Brymen BM869 review – By the eye of the Industrial electrician.

[ebastler]

Why do you assume the relationship is linear? It might not be. A simple test would put the matter to rest.

Because of the linear accuracy spec, which is +- (0.02% of value + 2 digits).

Why do you assume the spec is wrong, without even a single data point to suggest that?
There is no "matter" to be put to rest, as far as I am concerned.

[Wytnucls]

A simple look at Joe's 1mV pictures, will show that the relationship is not linear and points to a dubious 0.02%+2 accuracy in the mV range, at least for negative values.

[ebastler]

A simple look at Joe's 1mV pictures, will show that the relationship is not linear and points to a dubious 0.02%+2 accuracy in the mV range.

Ah, sorry -- I had missed that post due to the page break. That does indeed look like it's out of spec. (Assuming that the non-existent spec for the 500,000 digit mode is not worse than for the regular mode, i.e. 0.02% + 20 digits.)

[joeqsmith]

I would assume the error in the 1mV is mostly due to the input bias current of the meters being combined.  Running them all in parallel like I show may not be valid if you were actually taking this sort of measurement.  I would need to run them all separate and was just too lazy.  It would make sense they would all swing the same with roughly the same voltages.  I would guess if they were separate, we would get tighter numbers and maybe a flyer meter. 

[joeqsmith]

Repeating +/-1mV from Fluke standard to each separate meter.

[evava]

Thank you for posting these results!

Oh, UNI-T is not bad, and temperature stability test on uV range was also fine.
I only hope that sometimes in future you uncover and describe your input mods to your UT181a (to withstand grill starter at least), if they are doable for common person.
Otherwise I like UT181a very much 

P.S. Kiriakos is certainly happy that his thread lives again 

[joeqsmith]

Thank you for posting these results!

No problem.

Oh, UNI-T is not bad, and temperature stability test on uV range was also fine.
I only hope that sometimes in future you uncover and describe your input mods to your UT181a (to withstand grill starter at least), if they are doable for common person.
Otherwise I like UT181a very much 

Of course the 181A has a non-standard battery that takes about a half day to charge and you can't use the meter while it is charging.  The charger does not come with standard AC adaptors.  They do not offer spare batteries so at least you could swap it out if one dies in the middle of you doing something.   The clear plastic lens is really showing it's age and the meter has only seen bench use.  Of course, no certification.  And yes, seeing it fail so easily like many UNI-T products on the stupid little grill starter is pretty sad.

I see absolutely no advantage to providing that information.  If the people who ask me about it would instead ask UNI-T, you may actually see a better product.  There is little I can do to change their designs beyond making videos that at least may make people aware of the shortcomings.

[evava]

I see absolutely no advantage to providing that information.  If the people who ask me about it would instead ask UNI-T, you may actually see a better product.  There is little I can do to change their designs beyond making videos that at least may make people aware of the shortcomings.

I understand you fully, I only thought that because you provided in full your input and other mods to UT61e.
I admire you how you always are able to revive each damaged meter.
Many thanks for your effort and sharing with us.

[Dr. Frank]

Wow! 2 counts is huge. Accuracy is supposed to be 0.05%+2 in the 5V range.

Unless I am mistaken, 0,05% of 5.00000V is 0.00250V, i.e. 250 counts in 500,000 count mode.
So 2 counts, or even 20 counts as you probably meant to write, is not bad at all.

Full scale is not the problem. The bottom of the scale might be.
Try measuring 600mV for example. If the difference is still 2 counts, the meter could easily fall outside of its 1 year calibration.
...
A simple look at Joe's 1mV pictures, will show that the relationship is not linear and points to a dubious 0.02%+2 accuracy in the mV range, at least for negative values.

I just can't understand your arguments.

The BM869 is specified at 4 4/5 digits, and this spec for DCV as usual contains accuracy for the reading plus accuracy for the range, that is a relative and an absolute error spec.
These errors are valid, independent of the reading, i.e. it fits to Full Scale, as well to low readings.
Both errors simply add up, but do not depend on each other, so there's nothing dubious, and the 2digits error can not violate the relative 0.02% reading error, even at lower readings.

Also, the question of linearity is not appropriate, here.

The absolute error, also often defined as 'error of range' is the dominant value for lower readings, covering effects like internal e.m.f. voltages, and also this roll-over error (difference in reading at input reversal).
So it's a bit convoluted, to distinguish e.m.f. and roll-over error, especially on low voltages as 1mV, and at low scale readings.
Clever e.m.f. cancellation is needed, or usage of the higher range, where e.m.f. does not play a role.


Anyhow, the BM869 obviously meets its specification in all cases, but more than that can not be expected..

