Having a few decade resistor boxes - thinking about building a DC source. Ideas?

[Ysjoelfir]

Hey everyone,

some time ago I got my greedy hands on a few old Hartmann & Braun decade resistors (single magnitudes, not your typical 000000 to 999999 Ohms thing). Those are 1% (0.1R, 1R, 10R) and 0.1% (100R, 1k, 10k, 100k) wirewound resistors on high quality ceramic switches, every magnitude in its own box. So, nothing too fancy.
Since I already have a more precise resistor decade and the cases of some of those are quite busted (have been used in university for the last 60 years) I thought about using those to build a simple DCV calibrator. I guess you would normally use a Kelvin Varley Divider setup, but maybe someone has build something simillar with single decades?

My basic thought was to create a more or less stable reference voltage, give that onto the serialized decades configured as a voltage divider, buffer the output voltage and boom, done. Sure, it's not that easy since as soon as you add a second decade to it it doesn't work anymore. Maybe someone has a good idea what could be done without reducing stability and accuracy too much? Summing several reference Voltages that were divided and buffered seems to be a bit over the top. Am I missing something? There should be an easier way, shouldn't it?

PS: Posting that in metrology since I would love to hear your ideas to get the best result possible

[TiN]

You should list your understanding for best result possible (accuracy target) and budget expectations. There are multiple ways you can go with such projects, and result would be quite different too. Since you don't provide details about your resistance decades and don't know their stability it may (or may not) be a good choice in such project.

Perhaps looking at the boxes like EDC MV106 can give you better idea, as these are quite easy (in concept) and have decent performance for what they are. Good for up to 40 ppm a year or so with 5 digit resolutions. Everything better will be more expensive and more difficult to maintain or verify.

[Ysjoelfir]

Hi TiN and thanks for your reply I actually read your article about the EDC MV106 quite some time ago. I guess that is the reasone that somewhere in my mind there was the idea that it shouldn't be to complicated, looking at the schematic it looks like an easy concept.


Sadly I don't have that many details about the decades besides what I already wrote: They are Hartmann & Braun "Technischer Dekadenwiderstand" for DC to 10kHz, accuracy is 1% for 0.1R to 1R and 0.1% for values above 1R.
They consist of 10 identical wirewound resistors on ceramic bobbins and a nice clunky 11 position switch. I have the values 0.1 (so 1R max) to 100k (so 1M max) available with 1k and 10R being available twice.
The only other thing that I can find in regards to accuracy is the number two in a star like in the accuracy class of analog panelmeters.
They are simmilar to those: https://www.ebay.de/itm/143221153191?ul_noapp=true
Edit: I just noticed that the seller posted a photo of a "datasheet" of those specific decades. If we assume that the resistors are as similar as the specs of the decades then they should be made from "Aurotan", which is actually a brand name for Manganin. So that woule mean a temperature drift of 10ppm per kelvin.

Regarding my personal goals in accuracy and the available budget:
Surely better is always... well, better. But since I don't own any too fancy lab equipment (yet.. the TEA virus is strong.) and my keithley 199 is the most precise thing I have. It would be cool to be able to build something that could potentially calibrate the DMM, but it is not neccessary to get too fancy. Let me state it like this: A LTZ1000 shouldn't be needed here It should be quite stable over time and temperature but I don't want to get to fancy and basically scrap the decades and buy a whole bunch new resistors. If those wirewound resistors prove to be not good enough - well, then they aren't and I have to work with what I have and reduce the spec. Price wise I am not that restricted, however, everything should be somehow reasonable.

[branadic]

You can duplicate something like the Burster DC-Standard, just take a look into the schematics that can be found here:
http://www.ko4bb.com/manuals/2.206.157.104/Burster_6401_DC-Standard_Operator_Manual-1980.pdf

-branadic-

[Ysjoelfir]

That is actually pretty similar to the concept of the EDC. I guess I will really try to replicate this concept, it is simple and straightforward.
Since I haven't been doing anything serious in metrology I don't know how I can calculate the theoretical resulting accuracy and stability using those Resistors (assuming they are indeed made from manganin with 10ppm tempco).
How stable should my reference be to fit to those things? Obviously a more stable ref isn't bad, but where is the sweet spot in regards of cost vs. complexity? How can I calculate this?

[branadic]

As Illya already suggested, start with a LM399, this is the simplest way of making a reference as you can use it as is, no need to fiddle around with zero t.c. current finding, no selection required (except for popcorn noise or low noise if you want to go to the limit) and cost are low. Simly use it with an appropiate resistor, that's it.

-branadic-

[beanflying]

Just watch for loading from the Meter, if you only have a 10Meg one you will also need to add it into the calculations if you are trying to use the higher resistances on the box in particular. Even some of the 6 1/2 digit meter won't stay in Gohm ranges on their higher ranges.

[Ysjoelfir]

As Illya already suggested, start with a LM399,

I must have missed that.