# Resistors - metal film or carbon film and what values?

I am buying a bunch of resistors as a start to my electronics hobby. These are the values I have chosen:

100, 470, 1k, 4k7, 10k, 33k, 47k, 100k, 470k, 1M

Most of my projects will involve microcontrollers, such as the Arduino and other AVR ones. I have two questions which I would appreciate replies to.

1. Should I purchase carbon film or metal film resistors, and why? As I understand from my research, metal film tolerate heat better, and are more precise. Price is not something to consider too much as they are essentially the same. Carbon film resistors are $0.01/resistor, and metal film are$0.012/resistor.

2. Should I add or remove any resistor values from the aforementioned list based on my intended use for them?

Thanks.

• You may want a few more values for driving LEDs from 5 V @ 20 mA (150 Ohm, 180 Ohm) – m.Alin Jul 29 '12 at 19:31
• This is the resistor set that I own: amazon.com/Joe-Knows-Electronics-Value-Resistor/dp/B003UC4FSS/…. It has pretty much every resistor I could need (as a hobbyist, not a professional) – Chris Laplante Jul 29 '12 at 19:33
• Stock up more on the decade values 1, 10, 100... as you can combine them (parallel/series equivalent) easily to come up with the other values when doing breadboard experiments. For example if you had run out of 330 Ohm resistor, you can substitute that with three 1K Ohm resistors in parallel. – shimofuri Jul 31 '12 at 20:06
• If you are breadboarding, be aware that lead stiffness varies. It might be good to get a few of a brand to try before ordering large quantities. – Chris Stratton Mar 9 '13 at 15:25
• @SimpleCoder - hey, that kit even includes the hard-to-find 0 ohm resistor! ("values from 0 ohm to 10M ohm") – Pete Becker Mar 9 '13 at 18:16

(1) Use metal film where possible. Fewer bad surprises. At 1 cents each either way the cost of bad surprises exceeds the component cost, even if the cost is only measured in frustration and wasted effort.

(2) Wouter (correctly (of course)) says "evenly spaced" but doesn't quite explain it. He means that the ratio between the eresistance adjacent resistors should be about the same. You shold aim to always include the powers of 10 values and then have as many as appropriate in between to fill in.

SO

1, 10, 100, 1000, 10000 ...
OK, that one was obvious.

But sqrt(10 ) = 3.16, so
1. 3.16, 10, 31.6, 100, 316 ... :-)
BUT they don't make 3.16 etc in sensible standard ranges, so using the nearest "E12" values:
1, 3.3, 10, 33, 100, 330, 1000, 3k3, 10k, 33k ...

The "obvious" thing to do may be to use
1, 4.7, 10, 47, 100, 470 etc
BUT the ratio of 47/10 = 47 (of course) BUT the ratio of 100/47 = 2.13.
So, if you had a fixed voltage and were connecting successively higher value resistors to ground the change from 100 to 470 would decrease the current by a factor of 4.7, but the next step from 470 to 1000 would reduce the current by a ratio of 2.13. As you went up the currents would change by factors of 4.7, 2.13, 4.7, 2.13, 4.7 ...

You usually get more than 2 steps per decade.
The smallest sensibl;e number has 12 steps per decade.
These are say 1, 1.2, 1.5, 1.8, 2.2, 2.7, 3.3, 3.9, 4.7, 5.6, 6.8, 8.2, 10 ...
If looked at by resistance difference the series seems uneven, The differences are.
0.2, 0.3, 0.3, 0.4, 0.5, ... 1.4, 1.8
BUT - when looked geometrically by ratio we see:
1.2/1 = 1.2
1.5/1.2 = 1.25
1.8/1.5 = 1.2
2.2/1.8 = 1.222
2.7/2.2 = 1.227
3.3/2.7 = 1.222
...
10/8.2 = 1.22

SO, within the resolution afforded by 2 significant digit numbers we see that the ratio of adjacent resistances is about 1.21152766 :-) . I use that "strange" value as it is the tweflth root of 10. If you multiply a number by 1.21152766 twelve times you get a result 10 times larger.
So if you space twelve resistors across a decade range with each a factor of 10^(1/12) larger than the prior one you get resistors which increase in value "smoothly" from a current flow point of view.

E12 - 12 resistors per decade spaced in value by a ratio of the 12th root of 10 .
E24 - 24 resistors per decade spaced in value by a ratio of the 24th root of 10 .
E48 - 48 resistors per decade spaced in value by a ratio of the 48th root of 10 .
E96 ...

More anon maybe .... brake pads to change, darkness fallen ...

• @capcom For a list of the E-series values have a look at: en.wikipedia.org/wiki/Preferred_values#E_series – ARF Jul 30 '12 at 8:09
• There is also an E6 series: 10, 15, 22, 33, 47, 68. – starblue Jul 31 '12 at 7:46
• @starblue, +1 for the E6 series. This is what I stock in my resistors box. It's a good compromise between value range, spent money, and also for storage box space. From 1R to 10MR, 42 (7 decades x 6 values in each one) boxes are needed, which is not too bad for the hobbyist. – Marcovecchio Jul 13 '16 at 12:07

If you are going to be doing a lot of analog electronics, you should buy metal film. Metal film produces less thermal noise than carbon. Metal film resistors also typically have a much lower inductance/capacitance than carbon so they(metal film) work better at higher frequencies. Carbon has no real advantage except that they are cheaper. If I was only going to work on digital stuff, I would buy the carbon comp.

