USA Cal Club Round 3

[CatalinaWOW]

I started warming up and checking test gear today.  Will be doing test runs tomorrow.  Hope to send the stuff back to Randall about three days after it arrives.  If I and the rest of us can keep this cadence up those at the end of the list won't have to wait more than a year.

I'd (please) not completely blast it from the front door straight to the back - I was under the impression a couple of weeks of using it is fine. The list is not that long, currently (and I assume it's being updated). I am second in line and would hope I'd get a bit of heads up before it's heading my way. Most people have actually asked to be pushed towards the end of the line.

I do appreciate the expenditure, I just hope we're keeping things on a similar timeline to what was presented.

My intent is a gentle nudge more than a push.  In Round 2 the dwell times were often very long.  How long it takes depends on what you are doing.  I am not adding new knowledge to the stability of the standards, merely checking my equipment.  I have set up GPIB to automate the data collection, and if 72 hours of data isn't enough it will hardly be improved by adding a few more days or weeks.  Experience in the first two rounds suggests that things will happen that make longer than two weeks happen for some folks and some folks have instrumentation good enough that it is worthwhile to have them gather extended data sets, so to achieve a two week average those of us who are able should do it a little faster.

By setting up, warming up and doing test runs with my local standards I avoid problems that have occurred to me and others in prior rounds.  Things like Windows updates that killed my automation, experiments making sure my instruments temperature was controlled adequately, misplaced equipment and finding unexpected noise and having to locate and squelch it.    It is only polite for me to find and correct these on my own time so that others aren't forced to wait.

 

[maxwell3e10]

Please add me to the list as well, I am also happy to be near the end of the round. I was near the end of round 2, so it would be interesting to see how the instruments have drifted over several years.

I think it would be good for the community if anyone who receives the kit is almost required to post some results here. Taking long data is fine, but even spot calibration numbers would be useful to get some statistics on stability of various instruments.

[1audio]

I'm interested. I have  Fluke 732A with no real history, 2 8506A's, a Prema 6001, Keithley 2015,  HP735 and Fluke 731B. Needless to say none agree but they all seem to disagree consistantly. I would hope this helps me get my bearings. Is there a schedule to manage the transitions? Are there others in the Bay Area on the list so the kit could be handed off without shipping?

[Rax]

Are there others in the Bay Area on the list so the kit could be handed off without shipping?

Is that a consideration? I know at least one person in the list that is my neighbor, and we're not contiguous on the sequencing.

[RandallMcRee]

I'm interested. I have  Fluke 732A with no real history, 2 8506A's, a Prema 6001, Keithley 2015,  HP735 and Fluke 731B. Needless to say none agree but they all seem to disagree consistantly. I would hope this helps me get my bearings. Is there a schedule to manage the transitions? Are there others in the Bay Area on the list so the kit could be handed off without shipping?

I go to the East Bay (Alameda) a few times a month to visit family, perhaps we could arrange a hand-off? I only suggest this because I am the facilitator of this round, it does not work in general.

The logistics of hand-offs is too complicated across an entire group. But if two people like Rax and X (whomever that might be) wanted to share a four-week slot we could have scheduled them together--had we known beforehand. As it stands Rax would have to query everyone who would be inconvenienced by now re-scheduling X up to where Rax is...still not impossible. As well, I doubt that anyone would object to Rax voluntarily moving down the list.

Regards,
Randall

[CatalinaWOW]

The lovely box of goodies arrived safely in the mail today, in spite of rain, snow and road closures along its path, and is doing its shakedown run.  Everything looks good as far as the voltage references go.  One thing that will be interesting to watch in the data will be the FX reference, which was still quite chilly from its travels when I started logging.  I will look at the resistances in a couple of days after collection on the voltage references.

