OK, let me climb up on my old soap box. I haven't been up here in awhile, but this is the kind of thing that probably pi$$ed me off the most about the AFMETCAL program.
Please, all you blue suiters, and all you lurkers from Newark, explain to me again why the AF continues to look at the calibration process with blinders and completely fails to consider the big picture when K procedures are written and/or edited.
The example I found today is 33K5-4-549-1, for the Thunder Scientific 2500 Humidity Generator.
First, just a few specifications from Thunder Scientific:
Humidity Generation accuracy = ±0.5% RH
Temperature accuracy = ±0.05°C
Pressure Accuracy = ±0.15% FS for each transducer
Now 33K5-4-549-1 still lists the Humidity Accuracy as ±0.5% RH, but the temperature accuracy has been expanded to ±0.06°C, but there is no requirement to limit the calibration for temperature. Step 3.6 limits the pressure accuracy because of the resolution of the TI, and the procedure DOES call for a Limited Calibration Label marked according to the limits of step 3.6 for pressure.
Now I have 2 questions so far....
1. What good does it do to put a limited calibration on a humidity generator in units of pressure? Isn't the intent of the Limited Calibration Label to clearly indicate to the user a limit in the accuracy of the item? And since this unit produces humidity, and not pressure, isn't the parameter of interest humidity? Wouldn't it be more pertinent and useful for the limitation to reflect humidity accuracy?
and
2. Was any uncertainty analysis performed to quantify the impact of this pressure limitation and also the decrease of temperature accuracy (even though no Limited Calibration for temperature was noted)?
Now to dig just a little deeper into the K procedure and analyze the Equipment Requirements a bit.
2.1 SPRT. Range: 0 to 70°C. Fine choice for temperature work, a little fragile, but very accurate.
2.2 Triple Point. It don't get much better.
2.3 HP 3458. A very good meter, a bit past its prime, but still a work horse. Unfortunately, the Range is listed as 100 to 128 ohms. If we think about the resistance of the SPRT in the range of 0 to 70°C it is in the neighborhood of 25 to 33 ohms. A little mismatch here. And the Accuracy is listed as ±0.02% indication. The 3458 is MUCH better than that.
Lets ignore all the others down to
2.9 Ice Bath (optional). Accuracy ±0.004°C
OK, now lets do our own uncertainty analysis with the equipment listed here. If I use the optional ice bath, the SPRT, and the 3458 at a listed accuracy of ±0.02% indication, we come up with an uncertainty (k=2) of 0.059°C at 0°C and 0.075°C at 70°C. And that's not even considering any non-uniformity or gradients in our temperature bath.
Since the TI temperature accuracy was listed as ±0.06°C, that seems to be impossible to achieve using standards of this accuracy.
It seems like nobody has taken a critical look at this (or a lot of other procedures) and really identified the things that truly make a difference and dealt with them accordingly. Like the necessity of the pressure limitation. So you don't have 3 decimal places of display resolution at 100 psi. The "computed" accuracy is ±0.225 psi. If the TI reads within 0.2 psi its good, if it reads greater than that then it's bad. Simple as that.
There is WAY too much emphasis placed on the little crap like the pressure limitation and none on the big picture of what affect does it have on the end calibration. The ability to analyze the calibration process has been lost. More accurately it has been stomped out by overzealous QA and auditors that expect us to act like little Lemmings and follow the leader blindly over the cliff.
Any comments? Any stones coming my way?
No stones. I've always been a proponent of the "big picture," so your take on the procedure is right on. Also your opinion of the 3458!
I'll be interested to see the response of a Newark metrologist for this one...
Pressure limited due to the PPC2AF accuracy.
The reason for the pressure limitation to be noted is that the humidity generated is a function of pressure within the 2500S. They rounded up the accuracy. I believe it was like +/- 0.055psia. However the use of the 2465 would eliminate the problem of the PPC, but more a PIA.
The 100-200 ohm range for the 3458A is (best a guess than anything) the fact that it has 0-10 ohms, then the next range is 100 ohms. Then the resistance at zero would be 25.5 ohms. however it is being used in ratio mode so the resistance would be 1.00000.....blah, blah.
Again best guess on it, being just the ranges.
As for the stones....none coming from me. I learned while working in PMEL that initiative is stomped and beaten out of you. If you go outside the box..........expect to be thrown back in. It seems that some think it is easier to just ship something somewhere elese instead of using their brains and actually thinking.
I used to do alot of lateral support for blue suiters, and guess what........I had the exact same equipment they did.
Yes, I know the accuracy of the 3458, and since you are using it in ratio mode the majority of the error is cancelled out. I was only quoting the K procedure requirements and wondering aloud what would happen if similar equipment was substituted with the same specifications as listed for the 3458 in the procedure. Aren't equipment substitutions still allowed if the specifications in Table 2 are maintained?
I'm also curious about the reasons for decreasing the temperature accuracy to 0.06°C from the Thunder specs of 0.05°C. And better yet....why does THAT deviation not require a Limited Calibration Label?
Actually it appears from the K procedure that the reason for the pressure limitation is not the accuracy of the PPC2AF (it's been so long since I used one I really can't remember the accuracy of it however). Let me quote step 3.6 in its entirety:
"Due to the resolution of the TI, the TI Low Range Transducer will be certified to ±0.08 psia and the High Range Transducer will be certified to ±0.23 psia from 0 to <100 psia and ±0.3 psia from 100 to 150 psia. Upon completion of calibration, annotate and attach a Limited Certification Label accordingly." Now THATS a mouth full....and so useful and meaningful when I'm generating humidity.
