Brymen multimeters - fault?

[joeqsmith]

What about Fungus's favorite meter, the Fluke 101?  Fungus recommended that meter to me for the very first $50 shoot em up test I ran and it cam out a clear winner.  The one I have has never been apart. Not even for a peak!  It has survived every thing I have thrown at it to date and works great. It's an averaging meter but again not a big deal for what we are doing.  Surely this is our perfect meter!

It has a mVAC range and appears to over range at 660.0mV AC RMS 60Hz sinewave.  Injecting 1.7VACp-p meter reads roughly 600mV as expected.  Adding 300mV DC of bias, the meter reads 13mV high.  With 600mV bias, the meter reads 10mV low.  At 1VDC bias, its now reading 480mV.  At 1.9VDC bias, its all the way down to 85mV and no indication there is anything wrong!  Big fail for the Fluke 101! 

[joeqsmith]

I do own one meter that does not have a problem with this bias stuff, well at the levels we are playing around with.  It's the YX-360TR.  This was may be the least amount of money I have spent on a meter next to the free HF ones.   The meter was damaged, repaired and then several mods were made to it to attempt to make it more  robust.  This one is almost bullet proof.

Did I mention it is analog?

So what makes this meter work.  They have a separate physical banana connector just to handle the DC bias case and this input is connected to the main input connector using a film blocking capacitor. 

[IanB]

Well, I just tried with my HP3456A. It behaves well in all my tests and I cannot find fault with it.

If I put it in the 100 mV DC range and feed it 100 mV DC it indicates correctly. If I now add a 1 V RMS AC signal on top the meter shows OL. However, if I switch to the 1 V DC range I again see 0.1 V.

Conversely, if I switch to the 1 V AC range and feed it 1 V AC it shows as such. It shows OL if the voltage is increased above 1.2 V.

Now if I keep the AC voltage at 1 V and add a DC bias, nothing happens to the AC reading. Even if I add 20 V of DC bias, the AC reading remains unchanged to the full display resolution.

So as far as I can tell, the 3456A either shows OL, or it autoranges if enabled, or it shows the correct reading. Nothing I tried could induce it to show an incorrect value on the display.

[evava]

Oh, it is like detective story, isn't it?
Joeqsmith, please please for the last time, Fluke 87V and 289, manual range!

[Wytnucls]

Fluke 185 50,000 count
500mV manual range
test waveform: Sine 480mVAC 9VDC offset 60Hz

Selection:
AC: 487mV 60Hz
DC: OL
AC DC:  OL   - mVAC
AC+DC: OL   - Hz (60Hz with < 1VDC offset)

Auto range mode has no discrepancies:
AC: 0.489mV 60Hz
DC:  09.07V
AC DC: 09.04V  00.49V
AC+DC: 09.06V 00.00Hz ( 60Hz <4VDC offset)

[2N3055]

As I said before it is a matter of overloading dynamic range of analog and/or digital path in these instruments..

I looked at schematic of Voltcraft 940, similar to UT 71 series... Input goes trough input attenuator and into multimeter chip all DC... there is AD636 TRMS converter that is run as a loopback from the chip itself.. Before AD636 there are two opamps serving as buffers.. On those buffers, there is a switch and DC decoupling capacitor.. This is the place where they decouple, not on the input.

So basically, composite signal (AC+DC) goes directly into multimeter chip and is always AC+DC, all the way through the second opamp buffer.  Complete signal path before that sees full voltage, AC+full DC offset.
(That is exactly the opposite what a good analog meter does.. )

So if you apply 200mv 50Hz sine wave with 1V offset, and try to measure it on 400 mV range, you are overloading analog path.. Where exactly it will happen, depends on dynamic reserve of said path..
But 9V batteries used and single supply op amps give maybe +-4V,  if that.. On 400mV range you're running at 10x amplification, so 400mV overload on input becomes 4V of dynamic range...

I didn't look any further and I can't be sure, but I'm guessing majority of multimeters mentioned here has architecture similar to this.. And it's a source of "error" we see here..

I agree that it is not best way to do AC/DC switch.. I'm also guessing it was done this way to have guaranteed frequency response (controlled impedances and such).

But I know about this, and measure with this in mind.. I think you can confuse any instrument if you try hard enough..

[Dr. Frank]

The fault of the BM869 in all ACV / DCV modes can be confirmed and also quantified:

As the peak voltage (AC + DC) exceeds about 1.095V, an internal amplifier clips the signal without overload indication, and the TRMS circuit / software creates increasingly high measurement errors, for both the Udc and Uac components of the signal.