I won't be blasting it back out again, but will try to avoid the long tail of intervals observed in round 2.  By my analysis of the thread documenting round 2 the intervals were (in no particular order) 6, 9, 9, 12, 13, 18, 19, 25,31, 36, 53,  65, 91, 96, 116, 125, 136 and 148 days.  There are mitigating circumstances, COVID, trips back to TiN for recheck, and stops that weren't documented in the thread to name a few.  Randall suggested a two week goal and I think that is generally achievable.  The thing is not to be insanely schedule driven, but mindful of keeping it moving.

[Rax]

Randall suggested a two week goal and I think that is generally achievable.  The thing is not to be insanely schedule driven, but mindful of keeping it moving.

Agreed and thanks for your responsible approach!

[Rax]

The logistics of hand-offs is too complicated across an entire group.
Regards,
Randall

Randall - thank you, I think we'll stick with the current order.
Looking forward to the kit!

[CatalinaWOW]

The package is boxed up and will be in the mail tomorrow AM to Randall McCree.

It took longer than expected, a combination of equipment failures, family emergencies, power failures at the house and more interesting stuff to measure than there was last time around.  I have just begun to reduce the data.  It will be posted at XDEV in the next few days (I hope), but due to the demand for sample answers I will post the following pretty raw data.  These are readings from my HP3457 or HP3456 and have not been corrected, tracked against prior measurements or anything else.  The stdev is only representing the noise in the measurements, no other uncertainties are included.    They appear to be relatively accurate, but how accurate will take further looking.

1.  FX reference with 3457 -     9.9999982 with stdev 6.73E-6
2.  FX reference with 3456 -    10.0001927 with stdev 4.5 e-6
3.  PX reference with 3457 -      7.0456920 with stdev 6.36E-6

PX 1k resistor - 1000.00908 with stdev 0.000261
PX 10k resistor - 9999.8767 with stdev 0.00215

L& N 10000 resistor - two measurements see comment/question following numbers

Correct connection  - 10000.055 with stdev 0.011
'wrong' connection  - 10000.084 with stdev 0.012

McCree Resistance Box
Resistance              St Dev
0.0004306              3.95E-5
10.0038942            4.821E-5
20.0013949            4.74E-5
100.0017945           3.25E-5
200.0101323           6.55E-5
1000.36141             0.0017183
1999.98289             0.0020138
10000.36141           0.0017183
19999.38331           0.0010730
99999.67856           0.040206
190002.132             0.05293
1000118.609           3.9216

It took me a while to find TiNs data sheet on the Leeds and Northrup resistor, so after thinking about it decided that I should put the current drive on the center (top) terminals and the voltage sense on the side (end) terminals.  The logic being that the only current flowing from the current terminals to the sense terminals would be that limited by the extremely high impedance of the meter. 

The data sheet recommends using the side terminals for two wire measurements, but recommends doing the voltage sense on the top terminals when doing four wire measurements.  I don't see the logic behind this, but ran a data set connected that way.  There is a difference, but it is unclear to me which is correct.  Well, actually, the way it was calibrated by L&N and presumably rechecked by TiN is 'correct' so that measurements can be compared, but why is it the better way to do it?  I would better understand the Leeds and Northrup recommendation if two wire measurements were also performed using the top terminals.  Does anyone know the logic?

[guenthert]

It took me a while to find TiNs data sheet on the Leeds and Northrup resistor, so after thinking about it decided that I should put the current drive on the center (top) terminals and the voltage sense on the side (end) terminals.  The logic being that the only current flowing from the current terminals to the sense terminals would be that limited by the extremely high impedance of the meter.

The error due to the additional current (some 10 to 100pA across a few tens of mOhm) through the meter is insignificant here, regardless how it is connected.

For a 10kOhm resistor measured by a 3456A or 3457A it won't matter.  But when measuring low impedance shunt resistors or when measuring at the edge of what's possible, then even the point of contact of the sense wires will make a difference (there's a minuscule drop across the nut).  There's a pamphlet of one of the resistor manufacturers on the 'net describing such.  If the driving current would be applied to the inner connectors, then the point of contact of those would matter.  I suppose the point of contact of the sense wires is more easily controllable.

