OT-ish: resistor value solver

[pete]

Does anyone know of an online utility that can calculate what combination of series and parallel resistors are needed to get a particular value? Specifically, I'm trying to calculate the best way to get close to 5250 Ohm, using E12 preferred values. Power consumption is not an issue and I'd like the value to be +/- 2% as that's the resistor tolerance.

I'm not looking for the answer, I'm looking for the way to find the answer. There are lots of websites where you can tap in resistor values and have it calculate the result, but that gets long winded. I've got a combination that gives

5253R with 4 resistors, but I'd like to do better

[Andrew Mawson]

..he..he - showing my age, but in the era of solid carbon resistors it was very simple. Choose a value slightly lower than you want, then with the Avo across it on Ohms file a nick in the carbon until it was spot on. Excellent for making meter shunts and the like.

AWEM

[Guy Dawson]

Given that 2% of 5250 is 105 I'd be happy with 5253!

Guy

-- -------------------------------------------------------------------- Guy Dawson I.T. Manager Crossflight Ltd [e-mail]

[Dave Baker]

I'm puzzled. Knowing nothing about electrickery I've Googled resistors, found out what E12 means, got a table of what values are possible which appear to be 10, 12, 15, 18, 22, 27, 33, 39, 47, 56, 68, 82 and so on in further powers of ten.

So what's wrong with the following four in series, 4700 + 470 + 68 +12 =

5250 or am I missing something obvious?

[Gordon Henderson]

Missing the "Countdown" music in the background ;-)

It's the way I'd do it - you subtract the biggest and so on, however the tolerances are cumulative - so if you used 1% resistors, you might be fine, but lower tolerance resistors and it might be out by too much...

Have to say, it's been many years since I tinkered in electronics, but it was rare to actually require something that precise (test system exepted)

- so a 4k7 or 5k6 would be tried...

Gordon

[Dave Plowman (News)]

Have you Googled for one? 'resistor calculator prog' seems to give plenty hits.

The one I use - on this Acorn - gives 82k and 5k6 in parallel at a 0.15% error. With just two in series the best it can achieve is 0.95% (1k3 and

3k9)

[Dave A]

Surely if you used, say, 2% resistors the maximum you could be out in total is 2% (high if they're all high or low if they're all low). In all probability some will be high and some low so the total error will be less than 2%.

Which brings me to the method where you find a nominal value of resistor (or a couple combined) close to what you want and measure a few until you find one where the actual value is near enough spot on.

[fred]

In article , Dave Plowman (News) writes

The general rule for parallel friggery being: Start with the standard value above the desired value and add a parallel resistor from around a decade (x10) up, 56000 gives 5091, 68000 gives

5174 and as DaveP says, 82000 gives the closest at 5242.

[Phil Addison]

Use a spreadsheet.

Put the E12 values in a row at the top, and again in a column. At each intersection of row/column put the formula for the combination you want to try, i.e. parallel Rp = 1/(1/Rr + 1/Rc), or series Rs = Rr + Rc.

Pick the best result by eye.

OR

Take any 4k7 out the box. Measure it. Add approriate series resistor/s. In the trade, this series resistor is labelled SOT (Select On Test).

Phil

[pete]

Yes, there are lots of downloadable ones (that I presume do what I want). However I'm looking for an online utility - one where I can go to a web page that asks: what value do you want to find? what family of values (E12, E24 ...) to use? how close do you want to be

and then it tells you that for a value of X, you need an A and a B in parallel and a C in series - or whatever the answer might be. The ones that google spits back merely calculate the result from values you type in. I've (easily) got this solution, but I'm looking for a more general solution in the future. Although, in practice the answer is to choose the next lowest value and add a preset (multi-turn for extra accuracy) and tune for maximum smoke.

[Al]

Phew! It's not just me then ;-)

Al.

[The Natural Philosopher]

theers a 5.1k resistor in E20 range.

Add a 150 and that's 5250 Add a 3 ohm and that's 5253

[Al]

Not an answer to your question, but sticking a 4k7 resitor in series with a

1k pot would be my solution.

Oh, and yes, programs to do exactly what you want have been around for years. I wrote one. In Fortran. In 1979. I'm pretty sure I have the punched tape somewhere around here ... ;-)

These days I'd just muck about in Excel and use goal seeking or whatever to get the answer.

Al.

[john]

There's lots of R calculators on the web but most just give one answer, when what most people want, is a list of toleranced result options, to pick and choose from. This prog is OK but has not a specific "within 2%" sort. (the best pair comes to 0.1% anyway!)

[Tim S]

Gordon Henderson coughed up some electrons that declared:

I thought the E12 scale was specifically developed so you could get close enough (TM) to almost anything you practically want with just two resistors - series or parallel.

Or is that another factoid in my myth database?

[Andy Champ]

Each one is 20% more (roughly) than the previous one.

Andy

[Dave Plowman (News)]

Right - sort of misread the question.

[NT]

People have suggested a main resistor plus a parallel tweaker resistor, which is a fine way to go. A close main R plus a relatively high value parallel tweaker can often get you there with just 2 Rs.

But one error to avoid is to calculate your tweaker r value using the nominal value of the main R. Instead use the real value of the main R to work out what tweaker R you need, and the result will be far more precise.

For some apps you can use basic 5% Rs to get 1% accuracy, but for some you cant. Wider tolerance Rs are so for 2 reasons. First is simply selection, its cheaper to stick them in the E12 range than E24. The other one is tempco, Rs do change value with temp, so if you keep your R temp steady you can use cheap Rs for quite good accuracy.

PS one of the key ideas of the E ranges is that every real value of R produced can be placed in a nominal value band and be sold.

NT

[<me9]

Or 4700+560 would give 5260, well within 2%.

[Chris J Dixon]

I have to say that when I first discovered this fact, I was struck by the sheer cunning of the arrangement.

Chris