[Born2bwire]

Quote:

Originally Posted by Denim You have the right idea here, but the explanation could use some additional info. It's possible that 2 resistors in parallel will achieve a final resistance closer to the desired value, but the errors need to cancel out each other. That is, one resistor having a higher reading, and the other having a lower reading. If they are both higher, or lower, then the total resistance can still be off as much as a single resistor. Here's the math -- Desired resistance - 50 ohms Circuit 1: 100 ohm (measures 110 ohms) & 100 ohm (measures 110 ohms) in parallel = 55 ohms. Resistors off by 10%, circuit off by 10% also. Circuit 2: 100 ohm (measures 110 ohms) & 100 ohm (measures 90 ohms) in parallel = 50 ohms. Resistors off by 10%, circuit achieves desired resistance because the offsetting differences cancel each other out. As for resistors being polarity sensitive, I haven't seen one yet.

Whoops, made a mistake in the math. The error should be:
abs(1+2*(E1*E2-1)/(2+E1+E2)).
So it would appear that the final error is going to be equal to or less than the larger error of the two resistors. In your second circuit though there still is an error of 1%.

Another reason to go parallel that I thought of is that you split the current running through each resistor by half while increasing the resistance by 2. So the power dissipated will reduce by a factor of two. So theoretically, you could use cheaper parts. For example, instead of a 1K 1/2 W 5% resistor, you could use a 2K 1/4 W 1% and a 2K 1/4 W 10% and achieve the same power specs with a maximum error of 5.7% Though in real life I doubt there is any such combination of two resistors that would actually be cheaper than the single resistor that you are replacing.

 

[NeilR]

Quote:

Originally Posted by Denim You have the right idea here, but the explanation could use some additional info. It's possible that 2 resistors in parallel will achieve a final resistance closer to the desired value, but the errors need to cancel out each other. That is, one resistor having a higher reading, and the other having a lower reading. If they are both higher, or lower, then the total resistance can still be off as much as a single resistor. Here's the math -- Desired resistance - 50 ohms Circuit 1: 100 ohm (measures 110 ohms) & 100 ohm (measures 110 ohms) in parallel = 55 ohms. Resistors off by 10%, circuit off by 10% also. Circuit 2: 100 ohm (measures 110 ohms) & 100 ohm (measures 90 ohms) in parallel = 50 ohms. Resistors off by 10%, circuit achieves desired resistance because the offsetting differences cancel each other out. As for resistors being polarity sensitive, I haven't seen one yet.

The problem with this is that resistors within a batch tend to match each other much closer than their spec'd tolerances. This is great for matching but not helpful for absolute tolerance to the specified value. Within a random lot it would probably be easier to find a single very close match to the specified value than to find two that are relatively close in absolute error, on opposite sides of the intended value.

Then again, there aren't many circuits that care about absolute values below 5% or so and most have much higher tolerances.

 

[Denim]

Quote:

Originally Posted by NeilR The problem with this is that resistors within a batch tend to match each other much closer than their spec'd tolerances. This is great for matching but not helpful for absolute tolerance to the specified value. Within a random lot it would probably be easier to find a single very close match to the specified value than to find two that are relatively close in absolute error, on opposite sides of the intended value. Then again, there aren't many circuits that care about absolute values below 5% or so and most have much higher tolerances.

I hear what you are saying. I was running with the idea that Born2wire submitted and stayed with the theory. This sounds like the problem I constantly run up against. Just because you can put it on paper, it doesn't mean you can go to the parts bins and wire it up. I have enough trouble just finding the right components.

 

[NeilR]

Quote:

Originally Posted by Denim I hear what you are saying. I was running with the idea that Born2wire submitted and stayed with the theory. This sounds like the problem I constantly run up against. Just because you can put it on paper, it doesn't mean you can go to the parts bins and wire it up. I have enough trouble just finding the right components.

You were working through a theory. Theory and practice are often two different animals