BK 5491B DC Accuracy

[Circuitous]

I’ll follow this up with a more detailed review and possible teardown of a BK 5491B DMM (50,000 count).
I want to make sure that I understand the specified accuracy of this meter on DC voltage.
This specification is slightly different from Dave’s video:
    5491B DC Accuracy = +/- ( 0.02% of reading + 0.008% of range) 

So, if I measure a 10.000 volt source, using the 50 volt range, the reading should be:
   +/- (10.000 x 0.0002 + 50.000 x 0.00008) = 0.002 + 0.004 =    +/- 0.006 volts

When measuring a 10.000 volt source, the reading should be between 9.994 and 10.006, right?

What I’m getting is a very, very repeatable:  10.018 volts.  When I reverse the leads, I get -9.982 volts.  This seems way out of spec, I originally questioned the accuracy of the voltage source, but not any more (I’ll post more on that later).  When I short the leads, the meter reads 0.000, could the internal voltage reference be way off? 

I appreciate any thoughts on this. 

[amspire]

I’ll follow this up with a more detailed review and possible teardown of a BK 5491B DMM (50,000 count).
I want to make sure that I understand the specified accuracy of this meter on DC voltage.
This specification is slightly different from Dave’s video:
    5491B DC Accuracy = +/- ( 0.02% of reading + 0.008% of range) 

So, if I measure a 10.000 volt source, using the 50 volt range, the reading should be:
   +/- (10.000 x 0.0002 + 50.000 x 0.00008) = 0.002 + 0.004 =    +/- 0.006 volts

Isn't the specification for accuracy +/- ( 0.02% of reading + 0.08% of range)  which means +/- 0.042 volts for the 50V range?

I think you are within the specs.

But there is also the issue of whether something is going on with the measurement. There is a +0.018 offset at both the negative and positive ends.

Have you tried swapping the two probes probes, or even use some freshly stripped plain copper insulated wire? Also how is the reference powered - by batteries or from a mains power supply? If it is mains pwered, try measuring a 9V battery in both directions.

Do you need to use the REL button to get 0.000 volts when the leads are shorted, or are you getting 0.000V with the REL annunciator off?

Richard

[Bored@Work]

DC Accuracy = +/- ( 0.02% of reading + 0.008% of range) 

...

So, if I measure a 10.000 volt source, using the 50 volt range, the reading should be:
   +/- (10.000 x 0.0002 + 50.000 x 0.00008) = 0.002 + 0.004 =    +/- 0.006 volts

You are using the equation the wrong way "real value -> displayed value", while the equation is for "displayed value -> real value".  But the good news is, the error in doing this is minimal for your particular values (the range offset dominates). Therefore,

When measuring a 10.000 volt source, the reading should be between 9.994 and 10.006, right?

is acceptable, although the correct statement would be "when reading a value of 10.000 V the real voltage is between 9.994 V and 10.006 V."

What I’m getting is a very, very repeatable:  10.018 volts.

If the meter would be within calibration this would mean the real value is somewhere between 10.012 V to 10.024 V.

When I reverse the leads, I get -9.982 volts.

If the meter would be within calibration this would mean the real value is somewhere between -9.988 V to -9.974 V.

Now comes the funny part. If you calculate the average of the voltages you measured (ignoring sign), you end up with 10.000 V. Did you do the measurement in relative mode?

[Circuitous]

Good questions.  I rechecked the accuracy in the datasheet, online manual, and the printed manual are:  0.02% of reading + 0.008% of range.  I can get out the calibration sheet tonight and check that as well.

I only used one set of probes, I’ll try another set (different maker) and see how that looks.

The meter zero’s when I connect the probes, without using REL.  I did try it with REL and got the exact same readings.  I will try something like giving it a 10.000 V input, then using REL and measuring another voltage, comparing that to what I would have expected to get if I hadn’t used REL.  Perhaps that will remove the 0.018V offset.

the equation is for "displayed value -> real value". 

Good point, I’ll try to correct my phrasing in the future.  I just don’t always think of it that way.

I haven’t had a chance to call BK and see if their $100 calibration will correct the meter, or just tell me how far out of spec it is.
Does anyone know if the calibration for this meter is kept in software, or is just a pot?

I'll run a few more tests and post some more data... might not be able to do that tonight.

