Mixing 5% and 10% resistor tolerances?????

[stompvulture]

Here's the deal. I am 99% complete on populating a fuzz clone board with standard 1/4 watt 5% carbon film resistors. I am short one resistor but found one, 10% carbon comp 1/2 watter. Will this work?

On the topic. I know the vintage gear used a lot of carbon comps, and amps are notorious for 1/2watters, what about populating a stompbox with them?

[blooze_man]

It will work. Fuzz Faces are very picky, though and most people will advise building it with 1% metal films.

[Electron Tornado]

You'll be fine with that part. Some people want to tweak and refine something down to the Nth degree, but remember that the old vintage stuff was built with whatever was easiest to obtain and cost the least. So use what you've got on hand to get you going. If you want to tweak it later then go ahead, that's part of the fun of this hobby.

Mixing tolerances is a non-issue, and for these small voltage pedals, mixing wattage is also a non-issue.

[stompvulture]

Excellent, thanks. I figured as much. I landed a bunch of old varying 1/2watt carbon comp resistors with 10% tolerances. Never got into amp building, thought of trying a stompbox build with them someday. For this project, I'll use the odd man out resistor to get the project off the ground and tweak it later.

[Processaurus]

The tolerance is just the manufacturer's guarantee of how close the part will be to the specified value.  A 10% resistor could conceivably hit the mark closer than a 5%.  You can measure the actual value to know.  Plenty of times I've needed a 1% resistor, and grabbed a handful of 5%, and tested them until finding ones that were within the 1%, voila, a 5% resistor just became a 1%.

It's funny, an old tech guy gave me a full collection of ancient 1% carbon comp resistors!  No idea what to use them for, but seems too cool to throw out...

[twabelljr]

You can measure the actual value to know.

+1. It might be dead on value!

[Electron Tornado]

The question about tolerances also gets asked about capacitors, and for effects pedals, it's even less of an issue. For example, a 0.1uf cap that has a 10% tolerance could be 0.11uf or 0.09uf. I can't think of an effects circuit where that would make a difference. You certainly won't hear a difference of +/- 0.01uf. Even with a 10uf cap, I doubt you can tell the difference between 9uf, 10uf, or 11uf. 

This all begs the question: when, in an effects pedal application, would parts tolerances make any difference, much less be critical?

[R.G.]

Tolerance is a more subtle issue than it looks. Parts are manufactured to be a specific value. The specific value is nothing more than a useful target to hit, so one can specify a target value, then buy a part that is close to the target value. It's like shooting an arrow at a bull's eye target. The target is placed somewhere convenient for you to use, and the degree to which the arrows hit the exact center of the center spot depends on the skill of the archer and how much precision is needed.

It used to be (c. 1950-1960) that manufacturing resistors to tolerances closer than +/- 20% was economically impractical. So they made them as close as they could, threw away the ones worse than +/-20%, and sold the rest. If someone demanded a +-10% part, they would hand-measure and select the ones which happened to be closer to the target value, charging a premium for the extra work involved. If someone needed a 5% or 1% part, they'd do the same. Later as the need for more precision parts increased, the manufacturers took to sorting parts all the time, removing the 10% parts from the 20% parts, the 50% parts from the 10% parts, and the 1% parts from the 5% parts. This left some funny results, as 10% parts could reasonably be expected to not be closer than 10% - everything closer had been removed.  Same for 10% and so on. The selected 10% parts were marked with a silver band to indicate their higher precision, and the 5% parts were marked with a gold band to show their even-higher precision.

As manufacturing got better, 10% parts were available all the time, and there was no need to select them out of a 20% distribution. In the last couple of decades, 5% parts are as economically available as the 10% used to be, so it's hard to find parts with only 10% precision any more.

This all begs the question: when, in an effects pedal application, would parts tolerances make any difference, much less be critical?