As far as what values, the values you have chosen are reasonable for digital stuff. If you think you may use op amps or transistors to amplify the signal, I would look into the E6 series of resistors.

• Great, thanks. Curious to know if metal would work well for the digital stuff too, even though you prefer carbon. Why do you prefer carbon for digital stuff? – capcom Jul 29 '12 at 20:06
• Carbon is usually much less expensive. I would check the wattage of the carbon vs the metal film you are looking at. When I buy metal film, they are typically about 10 times more for the same specification. – OhmArchitect Jul 31 '12 at 23:53
• Odd, in my case, that's not true. They're both rated at 1/4W. Check out Tayda Electronics if you want to. Thanks for following up. – capcom Jul 31 '12 at 23:56
• Who said carbon resistors are bad? They're good for the budding evil genius: quite handy as electronic cannon fodder. ;-) Better burn cheap carbon resistors when doing crazy experiments and leave the metal film for prototypes and final designs. If you don't like the thought of sacrificing poor resistors in the quest for knowledge, you can use computer-aided modelling/simulators before building out a circuit in hardware. en.wikipedia.org/wiki/… – shimofuri Aug 3 '12 at 20:08

(1) Go for metal film resistors. Typically, they are of 1% precision, compared to 5% precision for the carbon film ones. Also, their values vary less with temperature changes.

(2) For initial stockpiling, I'd suggest purchasing a broader set of values. Otherwise you risk making frequent trips to the local store, just for a pair of resistors, for the next cool project you found online. At 2AM on a Sunday, it spoils all the fun.

Here's a sample set of 25 values, from the E6 series:

100      150      220      330      470      680
1k       1.5k     2.2k     3.3k     4.7k     6.8k
10k      15k      22k      33k      47k      68k
100k     150k     220k     330k     470k     680k
1M


(You may want to add the 10 ohm ~ 68 ohm ones as well)

These can be further combined, to substitute for other values. Wolfram Alpha is my preferred calculator and there are many others online. For example: 314 ohm

• If an odd value, say 107 Ohms, is required and you don't have that in inventory (because it can only be found in the >=E96 series), you can easily find that resistor from a bunch of 100 Ohm resistors with low precision. Get an ohmmeter, measure, and look for one closest to it. It isn't really like finding a needle in a haystack. The lesson here is to get a bunch of resistors in stock (30 pieces) for a certain value. – shimofuri Aug 3 '12 at 20:22

The shipping costs would be more than the costs of the actual resistors you've suggested. As others have said, they're so cheap that you should get more.

Like I suggested above, check out this resistor kit:

860 resistors (10 of 86 values), metal film, 1/4 W for $17.99. Note that they are 5-band resistors - a little harder to read, but not much. Here is the product link. As @shimofuri points out, you should supplement your collection with lots of common resistor values (10k, 1k, 220 (or whatever you use for LEDs)). I recommend resistor busses in these values, which are helpful for driving LEDs and handling large numbers of pullups and pulldowns. • Actually, shipping is also very cheap. It's about$2 for every \$10 I spend. So the resistors actually are extremely cheap. Thanks for the link too! – capcom Jul 29 '12 at 20:04
• Oh, that's not bad. Still, the link I posted has the added benefit of the resistors being presorted into individual labelled bags :) – Chris Laplante Jul 29 '12 at 20:05
• Alright, I do suppose that is an advantage as well! Although the place I am ordering from has sent me them in labelled bags before too :D – capcom Jul 29 '12 at 20:08
• You might want to supplement that pack with quantities of resistors in the 100 to 10K Ohm range. Ten pieces for each value will not be enough in that range as they are commonly and frequently used. For example, driving 8 separate LEDs for a simple "Kit Scanner" evilmadscience.com/productsmenu/tinykitlist/152-scanner project would require 8 pieces of resistors at say 120 Ohms (depending on LED, and supply voltage). – shimofuri Jul 31 '12 at 19:58
• @shimofuri: Good point. I did exactly that, and I also bought some resistor busses (10k, 1k, and 220k) for driving LEDs and handling large numbers of pullups/pulldowns. I'll edit the answer to include this. – Chris Laplante Jul 31 '12 at 20:44

If you can get metal for essentially the same price as carbon go for metal. Your sequence is not evenly spaced, I would prefer 330, 3k3, 33k, etc instead of 470, 4k7, 470k. Both 33k and 47k seems silly, unless you have a special need for these two values. Resistors are cheap compared to most other components, so I would buy some more. Long ago I bought a 20-value box. For my hobbyits-like work I very seldom needed something that I could ot find in that box.

• Alright, that sounds good. They're so cheap, I'll just buy a wider selection, including the ones you recommend. Thank you. – capcom Jul 29 '12 at 20:06
• If you go for more values make them evenly spaced, ate least in the most usefull range (100 .. 100k). So either 1, 3.3, 10 (factor ~ 3), or 1, 2.2, 4.7, 10 (factor ~ 2), etc. – Wouter van Ooijen Jul 29 '12 at 20:14