Resistance Substitution Box (0 - 100 Meg Ohm Resistance)

I needed a variable resistor with a very broad resistance range, from 1 ohm to 10+ megohms.

I did have a set of decade thumb-wheel switches that were gathering dust for the last 5+ years (original plan was to make an analogue thermocyler where you would set the temperature and time values with the thumb wheel switches but then arduino happened!). Thought it is about time I used these for a decade resistance box. Turned out well, though was much more work than I had anticipated.

There is a lot out on the web about building these decade boxes so nothing really breakthrough about it except for tapping the resistors at 99 ohms, 99 Kohms and 99 Mohms giving me the ability to create 10x voltage dividers and to reduce the switching resistance at the lowest range.

For the low resistance values, I needed at least a few watts of power handling capability so that ruled out surface mount resistors (about 1/10 W or less) and the standard 1/4th W through-hole resistors. Ebay vendors from China had an assortment of 1% 3 W resistors from 0.1 ohm to 10 Mohm for about $10. I did not need 3 W for the higher value resistors but for uniformity I did decide to use all 3 Watts. I also did not use the 0.1 ohm resistors as the internal resistance of the switches was more than about 0.05-0.1 ohm.

So for an eight decade resistance box with the lowest range of 1-9 ohms followed by 10-90 ohms, 100-900, 1K-9K, 10K-90K, 100K-900K, 1M-9M, and 10M-90M, I would need 8 of my decade switches. Was worried about leakages on the 90M ohm range but thought would give it a shot.

And yes this will give 1% accuracy at best, so the last digits at every setting would be irrelevant. That is, if I selected 99,999,999 (almost 100 Mohms) the error would be nearly 1 Mohm that is the last six digits (99,***,***) do not have much meaning. Now if I had 0.1% resistors I could add an additional relevant digit but did not see any 0.1% resistors on ebay.

I have shown the switch that I had. Yours may differ so make sure you get the wiring right. The circuit diagram fr the 8-decade resistance box is shown in a simplified format.

The decade switch I had was designed to be PCB mounted but I decided to solder the wires directly to the pins of the switch. This turned out to be a lot of work due to the large 3 W resistors and the small pitch of the pins on the decade switches. After I did every switch I had to test with an ohmmeter for shorts and had to use the solder wick to suck up solder bridges.

By the time I was ready to solder the last three sets of resistors I was exhausted. So decided to put the last 27 resistors on a Vero board, wrapped the wire of one resistor around the previous resistor to create a serial chain and then soldered solid copper teflon coated wire from a Cat6 cable (typically 23 gauge) to the end of each resistor in the chain. The other end of these wires were connected to the switches.

I cut the pieces for a box from 1/4th inch thick plywood scraps for the top, bottom, front and back panels and 1/2 inch thick plywood for the sides. Glued and nailed the back and the bottom to the sides.

I had removed the audio 3-way binding posts from a previous house that had in-wall speaker cables. So in the front panel I drilled 4 holes for the 4 binding posts and then I measured and cut out the rectangle for the eight decade switches. Painted the partly assembled box and the other panels with a black acrylic paint that was not conductive (had tested this for a previous project).

Screwed in the 3-way binding posts into the hole. Was not sure that I could solder the wires into the binding posts so used a 24 gauge wire (brown in the picture) with a doubled end that I tinned. The tinned end was held by screws in the binding posts. The other end of the brown wires coming from the terminal was connected as shown in the circuit diagram to the wired decade switches. You can see the vero board for the last three sets of resistors (the 10 Meg, 1 Meg, and 100K resistors).

Everything was checked with an ohmmeter. Also made sure that the resistance values were as expected.

Added hot glue to the binding posts so that they would not swivel (I did not use nuts to hold the binding posts) and to the decade switch and front panel interface.

Finally attached the top panel with a couple of nails and the front panel with the switches and binding posts with screws.

At a setting of zero ohms I got a reading of 0.7 ohms. This dropped if I used the 0-99 ohm binding post.

The high values were quite accurate, in general got about 2-3% error rate with a non-calibrated multimeter.

So finally designed the labels ... and one more useful little device that was surprisingly laborious to put together.