Whether you need precision is a design issue. When you calculate or adjust what a circuit needs, you are working against two opposing factors. The equations which set up whatever you're doing could be solved for a precisely exact number for each component value. However, some components in a circuit cause changes in operation that are proportional to the value change, some cause disproportionately large changes, and some cause disproportionatley small changes. It depends on where the component is in the circuit and what function it does in the circuit. The amount of change in circuit function per change in component value is referred to as the Sensitivity Factor - how much what you're looking for changes with the change in a component value. Sensitivity Factor changes depending on what the part does in the circuit.

For instance, pullup and pulldown resistors have a SF of far less than one. It's not unusual for a resistor serving one of these functions to work fine and not matter for values from 10K to 10M. There are cases where one resistor affects the gain of one stage a lot, and in some cases the gain of multiple stages. In cases like that, the SF of gain to that resistor may be an exponential factor on the circuit operation. The same applies to the AC operations of a circuit when an inductor or capacitor is involved.

The frequency of an LC resonant circuit, for instance, varies by the square root of the value of the inductor or capacitor. This is why the range of a Vox wah circuit is hard to make vary much bigger than it already does - the virtual capacitance is what varies, and this changes the resonant frequency less than linearly.

The moral to all this is that the exact value of a part may matter hugely, linearly, or almost not at all. It depends on the circuit. Quoting one of my favorite authors, of course one horse can run faster than another - but which one? Differences matter. To make good guesses, you have to know the circuit operation, or trust the designer if that's not you.

[familyortiz]

The question about tolerances also gets asked about capacitors, and for effects pedals, it's even less of an issue. For example, a 0.1uf cap that has a 10% tolerance could be 0.11uf or 0.09uf. I can't think of an effects circuit where that would make a difference. You certainly won't hear a difference of +/- 0.01uf. Even with a 10uf cap, I doubt you can tell the difference between 9uf, 10uf, or 11uf. 

This all begs the question: when, in an effects pedal application, would parts tolerances make any difference, much less be critical?

Most respectfully, I would not agree with the capacitor tolerance statement. A common 0.1uF ceramic with X7R dielectric and , used for noise decoulping, can have a +/- 20% tolerance with no problem. But if your trying to keep a design's performance stable in critical signal processing, like an active or passive filter or any gain stage, a tighter tolerance cap with stable temperature performance like a film cap would be a much better choice.
In your case, the 5% carbon resistors will be perfect and a metal film or tighter tolerance resistor will not help other than making the layout smaller.

[ayayay!]

It's funny, an old tech guy gave me a full collection of ancient 1% carbon comp resistors!  No idea what to use them for

Uhh, pedals maybe?  I guess I don't get this, and never will.  Send 'em my way.    I don't like ALL metal film resistors in a pedal.  Some CC's are just fine with me.  (No disrespect intended, I just hear more a difference between MF caps vs CCs.)  I think CC resistors sound less harsh. 

[Electron Tornado]

The moral to all this is that the exact value of a part may matter hugely, linearly, or almost not at all. It depends on the circuit. Quoting one of my favorite authors, of course one horse can run faster than another - but which one? Differences matter. To make good guesses, you have to know the circuit operation, or trust the designer if that's not you.

Most respectfully, I would not agree with the capacitor tolerance statement. A common 0.1uF ceramic with X7R dielectric and , used for noise decoulping, can have a +/- 20% tolerance with no problem. But if your trying to keep a design's performance stable in critical signal processing, like an active or passive filter or any gain stage, a tighter tolerance cap with stable temperature performance like a film cap would be a much better choice.
In your case, the 5% carbon resistors will be perfect and a metal film or tighter tolerance resistor will not help other than making the layout smaller.

Thanks for the great replies. I understand that it depends on what the component is doing and the degree of precision desired or required. Learning where and to what degree precision can make a difference would be a great exercise. I'm also not knocking precision or anyone who wants to tweak something just right. Both are part of the fun of this hobby, and goodness knows I've spent some time tweaking various parts of circuits. However, I think many beginners can get caught up with the feeling that they somehow "must have" metal film resistors with 1% tolerance, or "can't" use a cap that is 0.022uf in place of one that is 0.02uf, etc.

