potentiometer resistance selection

[RyreInc]

We will be using a linear potentiometer in an industrial application. The product this goes into may be used in a range of conditions, and we need to try to accommodate noisy environments.

The pot will be supplied with a 10VDC source and output to a (typically) 1MΩ analog input.

We plan on having a 6" unshielded lead with twisted wires, which will terminate to an M8 connector. From there a shielded cable will be used for the remainder of the length (up to maybe 3-5m long).

The resistance options we have available are:
1k
2.5k
5k
10k
20k
50k

Which of these would you choose, and why?
Is there other noise-mitigating techniques we should employ?

EM interference and how that relates to impedance selection is definitely a blind spot for me!

 

[RyreInc]

One other relevant piece of info: The response time of the analog input can be relatively slow: a digital low pass filter with a knee around 50-100 Hz, or possibly less, would be just fine.

 

[IRstuff]

But, you've not really anything that describes how noisy you have to tolerate, nor what kind of voltage error you can tolerate nor how it's supposed to couple in

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[VEBill]

Gut reaction = 10k. It's a very common value for such applications, based on my experience.

It's not such a critical value that you can't stick to 'even' decade values: 1, 10, .... Having an 'even' decade value makes the voltage division ratio obvious, especially with 10 volts input. Someday it'll save a minute when troubleshooting. A trivial detail, but why not?

1k means 10ma and 100 mW dissipation. That's low power, but could be lower.

So 10k seems about right.

1M input is fairly high, so the non-linearity introduced by the wiper load is negligible. If the input Z was lower, then you'd have to double check the linearity. Might be worth calculating it anyway, and comparing it to the pot's linearity and your requirements.

Since it's a pot, it's presumably not moving very quickly. You could put a capacitor from the wiper to ground to reduce EMI noise before the noise confuses your digital filtering. Perhaps a series L too, if the environment is RF noisy.

Critical applications (automobile throttles) may use two pots (mechanically linked, measured separately) to provide failure detection. That's something worth considering carefully.

In formal design processes, one could end up with a 5 or 10 page Design File Memo to document this sort of small decision.

 

[RyreInc]

Good points IRStuff.

I do not know how noisy the environment will be. This will be used in a product we provide to customers who will use it in a range of industrial machinery in manufacturing environments. Typical sectors are automotive, medical, and consumer products.

The voltage tolerance must be 100 mV or less, and preferably 20 mV or less.

Thanks for your response VE1BLL, some good considerations there.

 

[VEBill]

"...voltage tolerance..."

The pot's linearity (resistance vice position) might be the deciding factor for that.

20mV on 10V is very tight.

If that's what you meant.

--

If you're 'buying in' the potentiometer and assembling it into the assembly, then have a conversation with the application engineer at the pot vendor. Typically they'll be a vast store of useful knowledge and advice. They might be aware of some unexpected detail about advantages and disadvantages of various part numbers.

 

[RyreInc]

The linearity is indeed worse than that (±5%), which affects accuracy, but that's not very important. As long as the precision is there we're in good shape.

Our design engineer has had some higher-level talks with the manufacturer but I think you're right that it's time for me to get involved and dive into these details.

 

[daveross100]

It is not clear what is the level/frequencies of the noise
BUT the RIGHT VALUE SHALL be less than 1 Kilo Ohms

if 10 kilo Ohms is a common value so what???
For the reason go to : youtube.com and search for :daveross problem solver
GOOD THINKING=TO RESULTS

 

[IRstuff]

Realistically, it would seem to me that what you are asking is not achievable with a conventional single-turn pot. Assuming you get 330 deg of rotation. 20mV translates to about 1/3 of a degree. Even with a large knob, say 2-in diam, that's a positioning tolerance of 12 mils, which is barely supportable with my engineering ruler for measuring something. I doubt that I could draw a tick mark that repeatably, even with a 0.3 mm pencil without a magnifying glass.

A 10-turn pot might be more practical.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[RyreInc]

daveross, care to elaborate? Why should the value be 1kohm or less?

IRstuff, this is a linear pot, not a rotary. And 20mV is the ideal number, 100mV the more realistic one.

 

[IRstuff]

Even so, say the slide is 2 inches long, 100mV/10V = 1/100, so that would be 20 mils, which is about 5 sheets of paper, which is just barely possible to repeatedly position by hand.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm