Fluke 732A DVMP - Splitting Hairs

Started by Hawaii596, 02-24-2016 -- 11:22:29

Previous topic - Next topic
Print
Go Down Pages1 2
User actions

Hawaii596

02-24-2016 -- 11:22:29
Doing a Fluke 10 VDC DVMP cal with Fluke provided 732B.  This is for any who have done this cal and are familiar with it.

So most of my readings are around (absolute value) 2.2 microvolts difference (plus and minus with all the various forward/reverse and lead swapping.

But I am noticing that some of them run up to about 2.7 to 2.8 uV or so.  I reseat the connectors, clean the terminals with Deoxit low thermal red oil, and it calms them down back to about the correct readings.

I noticed that inserting the Fluke supplied leads into their Fluke 732B, and the fit is looser (like the 732B has spent a LOT of time on the road, and the contact dimensions are worn a little).

Just wondering if anyone has experienced this.  I want GOOD readings, which optimize my uncertainties.  I know we are talking about a 0.05 ppm shift (compared with my most recent MU of 0.51 ppm (from last year).

Just wondering if anyone else has had this observation, and if so, thoughts about whether I am splitting hairs (due to the low uncertainty contribution of this delta).  Or is this something to be worried about.

I just went through an iteration, and it dropped a number from -2.755 down to -2.274 (which I consider my good number).

Or is this just normal, and I should not worry, and do the same measurand maintenance I am doing?
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883

metrologygeek

#1
02-24-2016 -- 11:27:05
Stop handling your connectors.

Hawaii596

#2
02-24-2016 -- 11:38:35
I am presuming thermals from my handling.  So I let them sit for a few minutes before taking readings.  Does this sound like maybe it is just thermals?
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883

silv3rstr3

#3
02-24-2016 -- 12:40:33
We actually have a set up now with ten 732A primary standard cells connected to two different voltage scanners.  Once we place the UUT in line with our set up an automated program takes one reading a day for 10 days.  I've been told it's a lot more accurate the more cells you can compare at once.  Only thing we are missing is the J junction that Fluke has.  Anyone familiar with what is the actual J Junction?  Just curious. 
"They are in front of us, behind us, and we are flanked on both sides by an enemy that out numbers us 29:1. They can't get away from us now!!"
-Chesty Puller

Hawaii596

#5
02-25-2016 -- 09:00:42
I probably have around 10 cells or so.  The 732A done with DVMP is the house reference.  I have two 732B's that are working standards that get cal'd against the 732A.  I have a Dataproof 160B that I have yet to put to work in this setup.  A Datron 4-banger (4910??), a couple of other 732A's, etc.  My goal is to achieve whatever I need to do 5720A's, 8508A's, 3458A's (90 Day), and upcoming capability to do 5790A's (working on getting the 792A online now - has 20+ years of history).  Lots of fun.
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883

silv3rstr3

#6
02-25-2016 -- 10:17:18
Nice!!  I just got us back up and running again doing 3458's automated.  I was having trouble running the 5720A characterization for AC Voltage and DC Current.  I over looked the fact you have to add "self_cal=yes" to the metcal.ini file to allow the 5720A to be characterized.  Let me know if you end up being able to do 5790's at a primary level cause Fluke is raping us on cost each year.
"They are in front of us, behind us, and we are flanked on both sides by an enemy that out numbers us 29:1. They can't get away from us now!!"
-Chesty Puller

DiligentMinds.com

#7
02-29-2016 -- 18:15:58
@Hawaii596:

Can you reveal how much it cost for Fluke's DVMP program?

-Ken

Hawaii596

#8
03-01-2016 -- 09:57:54
As I don't pay the bills, I'm not sure.  I want to say it is between $1K and $2K, but not sure.  I can say that I just got back a courtesy PDF copy of my certificate, etc. and in my proficiency test portion, I was within 0.05 ppm of their value, and my MU is now down to 0.36 ppm.  I like both of those numbers.
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883

BamaKid

#9
03-01-2016 -- 18:27:28
I asked my Primary Standards Metrologist to weigh in on your original posting and here is his response:

"Best case, the individual is splitting hairs and worst case is biasing his measurements. The measurements represent differences between a 10V standard and a 10V DUT, and therefore will vary depending on the values of those units. One microvolt represents a difference of 0.1 ppm, and I have had measurements of over 50 uV that occurred when Fluke sent me a reference that was about 2.5 ppm higher than nominal and my units were about 2.5 ppm below nominal. Fluke's provided connectors and switch do contribute some thermal EMFs, but the measurement system uses forward and reverse measurements to wash that out, and Fluke's uncertainty accounts for it. That said, I use my Dataproof scanners with solid copper low-thermal wires clamped under binding posts. Fluke's manual switch and banana plug leads are fine however."

