Fluke 731B calibration results

[VintageNut]

There are many on this forum who own the venerable 731B. Posted below are results obtained from the primary lab that I visited this week. These two 731Bs were installed into an array of 732Bs. There are more than 4 732Bs in the array as well as two other type of 10V standards in the array.

I drove 10 hours to visit this lab. My 731Bs were self powered in my car all day. In my hotel room overnight I plugged in the 731Bs. The next day I delivered the 731Bs. The 731Bs were plugged into AC power and were measured over the rest of the week.

One of the 731Bs was stable all of the time. The other 731B moved 3 uV after a day of settling.

[splin]

Interesting. What were they using to get readings to 9 digits of precision?

The average figures are not those of the 6 values shown - were these just some selected measurements?

Any idea how the uncertainty figures were calculated?

[VintageNut]

Interesting. What were they using to get readings to 9 digits of precision?

--The standards are known to that precision. 4 of the collection of standards are sent to Fluke once a year. The measurements are all differential variance from exactly 10V. There are more than 10 of the standards. The entire collection is measured against each other once a week and charted. ----

The average figures are not those of the 6 values shown - were these just some selected measurements?

---For the 731B that moved 3uV during the first day, the first two measurements were discarded (by me) for the average calculation. ----

Any idea how the uncertainty figures were calculated?

---No. I think that it is a statistical sum of squares. I would like to read an idiots guide to metrology uncertainty calculations.---

[CalMachine]

Interesting. What were they using to get readings to 9 digits of precision?

--The standards are known to that precision. 4 of the collection of standards are sent to Fluke once a year. The measurements are all differential variance from exactly 10V. There are more than 10 of the standards. The entire collection is measured against each other once a week and charted. ----

The average figures are not those of the 6 values shown - were these just some selected measurements?

---For the 731B that moved 3uV during the first day, the first two measurements were discarded (by me) for the average calculation. ----

Any idea how the uncertainty figures were calculated?

---No. I think that it is a statistical some of squares. I would like to read an idiots guide to metrology uncertainty calculations.---

https://www.a2la.org/guidance/est_mu_testing.pdf

[VintageNut]

Interesting. What were they using to get readings to 9 digits of precision?

--The standards are known to that precision. 4 of the collection of standards are sent to Fluke once a year. The measurements are all differential variance from exactly 10V. There are more than 10 of the standards. The entire collection is measured against each other once a week and charted. ----

The average figures are not those of the 6 values shown - were these just some selected measurements?

---For the 731B that moved 3uV during the first day, the first two measurements were discarded (by me) for the average calculation. ----

Any idea how the uncertainty figures were calculated?

---No. I think that it is a statistical some of squares. I would like to read an idiots guide to metrology uncertainty calculations.---

https://www.a2la.org/guidance/est_mu_testing.pdf

Thanks.

[Assafl]

Do they also measure the 1.018v output? I ask because if it keeps the 0.38ppm (~3.8uV) ratio it will be ~380nV for the 1.018v which may be useful for verifying calibration at the low range of DMMs.

[VintageNut]

Do they also measure the 1.018v output? I ask because if it keeps the 0.38ppm (~3.8uV) ratio it will be ~380nV for the 1.018v which may be useful for verifying calibration at the low range of DMMs.

No. I want to keep the 731B only at 10V. I believe that switching from 10V to 1.018V and back to 10V will introduce more uncertainty into the 10V value.

The change of each 731B over a period of three days in the cal lab was less than 1ppm. The 30 day spec for the 731B is 10ppm. I believe that if the goal is to have the best 10V stability, the user cannot switch from 10V to 1.018V amd back to 10V.

[Conrad Hoffman]

Playing devil's advocate, I'd suggest that not using the switch periodically to go between the ranges could introduce more uncertainty, as that keeps it clean. I've got three of the things, one of which I've tweaked the reference current for zero TC around room temperature. It's extremely stable, having had it's last official cal over ten years ago. The other two are good, but do move a bit with temperature, though I can't give you a number.

[Assafl]

I thought the same. But now that VintageNut hasn't worked the switch it may be a bad idea to work the switch. It would be better to clean the contacts and work the switch prior to calibration.

[VintageNut]

Playing devil's advocate, I'd suggest that not using the switch periodically to go between the ranges could introduce more uncertainty, as that keeps it clean. I've got three of the things, one of which I've tweaked the reference current for zero TC around room temperature. It's extremely stable, having had it's last official cal over ten years ago. The other two are good, but do move a bit with temperature, though I can't give you a number.

Its a valid point that working the switch regularly or at least occasionally will keep the switch contact cleaner than not working the switch.

For now, I am going to leave both of the 731Bs exclusively at 10V and see how they behave over several years.
Over the period of several days at the primary standards lab, both 731Bs moved less than 1ppm per day.

If these 731Bs are stable year-on-year, I may just bypass the switch and permanently place them at 10V.