Using the filter example - if I want a filter for a simple tone control with a cutoff frequency of 750Hz, but with the parts I have on hand I can only get a 730Hz cutoff, how critical is that? Would a 20Hz difference be noticeable to my ears in a side by side comparison? Would it be noticeable to my ears on its own without that comparison? Is 730Hz good enough to get my circuit going and in the ball park where I can refine it later?

[calpolyengineer]

It used to be (c. 1950-1960) that manufacturing resistors to tolerances closer than +/- 20% was economically impractical. So they made them as close as they could, threw away the ones worse than +/-20%, and sold the rest. If someone demanded a +-10% part, they would hand-measure and select the ones which happened to be closer to the target value, charging a premium for the extra work involved. If someone needed a 5% or 1% part, they'd do the same. Later as the need for more precision parts increased, the manufacturers took to sorting parts all the time, removing the 10% parts from the 20% parts, the 50% parts from the 10% parts, and the 1% parts from the 5% parts. This left some funny results, as 10% parts could reasonably be expected to not be closer than 10% - everything closer had been removed.  Same for 10% and so on. The selected 10% parts were marked with a silver band to indicate their higher precision, and the 5% parts were marked with a gold band to show their even-higher precision.

As manufacturing got better, 10% parts were available all the time, and there was no need to select them out of a 20% distribution. In the last couple of decades, 5% parts are as economically available as the 10% used to be, so it's hard to find parts with only 10% precision any more.

It's pretty similar today, except all automated. A batch of resistors is manufactured to a nominal target value. Naturally the actual resistances will be normally distributed about that target value. Then each resistor is tested and sorted by tolerance. So, it turns out that when you buy a 5% tolerance resistor, there is only a very small chance that the actual value is within 2%. And when you buy a 2%, it's unlikely that the actual is within 1%, etc, etc. And to cut waste even more, the resistor series was spaced so that a resistor that is out of tolerance for the value it was manufactured to be, is within tolerance for the value above or below.

In actual practice there are production targets for each tolerance level as well. So if the process is producing too many 1% ers, some of those would be labeled as 5% for instance.

And that, my friends is how resistors are made.

[Taylor]

Using the filter example - if I want a filter for a simple tone control with a cutoff frequency of 750Hz, but with the parts I have on hand I can only get a 730Hz cutoff, how critical is that? Would a 20Hz difference be noticeable to my ears in a side by side comparison? Would it be noticeable to my ears on its own without that comparison? Is 730Hz good enough to get my circuit going and in the ball park where I can refine it later?

In an RC lowpass, with a 500r R and a 1uf cap with +-10% tolerance, the cutoff difference can be about 1/3 of an octave. That's certainly a noticeable difference, but not necessarily a bad one.

[R.G.]

Using the filter example - if I want a filter for a simple tone control with a cutoff frequency of 750Hz, but with the parts I have on hand I can only get a 730Hz cutoff, how critical is that? Would a 20Hz difference be noticeable to my ears in a side by side comparison? Would it be noticeable to my ears on its own without that comparison? Is 730Hz good enough to get my circuit going and in the ball park where I can refine it later?

The huge screaming advantage to building your own is that you can nearly always use whatever is close, and put in the perfect thing later as long as the thing doesn't go up in flames. Yes, use what you have and do the suggested thing as you can. The person who suggested the value may or may not have had a good reason for picking that value, including not having the perfect value on hand themselves.  

It's pretty similar today, except all automated. A batch of resistors is manufactured to a nominal target value. Naturally the actual resistances will be normally distributed about that target value. Then each resistor is tested and sorted by tolerance. So, it turns out that when you buy a 5% tolerance resistor, there is only a very small chance that the actual value is within 2%. And when you buy a 2%, it's unlikely that the actual is within 1%, etc, etc. And to cut waste even more, the resistor series was spaced so that a resistor that is out of tolerance for the value it was manufactured to be, is within tolerance for the value above or below.

