Weston Cells

[JamesBee]

Hello, my first post.
Recently purchased some electronic parts on Facebook and received these items in the photos.




I didn't have any knowledge until I looked up but can no information on the "admiralty pattern" device.
It seems to consist of a row of cells in series, a group of precision resistors, the left hand block are all 10k, and a device on the right of which I have no idea.
Can anyone offer any insights.

There were also several loose cells, I measured one at 1.0185 volts.

A little too complicated for my first post!

[Gyro]

Welcome,

Are they what you were expecting to receive, or just thrown into the mix?

Firstly you should know that the contents are highly toxic - Mercury and Cadmium so if you don't have the skills, or intention, of using them then they need to be disposed of at the appropriate disposal facility.

Muirhead D-845s are Saturated Standard cells, so they don't significantly degrade over time like unsaturated ones. Their lifetime is based on the finite energy drawn from them. Assuming that they haven't been exhausted, shorted even for a few seconds, or measured on a low impedance meter, they should still be ok.

I've attached the datasheet (courtesy of another member)...


EDIT: As you've added more photos, the main bank of cells appear to be D-550-As. I have attached the datasheet for those too...

[alm]

A 10 MOhm DMM is probably already borderline re impedance, so I'd only measure it using a DMM with an impedance in the GOhm range. 5.5+ digit bench meters will often have a high impedance mode up to 3V or 10V. See the manual or datasheet.

[JamesBee]

Many thanks for the quick replies, much appreciated.
I have no test gear with that spec, though this very week I came by a Fluke 883A Differential Voltmeter, but a quick test with both 4 1/2 meters produced identical readings of 1.0185 volts on one of the free cells.
The resistors in the block seem to be in spec, all exactly 10k.
Better photo of the device.

[Gyro]

That was lucky! Differential voltmeters, and other types of bridge measurement devices that have infinite impedance at balance, are indeed the traditional method of measuring and comparing cells.

It looks as if you have quite a cell bank there!


P.S. Note that the temperature coefficient of Saturated cells is relatively high - you need to take ambient temperature into account for precision measurements. The formula is in the datasheets. They will probably take a while to fully settle after transport too.

[JamesBee]

The cells I got last night, the voltmeter on Sunday last though I've not powered up yet, busy week and no US lead that fits though I do have a Step up/down Xfmr.
A quick inspection revealed no obvious problems and it seems to be as originally manufactured. I can't find a spec on power consumption of the meter.
Temperature in this room where the cells are currently 19.8C and will probably not vary by much on any one day. That's lower than their calibration figure I'm sure.

[Gyro]

One other thing to try. The standard way of comparing cells is to connect them back to back (negative to negative) and measure the voltage between the positive leads on your lowest measurement range. That should help you to weed out any cells that are suspect - again, assuming that they have stabilised to the same temperature.

[HighVoltage]

A few years ago I was lucky to have found a brand new (Never used) Muirhead cell.
It even came with a certificate of test.

In 1971 the measured voltage was 1.01864 V and when I measured it first, it was 1.01858 V

[tggzzz]

A few years ago I was lucky to have found a brand new (Never used) Muirhead cell.
It even came with a certificate of test.

In 1971 the measured voltage was 1.01864 V and when I measured it first, it was 1.01858 V

Well, 60µV is only 1.5ºC temperature change.  https://www.eevblog.com/forum/repair/solartron7061/msg3008786/#msg3008786

Didn't the volt definition change by 9ppm since 1971?

[edpalmer42]

I'm surprised that nobody has linked in this document.

NBS Monograph 84
Standard Cells - Their Construction, Maintenance, and Characteristics

https://nvlpubs.nist.gov/nistpubs/Legacy/MONO/nbsmonograph84.pdf

Everything you ever wanted to know about the care and feeding of your Standard Cells.

Ed

[doktor pyta]

Hi guys,

I own Guidline voltage standard built as 12 weston cells inside the oven.
I have it with some historical data. Each cell was tested for internal resistance and the values are close to 600 ohm as they should be.

However i don't find this setup useful .
I thought about connecting 10 of cells in series and dividing down to 10.00000V.

Is this idea reasonable?
Is there a better opamp for this purpose than OPA140 or ADA4522 or older LT1012 (i'm talking about the first stage) ?

I'm curious what do you  think about my idea.

[Andreas]

Hello,

I would use a low input bias current OP like the ICL7650 / LTC1050 or something like that as first stage.
I think for power off protection you will need a N-Channel MosFet like BSS138 or similar.

with best regards

Andreas

[doktor pyta]

Andreas,

thanks for the suggestion.