I know the 2500 produces humidity as a function of temperature and pressure. And if the Thunder Scientific specifications of ±0.05°C and ±0.15% pressure are met, then the 2500 is capable of producing humidity from 10 to 95% RH at an accuracy of ±0.5% RH.
So, how do these increases in temperature and pressure limits affect the ability of the 2500 to produce humidity accurately? Has anyone at Newark performed the uncertainty analysis? If it has no affect on the ability to generate humidity, then why bother with a limited certification. If it DOES affect the accuracy, then shouldn't that be clearly and unambiguously stated on the Cal Label instead of some cryptic limitation for pressure?
You sure hit the nail square on the head about having the initiative pounded out of you MIRCS. And that's a shame, there are so many absolutely brilliant people in PMEL that if everyone was encouraged to question, learn and excel.... Oh never mind, that might confuse some of the engineers at Newark and make them work a little too much.
What I don't understand is why your uncert for the SPRT and Meter is so wide ( 50mK). My guess is that they meant for a PRT and not an SPRT, that would account for a uncert of +/- 50mK and the 100 ohms at TPW. I think the typo was the SPRT as the standard. However, a SPRT with a Hart 1590 should give you about 10-15mK uncert and would give you a 4:1 T.U.R. @ +/- 0.06Deg.C.
Who knows? AFMET is notorious for having to rewrite K-Pros. :|
Quote from: flew-da-coup on 12-13-2006 -- 09:36:15
What I don't understand is why your uncert for the SPRT and Meter is so wide ( 50mK). My guess is that they meant for a PRT and not an SPRT, that would account for a uncert of +/- 50mK and the 100 ohms at TPW. I think the typo was the SPRT as the standard. However, a SPRT with a Hart 1590 should give you about 10-15mK uncert and would give you a 4:1 T.U.R. @ +/- 0.06Deg.C.
Who knows? AFMET is notorious for having to rewrite K-Pros. :|
Well considering that only IIA's have the 1590 if all of em. then I guess we will have to continue with the 3458A.....though I would like to see them get 8508A's.
I'm guessing the uncertanity of 50mK will go back to every argument I have had with Kevin John...............the tolerances have to be decided for every lab and to cover the entire the range of the temperatures.
Quote from: MIRCS on 12-13-2006 -- 11:02:20
Quote from: flew-da-coup on 12-13-2006 -- 09:36:15
What I don't understand is why your uncert for the SPRT and Meter is so wide ( 50mK). My guess is that they meant for a PRT and not an SPRT, that would account for a uncert of +/- 50mK and the 100 ohms at TPW. I think the typo was the SPRT as the standard. However, a SPRT with a Hart 1590 should give you about 10-15mK uncert and would give you a 4:1 T.U.R. @ +/- 0.06Deg.C.
Who knows? AFMET is notorious for having to rewrite K-Pros. :|
Well considering that only IIA's have the 1590 if all of em. then I guess we will have to continue with the 3458A.....though I would like to see them get 8508A's.
I'm guessing the uncertanity of 50mK will go back to every argument I have had with Kevin John...............the tolerances have to be decided for every lab and to cover the entire the range of the temperatures.
Of course that goes without saying, but usually with an SPRT & 1590 you will be able to get a mesurement uncert. between 10mK &15mK. We have ours @ +/- 14mK.
Ha....ha...ha................I just located a copy of a AFTO 22 I submitted on 9/25/03
ready for it.................are you sure.................. :-D
Block 19 PART IV – DEFICIENCY]: The 0-50psia and 150 Pressure Calibration no downgrade listed due to resolution. The actual error is 0.075psia or 0.225psia and the tolerance given in Table 3 is ±0.08psia and Table 4 is ±0.23. PPC2AF has the wrong accuracy listed.
[Block 20 PART V – RECOMMENDED TM CHANGE]: Change accuracy of PPC2AF to ±0.01% FS of Range. Add due to resolution of TI Pressure is limited to ±0.08psia for 0-50 and ±0.23psia for 0-150 transducers.
OK, And what exactly did all that prove?
It's a HUMIDITY generator. How the world does annotating the Cal Label with that pressure limitation benefit anyone? Does that reduce the accuracy of the 2500 to ±0.51% RH?
The point I am trying to make is the information on the Cal Label should be USABLE information. Not BS that means nothing.
Quote from: OlDave on 12-13-2006 -- 18:47:08
OK, And what exactly did all that prove?
It's a HUMIDITY generator. How the world does annotating the Cal Label with that pressure limitation benefit anyone? Does that reduce the accuracy of the 2500 to ±0.51% RH?
The point I am trying to make is the information on the Cal Label should be USABLE information. Not BS that means nothing.
No sir I just found it funny that I submitted that 22 over 3 years ago on this.
Yes the special block should have info that is usable, but it seems more to be a block to cover thy ass
Yeah I guess it is kind of ironic that I'm in here pissing and moaning about it 3 years later to the culprit that helped create the mess.
Do GIs at some bases still get money for submitting 22s? I always thought that was bogus.
Quote from: OlDave on 12-13-2006 -- 19:58:43
Yeah I guess it is kind of ironic that I'm in here pissing and moaning about it 3 years later to the culprit that helped create the mess.
Do GIs at some bases still get money for submitting 22s? I always thought that was bogus.
Last I knew they did.........which really sucked for me as a contractor....we didn't.
Hey it wasn't a mess..............I called a spade a spade and let those in Newark decide if it was a spade.........really I am sorry for it.
Though the one thing I will always be proud of is the note in 427 about diver gauges.....................that took many years