DIY Cheap resistor decade substitution box

[vtl]

After seeing these adjustable resistor substitution boxes I decided that they would be a nice addition to any person fiddling with electronics regularly. I don’t particularly like potentiometers for this purpose as you usually have to have a pretty decent quality one otherwise it can be hard to dial in a precise resistance.
After looking on the interwebs, it became clear that these things are quite expensive, usually utilizing expensive rotary switches. We’re talking on the order of $50-$200 for what really is just a resistors hooked up to some switches. Surely there’s a cheaper way?

I saw these thumbwheel switches on eBay for a reasonable price, just $5 for 10. These can be stacked and locked together, with the numbers ranging from 0-9. The clips aren't particularly strong so glue may be needed to keep them together.



Design
 
Looking at the spacing it seems like it could be ideal for 0603 resistors. I decided not to go this route for now, I had a look at element14 and when you take into account minimum order quantities, the costs can rise pretty quickly when buying precision resistors.



The biggest challenge to a resistor box is making it accurate. The box just puts the resistors in series to make up the value so the error of the resistors can add up pretty quickly and give you pretty wild deviation from what number you dial in. Now accuracy was not a large factor for me, as long as it was somewhat correct that would be fine for me.
I wanted a box that would do 1R, 10R, 100R, 1K, 10K, 100K and 1M. Even if I had got 0.05% resistors, at 100k this could produce an error of 50ohms, quite large considering this device is supposed to do 1R steps. Clearly the larger resistors require a very good tolerance or you’ll end up with large offsets from the value entered.
So I concluded that this box would not be intended on being a precision instrument. It would be just a nice thing to have for circuits that require irregular resistances and I could just hook it up to a multimeter if I needed more accuracy. It would be similar to a pot but with a huge range and very fine and precise adjustment.
On eBay they sell resistor kits pretty cheaply. A while ago I bought 2000 1% assorted resistors for $11 which is quite nice. I decided the accuracy would be terrible no matter what tolerance I use so I figured 1% should be good enough for hobby usage.
These thumbwheel switches are much more compact than those huge resistor decade boxes I’ve seen. Instead of making up some kind of housing, the resistors should fit directly on the PCB and should not increase the overall size at all. The wiring is fairly simple but it may not be immediately apparent when looking at the switch. You have a common pin and 10 other pins corresponding to each digit. For a 10M (9,999,999R), 7 digit box this will require 63 resistors, with 9 resistors per digit.




 
This is a simplified drawing of the switch with only 6 contacts drawn. You would measure the resistance between the common and pin 0. In this case the wiper is switched to 3 so the measurement would be 3xR. You can daisy chain these up to provide more resistance, just attach the common to pin 0 of the next decade.







 
And here is the end result. It can be quite time consuming I guess if you work fast you can do 1 switch every 10-15 minutes or so. Its probably a good idea to check each switch before combining them up, I ended up gettign a few soldering errors and reworking it when its all glued and soldered together can be a bit tricky.
 

Accuracy
The accuracy will depend greatly on what value is shown on screen. If you use only 1 resistor from each decade then this will be highly accurate, but if you use many resistors then the error will add up. So I’ve tested by setting each range to 9 and recording the reading. REL was used when the resistor box was set to 0 to null out the resistance of the test leads




 
The percentage error looks quite bad at the lower resistances but the actual absolute values are quite small. We’re talking less than 1.5ohms at the errors of greater than 1%.
While probably not lab grade equipment I’m pretty pleased with the accuracy results. In general, when dialling in high values you tend to need to trim it out slightly. For this device, most of the time the device will give a value less than the desired value. I suppose it is possible to improve the device by adding some trimmers for each decade and might be handy if you’re just building up a 1K box.

Overall cost estimate:
The parts cost is pretty cheap for this:
$5.45 for 10 thumbwheel switches
$.0055*63= $0.35 for 63 1% resistors
Prices in AUD

So surely this beats the hell out of those crazy expensive rotary switch boxes in terms of price and portability.

So go out and build your own, I’d be interested to see what kind of error if 0.05% or better resistors were used.

[gregariz]

If only you had posted a day earlier! I finished a couple builds last night.

