EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: picclock on April 19, 2022, 08:18:53 am
-
I am looking for a portable multimeter which will give fairly accurate DC current readings. Some of the Brymen meters (BM821s, BM822s, BM827s) offer 0.2% on the 6 Amp range, but I think that is the best. I am typically measuring between 0.2 and 1.5 DC Amps, and most meters offer really poor precision in this area (ie testo 760-2 1.5%, Brymen bm869 0.5%, Fluke 117 1%[the same as an Avo model 8 :scared: ]).
The specs are also pretty vague as to whether the % accuracy is of maximum range value, or reading. For instance, 0.2% on the 6 amp range is 12mA, but that error could be 12mA 600mA, which would give a reading error of 2%. Alternately if the 0.2% error is of the read value it would only be 1.2mA.
Any help much appreciated.
Best Regards
picclock
-
Instead of Chinese meters suggest you use Western brand, 5 1/2 digit DVM
Alternative is an accurate V function with a meter,shunt
Jon
-
DMM accuracy specifications will generally be of the form X% + Y digits. This means the accuracy is X% * reading + Y least significant digits of the range. Based on this you can calculate the uncertainty for any reading. For a BM827s in the 6A range, the accuracy is 0.2% + 4 digits, and the least significant digit at that range is 1 mA. So the uncertainty of a reading of 1.000A is 2mA + 4 mA = 6 mA.
Dave did a video about this a long time ago. Search the channel for accuracy and counts.
I think you might need to go up to 4.5 digit meters, ideally 20000 count meters, to get the accuracy you want. I don't have specific model recommendations, but I'm sure others do.
-
Build yourself one or more external current shunts. Close tolerance resistors aren't that hard to come by.
If you're interested in accurate current measurement then, almost by definition, you must be concerned about voltage burden too. You can put external shunts very close to the 'scene of the action' leaving the longer leads to the meter to be voltage sense only. You can also optimize the shunt values to your needs, rather than relying on what the meter manufacturer happens to have chosen for a particular current range. Voltage burden isn't something that is necessarily solved by buying a higher resolution bench meter (although it helps).
Here's my collection (maybe you're probably only interested in one value at the moment, I built my collection slowly, as I needed them)... https://www.eevblog.com/forum/testgear/my-shunt-collection-(ucurrent-passive)/ (https://www.eevblog.com/forum/testgear/my-shunt-collection-(ucurrent-passive)/)
The only cautions with this approach are:
- Don't use them in hazardous energy level situations, although if you're interested in sub 1% accuracy, you're unlikely to be measuring mains circuits!
- Fuses cause significant voltage burden. I have omitted them on my shunts and substituted care instead! Higher value, lower current shunts can be protected by back to back diodes (low leakage when applicable).
(https://www.eevblog.com/forum/testgear/my-shunt-collection-(ucurrent-passive)/?action=dlattach;attach=198656;image)
(https://www.eevblog.com/forum/testgear/my-shunt-collection-(ucurrent-passive)/?action=dlattach;attach=198658;image)
-
If you want to do it without using an external shunt (ie if you need traceable cal for ISO certification purposes) you'll have to shell out a fair wedge for something like a Gossen M248A, Fluke 289, or Keysight U1253B.
If you're needing something as a hobbyist, the external shunt is probably the easiest and cheapest route. Used F289's are usually £200-300 in the UK, when they appear for sale at all, and can have issues with supercaps. I don't own an M248A, or U1253B, but I do own a M242A and a U1401B, both used (as was the F289) and find them to be excellent meters. The U1401B has a DCA spec of 0.03% + 5 counts, but only a 500mA max.
-
@AVGresponding
The Fluke 289 is 0.3% and the Gossen M242A 0.9% .
@Gyro
Shunts are good, but they hamper portability. I'm not concerned about voltage burden (within reason) as its easy to compensate for. Voltage around 40-80v, no safety issues.
Thanks for your time and assistance but it looks like I am running out of options.
Best Regards
picclock
-
None of the handhelds will have exceptional accuracy at 10A range...
Mostly because of the self heating of the current shunt in small enclosed space....