6 1/2 bench DMMs probably perform much better, concerning this roll-over error. For example, the HP DMMs feature an additional -10V vs. +10V calibration item, which the handheld DMMs often / mostly do not have. But that comes at a cost..

For 5 4/5 digits of the BM869, the specification should be valid analogously, that means that the absolute error is 10 times higher, .. + 20 digits.
As the integration time is longer for this higher resolution, that implies a different integration algorithm, or maybe a different circuit, so results may be slightly different, but hard to be detected, due to the masked digit in 4 4/5 mode.

Just for completeness, I have measured the performance and linearity of the BM869 a few years ago:
https://www.eevblog.com/forum/testgear/bymen-bm869-measurement-of-dcv-linearity/msg666100/#msg666100, but unfortunately not measured both polarities. I will repeat this experiment accordingly, when I check my BM869 the next time.
As a first shot, the roll-over error on my BM869 at 5V F.S. is about 5 digits in 5 4/5 mode, so well below the +2digits specification.

Frank

[ebastler]

Repeating +/-1mV from Fluke standard to each separate meter.

Many thanks, Joe -- and a good catch regarding the unwanted interaction between the meters connected in parallel. It is reassuring to see that the BM869 is very comfortably within its stated accuracy, once the external bias of the other meters is removed.

[joeqsmith]

I see absolutely no advantage to providing that information.  If the people who ask me about it would instead ask UNI-T, you may actually see a better product.  There is little I can do to change their designs beyond making videos that at least may make people aware of the shortcomings.

I understand you fully, I only thought that because you provided in full your input and other mods to UT61e.
I admire you how you always are able to revive each damaged meter.
Many thanks for your effort and sharing with us.

You give me FAR too much credit.  Very few meters I damage can be repaired.  Many of them use chip on board.  Most you can't get replacement parts for.  So if the meter it not designed to at least protect the control IC, they are pretty much scrap.  The TPI is a rare exception. Two pins on the control IC are partial damaged.  I could not find a replacement IC but continue to use the meter. 

I have been looking into repairing the 121GW and it's been a nightmare trying to find parts as well.

[Bud]

Wheeeeh...what a horrible color 

[joeqsmith]

I checked the pre-production 121GW before dissecting it today.   Again it may not be aligned as well as production and it IS damaged. 

[Crumble]

The value is inverted, but the connector does not seem to be reversed. How did you change the polarity?

PS Sorry for being a whine, just wonderin'.

[joeqsmith]

The value is inverted, but the connector does not seem to be reversed. How did you change the polarity?

PS Sorry for being a whine, just wonderin'.

A) Its magic!
B) Photoshop 
C) Discovered my experimental ubeam project has a side property where the backside radiation can invert any signal and I thought I would demonstrate it 
D) I swapped the connector at the reference standard 
E) All of the above

Updated per CustomEngineerer.  Now we need a poll. 

[CustomEngineerer]

I vote E) All of the above!

[nanofrog]

FWIW, I suspect the proper answer is D.  That said however, I think E would be a lot more fun. 

[Crumble]

The value is inverted, but the connector does not seem to be reversed. How did you change the polarity?

PS Sorry for being a whine, just wonderin'.

Now we need a poll. 

You may guess I got a little suspicious after you tested us once by not showing the reverse pulse on one of the meters (forgot which one it was).

[RedSky]

Stupid question; Can the Brymen BM869 measure frequency and duty cycle of a DC voltage in the 20 volt range?  It notes only Hz at 'logic level' that i can see mentioned in the DC range.

On a side note;  I am quite excited by all these more affordable meters that can measure down to 1 micro volt, including the Brymen's and the 121GW (when it actually gets released) they are excellent to use to measure higher amperage shunts (100A, 100mV) at high resolution.   
I've had a Uni-T UT70d for ages which has a 80mV range (80,000 count) and can measure down to 1 micro volt and the resolution has been very useful indeed although I can't really measure it's absolute accuracy.  It's like most Uni-T meters - poor input protection, but has been fantastic in general.
Just about to invest in the BM867s I think as a second opinion.

[RedSky]

For reference;  I just got a Brymen 867s and the answer is yes, it can measure duty cycle / hz when measuring DC signals in the 20V range (and likely higher still).   This was of course expected, it's just the unusual way it's described in the manual and being that the function works on the DC millivolt range that had me second guessing.

Also of note.   I have one of those cheap Chinese Voltage references that has an aged 584LH in circuit.   When tested the Brymen 867s readings deviated by only 0.004% of the supposed Agilent 34401A desktop meter tested voltages.  At best it was within 1 count 5 decimal places accurate to the reading.   Very impressed.