The only tip I have is make sure you clean the flux off well as you'll see it on the Meg Ohm switch settings.

[IanB]

Very neat idea and good job--but I'm a bit confused about your accuracy estimations.

1% is 1%, whether it be 1% of 1 ohm or 1% of 100k. The circuit doesn't know the difference, and the 1% 100k resistor is no less accurate than the 1% 1 ohm resistor. (Consequently, a resistance value of 134,968 ohms, say, is rather meaningless and also not likely to be useful in any practical application.)

Also, the error doesn't "add up" when you put resistors in series. It remains exactly the same at 1% no matter how many resistors you combine. (The variance changes, but the mean error doesn't.) This means that if you put ten 1k, 1% resistors in series you will get a 10k, 1% resistor. Therefore, I would say that if you have an error of 2% on your 9 ohm value, it is either because the resistors themselves are out of spec or you have an inaccurate measurement. (At low resistance values you also have to worry about extra circuit resistances like switch contact resistances getting in there, but your measurements don't indicate that problem.)

[vtl]

Very neat idea and good job--but I'm a bit confused about your accuracy estimations.

1% is 1%, whether it be 1% of 1 ohm or 1% of 100k. The circuit doesn't know the difference, and the 1% 100k resistor is no less accurate than the 1% 1 ohm resistor. (Consequently, a resistance value of 134,968 ohms, say, is rather meaningless and also not likely to be useful in any practical application.)

Also, the error doesn't "add up" when you put resistors in series. It remains exactly the same at 1% no matter how many resistors you combine. (The variance changes, but the mean error doesn't.) This means that if you put ten 1k, 1% resistors in series you will get a 10k, 1% resistor. Therefore, I would say that if you have an error of 2% on your 9 ohm value, it is either because the resistors themselves are out of spec or you have an inaccurate measurement. (At low resistance values you also have to worry about extra circuit resistances like switch contact resistances getting in there, but your measurements don't indicate that problem.)

Ah yeah thats right regarding the numbers, but what I meant was that the absolute values in ohms do add up. I didn't say 9 1% resistors in series = 9% tolerance.

Percentage errors will roughly stay the same but when youre building a device thats supposed to have 1ohm resolution, youll never get near that kind of precision at the higher values which is the point I was trying to get across.

[hacklordsniper]

Just a quick note. I bought few of these "resistor kits" and some of them had all resistors labeled wrong and i guess they were factory discarded  and resold by clever Chinese guy

[Kiriakos-GR]

I am aware that this thread are about cheap decade boxes.

But I have to add a link of my own decade box.
It is a good example  about making with your hands, one box that in the market retails at 1500$,
with the lowest possible cost.

The low in Watt decade boxes, they must be easy to repair too.
The low in watt resistors, they can easily get damaged for high current.

And so its not wise to follow dirt cheap solutions, but to play safe by making something better.
And even by looking at your home made box, you will have with it and all the memories by the experiences that you earned by making it, with your own hands.   

https://www.eevblog.com/forum/index.php?topic=1148.0

[IanB]

Percentage errors will roughly stay the same but when youre building a device thats supposed to have 1ohm resolution, youll never get near that kind of precision at the higher values which is the point I was trying to get across.

Actually you will have 1 ohm resolution, but you won't have 1 ohm accuracy. Although as I was saying, I can't think of a 100k resistor where 1 ohm accuracy is of practical use. (Maybe a high precision voltage divider or bridge circuit, but in that case a trim pot would likely do the job.)

[ejeffrey]

Actually you will have 1 ohm resolution, but you won't have 1 ohm accuracy. Although as I was saying, I can't think of a 100k resistor where 1 ohm accuracy is of practical use. (Maybe a high precision voltage divider or bridge circuit, but in that case a trim pot would likely do the job.)

Not quite.  Unless you take special precautions, 10,000 ohms will not be within 1 ohm of 9999 ohms, nor will it even be greater.  In the language of DACs, typical resistor decade boxes have poor DNL and are not monotonic.

[george graves]

Looks great!  Nice work.

