0
\$\begingroup\$

I have quite a few resistors but occasionally I run into a situation where I need some specific value I don't have. in those cases I just select a a starting point (more or less at random) and then add resistors in series or parallel to get the desired value (I'm just calculating of course) but if I need some really specific value this stuff gets quite messy pretty quickly and I end up with either lots of resistors or having spent way too much time essentially just guessing (with a little bit of common sense) which combinations work the best.

So I though to myself: Why not automate the process? This, as far as I can tell, is really problematic though: The number of options here is scaling exponentially and a quick and dirty program I wrote (even utilizing CUDA) takes ages to brute force all options. Brute forcing is the only way I can think of.

So here's my question: Is there any "premade" solution for this? I'd imagine this can't be that unique of a use case.

I hope this is the right place to ask this question. I'll also post a question regarding the computational aspect of the problem on Stackoverflow assuming no one has any ideas in terms of a premade solution.

\$\endgroup\$
9
  • \$\begingroup\$ jansson.us/resistors.html \$\endgroup\$ Commented Aug 21, 2024 at 18:26
  • \$\begingroup\$ Thanks @G36! This website doesn't do exactly what what I was looking for but it's definitely pretty good and a lot better than nothing! \$\endgroup\$ Commented Aug 21, 2024 at 18:31
  • \$\begingroup\$ Maybe the problem can be approached differently. For example, for what purpose do you need a specific resistance, and why for example a 10% tolerance is not suitable? Also if you need a specific resistor for a specific purpose, you maybe should buy it, for example if a a specific chip needs a specific reference resistance, you bought the specific chip so you know you need that specific resistance too. \$\endgroup\$ Commented Aug 21, 2024 at 18:32
  • \$\begingroup\$ This question Does anyone remember this article about the Euclidean Algorithm? covers the case of building an arbitrary value from a network of resistors having a single value. It can be extended to cover having multiple available values. \$\endgroup\$ Commented Aug 21, 2024 at 18:33
  • \$\begingroup\$ But keep in mind that despite having a resistance that has the exact nominal value that you want, its accuracy won't in general be any better than that of the resistors you use. If you use ±5% parts, your composite resistor will also have an accuracy of about ±5%. \$\endgroup\$ Commented Aug 21, 2024 at 18:45

2 Answers 2

Reset to default
1
\$\begingroup\$

This website calculates the optimum series or parallel combination of two resistors to get near the desired value.

That combined value then has the tolerance of the resistors forming the combination.

\$\endgroup\$
1
  • \$\begingroup\$ And (off topic) if you want to calculate the best resistor values for the divider used with and LDO or DC-DC to set the output voltage, this site can help: syncbuck.com \$\endgroup\$ Commented Aug 21, 2024 at 19:49
0
\$\begingroup\$

I often series/parallel resistors when I don't have the right one to hand, or want to trim one or more to an exact value or match.

I try to keep the process manageable by starting with a resistor that's the nearest size above. If it's 10% above, then that means a need a roughly 10x value that resistor in parallel with it. Note that if this extra resistor is 10% out, it contributes only 1% error to the final parallel set. It may be worth making this extra resistor with two series resistors.

Depending on the tolerance I'm going for, one iteration may be enough. If not, I stop slightly above where I want, and now the third resistor may be 100x my target, with an even smaller contribution to the error.

Three resistors is usually enough. Even if I try to trim to 0.01% tolerance, the tempco from ordinary grade resistors will make that pointless.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.