelling, safe generic charging could be much easier AIUI.

Why do they go so far off then when the battery runs down?

they all go wrong when the battery is flat, not only the cheap or analog ones. Op amps and comparators are sensitive to their Vcc.

You can get 20,000 count true RMS DVMs for about £20-£30.

Not in my experience. Lots of multimeters I've tested from a range of manufacturers are not accurate enough for our purposes over the 200 V range, regardless of the number of digits. +/-1% is a typically quoted accuracy level, which can be 0.42 V at 42 V. Some are only +/-2% and some I've tested have been more than a volt out at 42 V. Accuracy costs money to implement. Just one of these reference ICs with 0.02% accuracy costs more than some cheap multimeters. These will have cheaper reference ICs, maybe just a 431 chip, it costs pennies, but with an accuracy of only about 1%.

Most of the time, the measurements are on the same target device, a battery for example, so they are differential in nature. It's similar to using a tare. As far as I can see, a true RMS DVM with full 4 digit reading is good enough without needing calibration.

At best, I have been able to get reasonable cheap meters with a consistent error of about 0.1 V. But even with that, is it under or over reporting? 42 V may be reported as 41.9 V, so balancing ok - under the 42 V threshold, or it might report 42.1 V and causing some unnecessary concern if you don't know the amount of error. Or, if the charger is outputting 42.1 V and the meter reports it as 42 V you might think all is fine, when actually you are at risk of over charging. I think you need an accuracy of better than 50 mV at 42 V to have peace of mind.

My super cheapo Rolson multimeter from Halfords seems to match expected voltages on the 500V range 54.6V on a fully charged 48V battery and 42.0V on a fully charged Wisper 36V battery (the Woosh 36V battery is 41.5V which I think is what the SANS charger charges to)

they all go wrong when the battery is flat, not only the cheap or analog ones. Op amps and comparators are sensitive to their Vcc.

You can get 20,000 count true RMS DVMs for about £20-£30.

 

Can you give an example of a £20 to £30 DVM with 0.01% accuracy in the 200 V range?

Can you give an example of a £20 to £30 DVM with 0.01% accuracy in the 200 V range?

 

why do you need a 0.01% accuracy for your bike? You only need a relative, not absolute, accuracy for most application. If you measure for example the output voltage of your battery in an hour or two, the error bias is the same, you compare the reading with previous readings, the difference does not depend on the absolute accuracy. The margin of error should be within +/- least significant digit on voltage or frequency measurements, +/- 5 for amps.

With a true RMS, intergrating multitester, you already remove the bulk of the causes for errors like supply voltage, ambient temperature in the first place.

why do you need a 0.01% accuracy for your bike? You only need a relative, not absolute, accuracy for most application. If you measure for example the output voltage of your battery in an hour or two, the error bias is the same, the margin of error should be within +/- least significant digit on voltage or frequency measurements, +/- 5 for amps.

With a true RMS, intergrating multitester, you already remove the bulk of the causes for errors like supply voltage, ambient temperature in the first place.

 

 

No, you do need an absolute, not relative value. You have only a 200 mV window between not balancing (less than 41.8 V) and cell overcharging (over 42 V). My feeling is that you want an accuracy of at least 40 mV to be confident you are in the right range. So then if your DVM measures a reading of 42 V, at the low end it could be 41.96 V - ie balancing ok but not over charging, or at the high end 42.04 V so a bit over, but enough probably to be taken up by shunts and other losses in the BMS (eg MOSFET Rds on resistances etc).

the balancing function does not have to be triggered at precisely 41.8V. A lot of battery control ICs check the voltages of individual banks all the time. When any bank shows a difference in voltage with its adjacent bank more than a set value (eg 35mV), the balancing function is triggered until it is brought to less than 15mV. That's why when a battery is out of balance, its days are numbered.

the balancing function does not have to be triggered at precisely 41.8V. A lot of battery control ICs check the voltages of individual banks all the time. When any bank shows a difference in voltage with its adjacent bank that is more than a set value (eg 35mV), the balancing function is triggered until it is brought to less than 15mV. That's the reason why I have not come across a battery out of balance for several years.

 

There are some fancy ones around, some where you can even customise your own voltage for balancing with an app etc, but most don't. Look at the specs of your average BMS. Balance starts at 4.18 V. Why don't you pull a pack apart, deliberately discharge one bank slightly, charge the pack to less than 4.18 V/cell and see if it balances? Of course, you will need a calibrated DVM to see that you are actually charging at less than 41.8 v

 

Plus, if you have a DVM which is one volt out on the 200V range and you measure 42 V from your charger, your battery could be getting 43 V, reducing it's life and increasing the risk of fires etc - and you would be none the wiser. So you do need an absolute not a relative measurement.