Thanks for posting that up.  I just ordered 20 of the switches, and will make a box or two when I get the chance.  Seems like a really great budget/alternative.  I could never pull the trigger on the $100 worth of parts.  This is great.

[gregariz]

Looks great!  Nice work.

Thanks for posting that up.  I just ordered 20 of the switches, and will make a box or two when I get the chance.  Seems like a really great budget/alternative.  I could never pull the trigger on the $100 worth of parts.  This is great.

If you don't need so many resistor levels, or precision resistances.. another cheap option is

http://www.elexp.com/tst_subk.htm

[IanB]

Not quite.  Unless you take special precautions, 10,000 ohms will not be within 1 ohm of 9999 ohms, nor will it even be greater.  In the language of DACs, typical resistor decade boxes have poor DNL and are not monotonic.

Oh, you are correct of course, thanks for pointing that out. Oops.

(Goes off to think about putting 10,000 one ohm resistors in series...)

[Alex]

Relevant, if not mentioned already:

http://www.bourns.com/data/global/pdfs/3680.pdf

[Kiriakos-GR]

Tell us the price Son !!   

[Conrad Hoffman]

I use resistance boxes constantly and have acquired quite a few over the years. Switch costs are always the primary problem when you build one, followed of course by the resistors. General Radio Corp made their own switches and wound their own mica card resistors- very precise and handle quite a lot of power if you don't mind the inductance. For low values it's easy to wind your own resistors if you get a roll of constantan or other solderable resistance wire.

You might get some more ideas from some stuff I've written on KVDs and precision resistor dividers. Look at the Mini Metro Lab and the Hamon Divider. With nothing special you can do ratios at the 10-100 part per million level. Absolute resistance value is a bit more difficult because you need at least one known standard. I also like the BCD methods because it takes fewer parts for both an R box and a C box. I suppose nobody builds L boxes, but you certainly can.

[tomlut]

Very neat idea and good job--but I'm a bit confused about your accuracy estimations.

1% is 1%, whether it be 1% of 1 ohm or 1% of 100k. The circuit doesn't know the difference, and the 1% 100k resistor is no less accurate than the 1% 1 ohm resistor. (Consequently, a resistance value of 134,968 ohms, say, is rather meaningless and also not likely to be useful in any practical application.)

Also, the error doesn't "add up" when you put resistors in series. It remains exactly the same at 1% no matter how many resistors you combine. (The variance changes, but the mean error doesn't.) This means that if you put ten 1k, 1% resistors in series you will get a 10k, 1% resistor. Therefore, I would say that if you have an error of 2% on your 9 ohm value, it is either because the resistors themselves are out of spec or you have an inaccurate measurement. (At low resistance values you also have to worry about extra circuit resistances like switch contact resistances getting in there, but your measurements don't indicate that problem.)

Ah yeah thats right regarding the numbers, but what I meant was that the absolute values in ohms do add up. I didn't say 9 1% resistors in series = 9% tolerance.

Percentage errors will roughly stay the same but when youre building a device thats supposed to have 1ohm resolution, youll never get near that kind of precision at the higher values which is the point I was trying to get across.

I may be wrong but i think the total tolerance is the geometric mean of the individual tolerances.

So %Total = sqrt(%1^2 + %2^2 + %3^2 + ... %n^2)

Or if all the tolerances are the same:

%Total = sqrt(n.t^2) where

n = number of resistors
t = resistor tolerance

So for 10 x 1% resistors in series you are looking at a total tolerance of 3%.

[IanB]

I may be wrong but i think the total tolerance is the geometric mean of the individual tolerances.

So %Total = sqrt(%1^2 + %2^2 + %3^2 + ... %n^2)

Or if all the tolerances are the same:

%Total = sqrt(n.t^2) where

n = number of resistors
t = resistor tolerance

So for 10 x 1% resistors in series you are looking at a total tolerance of 3%.

Maybe this depends on the definition of tolerance?

My understanding of resistor tolerance is that it is the maximum allowed deviation from the nominal value. So a 100 ohm resistor at 1% tolerance is guaranteed to be in the range 99 to 101 ohms with a degree of statistical certainty. (I don't know what that degree is.)

