Dave feels a tad nostalgic, so takes apart his 25+ year old DIY decade resistance substitution box. And some info on how you can build your own. DIY thumbwheel box: www.eevblog.com...
Back in the mid-70s as a teenager I once did something broadly similar: a friend who was into photography wanted a timer for his negative/print copy enlarger. It was to set time up to about one minute, with one-second resolution all the way. So I built a 555-based timer with two different trimpot banks in series, one set to 0, 10, 20, ... 50 seconds and the other to 0, 1, 2, 3... 9 seconds. Rotary knobs labeled at tens and unit seconds would add the selected trimpot values resulting in surprisingly precise timed intervals from 0 to 59 seconds, with 1-second resolution.
@hitachi088 Suppose you know that you will need some 1% resistor to trim a circuit, but don't know what the value will be (can't calculate it due to unknowns, lack of knowledge, etc.) . You substitute the box for the resistor, dial it until the ckt works and replace the box with the dialed value. A box made with cheap 5% resistors (called a resistor substitution box) can be used in the same way when you don't need the accuracy and don't want to risk damaging your expensive decade box.
@hitachi088 Another use for a decade box is to act as the precision adjustable part of some measuring instrument. Many of these old-school (pre solid-state) instruments (Wheatstone bridges, etc.) had a built-in decade box, making them expensive. By moving an external box between different instruments, the instruments can be made cheaply and still have the desired accuracy.
@Deckardsvr Whenever you interpose a solid-state device between the resistors and the binding posts, the output resistance is affected by the on-resistance of the device and the device's ratings. Unless you can design some really whiz-bang circuitry that incorporates the device's characteristics with the desired accuracy, you will just be asking for problems. Using the MCU to switch some husky relays with well-maintained (cleaned) contacts could work, but that's a lot of trouble.
neat idea, i might also make one myself. I usualy use some of the 25 turns presets (the tiny blue ones with the golden screw), calibrate them with the multimeter and then use them. but yeah, it's way better to just dial the value in instead of having to turn everything off to change it. plus even though they have 25 turns of precision, the low edge isn't that good so a 1KOhm pot will jump go from 24 to 0 ohm, and that can fry some stuff, the rest of the range will do 1ohm precision just fine.
10:00 Not sure on that argument. Assuming the resistor values are statistically independent of each other you need to treat the tolerances as you would standard deviations to calculate total tolerance.
I was looking up decade resistors on eBay, and now there's a couple of different boards for 7-decade ones, where they used IDE headers and jumpers on the board, and you simply put the jumper on the value you want to use. Guess it could be easily adapted to use thumbwheel or rotary switches by using a compatible connector for these headers. And they're like 5 to 15 bucks a pop, depending on the model you get, so I'd say it's worth a try at least!
@CampKohler Correct. "Dicky" is a slang word for something that's "a bit broken" or "playing up" etc. Dick Smith was commonly referred to as "Tricky Dick" and the store as "Tricky Dick's"
I thought it over and realized that you are really using the percentage key in that calculation...never mind then. You remembered, but used a method different than I'm accustomed to.
Any ideas for heatsinking resistors? It might be nice to build one of these in an aluminum box and mount the resistors on the box so it can heatsink them, but how to get good thermal contact without shorting the resistors?
>MrMistery (I do not get a reply button for you for some reason) Worst case is what tolerance is. At worst, a 1% 100 Ohms resistance will be 99 or 101 ohms. What you are talking about is what you might measure which is what the calibration sheets showed for the IET substitution box.
+SlyPearTree I still don't understand. Forgive me if I'm being thick here, but by adding uncertainties in quadrature the overall percentage that it might be off from ideal value will decrease from the original value per resistor, won't it? In this case [ 09:50 ], uncertainties for each resistor is 100 ± 1% but when in series as drawn, it would be [math] 100Ω + 100Ω + 100Ω +100Ω ± \sqrt{1^2 + 1^2 + 1^2} = 400Ω ± 2Ω = 400Ω ± 0.5% [/math] am I wrong in treating them as random values? that is some damn luck that I came back here almost half a year later.
@CampKohler you're right, the adding Ron would be a catastrophe especially in the lower decades, that's a shame, i thought 'why there's no digital decade box anyway ?' and the 'relays' approach as you said would be a terrible hassle, thanks for the answer.
That's a cool decade box! Reminds me of one I made as a teen. If you need hand made Resistors, shunts, or sensors, PM me and I will send you some on the house to use in future videos. Cheers.
The screwdrive has white in the bit grooves. Poking holes in dry wall? If you are going to use inexpensive rotary switches (while not ideal as the heavy-duty monsters used in a traditional precision decade box, though still useable for bench work), it would be better to use switches with an open-air design so that you can periodically spray contact cleaner to deal with oxidation and prevent wearing of any contact plating.
I may be mistaken here; but when calculating the tolerances, isn't he taking it as worst case scenario? Wouldn't uncertainties add up as independent uncertainties in quadrature and therefore go down?
Huh, never thought of having a bunch of equal value resistors be in series like that. Shame there’s no simple shortcut like that for capacitance boxes.
