Vintage Weston Cells and an HP 419A = too much fun + scanned Eppley 1968 catalog

[kj7e]

Okay, just have to share.  I had been waiting for an HP 419A for years, finally found one last year and totally rebuilt it (after trying to blow it up).   See here;
https://www.eevblog.com/forum/other-blog-specific/great-meter!-the-hp-419a-dc-null-voltmeter-(restoration)/msg3565331/#msg3565331

Someone posted a photo of a Weston Standard cell, which lead me down the road to wanting to find one, just to play with or set on my shelf with other old odds and ends.  I found one, then started reading all the old NBS bulletins and became fascinated.  The first sample I got was an old square Leeds and Northrup box that still measures 1.01747 so the cell is no longer good or stable, but makes a nice show piece on the shelf.  Just out of fun, I decided to pickup another I found for cheep, which then lead to a third before the second even arrived. 

The new boxed both arrived on the same day, but I was headed out of town, so they sat in my shop for a few days before I could play with them.  Got home today and measured them for the first time using the DMM7510 at 10G , they both measured very close to 1.018440v.  Next, I wanted to measure the difference between the two on my HP 419A, I was surprised to see only about 500nv diff once the 419A was nulled!  These are just toys, something to play with.  Oh, one of the cells came from Honeywell where it had been on a shelf in the lab for 30+ years, this one came with a mint 1968 Eppley manual/catalog with data and price sheets.

Scanned Eppley 1968 manual/catalog with price sheets;
https://drive.google.com/file/d/1eBw50tLOK7VMq5tzayuk0NW7EneqIlTO/view?usp=sharing

[alm]

The 419A isn't that high impedance. Its claim to fame in metrology is to measure if the voltage between two points is (extremely close to) zero. In that case, there should be very little current flowing through it (just the bias current).

If you rely on it to measure voltage between two different points, the limited input impedance comes into play, and it's accuracy is merely that of an analog voltmeter.

The way it would have been used would have been in a bridge. Adjusting something so two voltages are equal, using a Kelvin Varley Divider or potentiometric techniques. I imagine if you read old documents on how to compare standard cells, this is what you'll find.

[HighVoltage]

Weston Standard cells are somehow fascinating
A few years ago I bought a brand new one, that was never used or out of the original package.

Can you scan the "mint 1969 Eppley manual" for us to PDF?
I would really appreciate this, others probably as well.

[Kleinstein]

The 419 is not high impedance, it is only low current of used with a low voltage. In the low ranges it is even rather low impedance (100 Kohm). Even at zero voltage there may be a tiny bias current, similar to moder AZ OP.  Just do the check and compare the zero reading with shorted inputs and open inputs. They are not necessary the same and 1 µV of difference would be 10 pA of input bias.

The specs for the DMM7510 is >10 G and lager here may mean a lot larger. The more problematic case is the bias current, not the input impedance.
 Changes are the input current to the 419 is higher, onece  the measured voltage is more than some 5 µV.

[kj7e]

The 419A isn't that high impedance. Its claim to fame in metrology is to measure if the voltage between two points is (extremely close to) zero. In that case, there should be very little current flowing through it (just the bias current).

If you rely on it to measure voltage between two different points, the limited input impedance comes into play, and it's accuracy is merely that of an analog voltmeter.

The way it would have been used would have been in a bridge. Adjusting something so two voltages are equal, using a Kelvin Varley Divider or potentiometric techniques. I imagine if you read old documents on how to compare standard cells, this is what you'll find.

True, the 419a has a 100K input impedance 3uv to 3mv ranges, without being nulled.  Once nulled the input impedance is nearly infinite.  If you take a look at the first photo, you will see I am comparing two Weston cells in series opposition with the 419A nulled, which adds its own internal series opposition force raising the input.  Effective, nearly zero current will flow when nulled.

In the case above, I only wanted to compare the difference between the two cells, which came to 400-500nv.  In another test I balanced each of the individual Weston cells in series opposition with a PDVS2 voltage source (as close as I could with the given resolution of the PDVS2) the finished the balance with the 419A internal null opposition.  With the DMM7510 measuring the PDVS2, I added or subtracted the null differential to derive the Weston cell voltage without measuring the cell directly.  Its the best I can do with what I have and fun to play with.  Not going to get sucked in to a Kelvin Varley Divider, nope, stopping here haha.

[alm]

True, the 419a has a 100K input impedance 3uv to 3mv ranges, without being nulled.  Once nulled the input impedance is nearly infinite.

I know that's been written in some marketing materials, but it's nonsense. The impedance with 0V across it is still 100k, it's just that with no voltage across the input, the current flowing through the meter would be zero in theory. In practice you get some finite current due to bias current.

[kj7e]

Weston Standard cells are somehow fascinating
A few years ago I bought a brand new one, that was never used or out of the original package.

Can you scan the "mint 1969 Eppley manual" for us to PDF?
I would really appreciate this, others probably as well.

Done, link provided in my original post below the text.  The file was too big to upload as an attachment.

[guenthert]

In practice you get some finite current due to bias current.

      That's why you adjust the buck voltage (the HP419 has for that a mercury battery) with input shorted.  After the meter is nulled and zero'd (offset voltage and bias current compensated) the current ought to be zero.  Well, it would be if it wouldn't be for noise.  I found it difficult to adjust zero to less than +/-0.5uV which corresponds to +/-5pA.   However even for sensitive Weston cells with their high series resistance of 1kOhm and limited capacity, that's of little concern (just don't let it connected for weeks).