I like to look at this by taking an example. Suppose we have 10x 100 ohm resistors that we are going to put in series to make 1k, and suppose each of these resistors is at the lower limit of tolerance at 99 ohms. We will therefore have a nominal value of 1000 ohms and an actual value of 990 ohms. The deviation between actual and nominal will be 10/1000 = 1%. In the worst case, the accuracy of 10 resistors in series is no worse than one resistor. However, if we suppose the individual resistors have a random deviation from the nominal value, then some negative and positive errors will cancel out when ten resistors are combined. If so, the likely error on the combination will be less than the error individually. I think statistically you end up with a smaller variance around the mean value when many resistors are combined in series compared to a single resistor of the same value. But this presumes an equal and random distribution around the nominal value of each resistor. If there is a systematic error in the individual resistor values, then this systematic error will be reinforced when resistors are placed in series and the mean value will assert itself as being different from the nominal value.

[johnboxall]

Congratulations - you made it on 'hackaday.com'
http://hackaday.com/2011/07/22/resistor-substitution-box/

[Wartex]

vtl, thanks for providing an eBay link so everybody else can buy those.

[johnboxall]

vtl, thanks for providing an eBay link so everybody else can buy those.

Try http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&item=350476494666

[vtl]

vtl, thanks for providing an eBay link so everybody else can buy those.

Yeah sorry about that, I'd normally put the link up but the links are generally dead within a month. I used the keywords "thumbwheel switch"

This is the exact auction and arrived in a week which is quite fast for something on ebay:

http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&item=220806571911&ssPageName=STRK:MEWNX:IT

The pcb traces are actually different in the images than the one I got, the switches I bought look the same as the ones that tronixstuff posted

[insurgent]

FWIW, Looking around it appears resistors in series you would use the weighted average:
(R1*T1 + R2*T2 + ... Rx*Tx) / R1 + R2 + ... Rx

For parallel:
((1/R1) + (1/R2) +... (1/Rx)) / ((1/R1*T1) + (1/R2*T2) +... (1/Rx*Tx))

[alm]

In other words, for identical values, the variance of the resistors adds up, but since the standard error of the mean is s/sqrt(n), this neatly cancels out. For resistors with the same tolerance, SEM_series=sqrt(n)sqrt(s²)/sqrt(n).

[mitpatterson]

hmm, the ones i ordered(think from the link tronix gave) and i got mine today, but they only have 5 contacts, it appears for like 3, it connects common to both 1 and 2, like 7 it connections common to 4 + 2 + 1. definitely odd/different

[Uncle Vernon]

i got mine today, but they only have 5 contacts, it appears for like 3, it connects common to both 1 and 2, like 7 it connections common to 4 + 2 + 1. definitely odd/different

That's a BCD coded switch, an entirely different animal. Check the specs of what you ordered, there is an error somewhere in the process if you ordered a decimal switch.

[tekfan]

Did you by any chance make this?

http://www.electricstuff.co.uk/capbox.html

[KJ6EAD]

Those switches appear to be from this Chinese company.
http://yqhrie.en.ec21.com/Coded_Switch--1512901_1518592.html

They look like a clone of the Omron A7 series switches.
http://www.ia.omron.com/product/42.html

You might be able to complete your switch assembly by using Omron end caps.

[metalphreak]

hmm, the ones i ordered(think from the link tronix gave) and i got mine today, but they only have 5 contacts, it appears for like 3, it connects common to both 1 and 2, like 7 it connections common to 4 + 2 + 1. definitely odd/different

I just received mine today, and they are the same They are binary encoded versions (BCD) instead of decimal. The description clearly says decimal, and the pictures all show decimal contacts. It's a shame because there's only the one person selling them on ebay. Don't be fooled by all the usernames... they're all UXCELL. I bought mine from 'milancentral' and it has a UXCELL coupon in the package xD

I've emailed to see if they can send the correct ones. Shall see what happens.