The data sheet recommends using the side terminals for two wire measurements, but recommends doing the voltage sense on the top terminals when doing four wire measurements.  I don't see the logic behind this, but ran a data set connected that way.  There is a difference, but it is unclear to me which is correct.  Well, actually, the way it was calibrated by L&N and presumably rechecked by TiN is 'correct' so that measurements can be compared, but why is it the better way to do it?  I would better understand the Leeds and Northrup recommendation if two wire measurements were also performed using the top terminals.  Does anyone know the logic?

I suppose the different recommendation refer to the intended application, not measurement procedure.  When the resistor is used as a standard then you'd use 4W measurement (for best results).  When the resistor however is meant to be used in an assembly with other resistors (e.g. serial-parallel configuration of a voltage divider), then 4W might not be possible (measured resistance won't be matching the actual used one).

[CatalinaWOW]

It took me a while to find TiNs data sheet on the Leeds and Northrup resistor, so after thinking about it decided that I should put the current drive on the center (top) terminals and the voltage sense on the side (end) terminals.  The logic being that the only current flowing from the current terminals to the sense terminals would be that limited by the extremely high impedance of the meter.

The error due to the additional current (some 10 to 100pA across a few tens of mOhm) through the meter is insignificant here, regardless how it is connected.

For a 10kOhm resistor measured by a 3456A or 3457A it won't matter.  But when measuring low impedance shunt resistors or when measuring at the edge of what's possible, then even the point of contact of the sense wires will make a difference (there's a minuscule drop across the nut).  There's a pamphlet of one of the resistor manufacturers on the 'net describing such.  If the driving current would be applied to the inner connectors, then the point of contact of those would matter.  I suppose the point of contact of the sense wires is more easily controllable.

The data sheet recommends using the side terminals for two wire measurements, but recommends doing the voltage sense on the top terminals when doing four wire measurements.  I don't see the logic behind this, but ran a data set connected that way.  There is a difference, but it is unclear to me which is correct.  Well, actually, the way it was calibrated by L&N and presumably rechecked by TiN is 'correct' so that measurements can be compared, but why is it the better way to do it?  I would better understand the Leeds and Northrup recommendation if two wire measurements were also performed using the top terminals.  Does anyone know the logic?

I suppose the different recommendation refer to the intended application, not measurement procedure.  When the resistor is used as a standard then you'd use 4W measurement (for best results).  When the resistor however is meant to be used in an assembly with other resistors (e.g. serial-parallel configuration of a voltage divider), then 4W might not be possible (measured resistance won't be matching the actual used one).

After more thought I do agree with the first comment.  The calculated resistance between the two terminals is about .3 milliohms.  That resistance combined with the 10 GOhm input impedance of the HP3457 makes the difference about 3 parts in 10^11, unmeasurable, even with instruments much better than mine.

I don't understand the recommendation for two wire resistance connections.  The resistance between the two center terminals is "known", that is calibrated and verified.  The resistance between the outer two terminals is technically unknown.  The theoretical difference is far lower than specified tolerance of the resistance, but still, in principal the center two terminals would be more accurate even in two wire connections.  The only reason I have been able to come up with is to reduce wear on the center connections, but this seems a labored justification.

[CatalinaWOW]

Data is posted to the xDevs site.  I again was unable to use the directions given for web based access, but using an ftp manager was easily able to transfer the data.  I used WinSCP under Windows 10, but I am sure that other programs will work as well or better.

I have done little deep analysis of the data, but am pleased that it appears that my HP3457 is tracking the FX reference to about 4 parts per million.  The agreement on the 10000 ohm reference resistor is not as good, about 12 ppm. 

The rest of my analysis will take quite a bit longer, and may or may not be published.  The quality of my instruments doesn't really add to the knowledge of the traveling standards so there is little urgency to post them.

[Rax]

Data is posted to the xDevs site.  I again was unable to use the directions given for web based access, but using an ftp manager was easily able to transfer the data.  I used WinSCP under Windows 10, but I am sure that other programs will work as well or better.

