# Exact resistance output?

1. ## Exact resistance output?

Hello everyone,
I am facing a problem on creating some resistance. I want to create 100 ohm, 150 ohm, 220 ohm, 470 ohm, 680 ohm and 1.5 kohm using a digital potentiometer. The value will change with the data give to the digital potentiometer. The requirement is to provide this resistance value to two pin output. The Digital potentiometer which is available here is 10 kohm, and 1 kohm will available after few weeks. I am unable create exact values of resistance using this 10 kohm potentiometer.
IC I am using now = AD5254BRUZ1

I have tried to make a parallel connection of these potentiometer like the following image:

Still after the calculation of resistance as per the datasheet I am leaving with some error of 2%-3%.

Is there any better solution to this problem, to get exact value of resistance in two pins. Like multiplexing of resistance or something else?

Thank you.

•

2. ## Re: Exact resistance output?

Hi,

Using a 10k pot for a 100 Ohms to 1500 Ohms (1400 ohms)range just uses 14% of it's range.
Try to find a pot with more suitable range.

Or use some parallel resistors to use almost full pot range.

Still after the calculation of resistance as per the datasheet I am leaving with some error of 2%-3%.
% is a relative value. You need to say how much % "from what".
2% from pot FS is 2% × 10k = 200 Ohms
2% from output FS = 2% × 1500Ohms = 30 Ohms
Or do you talk about 2% deviation from the expected value which then is 2Ohms .. 30Ohms.

And complaining about 2% means nothing when you don's say where your target is.

Klaus

3. ## Re: Exact resistance output?

If the resistance desired is "end to end" and the e-pot likewise
(that is, not a shunt-to-ground or something like that) you could
go for a higher bit count e-pot, or you could cascade two very
different value e-pots or parallel them (depending on what's
available; suspect that very low values are not realizable due to
switch-FET size requirements) - a 500Kohm 8-bit shunt to a 2Kohm
8-bit "master" could take you from 8 bits all the way to 16-bit resolution
(maybe, or not, as monotonic as you'd like; fix that in software
perhaps?).

4. ## Re: Exact resistance output?

you have a 10k pot with 256 steps, one of which is 0 and the other is 10k
the step is approximately 39 ohms (10,000/256)

the available values closest to your desired resistances:
you can have 78 or 117, not 100
you can have 117 or 156, not 150
you can have 195 or 234, not 220
you can have 469 or 508, not 470
you can have 664 or 703, not 680
you can have 1484 or 1523, not 1500

unlike a real potentiometer, the digital pot takes digital steps
they may not all be exactly the same size, but it is still a step response
not a rap response

even with a real analog pot, it will not be easy to stop turning at exactly the correct spot,
and there will be backlash

maybe you want to use a digitally controlled analog switch
like TS3A24159DGSR from texas instruments
or 74HC4052 from nexperia

there are lots of others

•

5. ## Re: Exact resistance output?

Hi,

maybe you want to use a digitally controlled analog switch
I agree.
Especially when you want only a couple of fixed, precise values then a MUX and fix resistors is the better solution.

Klaus

6. ## Re: Exact resistance output?

Thank you so much for your reply. I do not know where this will be used, but I was also thinking the same way.
okay, then I have to select exact value resistance and 8:1 MUX to complete this job.

I have a query regarding the type of resistor, a through-hole or surface mount, is there any problem while selecting the type? Will this affect the circuit? Because right I can go with Through-hole, but SMD for later.

Thank you.

7. ## Re: Exact resistance output?

analog MUX, not digital MUX
pay attention to the resistance of the switch itself

through hole or surface mount does not matter, provided the components do not overheat
and it will fit in the desired box

tolerance and temperature coefficient are the primary things to pay attention to

you keep saying "exact" resistance
there is always a tolerance,
there is always a temperature coefficient
and there is a very minor resistance variation with voltage that is usually ignored because it is so small

you will have variations in resistance over manufacture, time, temperature and voltage
you need to design the circuit so these do not matter

1 members found this post helpful.

8. ## Re: Exact resistance output?

Your MUX will also have some resistance. And a tolerance.

9. ## Re: Exact resistance output?

You could have a switched array of resistors, 1,2,4,8,16,32,64,128,256,512,1048 ohms

form a simple digital drive and use low Rds-ON mosfets, you can then get a precision and accuracy of 1 ohm, assuming the resistors are 0.1% ...

e.g. 680 = 512 + 128 + 32 + 8

If you really only want: 100 ohm, 150 ohm, 220 ohm, 470 ohm, 680 ohm and 1.5 k

you can use those exact values with mosfets to connect them to gnd as appropriate ... 6 way mux.

•

10. ## Re: Exact resistance output?

I have ordered all the resistors and an analog MUX (MAX4781). This analog multiplexer is having a very low ON resistance of 0.7 ohm. I will tell you whether it works perfectly or not, It will take 9-10 days to meet those ICs.
Thank you all for your suggestion..

you can use those exact values with mosfets to connect them to gnd as appropriate ... 6 way mux.
Pardon me. Please can you refer me a circuit and some more light to how to achieve what you are suggesting. I want to give it a try.(with mosfet)

--[[ ]]--