Selectable 4-Terminal Resistance Standard

[Nogtail]

I've repaired a few multimeters recently and needed something to check the resistance ranges. I wanted a reference with values between 10Ω and 1MΩ with 0.01% tolerance, which should cover most multimeter ranges.

All the resistance standards I've seen use separate terminals for each resistor. I'm lazy and didn't want to move cables around to select resistors, so I used a rotary switch instead. Using a switch also ended up costing a bit less - gold-plated Pomona banana jacks aren't cheap!

The switch introduces some extra resistance, but from my measurements, it's only around 10mΩ, which is fairly negligible. I'll be using 4-wire sensing most of the time, which should eliminate switch resistance.

Are there any obvious downsides anyone can see with the design/construction? My main worry was mechanical stress on the switch influencing the resistors. I mounted the resistors a couple of millimetres off the PCB to give them some mechanical isolation. So far I haven't been able to measure any difference in resistance when applying force to the switch - at least with a 6.5 digit meter. I'm working on an SMD version, so perhaps it would be more of an issue there.

[enut11]

Very professional looking instrument. Well done. In my view, 4W is only needed below 1Kohm. However, I acknowledge the simplicity and ease of use of your design. The test leads in most hand-held multimeters introduce more error than calibrated resistors so it should be fine. Would you share how you made it look so good?
enut11

[Gyro]

A very neat implementation - I can say that because I went for the air wired approach! ( https://www.eevblog.com/forum/testgear/my-shunt-collection-(ucurrent-passive)/ ). Remember that you can also use the resistance standard as a multi range current shunt with lower voltage burden than typical DMMs (but without the protection). In my case, I went for 10R to 1k as I was targeting current shunt use.

[slavoy]

Very neat work 
May I ask how you made the labels on the enclosure? Is this a PCB?

[Nogtail]

Very professional looking instrument. Well done. In my view, 4W is only needed below 1Kohm. However, I acknowledge the simplicity and ease of use of your design. The test leads in most hand-held multimeters introduce more error than calibrated resistors so it should be fine. Would you share how you made it look so good?
enut11

Thanks! 4W definitely isn't needed for the upper couple of values, but it came for free with the design. It's convenient to have though, as you can leave the multimeter in 4W mode for all ranges.

I designed the front panel in Inkscape then imported it as a vector into KiCad. I really like the look of matte black PCBs with gold plating, but matte solder mask is a huge pain to get clean. Usually I'd throw boards in the ultrasonic cleaner after soldering and let it sort them out, but I didn't want to risk affecting the precision resistors so I cleaned them by hand.

A very neat implementation - I can say that because I went for the air wired approach! ( https://www.eevblog.com/forum/testgear/my-shunt-collection-(ucurrent-passive)/ ). Remember that you can also use the resistance standard as a multi range current shunt with lower voltage burden than typical DMMs (but without the protection). In my case, I went for 10R to 1k as I was targeting current shunt use.

Your 10R-1k box looks very similar, I'm surprised I haven't stumbled across it already! I'd like to build myself a 10M/100M box in the future - probably with an air wired approach like yours, or at least some isolation slots to cut down on PCB resistances/contamination throwing off the values. I was planning on using the some of the higher values as current shunts to get measurements of photodiodes in low light environments.

Very neat work 
May I ask how you made the labels on the enclosure? Is this a PCB?

Yep, it's a PCB as a top panel for the enclosure.

[Nogtail]

I've made a version with SMD resistors to see if I could get the cost down. I had to make some sacrifices with the tolerance at both ends - 10Ω and 1MΩ are only 0.05% in this version, but the other values are still 0.01%. I swapped the gold-plated jacks out for some nickel-plated ones, and the ENIG top panel is replaced with silkscreen. I also used a gloss solder mask instead of the pain-in-the-ass-to-clean matte black stuff.

If anyone knows of any 10Ω/1MΩ 0.01% resistors for less than $10 in low quantities, I'd love to make a low-cost version with tighter tolerance.

Mechanical stress seems to have a minimal influence despite the lack of mechanical isolation for SMD resistors. I was able to measure a difference this time - about 1 PPM while applying 20N to the switch, but it returns to the previous value once the force is removed. Not bad, considering the temperature coefficient is 5-10PPM/°C, and it's unlikely you'd be popping handstands on the switch while taking a measurement.

[donlisms]

Huh.  Hmm.  Pondering.

My first thought was that at first glance, 10mΩ shouldn’t be an issue in a 4-wire connection, since the whole point is to eliminate the influence of resistance in the connections.

Then it seemed to me that you might actually have a 4-wire connection to a switched array of 2-wire connections, where the switch adds resistance and uncertainty, as it’s part of each resistor.

Then it occurred to me that if you had enough switch layers, three I think, you actually *could* switch a bunch of 4-wire resistors, eliminating not only the resistance of the switch, but the variability of it.  It seemed to me that it might result in a more confidence-inspiring standard.

But that’s just pondering!

[enut11]

Hi @Nogtail
I like the 4mm sockets with the threaded terminals. Where do you get them?
enut11

[Nogtail]

My first thought was that at first glance, 10mΩ shouldn’t be an issue in a 4-wire connection, since the whole point is to eliminate the influence of resistance in the connections.

The switch resistance should only be a problem for two-wire measurements. Both red connections take separate paths through the switch and only meet at the legs of the resistor. The black connections pass through all resistors to ensure they always meet at the base of the selected one. The layout is roughly like this, but with a DP6T switch and more resistors:



In reality, it's a little bit more complex as I've got two resistor footprints, but it should always allow true four-wire sensing.

Hi @Nogtail
I like the 4mm sockets with the threaded terminals. Where do you get them?
enut11

The sockets are Pomona 72912-0 (black) and 72912-2 (red). I ordered them from either Digikey or Mouser.

[donlisms]

Ah, yes - pardon my interruption!  I assumed wrongly. 

[coromonadalix]

not bashing in anyways
as @NOGTAIL assumed

normally test jigs / rigs, tend to avoid any switch contact(s)   who can oxidize over time or become more resistive, or bad contacts impedance's  ... 
anything who could create falses values or add value(s)  ..