How to trim/adjust a decade resistance box?

[TERRA Operative]

I recently got myself some really nice big old rotary switches from an old wheatstone bridge thingo, and I am using them (and a stack of the 0.5% resistors) to make a decade resistance box.

One thing I am stuck on is how to trim or adjust this thing so it reads as accurately as possible?
Do I need trimmer pots on each resistor each rotary switch? Solder in bypass resistors where needed?

What is the usual way of adjusting these things?

[TiN]

I believe you should change the topic from "How to calibrate" into "How to trim/adjust". Because precisely speaking, resistance calibration is a measurement of your resistors with known uncertainty, but your question is about how to trim and provide adjustment options for better accuracy (usually at cost of worse stability).

Maybe you'd want to move discussion into this fine thread instead? 

[TERRA Operative]

Ah, yep, changed the title. Not up with the nuances of the terminology...

Looks like I might need to trim each individual resistor...

[Harfner]

I have one of these old decade resistors. I did not find any way to adjust the value inside. I have to assume that they tweaked the individual wirewound resistors. By filling away a little bit of the wire it is possible to change the value up. That is not possible with metal film resistors, at least not in a reliable way.
Trimmers are often not very stable. You may be better off selecting your resistors or solder in correction resistors.
What kind of precision are you trying to achieve?

[TERRA Operative]

I just need it close enough to be reasonable. Not super accurate but as accurate as easily and simply possible, if that makes sense.
I don't need a resistance standard, just something vaguely reliable.
I'm still constructing it, so I'm not sure how far out it is yet, I just thought I should think about it now rather than later when it might be more difficult to change.

Maybe adding series/parallel resistors just where needed is the best idea? I'm sure not all of them will be too far out of spec, I'm using 0.5% and 1% +/-50ppm Vishay MBB series resistors, so they should be ok enough for general use.

[Conrad Hoffman]

FWIW, in my Julie Research Kelvin-Varley divider, they started with all the resistors very slightly under nominal. Each resistor has a short piece of manganin wire soldered in series to correct the value. I'd do something like that, plus bring down high values with a parallel resistor. The parallel resistor doesn't need to be very good, since a slight drift will have almost no effect on the total.

I believe mica card resistors were adjusted with a draftsman's power eraser, using an ink rubber. That's more aggressive than the soft ones, which might not do anything at all. In general, resistance boxes are usually built with decent tolerance resistors and that's that. I've never seen one that was or could be trimmed. I have seen a few with black overheated resistors!

[Vgkid]

Post pics , once you get everything.
For values below 1000ohms it might be easier to select ones that are higher than the needed value , and put a resistor+trimmer in parallel. For values 10k and above you can still use the above method , or use a series resistor+ resistor/trimmer in parallel.

[Harfner]

You mentioned: ... (and a stack of the 0.5% resistors)

So you want to recycle the old resistors? Are you sure, they are worth it? Resistors with 0.1% can be quite cheap.

Post pictures of what you already got, maybe things become clearer then.

[TiN]

I've never seen one that was or could be trimmed.

ESI makes RS925D decade, which can be user-trimmed. There is PCB with trimmer pots to adjust top 4 decades.
Fluke 720A also allow trim on top 3 decades. It's a KVD, but idea is essentially same.

[Doctorandus_P]

Have you considered buying some extra resistors for this project?
Resistors with a tolerance down to 0.1% are reasonably cheap. (About 20ct each).

[IconicPCB]

Why not measure the value of resistor in situ without disturbing the assembly and relabel actual values on the instrument fascia.

[TERRA Operative]

Yeah, I reckon I'll go throw a bunch of 0.1% resistors into my next Digikey order.
Once I have things in a slightly more assembled state, I'll take some photos too.

I'll check out those resistance boxes you mentioned TiN, I might steal some ideas from them.

[David Hess]

If I was building one from scratch and could measure the precision resistors to a tolerance better than their absolute specification, I would include low value series resistors to raise their value a controlled amount, measure the series value, and then calculate the add the high value parallel resistance needed to bring them to as close to the desired value as I could measure.

[TERRA Operative]

Alrighty, assembly has begun, pictures incoming!

First pic: Markings on rotary switches. Anyone know who makes/made these?

Second pic: Side view of rotary switch.

Third pic: Front panel (Knobs waiting for shaft extensions) Does anyone know who makes these knobs? It'd be nice to get some that are numbered correctly (0-9, instead of 0-10 etc)

Fourth pic: Backside of front panel.



I have used 0.1% resistors where I could, 0.5% elsewhere (where 0.1% were not available).
Chunky 5W for the 0.1ohm, and 1W for the 1ohm and 10 ohm, 0.4W elsewhere up to 10GOhm 10Meg ohm.


Also, I have some questions....

I added the extra binding posts for 4-wire measurement, just because. Should I just hook the binding posts together, or run seperate wires to the first and last rotary switch? What is best practice here?

Also, for trimming the resistors, should I measure at the binding posts to take all the internal wiring into account? Set everything to 0ohm, then select 0.1ohm, measure trim, select 0.2ohm, measure, trim, select 0.3 ohm, measure trim, etc etc?

[TiN]

0.4W elsewhere up to 10GOhm

Gohm? To get any meaningful readings in Gohm region, your design wouldn't apply. That is area where guarded boxes, shielding and teflon insulators are required...

[TERRA Operative]

Whoops. I meant Meg ohm. haha