BTW, I found all the specifications here: http://www.easternradio.co.in/switches.html


FYI, the BCD ones are actually suitable for building a capacitance box

[Bored@Work]

FYI, the BCD ones are actually suitable for building a capacitance box

Or a constant current source http://www.edn.com/article/470718-Programmable_current_source_requires_no_power_supply.php

[alm]

Always try to verify important specs in the datasheet based on the part number. Vendors are sometimes lazy and copy the description from one member of the family to all, or copy the wrong specs. The part number is usually correct, or at least you have a good chance to receive a refund if it's wrong.

[amspire]

Love all the equations for the accuracy!

All inaccurate unfortunately.

If the decade box is made from 1 % resistors, then ignoring contact resistance, the accuracy of the decade box is 1%, the linearity is 2%.

The use of statistical tools are not really relevant, because 10 resistors made sequentially in a production run could all have similar errors and not random errors over the full +-1%, and more importantly, the resistance dialed up in the decade box will in many cases be a single resistor (1K, 10K, 100K, 1M, etc) and the accuracy of that single resistor is 1%, even if all the other resistors in the box had perfectly random error distributions. So you cannot quote less the 1% accuracy.

The accuracy cannot be over 1%.  Say every resistor was exactly 1% high - the worst case situation. This will give the maximum possible error.

If you dialed up 99,999,999 ohms, the what would the error be?

1% high.

So the accuracy cannot be over 1% and it cannot be under 1%.

[IanB]

If the decade box is made from 1 % resistors, then ignoring contact resistance, the accuracy of the decade box is 1%, the linearity is 2%.

As other people have pointed out, that is not the whole story.

Suppose you dial in 999 ohms, which is made up from 9 x 100 + 9 x 10 + 9 x 1, and suppose this has the maximum positive error of 1%, then the  actual resistance will be 1009 ohms. Suppose you now increase the resistance setting by 1 ohm to 1000 ohms, only this is now made up from 1 x 1000 and it has the maximum negative error of -1%. The new actual resistance will be 990 ohms. Increasing the nominal resistance by 1 ohm caused a decrease in actual resistance of 19 ohms, a relative error of -1900%!

[gregariz]

Love all the equations for the accuracy!

... I think I could count on one hand the number of times in the last 15 years I needed a precision resistor box. 5% resistors will be fine thanks.

[ejeffrey]

If the decade box is made from 1 % resistors, then ignoring contact resistance, the accuracy of the decade box is 1%, the linearity is 2%.

As other people have pointed out, that is not the whole story.

Suppose you dial in 999 ohms, which is made up from 9 x 100 + 9 x 10 + 9 x 1, and suppose this has the maximum positive error of 1%, then the  actual resistance will be 1009 ohms. Suppose you now increase the resistance setting by 1 ohm to 1000 ohms, only this is now made up from 1 x 1000 and it has the maximum negative error of -1%. The new actual resistance will be 990 ohms. Increasing the nominal resistance by 1 ohm caused a decrease in actual resistance of 19 ohms, a relative error of -1900%!

This is what is meant by 2% (differential) nonlinearity.

[JuKu]

Talking about accuracy: Do you know do they make 0.5% or 0.1% resistors? They don't, they just make resistors and measure what they got. In other words, if a factory makes 5%, 1% and 0.5% resistors, then 5% parts are pretty much guaranteed to be at least 1% off. Likewise, in this case, 1% resistors will have 0.5% to 1% error. As said, a 1% resistor box is pretty handy, and if you ever need better than 1% in a decade box, just find a trim pot:-).

[alm]

Talking about accuracy: Do you know do they make 0.5% or 0.1% resistors? They don't, they just make resistors and measure what they got. In other words, if a factory makes 5%, 1% and 0.5% resistors, then 5% parts are pretty much guaranteed to be at least 1% off. Likewise, in this case, 1% resistors will have 0.5% to 1% error.

Funny, since the majority of through-hole 5% resistors are carbon film, and the majority of 1% resistors are metal film. I'm sure selection is taking place, but the production processes have improved enough that just producing resistors with 1% tolerance has a good yield. More accurate resistors often have a better tempco and sometimes different materials.

As said, a 1% resistor box is pretty handy, and if you ever need better than 1% in a decade box, just find a trim pot:-).