I made a measurement sequence, using an HP3325B signal generator for ACV plus DC offset at 60Hz, and the BM869 only.
I configured the DMM to 5V Udc, DCV and ACV dual display, and the 500mV Udc range likewise.



On the 5V range, the DMM would always display correct and precise values, checked by the HP3458A in DCV, ACV and DCV+ACV mode.
I compared the 5V range readings with the related 500mV range  readings, and calculated the % error between them, and the resulting Vpeak.

Errors of about < 0.2 .. 0.4% between both ranges indicate correct readings; this limit is due to the 10 times smaller resolution of the 5V range.
Every error above that limit indicates the described clipping effect, which sets in sharply at about 1.10V.








In the extreme case of overloading with 0.56Vdc, the DMM still displays no OL, and a really big error of -15% occurs.

As you see, the battery was already a bit low at 7V, and 6.2V with lit display.
This did not affect the accuracy greatly, though I turned off the light at the crucial limit values.
This 1.095Vpeak limit is valid as well for a fresh battery @9.3V.


This effect will also appear on higher ranges, I assume. So that's a really bad characteristic of the BM869, to create such gross errors w/o any warning or OL indication.

Anyhow, an unknown DC/AC signal can always be analyzed correctly by simply switching to the next higher range.

Frank

[2N3055]

I did a post earlier but deleted, as it was unclear and convoluted.. I also did some checking with UNi-T 71C, but it doesn't have 500mV AC range..  On 4V no problems, both my siggens couldn't apply enough DC offset to make it go into error..

[joeqsmith]

Oh, it is like detective story, isn't it?
Joeqsmith, please please for the last time, Fluke 87V and 289, manual range!

None of this should be a surprise based on the original posts.  Well, maybe that Brymen was being isolated for what ever reason.  I do not own the meters you mention.  Someone else would have to check them. 

[IanB]

I wonder if that "discovery" by user hgg in thread "A-look-at-the-uni-t-ut210e" regarding Brymen meters is truth, can someone confirm that (Lightages ?)
I dont want cross-link, but I am talking about "Reply #304 on: October 28, 2016, 08:30:20 PM" and following posts (where he shows that mV AC range does not indicate an overload at specific AC signal with DC component).

I was sort of before buying BM869s for my Christmas gift  , but now I hesitate a little.
Would it be possible to fix that eventually by firmware upgrade at home?

Or is it not problem at all?

I think now you have your answer?

The problem exists with many meters, not only with Brymen.

With knowledge of where and how the problem can occur, you can make sure to avoid it (for example, by always measuring on a higher range first and then down ranging when you know the voltage magnitudes).

This would not be a good reason to hesitate about buying a BM869s.

[Lightages]

I have emailed my contact at Brymen and I am sure he will respond early next week.

[joeqsmith]

Well, I just tried with my HP3456A. It behaves well in all my tests and I cannot find fault with it.

If I put it in the 100 mV DC range and feed it 100 mV DC it indicates correctly. If I now add a 1 V RMS AC signal on top the meter shows OL. However, if I switch to the 1 V DC range I again see 0.1 V.

Conversely, if I switch to the 1 V AC range and feed it 1 V AC it shows as such. It shows OL if the voltage is increased above 1.2 V.

Now if I keep the AC voltage at 1 V and add a DC bias, nothing happens to the AC reading. Even if I add 20 V of DC bias, the AC reading remains unchanged to the full display resolution.

So as far as I can tell, the 3456A either shows OL, or it autoranges if enabled, or it shows the correct reading. Nothing I tried could induce it to show an incorrect value on the display.

Good to know.  Of course, now I need to run my old bench meter like you asked me to in the first place.     This meter is the same one I show in: 

https://www.eevblog.com/forum/testgear/bm235-variability/ 

Again, bought it brand new from HP and has never been aligned.  The DC is effected by the AC level.  But the AC is not effected by the DC level.   

This meter replaced my HP3468A which I also bought brand new from HP with that old HPIL bus.  I had it cal'ed in 2002 but it has never been aligned.  This meter for what ever reason does not appear to have this problem.  In manual range it just over ranges.    So the cheapest meter I think I have bought and the most expensive one don't have a problem.     I wonder about that HIOKI....

