EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: smoothVTer on May 27, 2020, 09:16:49 pm
-
Given that the cheapest pots are 20% tolerance, what are techniques used in industry to design around this wide range of potential resistances?
I ran across this problem designing with a LED driver that takes an analog level of 0.140V - 2.25V to control the LED brightness. Working with a voltage reference of 3V and a 10k pot, I have to add a top resistor = 25% of the span and a bottom resistor = 4.67% of the span. The circuit is shown below with the pot shown as 2 separate resistances split at the wiper:
(https://photos.smugmug.com/Pics-for-links/i-TVWHKVz/0/e51de203/L/pot1-L.png)
...for the pot wiper at the left extreme...
(https://photos.smugmug.com/Pics-for-links/i-6j6HsDh/0/60c64d54/L/pot2-L.png)
...for the pot wiper at the right extreme...
This is all fine and dandy if the pot is exactly 10k, but I've got pots of the same part number that range from 8.56k - 11.2k. This would throw off the output range of the pot from designed conditions. I need to use all of these pots that can vary by +/- 2k in the same circuit without changing part values.
If I were strictly in the digital domain, I would put a constant current through the string and read 3 voltages with a multiplexing ADC, subtracting out the errors and relying on a ratiometric technique to obtain the final value. However I don't have the capability to add digital functionality at this time ( there is no uC on the PCB )
Is there any way to improve the circuit's immunity to tolerance variations of the pot using only passives?
Can the addition of an op amp somewhere help the situation?
Is there another circuit that would be of better use in this situation?
-
If you need accuracy for something, you don't use pots. It is that simple.
Applications where classic mechanical potentiometers are still used usually don't care about a 20% tolerance - amplifier gain/volume control (your ear isn't calibrated neither), feedback in hobby servos, mixing console pots, etc.
Everything else would use a microcontroller and a DAC today.
-
2 ideas...
If you're OK with using some extra current then scale your resistors by a factor of at least 10.
e.g.
R1=35.55R
R2=66.4R
R3=1.11K (instead of 1K) // 10K pot connected across R3
Alternatively buffer the low and high voltages with opamps and connect the 10K pot between the two opamp outputs
-
Yes . sort of . Industry works within the tolerances of the components supplied. How much accuracy do you need?
You can't change the parameters of the potentiometers but you can work within them.
You can narrow the range of you voltage divider by setting the highest range pots to match the lower range pots with series resistors or trim pots.You can fine adjust a higher value pot with a smaller value pot in series.But that requires more components and more adjusting.
You can use a potentiometer at a lower value. Remember a potentiometer is just a resistor . The higher the resistance the more variation you get from the tolerance. A 10KΩPot ± 2KΩ as you have observed .A 1K±20% would give you ±200Ω. A 100Ω±20% in with give you 100Ω ± 20Ω . But that also narrows the range.
You can use a resistor or or trim pot in parallel of each potentiometer to narrow the range to match the lowest range you can work with. You can make a 10K pot act like an 8K pot. This requires careful measurement of each resistor and potentiometer parallel pair. Search resistors in parallel.
Or you could use a low value potentiometer to adjust an opamp with a set gain to give a higher variation at the output. Requires careful adjustment of opamp gain and a lot more components.
-
Thanks ... trim pots is what I think i will use. Looks like Bourns makes tiny trim pots for $0.28 ... perfect.
-
If you need accuracy for something, you don't use pots. It is that simple.
Applications where classic mechanical potentiometers are still used usually don't care about a 20% tolerance - amplifier gain/volume control (your ear isn't calibrated neither), feedback in hobby servos, mixing console pots, etc.
Everything else would use a microcontroller and a DAC today.
I guess I would need to see amplifier volume control or better yet -- mixer console schematic -- in order to understand how it could be a "don't care" situation. Off the top of my head, I have a notion that a 20% difference in R manifests itself as the volume knob being unable to go down to a "zero volume" condition. I'm aware my intuition on this is wrong...
-
If the entire voltage or signal to be controlled is applied across the potentiometer, then the tolerance on the pot value is not important.