How's the stability, settability and tempco of that trim pot? One thing that does work is put a small value trimpot in series with a resistor that's pretty close (or a large value one in parallel). Because of the small contribution to the equivalent resistance, the error also has less of an effect.

[metalphreak]

I measured about 25 or so 5% SMD resistors and there were plenty that fell within the 1% range. The really high accuracy ones are normally produced the same as 1% resistors, but they individually test and laser trim them to the correct value. Higher accuracy ones are usually made from different materials that have less variability due to temperature (metal vs carbon film for example).

[Vertigo]

couldn't one simply measure all resistors one by one and cherry pick the accurate ones?

[ejeffrey]

couldn't one simply measure all resistors one by one and cherry pick the accurate ones?

Yes, assuming you have a sufficiently good meter -- which just translates to a set of known good resistors to compare against: ohmmeters work by comparing the DUT to a set of precision resistors.

However, it isn't as good as you would like.  Resistors change with time, temperature, applied voltage, and assembly technique: the heat from your soldering iron or stress on the leads can change the resistance permanently.  If you select resistors then solder them into a circuit, they may no longer be matched.

Temperature coefficient of resistance is also a big deal.  Lets say you select some 1% metal film resistors for .1% accuracy.  They still have a roughly 100 ppm/K temperature coefficient, and will change by 0.1% with a 10 degree temperature rise.  No problem you think: I will only use my decade box indoors! I won't seem more than +/-2 C.  Well, keep in mind that resistor power ratings are based on heating to some maximum temperature.  Lets say you have a 1/4 watt resistor with a maximum operating temperature of 125 C -- about 100 degree temperature rise over room temperature.  25 milliwatts of power will be enough to raise the temperature 25 degrees and change the resistance by .1%.

High precision resistors are usually made from more stable materials, built so that the body protects the resistive element from strain during assembly, and thermally oversized to keep the temperature rise to a minimum under operating conditions.

[Guppzor]

Just would like to confirm that the switches listed on eBay here do indeed have 10 contacts (0-9) plus Common. I checked with the seller prior to purchasing, and the units that arrived today are as per the image. Now to get a bunch of resistors...   

[kripton2035]

Just would like to confirm that the switches listed on eBay here do indeed have 10 contacts (0-9) plus Common. I checked with the seller prior to purchasing, and the units that arrived today are as per the image. Now to get a bunch of resistors...   

it's ok : it says
Digital   0,1,2,3,4,5,6,7,8,9
Code Method   Decimal

[IanJ]

Hi all,

Found this thread AFTER I'd started building my own box.........and using the same switches via Ebay........

I'm not so much interested in accuracy but the step change resistance......so at the bottom end are 0.1 ohm, and the top end 10M ohm resistors. 9 switches in total.

As VTL noted, the clips are pretty hopeless so I'd elected to use some thin Alu tape.

Testing done, just got to box it up now.

Ian.

[amspire]

If anyone comes up with a good way of mounting these switches without the endplates, please let us know.

Other then gluing, of course - the last resort of the desperate.

I have not found anyone selling the endplates yet. If the true manufacturer could be located, it would help.

Richard

[IanJ]

Is this them?

http://www.tootoo.com/d-cu14047240-Yueqing_Huari_Instrument_Electron_Co_Ltd_/

Ian.

[kripton2035]

I ordered my switchs (on ebay) also before reading this thread...
they arrived this week and I use 0805 resistors, they fit perfectly between the tracks...
works nice, but I have to put them in a box like Ianp, must look for a box that fits the decade box and 2 banana plugs

I used standard resistors, and the precision is not here... but enought to search for a precise resistor in a filter for example...

[amspire]

Is this them?

http://www.tootoo.com/d-cu14047240-Yueqing_Huari_Instrument_Electron_Co_Ltd_/

Ian.

Ian thanks. They don't seem to have a website, but I tracked this down:

http://shlemore.en.alibaba.com/product/215208027-200521245/KM1_2_3_push_wheel_switch_22x_8.html

They are HKM2 decade switches.

Richard

[IanJ]

Hi,

Here's mine finished, just awaiting a label or two (haven't thought about that yet).

The switches are a tight fit, and with the silver tape the switches aren't going anywhere.