[tggzzz]

Weston Standard cells are somehow fascinating
A few years ago I bought a brand new one, that was never used or out of the original package.

Can you scan the "mint 1969 Eppley manual" for us to PDF?
I would really appreciate this, others probably as well.

Done, link provided in my original post below the text.  The file was too big to upload as an attachment.

Nicely scanned.

Can I suggest uploading it to http://bama.edebris.com/ and http://www.ko4bb.com/ since those are the two traditional starting points for manuals.

[kj7e]

Weston Standard cells are somehow fascinating
A few years ago I bought a brand new one, that was never used or out of the original package.

Can you scan the "mint 1969 Eppley manual" for us to PDF?
I would really appreciate this, others probably as well.

Done, link provided in my original post below the text.  The file was too big to upload as an attachment.

Nicely scanned.

Can I suggest uploading it to http://bama.edebris.com/ and http://www.ko4bb.com/ since those are the two traditional starting points for manuals.

I cleaned up the white margins first, then sent it to both, also to TIN's xDev site.

[Conrad Hoffman]

Good scan! I love to play with this stuff and find the HP null meter is very nice to use. The low impedance never bothered me for use as a null meter, which is the only way it makes sense to use one. Remember that the last un-saturated cells were made long ago, much longer than the expected life. Unlike saturated cells, they slowly die and become unstable. I've had quite a few, all of which went unstable and have been disposed of. But, I only sent the internals off to hazardous waste. I kept the very nice thermally lagged housings. Why? What better thing to mount a modern voltage reference inside, powered with a wall wart from the back or underside. The Eppley #100s are nice but IMO, the round Weston model 4 are the prettiest of all. 

[kj7e]

Conrad,

Can you help me understand how to use a Kelvin–Varley divider to measure the Weston Cell EMF properly?  Is the function of the divider to simply to ratio down a known standard voltage to be nulled to the unknown (Weston Cell for example).  If so, I think I understand and will now be forced to find a Kelvin-Varley divider, damn it.

[HighVoltage]

Your scan came out very nice, thank you.
The price list is also very interesting!

[TimFox]

Conrad,

Can you help me understand how to use a Kelvin–Varley divider to measure the Weston Cell EMF properly?  Is the function of the divider to simply to ratio down a known standard voltage to be nulled to the unknown (Weston Cell for example).  If so, I think I understand and will now be forced to find a Kelvin-Varley divider, damn it.

Yes, you must never draw a substantial current from the standard cell.  Obtaining a null voltage between a K-V divider (or similar potentiometer) driven from a known (higher) voltage is the proper method to measure the open-circuit voltage of a Weston cell.

[HighVoltage]

These days we can measure a Weston Cell very well with a DMM set to High-Z.
 

[Gyro]

Conrad,

Can you help me understand how to use a Kelvin–Varley divider to measure the Weston Cell EMF properly?  Is the function of the divider to simply to ratio down a known standard voltage to be nulled to the unknown (Weston Cell for example).  If so, I think I understand and will now be forced to find a Kelvin-Varley divider, damn it.

It doesn't need to be a Kelvin-Varley divider, it can be a Potentiometer. The term spans anything from a single turn pot, through precision multi-turn, right up to the big old boxes with lots of dials and a slide-wire fine adjust. A multi-turn pot with turns counter dial will do, maybe 10k ohms. You just need a stable source, say, 10V to drive it. Then set the pot to approximately what the standard cell will be. Then connect a null meter (DMM on lowest range will do) between the pot wiper and standard cell and accurately null the pot for zero offset. You can then measure other voltages without loading them by adjusting the pot wiper to null against them.

The pursuit of accuracy will of course lead to seek greater and greater pot resolution and linearity, temperature stability etc, but you can start with a simple pot. Look on ebay for 'Potentiometer' filtering on the test and measurement section to avoid getting hits for every single turn pot in existence(!) and you will probably find some old multi-dial ones. Far cheaper than KVDs!

[kj7e]

Thanks, I do have some nice 10K 21t pots laying around, Ill play with it.  I found the 1965 National Bureau of Standards Monograph 84 on Standard Cells - Their Construction, Maintenance, and Characteristics.  Lots of info in this doc.

https://www.govinfo.gov/content/pkg/GOVPUB-C13-c28efff2f98ed0db9f802de89a95e733/pdf/GOVPUB-C13-c28efff2f98ed0db9f802de89a95e733.pdf

[alm]

It doesn't need to be a Kelvin-Varley divider, it can be a Potentiometer. The term spans anything from a single turn pot, through precision multi-turn, right up to the big old boxes with lots of dials and a slide-wire fine adjust. A multi-turn pot with turns counter dial will do, maybe 10k ohms. You just need a stable source, say, 10V to drive it. Then set the pot to approximately what the standard cell will be. Then connect a null meter (DMM on lowest range will do) between the pot wiper and standard cell and accurately null the pot for zero offset. You can then measure other voltages without loading them by adjusting the pot wiper to null against them.

A trick they explain in the NBS paper is to instead of using the potentiometer to generate a voltage equal to that of the standard cell, use a standard cell with a known value A plus a voltage deltaV from a potentiometer to equal the voltage of standard cell B, and adjust the potentiometer so VA + deltaV = VB. This way the contribution of the uncertainty of deltaV to the accuracy of the determined VB is much less.

The old Fluke 731A/731B voltage standard could generate voltages between 0-1 mV in 1 uV increments for exactly this purpose.