-
The specs are also pretty vague as to whether the % accuracy is of maximum range value, or reading. For instance, 0.2% on the 6 amp range is 12mA, but that error could be 12mA 600mA, which would give a reading error of 2%. Alternately if the 0.2% error is of the read value it would only be 1.2mA.
The specifications of any decent meter should be pretty clear--it will be expressed as (ppm/% of reading) + (ppm/% of range OR counts). However, there is probably some variance between manufacturers of actual performance vs specification.
What accuracy are you hoping for in your 0.2 -1.5A range? If 0.2-0.3% isn't good enough, you're in for a struggle. Measuring larger currents always lead to precision issues due to tradeoffs between shunt heating and very low shunt resistances. That and current precision isn't usually the multimeter designers highest priority.
-
@AVGresponding
The Fluke 289 is 0.3% and the Gossen M242A 0.9% .
@Gyro
Shunts are good, but they hamper portability. I'm not concerned about voltage burden (within reason) as its easy to compensate for. Voltage around 40-80v, no safety issues.
Thanks for your time and assistance but it looks like I am running out of options.
Best Regards
picclock
The F289 is 0.15% in the 400mA range, and yes, 0.3% in the 5A range. In reality it's probably better than that, but I haven't checked it against any of my bench DMMs.
Yes, I know the M242A doesn't meet your requirements, the M248A is the one I listed that does. I was merely relaying my personal experience with the brand.
Also another aspect regarding external shunts is they can be made to meet the use case more precisely; you haven't specified what that is, but it's worth pointing out that measuring current is pretty horrible for accuracy if you're doing it constantly as the heating effect on the shunts, particularly in handheld DMMs causes big drops in accuracy. With an external shunt you're at liberty to use much a bigger one that isn't affected as much.
-
An accurate shunt is going to be far cheaper than an accurate meter.
(plus you can easily adjust them with a file...)
-
An accurate shunt is going to be far cheaper than an accurate meter.
(plus you can easily adjust them with a file...)
That's all presuming you have the ability to calibrate them and verify their performance over the entire anticipated range of measurement and environment. That is not a simple issue.
-
You do certainly need to select a good resistor with specified maximum tolerance, temperature coefficient and generous power rating. Not too hard for <0.2% however when you have a reasonable amount of money to spend.
-
The specs are also pretty vague as to whether the % accuracy is of maximum range value, or reading. For instance, 0.2% on the 6 amp range is 12mA, but that error could be 12mA 600mA, which would give a reading error of 2%. Alternately if the 0.2% error is of the read value it would only be 1.2mA.
I think they assume you're using a decent amount of the range, ie. enough that the "0.2%" part of the specification is bigger than the "+/- 20 digits" part of the specification.
-
You do certainly need to select a good resistor with specified maximum tolerance, temperature coefficient and generous power rating. Not too hard for <0.2% however when you have a reasonable amount of money to spend.
If you can afford to get a 4-lead (internally Kelvin connected) resistor with all three characteristics (%/TC/W) guaranteed well within your expected tolerances and current rating, then perhaps it works out. But you're never sure, which may or may not be OK. And I've had some so-called precision resistors that just barely--or just barely not--met their specs, especially tempco. As RR would say 'trust, but verify'.
b/t/w, what type of boxes are your units installed in?
-
The problem is more if you need traceable measurements (it's not clear to me if that's necessary), you either need a cal lab to certify your shunt, or you need to have your own procedure where you use a calibrated bench dmm to characterize and validate the shunt. But depending on your requirements you might need a whole bunch of paperwork to qualify the shunt. At that point lugging around a bench DMM like the HP/Agilent 34401A (0.1% of reading + 0.01% of range for 1A range , 0.12% of reading + 0.02% of range for 3A range) or better Keithley DMM6500 (0.04% of reading + 0.005% of range 1A, 0.08% + 0.010% of range 3A) might be simpler and cheaper. I'm not aware of any of the current bench DMMs coming with a battery option.
If you don't need any of that, than building a stable shunt with either a very low temperature coefficient or otherwise designed so it doesn't heat up much is probably the simplest and cheapest option.
I think they assume you're using a decent amount of the range, ie. enough that the "0.2%" part of the specification is bigger than the "+/- 20 digits" part of the specification.
