Author Topic: Accomodating resistor tolerance for reasonable accuracy ?  (Read 738 times)

0 Members and 1 Guest are viewing this topic.

Offline SarvesaaTopic starter

  • Contributor
  • Posts: 36
  • Country: in
Accomodating resistor tolerance for reasonable accuracy ?
« on: April 28, 2023, 03:07:38 pm »
Hey friends,
 
I am building a lab power supply. I am using a LM358 as a current sense amplifier and feeding the output to a micro-controller(Atmega328p).
Sad thing is that I cannot find 1% resistors locally in the market, they only have 5% tolerance resistors.
The gain of the current sense amplifer is 1.
I measured the resistance of the resistor and it was about 17.5k instead of 18k as expected, could I substitute the actual read resistance in the gain formula and change the code accordingly ?.

Need your help on this. And the actual current sense cirucit is based of EEVBlog usupply Rev C.

Thanks
 

Offline PartialDischarge

  • Super Contributor
  • ***
  • Posts: 1625
  • Country: 00
Re: Accomodating resistor tolerance for reasonable accuracy ?
« Reply #1 on: April 28, 2023, 03:23:30 pm »
Yes you can, but you must be sure that your multimeter is reading correctly.
That is all you need, after that with 5% resistors you can put them in series/parallel to obtain mostly any value.

 
The following users thanked this post: Someone, Sarvesaa

Offline tom66

  • Super Contributor
  • ***
  • Posts: 7336
  • Country: gb
  • Electronics Hobbyist & FPGA/Embedded Systems EE
Re: Accomodating resistor tolerance for reasonable accuracy ?
« Reply #2 on: April 28, 2023, 03:33:54 pm »
One thing to be aware of is temperature drift.  Metal film resistors (most likely that is what you have) are quite stable over long periods of time, at least for most applications, but can have a large enough temperature coefficient that can impact the performance of a system.  This means if room or device temperature varies, the gain could change.

You could minimise the effect by keeping the resistor away from hot things like the heatsink for the main transistor/regulator and, if necessary, having a cooling fan in the product.

Though at the end of the day unless you are building a precision product the tolerance is likely fine. A typical worst-case figure is 100ppm/degC worst case which means a 30C rise means a 0.3% additional error.  If you have designed the product around a 1% toleranced 18kohm resistor, then adjusted it for a real world 17.5kohm resistor, consider if the additional error is consequential or not to your application.
 
The following users thanked this post: Someone, Sarvesaa

Offline SarvesaaTopic starter

  • Contributor
  • Posts: 36
  • Country: in
Re: Accomodating resistor tolerance for reasonable accuracy ?
« Reply #3 on: April 28, 2023, 04:06:54 pm »
Thanks tom66 and PartialDischarge

I thought of temperature coefficient and I will make sure it stays constant.
And I am using EEVBlog Brymen 256 multimeter.
 

Offline ejeffrey

  • Super Contributor
  • ***
  • Posts: 4034
  • Country: us
Re: Accomodating resistor tolerance for reasonable accuracy ?
« Reply #4 on: April 28, 2023, 06:55:00 pm »
One thing to be aware of is temperature drift.  Metal film resistors (most likely that is what you have) are quite stable over long periods of time, at least for most applications, but can have a large enough temperature coefficient that can impact the performance of a system.  This means if room or device temperature varies, the gain could change.

If they are axial leaded resistors they could also be carbon film resistors which have much higher temperature coefficient.  It's still not likely to be a problem, but it would be if the resistor were handling significant power and self-heating.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17428
  • Country: us
  • DavidH
Re: Accomodating resistor tolerance for reasonable accuracy ?
« Reply #5 on: April 30, 2023, 11:30:29 am »
could I substitute the actual read resistance in the gain formula and change the code accordingly?

You could do that, or even better most circuits rely on the ratio of two or a set of resistors, so you could go through some resistors of the same indicated resistance and select a set of two or four which closely match.  This removes the accuracy of your resistance measurement as a factor to be considered.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf