So for a 6838 Ω resistor you are planning to use 6980 Ω, 634.o kΩ, and 715 kΩ all in parallel. I agree that this will give you nominally 6837.96 Ω. But the question is whether there is any point. I think you said you were going to use 0.1% resistors. That means that the tolerance on the 6838 Ω resistor is 6.8 Ω all by itself (which is going to dominate the tolerance range). If all of the resistors are at the lower end of their limits then the resistance could be as low as 6831 Ω while if they are all at the upper end of their limits, then the resistance could be as high as 6844 Ω, giving a tolerance of right at 6.5 Ω or just under 0.1% (not too surprising).

What if you just put three 20.5 kΩ resistors in parallel? That would give you a nominal resistance of 6833 Ω with a tolerance of that same 6.5 Ω and at least you could buy all the same values. The value you want is within the tolerance range of this combination, so is there any real benefit to trying to get the ideal value any closer?

I would say the tolerance is secondary to the stability. I would buy 10-20 of ea value and they can be sorted to get closer match, and if they are stable used in an air conditioned setting they should exceed other variables of the meter without requiring rescaling. I saw that Vishay makes an ultraprecision bulk metal trimpot also but at $66ea I'm not planning on that soon.

 

How are you going to sort them? Do you have a meter that has an accuracy of better than 0.01%?

 

How are you going to sort them? Do you have a meter that has an accuracy of better than 0.01%?

An HP3456 it's a beast but was too good a deal to pass up.

 

OFF-TOPIC

I just realized that the link (Post #8) I gave to TI for a parallel resistance calculator works, but the application doesn't. Notified TI support and received a message that they had confirmed it was not working and are trying to fix it.

Unlike other parallel resistor calculators where you must enter values, usually by trial and error, and the result is calculated, the TI application lets you enter the desired result, and it calculates suggested resistors to use from the series you pick. I am not aware of any other on-line calculator that does that.

John

 

Hello,

A simple way of calculating a parallel is as followed:

Rparallel = 1/((1/Rwanted)-(1/Rnext higher value))
In example the 6838 , the next higher value is 7150 so Rparallel = 1/((1/6838)-(1/7150)) = 154286
When you calculate back the values of 7150 and 154000 it will give 6832.76

Bertus

 

When you are measuring down to the nearest ohm you have test lead resistance. Do you have access to a 4 wire ohm meter?

 

When you are measuring down to the nearest ohm you have test lead resistance. Do you have access to a 4 wire ohm meter?

Yes (HP3456A)