I have done little deep analysis of the data, but am pleased that it appears that my HP3457 is tracking the FX reference to about 4 parts per million.  The agreement on the 10000 ohm reference resistor is not as good, about 12 ppm. 

The rest of my analysis will take quite a bit longer, and may or may not be published.  The quality of my instruments doesn't really add to the knowledge of the traveling standards so there is little urgency to post them.

Good work, CatalinaWOW. I think all this is about empirical accrual of data, so all contributions with associated certainties baselines (meaning that loosely) are instrumental. I'm no bureaucrat of the US CAL CLUB, but I personally appreciate your diligence in generating this data.

[RoadDog]

I'd like to be added to the list. Currently just have an HP34401a but I'll be upgrading at some point this year. I'm building some references as well.

[RandallMcRee]

I'd like to be added to the list. Currently just have an HP34401a but I'll be upgrading at some point this year. I'm building some references as well.

Added in post #2. And we are on Rax, at the moment. He will have the kit mailed to him next week, all goes as planned...GWAGAP

[guenthert]

[..]
I don't understand the recommendation for two wire resistance connections.  The resistance between the two center terminals is "known", that is calibrated and verified.  The resistance between the outer two terminals is technically unknown. 

True, but L&N gives this as 0.00004 Ohm [1] (I suppose the variations due to manufacturing of those "horns" are much smaller than the uncertainty of the resistance element).

The theoretical difference is far lower than specified tolerance of the resistance, but still, in principal the center two terminals would be more accurate even in two wire connections.  The only reason I have been able to come up with is to reduce wear on the center connections, but this seems a labored justification.

If the application calls for two wire connection, you will want to make the connections as well as possible.  For that historically mercury cups have been used in which the ends of the horns were sunk.  I suppose the outer nuts give a smaller difference to the ends then the top ones, but this is just speculation on my part.

[1] https://www.surplussales.com/equipment/testequipment/pdf/eqp-ln-4025b-ln-4045b_PDF.pdf

[RandallMcRee]

[..]
I don't understand the recommendation for two wire resistance connections.  The resistance between the two center terminals is "known", that is calibrated and verified.  The resistance between the outer two terminals is technically unknown. 

True, but L&N gives this as 0.00004 Ohm [1] (I suppose the variations due to manufacturing of those "horns" are much smaller than the uncertainty of the resistance element).

The theoretical difference is far lower than specified tolerance of the resistance, but still, in principal the center two terminals would be more accurate even in two wire connections.  The only reason I have been able to come up with is to reduce wear on the center connections, but this seems a labored justification.

If the application calls for two wire connection, you will want to make the connections as well as possible.  For that historically mercury cups have been used in which the ends of the horns were sunk.  I suppose the outer nuts give a smaller difference to the ends then the top ones, but this is just speculation on my part.

[1] https://www.surplussales.com/equipment/testequipment/pdf/eqp-ln-4025b-ln-4045b_PDF.pdf

That pdf talks about using mercury for better contact. One of our members asked me about the possibility of mercury contamination on the L&N resistor.

Let me just say: neither Randall McRee, nor the USA Cal Club generally, guarantee the safety of this equipment. Everything you do is at your own risk. If you have doubts do not ask for or accept the equipment.

As one example, leaded solder has been used in the creation of the items in the equipment. There are well-known dangers associated with lead. Again, if you are concerned do not accept the equipment.

Regards,
Randall

[CatalinaWOW]

There is some residual cruft on the ends of the resistor.  But as long as you don't touch orr abrade it I suspect the danger is immeasurably close to zero.  Far less than including fish in your diet.

[Rax]

That pdf talks about using mercury for better contact. One of our members asked me about the possibility of mercury contamination on the L&N resistor.

Regards,
Randall

I'm one of those with reluctance to work with L&N resistors. Some of you have posted on other threads on this, as have I, and I personally don't feel my doubts and hesitance have been put to rest. Unless new pertinent information surfaces, I think I'll personally skip the L&N resistor.