[amspire]

Good questions.  I rechecked the accuracy in the datasheet, online manual, and the printed manual are:  0.02% of reading + 0.008% of range.  I can get out the calibration sheet tonight and check that as well.

I was looking at this spec sheet:

http://www.tequipment.net/pdf/BK/5491B_Datasheet.pdf

It says that 50V DC voltage is 0.08% of range while the 50mA current range is 0.008% of range. You are not looking at the current accuracy table instead of the volts table?



Richard.

[alm]

I would expect that to be a typo. It seems unlikely that the 500 mV range has a much lower offset error than the higher ranges. The offset error, even the absolute offset, is also worse than the lower-specced 2831E model. 40 LSDs of offset also seems quite bad for a 4.5 digit meter.

[Bored@Work]

The tequipment datasheet is outdated. The one at http://www.bkprecision.com/downloads/datasheets/2831Eand5491B_datasheet.pdf has 0.008% And that makes more sense, since the meter is actually a Tonghui TH1942 http://www.tonghui.com.cn/en/goods/index/20.html and has of course the same spec.

[amspire]

I would expect that to be a typo. It seems unlikely that the 500 mV range has a much lower offset error than the higher ranges. The offset error, even the absolute offset, is also worse than the lower-specced 2831E model. 40 LSDs of offset also seems quite bad for a 4.5 digit meter.

You are right. I just downloaded the latest spec sheet from B&K and it does say 0.008%.

Getting the specs wrong by a factor of 10 for the volts range is pretty big mistake.

The reason I was suggesting reversing the cable is that it is possible (though unlikely) that some kind of contamination or flaw is generating 18mV offset voltage.  It is unlikely that thermal voltages would cause a 18mV error, unless something has happened like a connection that has a thin conductive copper oxide layer between the probe wire and the probe metal. If there was a faulty cable, it would probably have a higher resistance then the other cable. Copper oxide can generate high thermal emf's and it can also have a rectification properties.

Richard.

[Circuitous]

since the meter is actually a Tonghui TH1942

Good info... wish I'd discovered that 10 months ago.

The reason I was suggesting reversing the cable is that it is possible (though unlikely) that some kind of contamination or flaw is generating 18mV offset voltage.  It is unlikely that thermal voltages would cause a 18mV error, unless something has happened like a connection that has a thin conductive copper oxide layer between the probe wire and the probe metal. If there was a faulty cable, it would probably have a higher resistance then the other cable. Copper oxide can generate high thermal emf's and it can also have a rectification properties.

That's interesting.
Last night I used a brand new(2 months old) set of cables for this set of tests... got the same results as before. But, I ran a few more tests that might be informative, or maybe not.
I setup a stable power supply to about 15 volts, the meter read 15.012 and (when reverse biased -14.976), leading me to believe the voltage is 14.994.  (that again would confirm the offset of 0.018)

I used the same 10.000 volt reference connected to the meter(which displayed as 10.018), and hit REL.  Then, there I connected it to a 15 volt supply (that powers the reference board). It measured +4.994, which if combined with the 10.000v reference,  indicates the 15v source is at the estimated 14.994.

The manual and datasheet indicate that the calibration is stored in software... does anyone know how to change the calibration settings?

I may just pay BK to calibrate it, but I don't have much confidence that they'll do any better at it than the original calibration.
I'll try to call their tech support tomorrow.

[amspire]

I have a 7 1/2 digit Advantest DMM and the calibration procedure does include a step to equalize the positive and negative readings of the A/D.

So the B&K could easily have same thing in its calibration procedure. Have you found a service manual or calibration procedure?

Richard.

[Circuitous]

Have you found a service manual or calibration procedure?

I've searched under BK and under Tonghui, but no luck.  I expect there's a power on sequence to gain access to the setup menu.  According to the manual, it's possible to change the calibration settings through the USB connection using the command protocol... but, they seem to have left out those commands(intentionally, no doubt).

I appreciate the feedback I've received on this (my 3rd post).  This is a great forum!
As I said initially, I'll follow up with a review when I can.

[amspire]

Section 5 in this document for the 5941A meter says how to get into calibrate mode, and it does say that all the calibration can be done from the front panel or via RS232.

Unfortunately for the detailed calibration procedure, it say to contact "factory or authorized distributor"

http://128.238.9.201/~kurt/manuals/manuals/Other/BK%20PRECISION%205491A%20Introduction.pdf