[joeqsmith]

Looking at my HIOKI DT4252, the AC does not appear to be effected by the DC level but the DCmV measurements can show inaccurate readings.  Note that the DCV does not appear to have a problem at least at the levels I have been running at. 

So my Made in USA old HP meter comes out on top of my heap for this test anyway.  Not a bad meter except I was ignorant about HPIB at the time and was not able to predict it's future..


Now who is going to call every other manufacture?   

[IanB]

Good to know.  Of course, now I need to run my old bench meter like you asked me to in the first place.     This meter is the same one I show in: 

https://www.eevblog.com/forum/testgear/bm235-variability/ 

Again, bought it brand new from HP and has never been aligned.  The DC is effected by the AC level.  But the AC is not effected by the DC level.

On the 34401A the DC readings are influenced by the amount of AC present? 

This meter replaced my HP3468A which I also bought brand new from HP with that old HPIL bus.  I had it cal'ed in 2002 but it has never been aligned.  This meter for what ever reason does not appear to have this problem.  In manual range it just over ranges.    So the cheapest meter I think I have bought and the most expensive one don't have a problem.     I wonder about that HIOKI....

So the 3468A behaves better than the 34401A?

Which is the most expensive meter, the 3468A or the 34401A? I would have guessed the 34401A to have cost more than the 3468A based on features and year of manufacture.

[IanB]

So my Made in USA old HP meter comes out on top of my heap for this test anyway.  Not a bad meter except I was ignorant about HPIB at the time and was not able to predict it's future..

Seems HP knew what they were doing and made good stuff.

The 3456A seems like a very nice instrument, if only it wasn't so damn big! It is hard to call it a bench meter, since it doesn't fit neatly on the average bench (or shelf, for that matter). It is more of a rack meter.

[joeqsmith]

Good to know.  Of course, now I need to run my old bench meter like you asked me to in the first place.     This meter is the same one I show in: 

https://www.eevblog.com/forum/testgear/bm235-variability/ 

Again, bought it brand new from HP and has never been aligned.  The DC is effected by the AC level.  But the AC is not effected by the DC level.

On the 34401A the DC readings are influenced by the amount of AC present? 

For the one I own, that appears to be the case.  I bought it when they first came out and newer versions may behave differently.

So the 3468A behaves better than the 34401A?

Which is the most expensive meter, the 3468A or the 34401A? I would have guessed the 34401A to have cost more than the 3468A based on features and year of manufacture.

For this example, would say yes. 

As for the cost, I would say its a wash.  In 92, the 3468A was $820 or $1412 in today's dollars.  In 97 the 34401A was $995 or $1498 in today's dollars. 

Let's contrast that with your 3456A which in 1992 without any options listed at $5696  or $9812 in today's dollars.   Would have been just a little overkill for my hobby work back then.    

[evava]

I think now you have your answer?

The problem exists with many meters, not only with Brymen.

With knowledge of where and how the problem can occur, you can make sure to avoid it (for example, by always measuring on a higher range first and then down ranging when you know the voltage magnitudes).

This would not be a good reason to hesitate about buying a BM869s.

+ for complete certainty try even use decoupling capacitor in series with the meter.

Thank you all again.

That thread https://www.eevblog.com/forum/testgear/the-uni-t-ut139c-and-my-power-supply-ripple/ concerns probably similar/same issue.
It seems, that higher rank Flukes (87V, 185, 189, 289) do not suffer from this issue.
In this respect Brymen does not stand out from the crowd of common meters, unfortunately.
Price difference tell that.

I probably will purchase BM869s, but am a little bit disappointed.

[IanB]

I probably will purchase BM869s, but am a little bit disappointed.

I don't think you should be that disappointed. Every tool has its strengths and weaknesses. It is important to understand that in the real world you cannot achieve something that is ideal in every possible way. There are always practical considerations about what actual hardware can achieve.

Let me close by showing you two pictures. I can assure you I am in no way disappointed with how this meter performs.

[joeqsmith]

Thank you all again.

That thread https://www.eevblog.com/forum/testgear/the-uni-t-ut139c-and-my-power-supply-ripple/ concerns probably similar/same issue.
It seems, that higher rank Flukes (87V, 185, 189, 289) do not suffer from this issue.
In this respect Brymen does not stand out from the crowd of common meters, unfortunately.
Price difference tell that.

I probably will purchase BM869s, but am a little bit disappointed.

They only talk about the 87V and are not specifically looking to see if they can cause the meter to show an invalid result.   