Hawaii596

#10
03-02-2016 -- 09:18:09
I kind of knew I was most likely splitting hairs.  And I also kind of knew that these values are a little down in the dirt.  However, as my stated MU from the Fluke cert for this round is at 0.36 ppm (uV/V for you ISO purists), I believe they were important hairs.

I would like to contend that all measurands are biased.  There is no such thing as a completely unbiased measurement.  So as someone who watches trending closely (in measurands), my goal is to get to the truth of the measurement.  I want the real value (as closely as possible).  Particularly, when I did the proficiency test (required prerequisite for doing DVMP cal on the 732A), and my measurand was within 0.05 ppm of the "true" value, that made me feel good that I was heading after truth in the value.

So when I get inconsistencies in the reversals, I presume an intermittent bias exists within the Fluke supplied lead set.  And, rather than introducing bias (actually, probably am - but bias that heads toward the true value), I am attempting to observe and evaluate the existing lead-based bias, and account for it the best I can.

I can presume based on my improved MU (2013=0.55, 2014=0.51, 2016=0.26 ppm), and based on my Proficiency Test error of 0.05 ppm (versus declared uncertainty of 1.07 ppm), that I am hopefully doing the right thing.

The challenge is that in values as tight as these in a remote, in-place cal such as the DVMP, that I have nothing better to check it against.

With the exception that I have two 732B's that are my cross check units (which I cal against my primary DVMP 732A immediately after each years cal).

End of ramble.
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883

DiligentMinds.com

#11
03-02-2016 -- 19:10:33
@Hawaii596:

I forgot to ask--  When they send you their 732B, does it come with the assigned value, or do they keep that secret until you ship the unit and send in your measurement data?

The reason I ask is that I have a "new to me" 732B that I would like to set to match as close as possible to 10V--  Since I have no history on it, there is no problem with changing it's output voltage.

-Ken

Hawaii596

#12
03-03-2016 -- 10:27:32
They keep it a secret.  That is, I believe, to prevent biasing the measurement values.  If the 732B is new to you, I would strongly recommend, though, unless you know it has a new battery, change out the battery and put a label on the unit with the change date.  And allow it to operate for at least a month uninterrupted before doing the DVMP cal on it.  Personally, I wouldn't worry too much about setting to near nominal.  I did that with two of my 732B's, but depending on their individual drift rate, it won't stay there for very long.  Regarding the two comparison 732B's I have with my 732A, one has a -0.30 ppm/year drift, and the other has a +0.30 ppm/year drift.  Seems like the drift rate drifts over time as well.  Not exactly linear, year over year.
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883

DiligentMinds.com

#13
03-04-2016 -- 13:49:40
Quote from: Hawaii596 on 03-03-2016 -- 10:27:32
They keep it a secret.  That is, I believe, to prevent biasing the measurement values.  If the 732B is new to you, I would strongly recommend, though, unless you know it has a new battery, change out the battery and put a label on the unit with the change date.  And allow it to operate for at least a month uninterrupted before doing the DVMP cal on it.  Personally, I wouldn't worry too much about setting to near nominal.  I did that with two of my 732B's, but depending on their individual drift rate, it won't stay there for very long.  Regarding the two comparison 732B's I have with my 732A, one has a -0.30 ppm/year drift, and the other has a +0.30 ppm/year drift.  Seems like the drift rate drifts over time as well.  Not exactly linear, year over year.

Thanks!  That's kind of what I thought they would do-- otherwise you could use their 732B to calibrate a lot of other things other than just the one voltage standard.  Still, they do seem to provide very good uncertainty for this.  All I have to do now is decide if the price is worth it.

I think the battery in my 732B is very old.  It seems to be working fine-- and holds up the reference according to spec [has not lost any ability to hold charge] which is quite surprising for a lead acid battery-- good heavens it is over 7 years old!  I will look into changing the battery.

Hawaii596

#14
03-04-2016 -- 14:10:07
I think the rule of thumb is about 5 years in a 732B, or so.  That could go longer for a well kept bettery.  I believe these Sealed Lead Acid (SLA) batteries don't like to be left discharged - which can be their demise.  But if they have been kept in charge, they do better.  I last changed mine in 2012, and think I should be considering changing them out soon. 

What I will have to figure is how I want to handle preventing the IN CAL from going out.
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883
Print
Go Up Pages1 2
User actions