Wavetek 9100 - When the equipment manual contradicts itself...

[AlbertPaglinawan]

...what can we do?

The service manual of wavetek 9100 said in section 10.4.6.2 Calibration Equipment Requirements, it suggested the use of Wavetek 1281 DMM. It always sugested 1281 DMM in other sections but this DMM is less accurate than the Wavetek 9100 in currents and some of resistance functions.

The manual also said there should be at least 3:1 TUR.

Please allow me to put these into numbers:

AC current to be measured: 3mA at 3kHz output

Accuracy of 9100: +/- 2.4 uA
Accuracy of 1281: +/- 132.9 uA

Accuracy of 9100:
= 0.07% output + 300 nA
= (0.07%) (3 mA) + 300 nA
= 2.4 uA

Accuracy of 1281:
(I used the range 10mA)
=300 ppm Rdg + 100 ppm FS + 130 ppm
=(300ppm)(3mA) + (100ppm)(10mA)(2) + 130 ppm
=132.9 uA

I wonder why the manual suggested the use of 1281 DMM in calibrating the current function of 9100 Multifunction Calibrator. I was under the impression that 1281 is 4 times or greater more accurate than 9100.

To those of you who uses this equipment, you might see in the calibration certificate that Fluke used 1271 DMM (an even lower model than 1281). How come?

I am sooo lost! My computations might be off but I really don't think so. Anyone who wants to correct me, please do so and I would be very much glad for it.

Hey, thanks in advance! I am happy that the forum is now back on-line.

cheers!

 

[Mike S.]

Lord I.,

Sounds like a call to Wavetek is in order. It doesn't make sense to me, either. It could be a mistake on Wavetek's part when they wrote the manual. In any event, I'd want to hear it from them. Let us know how it turns out, huh?

 

[AlbertPaglinawan]

Mike,

Thanks for your thought.

I just hoped the mistake is mine. Imagine our dear money lost when we purchased a DMM that is said to be able to calibrate the 9100 only to find out that it can't.

But in fairness, they might have used a shunt and then measured the voltage accross that shunt. Although their manual instructed a direct measurement of current in current lower than 1 A.

Then again, the manual can be correct, or rather, it may not be mistaken because on the last pages, the manual suggested the use of 4950 MTS for remote calibration and that the procedures and specs are conveniently not stated in the manual.

Then again again , the certificate issued by Fluke said that the instrument has been calibrated in accordance with the manufacturer's Instrument User's Handbook.... So we have to follow the manual.

However, the certificate didn't state what manual they used. Was it the manual of the 9100? of that of the 4950 MTS?

Boy! the only decent conversation I can have is arguing with myself! HAHA!

I am currently awaiting Fluke's explanation. Sad thing is, I am being passed from office to office and I can't help thinking that I am waiting for nothing.

I heard that Fluke and Wavetek merged. Is there.. i dunno... is there something like a pure Wavetek office that I can contact regarding this?

Cheers!

 

[Graeme]

sticking my left foot in ...

Lord I.,

re: Calibration of Wavetek 9100 --

As you mentioned, the 9100 manual states "A traceably characterized, long scale-length, Standards DMM" ... such as the Wavetek 1281. The key word is "characterized", because it implies a much higher degree of knowledge of the DMM uncertainty than normal calibration provides. Characterization is outwardly much the same as calibration, but is done much more frequently and the data is analyzed more rigorously. In a standards lab, characterization is a process that is performed very frequently – daily or weekly.

• Characterization of a standard allows it to be used with much lower uncertainty than is normally possible.
• Characterization relies on two things: the short-term (24-hour) performance of the standard, and data recording and analysis using various SPC methods.
• Characterization is normally performed by checking the standard against a group of high-accuracy transfer standards.
• When a characterized standard is used for a calibration, it is very often characterized before and after use on the customer’s equipment so that all error sources can be evaluated.

I have reviewed the calibration certificates for both of our 9100 units. I agree that both Wavetek and Fluke list the Datron 1271 DMM as one of the standards used. However, they also list the 4970 Multifunction Transfer Standard (MTS) on the certificate. I have not seen their calibration procedure for the 9100, but I strongly suspect that the majority of the AC/DC voltage and current tests, as well as resistance, are performed using the 4970. I am familiar with the 4970 because we used it (when it was still Wavetek in San Diego) to calibrate our 4808 standard. In that process, the Wavetek lab would characterize the MTS and ship it to us that day by overnight service. We used an automated procedure provided with the MTS to calibrate our 4808, and then shipped the MTS back to San Diego. They characterized it again, ran all the numbers through the computer, and then let us know what the results were.
(We do not use the in-place MTS process for our 9100 units, because we have several options that cannot be calibrated that way. Since we would have to send them in anyway, we do it all that way.)

In a lot of cases, the statement that "the instrument has been calibrated in accordance with its manual" has to be interpreted. Specifically, what is often meant is that the instrument has been calibrated to meet the performance specifications in the instrument manual. It often does NOT mean that the calibration procedure in that manual was followed verbatim. In the case of Fluke (and Wavetek before they were vanished), I know they use automated procedures wherever possible so strictly following the procedure in the manual is not possible. (My own lab does that as well.)

By the way, there is no longer a separate Wavetek contact that I know of. We used to deal with the Wavetek standards lab in San Diego, California but that was all moved early last year to the Fluke lab in Carrollton, Texas.

 

[Al Dyer]

Have to agree with contacting the supplier. Do it in writing and keep it on file to show any auditors that you are aware of the situation and have pro-actively handled it.

 

[AlbertPaglinawan]

Thanks guys!

If characterization relies on the short-term (24-hour) performance of the standard, and data recording and analysis using various SPC methods, is it possible to lower a characterized DMM's accuracy to less than it's 24-hour performance?