Ian.

[don.r]

I always advise builders of resistance boxes to make the lower ranges 1W or more as they are likely to be used in higher power applications. You also need to look at the switch's ratings. Most can only take 1/8 amp. Also be aware of contact resistance and how it affects the lowest range. This is around 0.1 ohm per switch so a tap at say around the deka-decade may make sense if you want some accuracy in the 0.1 ohm range. This would be about 1% so about the same as the resistors' tolerance.

[Conrad Hoffman]

Good advice on the power ratings; it's amazing how many used resistance boxes turn out to have burned parts when opened up.

[george graves]

Here's my take on them...

You can see the full tutorial here - with dozen of easy to follow pics for anyone not too sure how to wire things up.

http://www.digitalunderpants.com/?p=219

[mianchen]

Good tutorial

[Uncle Vernon]

Here's my take on them...

Nice work George. A good beginner tutorial.

[george graves]

Thanks - that's kinda what I was going for!

[Johenix]

First off, the lower resistance decades should be higher wattage resistors. (One hundred volts across one ohm would be one hundred amps or ten kilowatts, in a really ridiculous case.)

That said, let us get to the real heart of the matter.

So called "One percent resistors" are never accurate to 1%, they are just called that. They are made on 96 values in each step between 1, 10, 100, 1000, 10 000, 100 000, 1 000 000, and 10 000 000 ohms. They are spaced by the 96th root of 10, a sort of logarythmic progression, in a system that allows voltage divider ratios to be scaled up or down for more or less current.

Oh, by the way, "20% resistors" are seperated by the 6th root of 10, "10% resistors" are seperated by the 12th root of 10, "5% resistors" are seperated by the 24th root of 10, and "2% resisrors" are seperated by the 48th root of 10. 

[mzzj]

First off, the lower resistance decades should be higher wattage resistors. (One hundred volts across one ohm would be one hundred amps or ten kilowatts, in a really ridiculous case.)

That said, let us get to the real heart of the matter.

So called "One percent resistors" are never accurate to 1%, they are just called that. They are made on 96 values in each step between 1, 10, 100, 1000, 10 000, 100 000, 1 000 000, and 10 000 000 ohms. They are spaced by the 96th root of 10, a sort of logarythmic progression, in a system that allows voltage divider ratios to be scaled up or down for more or less current.

Oh, by the way, "20% resistors" are seperated by the 6th root of 10, "10% resistors" are seperated by the 12th root of 10, "5% resistors" are seperated by the 24th root of 10, and "2% resisrors" are seperated by the 48th root of 10.

What is this nonsense?

[amspire]

Johenix confused the explanation of the standard resistance ranges and the tolerance. They are two different things.

Just to be clear, a 1% resistor is definitely better then 1% accurate. Typically most of the resistors will be within 0.5% accuracy.

Accuracy doesn't always take account of long term drift, so you can have a 1% resistor that is within 1% at the time of manufacture, but after 2000 hours of use, it may be no longer 1% accurate. So it is important to look at the long term stability spec as well as the accuracy spec.

It is true that there is not much point making the full E96 range of 20% resistors but the selection of the resistance ranges is a choice for the manufacturer, and is not dictated by the tolerance.

For example, if you look at the very expensive, very accurate resistors (0.005% for example), they are often made with a relatively few standard values. The reason is that the customers often need very precise values (like 3,459.3 K) and so the customers often order custom the exact value they require rather then values from a standard range.

Johenix is correct in that the choice of standard values of the E12, E24, E96, etc ranges are approximately logarithmically spaced, but if you do the calculations, you see that the values are often a bit out - probably because someone thought that the calculated value wasn't as useful as a nearby value.

Theoretically, the E12 range should be 1, 1.2, 1.5, 1.8, 2.2, 2.6, 3.2, 3.8, 4.6, 5.6, 6.8, and 8.3.

However the actual values are 1, 1.2, 1.5, 1.8, 2.2, 2.7, 3.3, 3.9, 4.7, 5.6, 6.8, and 8.2

Richard

[babysitter]

About 1 year ago, I decided to make my own R substitution box which was basically a VTL clone, but using SMD resistors in the middle range soldered to the switches. It hasn't disappointed me yet. After my company is having a trainee who wants to go into electronics, I decided that a couple of such boxes are one of his first projects.

