Digital Decade Box

[mendip_discovery]

Wondering if anyone else has had a similar idea.

A 4&2 wire resistance decade box with precision resistors but the resistance values are displayed on a screen and you select the resistors using up and down buttons and using relays it switches between the different resistances. For me, the thumbwheels have contacts that degrade and suffer from contact resistance.

Use could be just metrology as a calibration tool but also for projects if you wanted to dial in a resistor but would like to do it digitally/programed.

[electr_peter]

There are specialised instruments already for calibration
It depends on the use case. For calibration purposes it is understandable, for semi-precission or regular use I am not sure on the added value - it gets very expensive and bulky quick. And all that just to get a digital indication and control.

[EPAIII]

I went through this exercise a little while ago. The first thing you must do is get a very clear statement of just what you want the resistance box to do.

You say precision resistors so you want a high precision box. But then you say switches and/or relays. Both switches and relays have contacts. And those contacts will degrade on relays just as they do on the switches.

So, is it going to be "precision" or not. There is no sense in using precision resistors if you use mechanical contacts that are going to have many milli Ohms of resistance and also will change over time. Perhaps mercury whetted, reed relays would get around the contact resistance problem. I don't know as I have not used them very much.

Then there is a box for use while designing circuits. In my humble opinion, such design tools do not need precision resistors; 5% values would be just fine. The idea of using a 1-2-4-8 sequence of resistor values is tempting. Thumbwheel switches can use such sequential values and you only need four resistors per decade, instead of 10. There are inexpensive, boxes like that available. But, when you spin those thumbwheel switches you are not guaranteed that the resistance value will switch smoothly from one value to the next. The 1 and 2 values may not switch out of the circuit at the same time so you may go from 3 to 2 to 6 and finally to 4. In many cases that may not matter. But in some it will. Perhaps it is just me, but I would prefer having the full ten values per decade and resistors, even precision ones, are the least expensive electronic parts available.

And as for an LED readout, WHY? OK, it does look neat. But with thumbwheel switches the value is already displayed on those switches if you took the trouble of lining them up. And you have the added trouble of either plugging it into the AC outlet or changing batteries when they run down. Again, to me, it seems like an unnecessary gimmick.

I wanted a resistance box that could be used to check analog VOMs. I choose 0.01% and one 0.05% resistors for good accuracy. But I did not use any switches. I brought all the needed connections out to gold plated, five way terminals. The resistors were each connected to a common and there are a number of terminals for the various values. Those five way terminals are my switch. It has only three values per decade in a 1-2-5 sequence and the only way that resistances can be combined is any two can be used in series so, for instance, I could go from 10K to 11K or 10.1K or 10.5K, but not 10.3K or 10.4K. But for checking and even calibrating an analog meter, this serves well. I have a set of test leads with gold plated banana plugs for making the connections to the meters.

And if I ever decide I need a switchable resistance box for circuit design, which is a totally different need, I will just buy one of the available ones - but probably a name brand instead of a cheap, no name import.

[EPAIII]

One idea for a decade box that I did a lot of pencil work on was a box that would allow a potentiometer circuit (three wire connection) to be set to precise values for several decimal places. This was done with wafer style, rotary switches with 11 or 12 (I forget the exact number) or contacts but with their rotation limited to only ten positions. Each successive decade would be in parallel with the two of the eleven resistors on the next higher decade's switch. And that successive decade would have a total resistance equal to those two resistors in series. The combination of two resistors on the higher decade with the successive lower decade in parallel would then be equal to the value of one of the 11 resistors on the higher decade. Thus, when the higher decade is switched to different positions, the total resistance over it's top and bottom terminals would remain the same.

This could continue for any number of decades with the final value being taken from a switch with only ten resistors. That would be the center tap of the many decade pot.

Such a decade pot could be used to obtain a very precise adjustment in many circuits while avoiding the need for 10, 20, 25, or more turn pots. I never built one, but it should work. And it could provide adjustments with three, four, five or even more decades of adjustment. And if wafer switches with shorting contacts were employed, there would not be any large jumps in the adjustment which can be very problematic in some devices.

[Dave]

EPAIII, what you described is known as a Kelvin-Varley divider. A must-have for any self-respecting voltnut. 

[Doctorandus_P]

Some time ago I thought of making a Kelvin-Varley divider with Nema17 stepper motors with magnets on their axles and reed switches. All the switches and mounting holes for the stepper motor can be put on a standard PCB relatively easy.

A bit more recently I thought of putting brass bolts in a circle around Nema 17 stepper motors, and make switches from several layers of stacked 0.2mm (or thicker) brass strips. The Idea was to make an adjustable high power transistor (10x 0.1R, 10x 1R , 10x10R) so adjustable between 0.1R and 111R.

It would of course need a microcontroller to control the thing, this allows any sort of user interface you would like and also remote operation.

Neither the electronics nor the mechanics for such a project is very difficult. For a Kelvin-Varley you have to think very carefully about the accuracy you want. It can get very expensive for the resistors, or you can spend quite some time matching resistors.

With the adjustable power resistor you can use relatively cheap resistors, and then do calibration in software to get within 100mOhm over the whole range.

[EPAIII]

You mean they ripped off my design? I should have applied for a patent.

EPAIII, what you described is known as a Kelvin-Varley divider. A must-have for any self-respecting voltnut. 

[mendip_discovery]

Hmm, some stuff to think about. I will have to take a look at the HP relays to see what is special about them and if it isn't way too expensive.

Part of my brain was on metrology mode, as I have a Tansmille at work but it has a selection of resistors (10, 100, 1k, 10k,  100k, 1M, and 10M) and I wanted to recreate this in another unit as a project.

The other part of my brain was thinking that if you have a sufficient number of resistors you could just specify the resistance you wanted and it would give you the closest resistance value to that. Rather than using a dial and mechanically switching it. There would be the ability to change the resistance values so that the ones stored on the device took into account the ones drifting up and down and calculated the best ones for the task while showing you the actual current limits. But that is more complicated than I would like.