The Brymen 82xs datasheet (https://datasheet.octopart.com/BM829S-Brymen-datasheet-151710208.pdf) clearly specifies 0.2% + 4 digits. For the OP it might be good to watch Dave's video on this topic (http://youtu.be/U4JFeU-o2kc).
-
b/t/w, what type of boxes are your units installed in?
They are grey crackle paint finish diecast alloy ones, I can't remember now but I think the smaller ones, I got as a cheap job lot from ebay. The larger one is a re-used one that happened to match. They might be Hammond, but no part numbers that I can see.
-
If you don't need any of that, than building a stable shunt with either a very low temperature coefficient or otherwise designed so it doesn't heat up much is probably the simplest and cheapest option.
Add a fan :)
-
In practice, it's not too hard to achieve <10% of power rating for most wirewound shunt resistors, using the meter's lowest mV range. Certainly easy to beat the 10-20s maximum measurement time low duty cycle shunts in a typical DMM. These tend to be effectively 2 wire too, with any kelvin sensing happens on the PCB, after the shared solder joints.
-
The Brymen 82xs datasheet (https://datasheet.octopart.com/BM829S-Brymen-datasheet-151710208.pdf) clearly specifies 0.2% + 4 digits.
I haven't seen those models before. They're not on brymen.eu (https://brymen.eu/). ???
That's surprisingly accurate, especially for a 6000 count meter - more accurate than a Fluke 289!
-
That's surprisingly accurate, especially for a 6000 count meter - more accurate than a Fluke 289!
Yes, as I said above "there is probably some variance between manufacturers of actual performance vs specification".
The stated accuracy spec is actually quite astonishing when you consider that given the stated burden voltage, they are likely using a 10mR shunt and therefore the 60mV range--which itself is 0.12% + 2d. So the current shunt and measuring circuit only add 0.08% + 2d, even at 10 amperes. That is......unbelievable? Or perhaps they have a different confidence interval in mind. Or something.
-
Thanks for all the info and assistance.
Since posting further research on Brymen has revealed 3 meter families which have a 0.2% DC Current accuracy.
The best for my use are the BM521 and BM525 models which also incorporate logging of data. Sadly these seem to be unobtanium. (stock in Poland but no ship UK)
The other ranges are the BM811,812,815,817, and the BM821s, BM822s, BM827s, BM829s, family.
The one thing that puzzles me is that if some of the above meter ranges can do this, why is it not implemented on the better quality meters like the 78 and 86 families ?
I am UK based so will likely have to import.
Many Thanks
Best Regards
picclock
-
I am UK based so will likely have to import.
Thank Brexit!
-
Both of my Fluke 189 and 289 are only 0.5% in the 5A range.
-
The best for my use are the BM521 and BM525 models which also incorporate logging of data. Sadly these seem to be unobtanium. (stock in Poland but no ship UK)
I can vouch that BM525S is a very good instrument..
-
Is Welectron an option for obtaining the BM525s? https://www.welectron.com/Brymen-BM525s-Multimeter_1 (https://www.welectron.com/Brymen-BM525s-Multimeter_1)
Published specifications are such a minefield. Some equipment just barely meets the spec, others exceed them by a factor of 10.
-
Published specifications are such a minefield. Some equipment just barely meets the spec, others exceed them by a factor of 10.
There are other considerations as well and there are many factors beyond calibration, drift and tempco that can come into play. Fluke intends their specification for the current ranges to reflect the maximum permissible error under the entire operating temperature range and with continuous operation (like logging). So if you put 10 amps through it, your initial reading will be in spec and an hour later, it will still be within spec. If the ambient temperature changes from 18C to 28C as well, it still will meet the spec. Accounting for all that means if you don't expose it to those widely varying conditions, you may get a much better result. Not accounting for all that means your meter may only barely meet the specifications under certain, ideal conditions.
For example, the published specifications for a Fluke 289 at 5A are 0.3% + 10 counts (out of 50000). This means that the specified tolerance comes out to 160 counts, so for exactly 5 amps a reading of 4.9840 to 5.0160 is 'in spec'. The calibration report on my F289 @ 5A shows 4.9994, or 6 counts error. That is over 26X better than spec and could be expressed as "0.01% + 2 counts". This isn't heroic over-performance, it is simply what it takes to ensure that it meets the published specs under all listed conditions 99% of the time. Comparing single value specifications or even worse, "basic DC accuracy", is not actually accomplishing much.