I was referring to a well known schematic from LT1012 datasheet with PMOS - without thinking too much.
Is it just a drawing error ? IMO the gate should be tied to -15V.

Your NMOS solution will work indeed if I add some capacitor to ground at the buffer's input.

BR

[guenthert]

[..]
I thought about connecting 10 of cells in series and dividing down to 10.00000V.
[..]

     Uh, why?  Consider yourself blessed to have access to a stable (saturated weston cells should still be superior to all solid-state references), virtually noise-free reference.  If you need 10.00000V, get yourself a lead-acid battery and a reference divider.

[Andreas]

Is it just a drawing error ?

Hello,

have a look at the LT1097 datasheet.
There it is a 4 terminal depletion Fet.
But it seems it is obsolete now.

with best regards

Andreas

[tggzzz]

Consider yourself blessed to have access to a stable (saturated weston cells should still be superior to all solid-state references), virtually noise-free reference.

Superior in some senses, not in others. It is true that a saturated cell will be better when measuring the noise of the DMM's reference, and when measuring temperature.

I've used my Solarton 7081 with two Trancell zener sources and a saturated Weston standard cell. Note the different vertical scales, and the 1500 reading is about 24 hours.

Trancell with 1.5ppm/15µV popcorn noise


Trancell without popcorn noise:


Saturated Weston standard cell acting as a thermometer, but without noise:

[doktor pyta]

@guenthert

the rationale for adding the buffer to 10 Weston cells is to:

1. provide low output impedance
2. provide protection against shorting the cells. It happens and it has bad effect.

I know that main advantage of my ovenized setup is the low noise.

Scaling 10.18 to 10.00000 should not introduce much of error /drift/ tempco.

According to the calibration reports my set of Weston cells drifts about 1ppm/ year.

[doktor pyta]

@Andreas

Good point.

The same application is in the LT1008 datasheet.
So it is not a mistake but a depletion MOSFET.

BR

[tggzzz]

According to the calibration reports my set of Weston cells drifts about 1ppm/ year.

And 40ppm/K

[HighVoltage]

Each cell was tested for internal resistance and the values are close to 600 ohm as they should be.

Interesting project!

How did you test the cell's internal resistance without loading them up?
Also, what temperature do you keep the cell at?

[Andreas]

The same application is in the LT1008 datasheet.
So it is not a mistake but a depletion MOSFET.

Hello,

its getting more and more mysterious.
according to page here it is a dual p-channel J-Fet:

https://www.darisusgmbh.de/shop/product_info.php/info/p8578_2N3609-----P-Kanal--Si-Feldeffekt-Transistor--350mW.html

my ECA-List lists the 2N4352 as replacement type
The replacement type should be a P-Channel MOSFET with substrate connection

http://www.calogic.net/wp-content/uploads/2019/03/2N4352_Datasheet_Rev_A.pdf

I for my part will stay with the logic level N-Channel BSS138 and the gate tied to a higher voltage than input voltage + ~2V.

with best regards

Andreas

[MegaVolt]

It is important to consider the resistance of a closed transistor. It should be comparable to the resistance allowed for the cell. 1GΩ is 1nA of leakage. It may be worth considering analog switches rather than individual transistors.

[Gyro]

Surely an analogue switch would load down the output of the cell(s) when the supply is removed.

[doktor pyta]

@HighVoltage

The procedure is measuring voltage across the Weston cell with and without 10Meg loading resistor. The loading phase should last few seconds and the cell will recover within hours without any harm.
The temperature is held at 30'C +/-0.1'C. Which not good for me because in summer months temperature in my lab is higher. But I can live with that.

It is interesting how Guildline made very stable oven using 5 transistors. It is a LC oscillator which operates at the edge of oscillation. The loop gain is determined by the divider made out of NTC and hermetically sealed resistor. See the attachment.

@tggzzz
fortunately it is ovenized

[Andreas]

It is important to consider the resistance of a closed transistor. It should be comparable to the resistance allowed for the cell. 1GΩ is 1nA of leakage.

Hmm,

I quickly measured a BSS138 (which I use to protect unbuffered LTZ1000 references).
With ~14V battery and in power off state I get ~19 mV across the 10 Meg input resistance of the 34401A.
So around 2 nA at room temperature.
So for a Weston cell I would perhaps try a BSS123 which is specced with a magnitude lower leakage current.

With best regards

Andreas