I should show off mine asap, but it is waiting at my desk at work to be the role model for this guys project

[kripton2035]

so should look like this one ?

[babysitter]

@kripton: very close to that, but the 1R, 10R, (100R IIRC?) and 1M as thruhole also.
The small values because of wattage, the 1M because available.

BR
Hendrik

[gerrysweeney]

If anyone is interested, I came up with an alternative low cost solution. Not perfect for everyone by any means but works for me.

http://gerrysweeney.com/seven-decade-programmable-resistor-a-low-cost-solution/

Gerry

[george graves]

First off, the lower resistance decades should be higher wattage resistors. (One hundred volts across one ohm would be one hundred amps or ten kilowatts, in a really ridiculous case.)

What kind of mad man would pump 100 volts into a substitution box - at any setting, let alone 1 ohm!?!?!?  That's crazy talk.

Anyways.  If you're interested in a pre-built unit - I have a few for sale - but the labor is a bit too much, so I can't keep them in stock for what little I'm charging:  https://www.tindie.com/products/redrocketlabs/resistance-decade-box-resistor-substitution-box/

[curiouscane]

Hello Everyone Out there,

'I too would like to build a cheap Decade Resistor Box ranging from 0.1 Ohm to 100kOhm . So i intended to purchase a "KSA-2 Decimal 10 Position Single Unit Thumbwheel Switch" but what was delivered to me was a "KM-1, 0-9 Pushwheel Thumbwheel Switch " from Ebay last week. Now i don't know how to exactly connect using a C,1,2,4,8 (pls refer the pic attached) !!! Is there any guidance you can provide on the connecting the resistors ? 

[Bored@Work]

There is no simple way to use BCD switches to build a decade resistor box. Use the switches to build a decade capacitor box http://www.electricstuff.co.uk/capbox.html or a simple constant current source / sink http://www.edn.com/design/power-management/4325377/Programmable-current-source-requires-no-power-supply

[nachbelichtet]

I've created an enclosure for 3D printing. You can grab it at Thingiverse:

http://www.thingiverse.com/thing:1079283

Thank you for the great ideal. It's really helpful!

Greetings
Markus

[timofonic]

Sorry for the necroposting

I have a ten or more of those switches from an going-to-be-repurposed CDROM SCSI box. I wanted to use them to make a resistor decade substitution box.

The big problem is that the buttons are very small, so I'm not sure if it's feasible to enlarge them or better to buy others,

Any ideas?

[kripton2035]

those buttons are so cheap from ebay that I doubt it will be interessant to try to modify yours...

[Old Grey]

The 1W is as tight as a fish's ahole.
I think I got Chinese-ed. The 1% resistors don't seam to measure 1%, but that seams to be only in the lower values.

Ah close enough, and anyhow I'll measure it and compensate with the actual value.

[Blastcap]

Here is another similar thread, for a SMD version.
https://www.eevblog.com/forum/projects/yet-another-resistance-decade-box/

[Bzzz]

Hey guys, I just published an article with a lot of BLAH about these dials in my blog, including resistance measurements of my 7-decade box. That plot is also attached for general convenience, and to spoil the results:
a) Each switch has about 1/8 ohms contact resistance, which adds roughly 1 ohm to my box -> deadly for the precision of the 1-9 ohm dial if not corrected
b) Once corrected in the first three decades, I'm able to achieve 10000 ppm over the entire range, which is the 0805 resistor spec of 1%. For values greater than 10 ohm, most of them are closer to 1000 ppm = 0.1%. And I have some lucky shots that are below 100 ppm (remember the temp co...!).
Stuff was measured with a Keithley 2700 DMM in 4-wire mode, 100 readings within 0.01% window and then rolling avg until the reading gets stable (don't breathe or move)

[advancednewbie]

I made these resistor adapter PCBs available on Tindie:  https://www.tindie.com/products/13690/

Makes it way easier to solder up all the resistors, and you can use higher wattage resistors as well.