Do you have data that shows the other meters you mention of "seems" is your feelings about it?

[evava]

I do not have any detailed data, my guess is based exclusively on that thread (https://www.eevblog.com/forum/testgear/the-uni-t-ut139c-and-my-power-supply-ripple/ and on that post by Wytnucls #56  https://www.eevblog.com/forum/testgear/brymen-multimeters-fault/msg1078871/#msg1078871 .

Yes, I agree with you that it would be more proper for all Flukes to undergo proper investigation like shown by you and others in this thread.
I already have asked you, I thought that (after your videos) Flukes 87V and 289 you have at least available from your friends.

Thank you in advance!
(I learned many things from this thread)

[joeqsmith]

I do not have any detailed data, my guess is based exclusively on that thread ...

I already have asked you, I thought that (after your videos) Flukes 87V and 289 you have at least available from your friends.

Thank you in advance!
(I learned many things from this thread)

Ok, it was a guess not a fact.  Just wanted to make sure you were not hiding some data on us.

[Wytnucls]

What data for the Fluke 185 did you not understand?

[joeqsmith]

What data for the Fluke 185 did you not understand?

Fluke 185 50,000 count
500mV manual range
test waveform: Sine 480mVAC 9VDC offset 60Hz

Selection:
AC: 487mV 60Hz
DC: OL
AC DC:  OL   - mVAC
AC+DC: OL   - Hz (60Hz with < 1VDC offset)

Auto range mode has no discrepancies:
AC: 0.489mV 60Hz
DC:  09.07V
AC DC: 09.04V  00.49V
AC+DC: 09.06V 00.00Hz ( 60Hz <4VDC offset)

I was not sure based on your data that you looked at the DCmV and increased the ACV.  You need to check both cases. 

I already have asked you, I thought that (after your videos) Flukes 87V and 289 you have at least available from your friends.

Thank you in advance!
(I learned many things from this thread)

The last 87V has moved from the area but I did talk with the 289 owner...

In the mV AC range it over ranges at 1.558p-p or 550mV RMS, again 60Hz sinewave
In the mV DC range it ORs at 550mV DC
With 500mV AC RMS and up to 4.2VDC bias, there was no effect.
With 400mV DC applied and with 5.467Vp-p the meter reads 550mV DC or an error of 150mV.  No errors or warnings.   I thought this meter wasn't going to have any problems?     
Now, if you select the AC and DC to enable the spit display (still in mV) the meter displays the correct information.  The owner was amazed.   

***************************************************************************************

You asked about the Fluke 189 and while I don't think I know anyone with that meter, I do know someone with a 179 which is a much lower end meter but still a good data point.  This meter has a DCmV range but was tested in manual range for ACmV.

In the mV AC range it over ranges at 660mV RMS, again 60Hz sinewave
In the mV DC range it ORs at 660mV DC
With 600mV AC RMS and up to 4.2VDC bias, there was no effect.
With 400mV DC applied and with 1Vp-p the meter reads 399.8mV DC.
With 400mV DC applied and with 10Vp-p the meter reads 198.7mV DC, about half what it should.  Again, no OR error. 

That's six Flukes I've looked at now that I can cause to read inaccurate without an over range.  Too bad not one member is willing to run their 87V to see.

[evava]

We started this thread by overloading AC mV range by DC, gradually we moved to overloading DC mV range by AC - interesting.
Some meters know how to decouple DC, but "decouple" AC is impossible/much more difficult.

Still, real problem is the first, overloading AC mV range by DC. Not problem for higher Flukes.
Second problem (overloading DC mV range by big AC) is solvable only by extremely huge dynamic range.

(damn it) What prevents all meters from recognizing overload and reporting it?

Your outcomes suggest that any meter may be affected, even the big expensive one, only at much higher levels (and thereby less likely to experience).
Very interesting.

[joeqsmith]

I thought you just wanted to know if there was any modes where a meter could give inaccurate data without some sort of warning to the user.  If we wanted to start adding constraints, seems we missed the point.  The meter should always give the correct data or nothing at all with some sort of error. 

When I ran my old HP which is one of the most expensive meters I have bought for home, I checked it both ways and it had no problems at all.  Demand Excellence, Demand HP.  Oh yea, I guess you can't anymore.  Too bad that little Keysight meter I had with it's fancy GDTs failed during my testing.  That was the only Keysight product I have purchased.

It will be interesting to hear what Brymen has to say.