If a trimpot is to be used for a smaller fraction of the applied voltage, then normally a voltage divider is formed with three resistors and the pot is in parallel with the middle resistor, which has a much smaller value than the pot. This allows a reasonable result with a wide-tolerance pot value and narrow-tolerance fixed resistors.
-
I have a notion that a 20% difference in R manifests itself as the volume knob being unable to go down to a "zero volume" condition.
often 1 end of the pot is connected to ground so at zero the signal,or input to the next stage is shorted to ground as showen in a typical master fader section of an old school mixer
(https://farm4.staticflickr.com/3906/14843798478_04b8536ac9_o.jpg)
-
If your pot is the wrong value, then turn the knob until it has the right value.
You could add a trimpot (or 2) for calibration of the max and min values, but why would you care?
-
You could use a zener or other means to produce a regulated voltage for the potentiometer to tap off.
-
If you need accuracy for something, you don't use pots. It is that simple.
Applications where classic mechanical potentiometers are still used usually don't care about a 20% tolerance - amplifier gain/volume control (your ear isn't calibrated neither), feedback in hobby servos, mixing console pots, etc.
Everything else would use a microcontroller and a DAC today.
I guess I would need to see amplifier volume control or better yet -- mixer console schematic -- in order to understand how it could be a "don't care" situation. Off the top of my head, I have a notion that a 20% difference in R manifests itself as the volume knob being unable to go down to a "zero volume" condition. I'm aware my intuition on this is wrong...
No, 20% tolerance of zero is still zero!
Also, remember, audio levels are measured in dB.
Our ears have a logarithmic response, so that a half power audio signal is barely perceived as a change
Say the pot variation caused a fall of 20% in audio output power --that would be -0.9691dB.
An increase of the same percentage would give you an increase of 0.7918 dB------neither would be perceivable.
Many circuits are designed so that they do not rely upon precise component values, ln fact, in earlier days, over dependence upon component value was regarded as the hallmark of poor design.
A micro & a DAC aren't the "royal road" to accuracy, if either the sample steps are too coarse, or the input to the DAC has a significant DC shift.
The poor old DAC can only work with what its given!
-
Hi, Leslie Green discusses potentiometer in his book ANALOG SEEKrets ( http://lesliegreen.byethost3.com/seekrets/seekPDF.pdf (http://lesliegreen.byethost3.com/seekrets/seekPDF.pdf) ) in chapter 5, page 52. On page 53 there is a circuit that might be useful to you...
-
A micro & a DAC aren't the "royal road" to accuracy, if either the sample steps are too coarse, or the input to the DAC has a significant DC shift.
The poor old DAC can only work with what its given!
If that's a problem, there are 16 and even 24 bit DACs. Yes, expensive and potentially requiring complex low noise design - but if you care about such resolution, you would need to do it regardless of whether you use a mechanical pot or a DAC.
DC shift can be handled.
The idea with the micro, DAC (and a good voltage reference) is that you have accurate and repeatable values. Good luck getting that out of a pot with 20% tolerance and who knows what tempco and worn out/dirty wiper ... Of course, it may be also a complete overkill if you don't really need that sort of precision. But only the OP would know that.
-
Use the pot as a voltage divider, this way absolute value doesn't matter.
BTW, speaking of precision, it's widely considered that trimpots have horrible tempco and stability (didn't check that, but I tend to believe), they are al susceptible to e.g., vibrations. For precision circuits, voltnuts use them only for very fine-grained adjustments. They also use expensive trimpots with specified tempco and mechanical stability.
-
Hi, Leslie Green discusses potentiometer in his book ANALOG SEEKrets ( http://lesliegreen.byethost3.com/seekrets/seekPDF.pdf (http://lesliegreen.byethost3.com/seekrets/seekPDF.pdf) ) in chapter 5, page 52. On page 53 there is a circuit that might be useful to you...
Just read about 20 pages of this book and already discovered tons of information I had neglected to learn in the past ... thanks for this ;)