[Rax]

Far less than including fish in your diet.

How do you know this?

[CatalinaWOW]

Far less than including fish in your diet.

How do you know this?

I don't know this to a certainty.  But any mercury on the resistor is a tiny quantity, physically tightly bound, and if the resistor is treated with the care it deserves not subject to being dissolved and transported from its current location.  On the other hand, most fish consumed by humans are apex predators and/or bottom feeders.  They concentrate any mercury in the environment.  Testing of fish shows widespread mercury contamination.

And then the really big reason.  You eat the fish, transporting any mercury they contain into your body.  Unless you touch the ends of the resistor and then lick your fingers before washing them there is virtually no way for any mercury on the resistor to be transported to your body.

For those worried about the resistor, I will point out that it is wrapped in a zip lock bag, which is then placed in a second zip lock bag.  You don't have to open either bag if the mercury worries you.  If you want to make measurements on the resistor you will have to open the bags, but wearing disposable latex or similar gloves would give you another layer of protection.

[guenthert]

There is some residual cruft on the ends of the resistor. [..]

Mine came thankfully with rubber sleeves covering the ends (not sure, whether they are there to protect the ends or me).  I haven't taken them off and see no reason to do so. 

[Rax]

Far less than including fish in your diet.

How do you know this?

I don't know this to a certainty.  But any mercury on the resistor is a tiny quantity, physically tightly bound, and if the resistor is treated with the care it deserves not subject to being dissolved and transported from its current location.  On the other hand, most fish consumed by humans are apex predators and/or bottom feeders.  They concentrate any mercury in the environment.  Testing of fish shows widespread mercury contamination.

Which is why I'm very carefully choosing the fish I (and my family) are eating, but that this is comparatively relevant is highly speculative and out of topic.

For those worried about the resistor, I will point out that it is wrapped in a zip lock bag, which is then placed in a second zip lock bag.  You don't have to open either bag if the mercury worries you.  If you want to make measurements on the resistor you will have to open the bags, but wearing disposable latex or similar gloves would give you another layer of protection.

This has been documented in other threads, but I personally don't regard the protection by your vanilla latex glove (or even lab grade in the hands of scientists) as sufficient (https://en.wikipedia.org/wiki/Karen_Wetterhahn). Admittedly, the case quoted had to do with organic Hg compounds, but what happens with Hg (salts and whatnot) on terminals which have obviously been handled by (organic) hands for decades is not something that I've been able to find studies on. I'm far from being a specialist or even comprehensive reviewer of literature, but I for one wll be making my own decisions about all this.

[CatalinaWOW]

Rax, the point on which we both agree is that you will make your own decision. Yours is undeniably the safer choice.  I doubt if either of us can numerically qualify the danger, and it is likely that you have not formally defined an acceptable risk.  These are thoughts that each of us should evaluate in choosing any activity.

[alm]

This has been documented in other threads, but I personally don't regard the protection by your vanilla latex glove (or even lab grade in the hands of scientists) as sufficient (https://en.wikipedia.org/wiki/Karen_Wetterhahn). Admittedly, the case quoted had to do with organic Hg compounds, but what happens with Hg (salts and whatnot) on terminals which have obviously been handled by (organic) hands for decades is not something that I've been able to find studies on.

Dimethylmercury is a well known neurotoxin (under chemists), but by this logic you could also be afraid of toothpaste because fluorine and hydrogen atoms can also be used to make https://en.m.wikipedia.org/wiki/Hydrofluoric_acid, which is an extremely potent acid that readily penetrates the skin and requires labs handling it to have a special first aid kit in case an accident happens.

Dimethylmecury is described as:

A highly volatile, reactive, flammable, and colorless liquid

How long do you think this liquid would remain on the terminals? Have a look at the synthesis section of Wikipedia to see what it takes to make this. The form of mercury you might find in fish, on the other hand...

You're obviously free to make your own decisions, but please don't exaggerate by suggesting how the terminals are likely to contain something like dimethylmercury.