[Vgkid]

Playing devil's advocate, I'd suggest that not using the switch periodically to go between the ranges could introduce more uncertainty, as that keeps it clean. I've got three of the things, one of which I've tweaked the reference current for zero TC around room temperature. It's extremely stable, having had it's last official cal over ten years ago. The other two are good, but do move a bit with temperature, though I can't give you a number.

Its a valid point that working the switch regularly or at least occasionally will keep the switch contact cleaner than not working the switch.

For now, I am going to leave both of the 731Bs exclusively at 10V and see how they behave over several years.
Over the period of several days at the primary standards lab, both 731Bs moved less than 1ppm per day.

If these 731Bs are stable year-on-year, I may just bypass the switch and permanently place them at 10V.

My guess is that your best bet is to bypass the switch. Even  when not moving frequently, switches can get noisy. On my cs-152 I can clean the switches, let them sit a few months, and they will become noisy.

[Conrad Hoffman]

FWIW, my 731s stay powered most of the time. I've never noticed much effect from shutting them down, but keep 'em powered anyway. Last cal was more than ten years ago, yet I can warm up my HP3455A, also not touched for a good few years, and be within 1-2 ppm at 10V. By intercomparison, the three 731s have stayed within a couple ppm of each other as well. I'm not suggesting this is good metrology practice, but I haven't had a reason to pay for a current cal. I don't think a 731 is good for any better than a couple ppm long term, but switching my switches or not hasn't made any noticeable difference.

[VintageNut]

FWIW, my 731s stay powered most of the time. I've never noticed much effect from shutting them down, but keep 'em powered anyway. Last cal was more than ten years ago, yet I can warm up my HP3455A, also not touched for a good few years, and be within 1-2 ppm at 10V. By intercomparison, the three 731s have stayed within a couple ppm of each other as well. I'm not suggesting this is good metrology practice, but I haven't had a reason to pay for a current cal. I don't think a 731 is good for any better than a couple ppm long term, but switching my switches or not hasn't made any noticeable difference.

Very good points. The 30-day spec is 10ppm. The 90 day spec is 15ppm. If the actual 731B in-hand is better, then you have a better than average specimen.

The only way to be certain is to have an independent calibration performed at least once per year. It will require some years.

The primary lab that I visited has some standards that date back to the late 1980s.

[guenthert]

[..]
Very good points. The 30-day spec is 10ppm. The 90 day spec is 15ppm. If the actual 731B in-hand is better, then you have a better than average specimen.

[..]

I think that's a fundamental misunderstanding.  The quoted uncertainty is a promise (the exact terms depend on the vendor) that a (any) given unit doesn't exceed those.  The vendor will likely just relay the uncertainty values from the manufacturer, how the latter arrived as such varies (e.g. for the HP34401 it is documented in its manual).  It does not necessarily mean that all or most samples exhibit an error less than the quoted uncertainty and most certainly it does not mean that any given unit exhibits the whole range during its lifetime.  A manufacturer might chose (because not all use cases could be tested to the fullest, to keep rate of returns low or out of marketing-strategic concerns) to post conservative uncertainties where all or almost all units happen to be well within that range and wander only in a small range within that posted error margin.

Then with old units the observation changes further, as the units which were marginal or tend to wander have been scrapped long ago and only the best survive.

[Edwin G. Pettis]

I just finished doing some servicing to my 731B, it came from the Navy's Metcal lab with a "special calibration' sticker over the 10V adjustment access, the inside components date it to 1976.  While I've had it, it has been almost always powered on but has had a fair number of shut downs for various reasons.  The outputs of this unit has been quite stable since I acquired it some years ago, shortly I will be shipping it off for a calibration check, partly because I got into it and did a little servicing on it.  I have switched the unit's output to the other voltages every now and then and have not seen any problems with the readings of which have also been very stable and accurate as well (I don't really need them).  The rotary switch in my unit is of a blue plastic with silver plated contacts and they do get a bit oxidized over time, they have not been difficult to clean.  About the only big difference between the 731B and the 732s is the noise level on the 10V outputs, the 732 being quieter.

[e61_phil]

About the only big difference between the 731B and the 732s is the noise level on the 10V outputs, the 732 being quieter.

I thought the 732x units are heated? Therefore, I would expect a much better TC from the 732x

[Vgkid]

About the only big difference between the 731B and the 732s is the noise level on the 10V outputs, the 732 being quieter.

I thought the 732x units are heated? Therefore, I would expect a much better TC from the 732x

With the 732, the zener diode, and the resistors are heated.
https://www.eevblog.com/forum/testgear/fluke-732b-dc-standard-

[VintageNut]

The primary lab that I visited has many of the 732B units that were compared as a group to my two 731Bs. The person who did the measurements told me that my 731Bs were noisier than the 732Bs. The standard deviation of my 731Bs were about 0.25ppm. To me, that is a very good standard deviation for what I paid for the 731Bs.