I would have said the same thing, except that in my recent past (five years or so) I have frequently found that 5% carbon comps were within 1% or even 1/2 % of the nominal value. This has happened so often that I think they may have quit selecting to tolerance buckets, at least for the high volume resistor makers I buy from - Speer, KOA, Xicon being examples. I have rarely found a 5% carbon film worse than 2%, although I hesitate to say that here, because the beginners will all start thinking they're all going to be that way.

I put this down to much-improved manufacturing.

Also, for a given material, it doesn't always make sense to sell very high precision parts. The aging and drift of the part may well be greater than the guaranteed tolerance. If you could get 0.1% carbon comp resistors, they'd drift maybe 5% over time. Silly to pay for precision when drift will get you in the end. I suspect maybe that's another reason you don't see commercial carbon resistors with tight tolerances. Better - and nearly as cheap, with automated manufacturing - to go to the more stable metal film, oxide, and wire wound for super-precision, so the precision will stay.

Even better to step over into digital where the precision is the precision of the clock crystal, not what the voltages drift to.

Precision, accuracy, and resolution are neat words, but at the end of the assembly line Mother Nature is waiting to see how well you understood the real-world requirements - or if you even noticed them.  

[stompvulture]

Wow, some good stuff here. An update to my question,  I took a reading off of a carbon comp 10% 1/2watt 120k resistor and it turned out to be 130k. I ended up running a couple 1/4 watt 5% in parallel to get around 123k. It'll do for now. Thanks again.

[PRR]

> threw away the ones worse than +/-20%

Ummmm.... that's why they changed from 100-200-500 sequence to 100-150-220... sequence. No matter what you made, it would fit one of the bins within 20%. Yes, some odd values sold less than others, so at inventory time you'd call the biggest buyers and offer a discount if they took some odd-value resistors... in MANY cases a 680K works just the same as 470K or 1Meg.

And there has never been a whole year when factory purchasing agents were not desperate to save a tenth-cent on a few parts, in markets where "radio is radio" and the $9.97 radio will out-sell the $9.99 radio.

Yes, I sure remember when a 10% resistor would NOT!! be within 5% of nominal. Your "1K" drawer had 940 ohm and 1,060 ohms and nothing in between.

I think solid high-yield 10% (and 5% at a price) were common by the 1960s, even into the 1950s. Depends on the industry though. There was always (until the US quit making cheap stuff) a market for mostly-almost-barely-good-enough. From the $9.97 radio right up through RCA's finest 1962 console, also Kay/Alamo, some purchaser was buying krapp you would not find in Fender or basic military gear (enlisted mess PA system; mission-critical was another step up).

With carbon-film, a mature process seems to hit 1% if they take care, so they go fast for 2% parts or real-fast and mark them 5%. Even the low-accuracy parts tend to be real-darn close through a whole reel. As you have observed.

> If you could get 0.1% carbon comp resistors

I have real doubt about the 1% carbon-comp. AFAIK, even A-B could not hold CC value within 1% though the Aging test which is pretty expected for supposedly high-precision parts. Drift may be near 1% for 99 out of 100, but the guy paying a lot extra for 1% precision does not want to see 1 in 100 units come back as "bad". (More, because typically more than one 1% part per product; less, because many customers won't notice or won't return when the product drifts 1.5%....)

> a simple tone control with a cutoff frequency of 750Hz, but with the parts I have on hand I can only get a 730Hz cutoff, how critical is that?

Not at ALL. A "simple" filter has a VERY broad "cut-off". We speak of -3.01dB at Fc because the math works out that way, but  at 750Hz a 730Hz filter's loss is 2.97dB, 0.04dB different, and you hardly hear 1dB different in a direct A/B comparison.

> the exact value of a part may matter hugely, linearly, or almost not at all. It depends on the circuit.

It's like songwriting. Some places in a song, any word roughly the right size will do. Other places, it is critical to have the exact-right word.