Those 10 position rotary switches are a bit hard to find and also kind of expensive... I'll build the 2nd option since I have some 4 Position DIP Switches and they're also dirt cheap. I can also get some metal film Resistors, 0.5W and probably 1% accuracy. I'll probably put all this together on a perfo-board / strip-board and 3D Print a case.
Nothing was said about the ratings of the resistors. Fuses could protect the Rs, but will introduce inaccurate/unstable variables. Has anyone devised a circuit that will open if the rating is exceeded, but introduce a known and stable resistance (that can be mentally subtracted while using the box)? Solid state with a reset button would be best. Maybe with a set of n.c. clean relay contacts gobbed with air-excluding silicone grease.
Actually, it is VERY unlikely for all of series resistances to be high or for all to be low; it is much more likely for some to be high and others to be low. Because of this the tolerances add in quadrature, dR = sqrt( sum(dRi^2)) = sqrt(N) dRi if Ri=Rj for all i, j. Or expressed as percent dR/R = sqrt(N) dRi / (N Ri) = (dRi/Ri) (1/sqrt(N)). The %Tolerance is lower by 1/sqrt(N).
You’d have a point if the resistors were like a random sample from some unimodal continuous distribution. But they aren’t. Resistances in resistors you buy aren’t Gaussian distributed (not even truncated Gaussian) because they don’t come straight from the production line. They are preselected. They likely come from the same batch and the resistors are usually not centered on the nominal resistance, they are more like higher precision rejects (you’re getting say 0.5% rejects as 1%). Say an entire 1% tape may be like 0.4% high - and be matched much better, say 0.05%, within a consecutive run on the tape. It’s much worse with 5% resistors. They are the 1% discards/overruns, almost always (at least the ones I usually buy). You measure them and they measure much better than 1% between each other - because they are manufactured to 1% tolerance but were slightly out of the acceptance range. Or even maybe not rejects but 1% overstock marked and sold as 5%. Sometimes there’s not enough demand for “less popular” values so the manufacturer is free to sell them for a bit less as lower tolerance parts. Or sometimes the production line has to make runs too long for the market, so they know ahead of time that only a small part of the batch will be sold at highest tolerance. Basically the tolerance range doesn’t imply anything about distribution. If you get enough 5% of one value from a manufacturer don’t be surprised if you get a multimodal distribution, with modes around better tolerance series nominal values :)
@@absurdengineering Actually, I DO have a point; no ifs, ands, or buts. The central limit theorem from statistics supports my point. As more resistors are added in series, the resulting total resistance distribution DOES become more and more Gaussian regardless of the individual resistance value distribution. The vast majority of commercially available resistors are manufactured to specification; this is no longer 1930. The materials are specifically designed and molded to yield the specified values and to remain within those specifications over all temperature, humidity, pressure, etc. variations typically expected from their environment. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-gFEYuaY35Vo.html
cool, I've never seen one of these before :P. I'm an EE undergrad kind of becoming disillusioned with all the seemingly pointless work I have to do. It's gotten to the point that I'm not learning much as I'm just struggling to cram the next homework or test in. Your videos are a great! Got any advice for a young EE in the making?
@09:30 abouts, so 4 of 1% resistors equals 1%, ok get that, but..... if you have 2 1% and 2 of 5%, then how do you work out the effective % ? is it 5%+5%+1%+1% = 12% / 4 (for 4 resistors) so 3%.... or is it more complicated ....
Anyone making a box with 10% or 20% resistors for repair work should realize that a resistor maker may pull out all the 5% resisters from the 10% or 20% you might buy. So if you want 5%, buy 5%, rather hoping the other tolerances might be that close by accident. Note that "Dick Smith" is an electronics parts shop and has nothing to do (I think) with Dave's use of the word "dicky."
Dave, is it possible to use a rotary encoder(s)+microcontroller(arduino) to switch the restistors?because the way i think it it would be a lot cheaper and same reliability as using high quality rotary switches.
I have a question : what about analog switches (like the old 4066) some modern smd models are cheap these days, i thought i could built a decade resistance controlled by an MCU with enough I/Os using 4 AN/switchs per decade using your second schematic. it would be great to dial the desired value on a keypad :) is there anything wrong about this approach ? by the way, you built that one at 12 ?, very impressive!
Dave 0.0025% x 60% does not equal 0.15%. The formula you want is: (R nominal/10E6) x (PPM) x (Delta Temp) which will give you the 0.15%. Just thought I would remind you, in case you have forgot how to calculate a resistor PPM/Centigrade change.
You made this when you were around 12 years old. And you don't know if it's 20 or 30 years old? So you're in doubt being around 32 or 42 years old? That sounds like you're at least 50 years old 😂😂😂
+seth altobelli Google says "Some caps -- such as nearly all electrolytic capacitors and tantalum capacitors -- are polarized. Such caps use some sort of chemical reaction between an anode and a cathode made of two different kinds of materials to form a thin insulating layer." So basically it's polarized for the same reason batteteries are polarized, different metals reacting with an electrolyte between them. :)
It's since yesterday I'm trying to figure out what are those black and red connectors on the box called: can anybody help me out? I'm trying to find some for my own box but I don't know what they're called.
Jesus Those are actually called 5 way terminals. They take banana plugs, and unscrew to take bare wire, fork terminals, eye terminals, and I forget what else.