Breadboard friendly Four Decade Programmable Resistor Board, 1%, 1/4W, KIT (WHL #25)
Got some more Open Hardware today! Not mine, though…it’s a four decade resistor adapter from luftek, who apparently comes from Slovenia. He has a github repository for his project called Dekada if you want to check it out (literal git checkout is also possible 
). I happened to buy Version 0.1 back in September 2016 for dirt-cheap 1.99 USD per unit – I bought two pieces via eBay, directly from luftek. Unlike the Wan Hung Lo guys, we in Europe generally have to pay a reasonable amount of money for postage…in this case 3.98 USD. So one of these puppies cost me just short of 4 USD/EUR at the time.
Currently, there are no more of these available. luftek decided to ramp it up to 5 decades in v0.2 and he just released v0.3 with minor changes a week ago. All versions are still available, so if you want to spin your own 4 decade board, you’re good to go with anything that is on the repo. Same is true for the 5 decade versions, which also could be available on eBay or the like as of now. If you’re interested, have a look before you send the PCB to a manufacturer yourself.
This is what I got in the mail:
So luftek included the PCB and anything that is needed (almost…), you just have to cut the pinheaders to size. I’d recommend soldering the SMD parts first, as this might be trickier if you have the THT stuff in place.
On a side note: One of the PCBs still has some of the panelization edges around, while the other was filed flat. No biggie – remove it using some pliers and smoothen it afterwards.
You might have noticed that one of the SMD resistor strips is shorter than the others. Initially I thought these had the same origin as my 123 resistor kit back from WHL#12, but they have straight edges, which means they have been cut using a machine. Not by hand. I guess one could order sections of nine from a Chinese manufacturer if your order size is large enough, but I doubt some Slovenian Open Hardware Guy will ever reach such numbers. So they may be 9 resistors long by mistake, which I will explain in a minute.
This is the fully assembled unit – it took me around one hour to do all the soldering and cleanup. The THT footprints are just perfect, the pinheaders almost snap in and are very easy to solder. I didn’t use the 4-pin single-line header of the breadboard connector, as I had some double headers around and thought it just looked nicer. But if you don’t – no problem, just make sure they align well. I also cleaned up the female headers a bit, as they were not super-clean cut and luftek most likely loses 2 pins per every 2 pins he produces with this method. Not sure if that is cheaper than buying 2-pin ones pre-cut from larger blocks or 2-pin individual molds.
That’s the bottom view (surprise!). In here, you can see the fully populated SMD footprints…including the 10th of the 10kΩ resistors. That’s the one that was missing from the SMD strip, and I replaced it with one of the 0805’s from my own order. Bite me if the lower 1/8 watt rating will limit this system – sinking 1.25W over 100kΩ would require 350 volts, and applying 350 volts on a breadboard with 2.54mm spacing…nope. 
It still looks a bit messy with flux residue, but this PCB was cleaned multiple times with isopropyl alcohol and a short ESD brush. I didn’t dare to used acetone, might mess up the silkscreen or the entire board (as well as the header plastic parts), you’ll never know until you try…
Also, I think my soldering isn’t THAT bad…leave me a comment if you believe otherwise After all, this is hand soldered, not done in a reflow oven.
So how does it work? Well, each decade obviously works the same and they are put in series. Which also makes it easy to produce 5 decade versions of it – just slap one additional decade block onto it. As you can see, there are ten settings for your jumper link – 0 to 9. The “0” setting creates a dead short from the input line to the next decade block. Shorting the “1” setting will have one resistor in line, setting “2” will have two, and so on. So one uses 9 resistors for every possible jumper link configuration.
What’s the last resistor good for? Well, initially I assumed this was for tweaking the values, just like a trimmer pot. But for such a crude device that’s already overkill, and the supplied 1% resistors aren’t that bad already. So having a closer look (use the picture of the unpopulated board above), the 10th pad is put in series after the others.
First, I didn’t populate it because I like clear malfunction of a device if I messed something up – like not putting in a jumper in one of the decades. But it dawned on me…if I adjust the resistor jumper while having something connected to it, it will actually go open circuit. That might be a bad idea…but if there is another resistor installed that is in series all the time, the worst that can happen is that the resistance goes up to the “1” setting of the next decade. It does NOT affect all other resistors because it is not in parallel with them – it is parallel to the jumper short that you are creating. And as we all know the lovely t-shirt that has the “resistance is futile” resistor vs. short print on it – it doesn’t really matter. But again, when adjusting the resistance, this is the life-saver for your DUT. So I retrofitted it. And that’s why you need ten resistors of each type
The additional resistor technically makes the silkscreen claim of “99.990R – 0R” wrong…as each decade can not only go up to 9, but to 10, so the highest achievable resistance is 10*10Ω + 10*100Ω + 10*1kΩ + 10*10kΩ = 111.10kΩ. Yeah, call me a smartass.
As usual, I couldn’t resist not only testing the device after full assembly, but also checking the specs. Remember, luftek supplied 1% resistors. I didn’t really expect specs down in the ppm range, so I kept the y axis in percent this time:
Well, doesn’t look this familiar? Above spec for the first decade and a little beyond, while the rest hovers around well inside the spec. Yep, that’s friggen contact resistance. 4% at 10Ω is just 0.4Ω – which isn’t that much for 4×2 contacts at the decade jumpers and two more contacts on the breadboard connectors (yeah, and 2 more in the multimeter, they are already corrected for in the Uni-T 203 calibration 
)
(the two data points at the beginning of each decade are the “1” setting of this decade, and an open jumper of the last one. For the 100kΩ line, these are 99.99kΩ (all jumpers max), 100kΩ flat (0 + 0 + 0 + last jumper off) and the already mentioned 111.11kΩ max setting)
If you remove those 0.4Ω offset, the first decade is absolutely spot on and the rest is at or below 0.5% – great. I could adjust for that in hardware by replacing the first resistor from 10Ω to 9.6Ω – but that would require a 9.53Ω one from the E48 series as a plug-in replacement, or 240Ω from E24 to be put on top of it. There’s no better combination when using readily available E6 or E12 series resistors (and not selecting high-tolerance resistors by hand)…so I couldn’t be bothered. If I need that kind of precision, I’ll
-
a) use my resistor box (#P1) that is more like <0.2% over a wider range
-
b) NOT use a friggen breadboard
So there’s that…