Because I think it can't. If we make a combination of uncertainties, it will be like this;

sqrt [(24-hour uncert )²+(data recording uncert)²+....]

then the uncertainties cannot be lower than the 24 hour uncertainty. So I think the lowest possible accuracy of a characterized 1281 DMM will be greater than;


Accuracy of 1281 @ 24-hour performance, 10mA range;

=150 ppm Rdg + 50 ppm FS + 130 ppm
=(150ppm)(3mA) + (50ppm)(10mA)(2) + 130 ppm
=131 uA

But 9100's uncertainty for this value is 2.4 uA. Characterization still needs a lot of trimming up to do to come up with at least 0.8 uA uncertainty (for a 3:1 TUR).

cheers!

 

[Ryan Wilde]

Okay, I've finally gotten the chance to reply on this. Sorry for the delay, we just went through another 17025 reassessment, and added 10 pages of prameters to our scope, so I've been somewhat busy.

Based on the 1281 1 Year Specification , assuming Selfcal has been performed within 24 hours and a change of <1 °C has occured since Selfcal. Lead error was not added.

Basic Spec (10 mA Range): (150 PPM Rdg) + (50 PPM FS) + (Std Unc. @ time of calibration)

FS = Range * 2
Rdg (from your example): 3 mA @ 3 kHz
Std Unc. (from my cert from Fluke): 0.004 mA (0.001 @ 1 kHz, 0.006 @ 5 kHz)

=2*Sqr(((3 mA * 0.000150) + (10 mA * 2 * 0.00005))/2^2 + (0.004 mA/2)^2)
=2*Sqr((0.000450 mA + 0.0002 mA)/2^2 + 0.004 mA/2 ^2
=2*Sqr(0.000325 mA^2 + 0.002 mA^2)
=2*Sqr(0.000000106 mA + 0.000004 mA)
=2*Sqr(0.000004106 mA)
=2*0.002026 mA
=0.004052 mA
Ue = 4.05 µA @ k=2


9100 Uncertainty (from your post)= 2.4 µA
9100 Uncertainty from Fluke website = 0.08 % + 32 µA
0.08 % of 3 mA + 32 µA = 2.4 µA + 32 µA = 34.4 µA
Ratio of 34.4 µA to 4.05 µA = TUR = 8.5:1

The 1281 should be fine for this application. I am not familiar with the 9100 (although I love our 9500), so I do not know about all of the ranges. However, it seems that you missed the noise floor specification. See Wavetek 9100 Spec book DS074

Ryan

 

[AlbertPaglinawan]

ryan,

why did you used that formula in calculating the accuracy of 1281? i think i used the formula stated in the manual (the +/- ppm reading + ppm FS + floor). am i missing something here?

also i can't seem to locate the 32 µA floor. I thought the floor is 300 nA.

In AC Current Accuracy (Sinosoiidal waveshape)
current output:0.32001 mA - 3.20000 mA
Frequency Band: 10 - 3k
Accuracy: +/- (%output + floor)
0.07 + 300 nA

i honestly don't understand the compliance voltages, total harmonic distortion and compliance error. will they be a factor in computing for the accuracy?

thanks in advance.

cheers!

 

[Ryan Wilde]

I got all of the tolerances/uncertainties from the Fluke website. If you are using the manual from Wavetek/Datron, that may be the difference. An old friend at Fluke told me that they completely recomputed the uncertainties of the continued Wavetek/Datron products.

A note on the AC parts in the specification for the 1281 (as done by Fluke) is that each AC range is only calibrated from 10% of range up, which removed the noise floor specification altogether. They did add, however, that the uncertainty of the calibration of the 1281 (they are on the Fluke certificate) must also be added.

Okay, now...

Compliance voltage is not something that you would add to uncertainty. Compliance Voltage is the maximum voltage that will be generated by the source to produce the current. It has no effect on accuracy or uncertainty.

Total Harmonic Distortion (THD) is another matter. Distortion is components of the sinusoidal waveform that vary from a perfect sinewave. What this does, in effect, is makes RMS voltage less accurate, since it is based on a perfect sinusoidal wave. But, if THD is less than 40 dBc, then you would never see it. An easy project to check with is to use a function generator. Using an oscilloscope, set up a sine wave output, then measure it on the 1281. Now, set up a square wave (technically, a sine wave with infinite harmonics) on the oscilloscope with the same amplitude, then measure it on the 1281. You will see a marked difference due to distortion. All of that said, the distortion on the 9100 is too miniscule to add any tangible uncertainty.

Ryan

 

[AlbertPaglinawan]

Oh! I just realized I added the ppm Rdg + ppm FS + ppm floor.
ppm Rdg unit: Ampere
ppm FS unit: Ampere
ppm floor unit: no unit!

so I have to cancel out first the Ampere unit BEFORE adding these values!

then, according to fluke, you make a Root Sum Square on the uncertainties which is strikingly similar to Ryan's equation . (You're the man Ryan! Born to be Wilde!)

THanks a lot!

cheers!

Let me add the calculations used by Fluke;

accuracy of 1281 @ 3mA
RSS [(300 ppm Rdg + 100 ppm FS) + 130 ppm]

Note: You must combine the two uncertainty using RSS = Root Sum Square
calculation method.

(300ppm)(3mA) + (100ppm)(10mA)
= 0.0009 mA + 0.001 mA
= 0.0019 mA

Convert to ppm:

0.0019mA/3.000mA
= 0.000633
= 0.000633 x 10+E6/10+E6 ppm
= 633.3 ppm

RSS [633.3 ppm + 130 ppm] ===> SQRT(633.3 x 633.3 + 130 x 130)
= 646.5 ppm

3mA x 646.5ppm
=3mA x 0.06465%
=0.00194mA
=1.94uA


cheers!