I was looking into how best to optimize my resistor values. I would love to have an entire kit of
E24 resistor values, but going from 10R to 4.7M is 5 1/2 decades at 24 per decade ... 132 different resistor values. So I thought I would see what I can get away with.
Starting with the E3 values (100,220,470) I wrote a program to exhaustively search all pairs of parallel resistors for value so see what could be obtained within the 5% tolerance of the resistors.
No dice... Up to a 15% error on the 130 value.
Requiv | E24 | R1 | R2 | %err |
100 | 100 | 100 | -1 | 0.00 |
110 | 110 | 220 | 220 | -0.00 |
110 | 120 | 220 | 220 | -8.33 |
150 | 130 | 220 | 470 | 15.27 |
150 | 150 | 220 | 470 | -0.10 |
150 | 160 | 220 | 470 | -6.34 |
180 | 180 | 220 | 1000 | 0.18 |
200 | 200 | 220 | 2200 | -0.00 |
220 | 220 | 220 | -1 | 0.00 |
235 | 240 | 470 | 470 | -2.08 |
235 | 270 | 470 | 470 | -12.96 |
320 | 300 | 470 | 1000 | 6.58 |
320 | 330 | 470 | 1000 | -3.11 |
387 | 360 | 470 | 2200 | 7.57 |
387 | 390 | 470 | 2200 | -0.70 |
427 | 430 | 470 | 4700 | -0.63 |
470 | 470 | 470 | -1 | 0.00 |
500 | 510 | 1000 | 1000 | -1.96 |
500 | 560 | 1000 | 1000 | -10.71 |
688 | 620 | 1000 | 2200 | 10.89 |
688 | 680 | 1000 | 2200 | 1.10 |
688 | 750 | 1000 | 2200 | -8.33 |
825 | 820 | 1000 | 4700 | 0.56 |
909 | 910 | 1000 | 10000 | -0.10 |
|
*note* a R2 value of -1, means you don't need it. Just easier to compute with this 'invalid' number.
I bumped up to the E6 values. Things are great, the biggest error is at the E24 value of 160, 3.12% error, within the 5% tolerance I was shooting for.
Requiv | E24 | R1 | R2 | %err |
100 | 100 | 100 | -1 | 0.00 |
110 | 110 | 220 | 220 | -0.00 |
123 | 120 | 150 | 680 | 2.41 |
130 | 130 | 150 | 1000 | 0.33 |
150 | 150 | 150 | -1 | 0.00 |
165 | 160 | 330 | 330 | 3.12 |
180 | 180 | 220 | 1000 | 0.18 |
200 | 200 | 220 | 2200 | -0.00 |
220 | 220 | 220 | -1 | 0.00 |
235 | 240 | 470 | 470 | -2.08 |
270 | 270 | 330 | 1500 | 0.18 |
300 | 300 | 330 | 3300 | -0.00 |
330 | 330 | 330 | -1 | 0.00 |
358 | 360 | 470 | 1500 | -0.59 |
387 | 390 | 470 | 2200 | -0.70 |
427 | 430 | 470 | 4700 | -0.63 |
470 | 470 | 470 | -1 | 0.00 |
519 | 510 | 680 | 2200 | 1.85 |
564 | 560 | 680 | 3300 | 0.68 |
618 | 620 | 680 | 6800 | -0.29 |
680 | 680 | 680 | -1 | 0.00 |
750 | 750 | 1500 | 1500 | -0.00 |
825 | 820 | 1000 | 4700 | 0.56 |
909 | 910 | 1000 | 10000 | -0.10 |
|
Why not more in parallel to get more? Or parallel/series combinations? I could, but I would like to keep this to a chart I can tape to the workbench.
As for laying out PCB's with these values, Surface mount components already have this problem solved, Just stack R1 and R2 on top of each other. ( I've seen this done, but I don't work in surface mount. If it's wrong please correct me.) For through hole, I can give all my E24 values the same footprint and only populate one resistor on the E6 values. I expect this will make layout simpler, as everything is consistent no matter the resistor value chosen.
The footprint is just a .400 and .500 footprint merged. The intent is to have one resistor stacked on the other, so, one value is on the inner .400 spacing, the other is on top using the outer .500 spacing.