-
For example, the published specifications for a Fluke 289 at 5A are 0.3% + 10 counts (out of 50000). This means that the specified tolerance comes out to 160 counts, so for exactly 5 amps a reading of 4.9840 to 5.0160 is 'in spec'. The calibration report on my F289 @ 5A shows 4.9994, or 6 counts error.
My Fluke 187 can beat that. The calibration report says that at 4.0A it read 3.9998 out of a permissible range of 3.979 to 4.021.
At 0.4A it read 0.4000 :-+
So, yeah, any of these high end Brymens/Flukes will probably read <0.2% error in a temperature controlled room. They just don't garantee it.
Is Welectron an option for obtaining the BM525s? https://www.welectron.com/Brymen-BM525s-Multimeter_1 (https://www.welectron.com/Brymen-BM525s-Multimeter_1)
If they stock it then, yes, Welectron is a good place to buy stuff.
I think they also sell their Brymens with calibration certificates for a bit extra money.
-
The specs are also pretty vague as to whether the % accuracy is of maximum range value, or reading. For instance, 0.2% on the 6 amp range is 12mA, but that error could be 12mA 600mA, which would give a reading error of 2%. Alternately if the 0.2% error is of the read value it would only be 1.2mA.
It better be 0.2% of the read value because that represents the gain error which is a combination of the error in the reference voltage and shunt resistance.
The datasheet says 0.2%+4d so up to 6 amps, the error could be 4 milliamps with zero input and 16 (12+4) milliamps at 6 amps, and 60 (20+40) milliamps at 10 amps.
The basic DC accuracy of this meter is 0.7%+6d, so not very good for a 6000 count meter. There are 4000 count meters with much more reasonable 0.1%+2d basic DC accuracy.
I become very suspicious when the amps accuracy is better than the basic DC accuracy.
The one thing that puzzles me is that if some of the above meter ranges can do this, why is it not implemented on the better quality meters like the 78 and 86 families ?
The optimistic view is that lower drift references, dividers, shunts, and ADCs cost more money and may require much more involved temperature compensation.
-
I become very suspicious when the amps accuracy is better than the basic DC accuracy.
It's definitely weird.
In theory the meter is using its DC measuring circuit to measure the voltage drop across a shunt. Even if the shunt is perfect it shouldn't be more accurate than the DC voltage accuracy.
OTOH it could be a marketing thing. Maybe the boss told them to make a meter with "0.5% DC accuracy" so that's what's written in the spec and the current accuracy slipped past the editors. :-//
-
The basic DC accuracy of this meter is 0.7%+6d, so not very good for a 6000 count meter. There are 4000 count meters with much more reasonable 0.1%+2d basic DC accuracy.
I become very suspicious when the amps accuracy is better than the basic DC accuracy.
If you're reading the same datasheet I am, that 0.7% is for the TRMS AC+DC function, the DCV spec is 0.12% + 2 counts. Which leaves you needing a very stable current shunt to only have 0.08% + 2 counts additional error. That is probably achievable if you only test the meter briefly under the exact same conditions as you set the calibration constants with plenty of rest in between current applications. What would be interesting is to get one of these meters with logging capability and then set it up with a calibrated current source and log it for an hour.
-
The basic DC accuracy of this meter is 0.7%+6d, so not very good for a 6000 count meter. There are 4000 count meters with much more reasonable 0.1%+2d basic DC accuracy.
I become very suspicious when the amps accuracy is better than the basic DC accuracy.
If you're reading the same datasheet I am, that 0.7% is for the TRMS AC+DC function, the DCV spec is 0.12% + 2 counts. Which leaves you needing a very stable current shunt to only have 0.08% + 2 counts additional error. That is probably achievable if you only test the meter briefly under the exact same conditions as you set the calibration constants with plenty of rest in between current applications. What would be interesting is to get one of these meters with logging capability and then set it up with a calibrated current source and log it for an hour.
I read the specifications wrong then.