Component Tolerance
-
- pluggin' in mics
- Posts: 30
- Joined: Sat Jun 30, 2012 9:38 am
Component Tolerance
Quick question about things like percentage tolerance. Is the only difference between say a 10% carbon comp. and a 1% metal-film resistor the fact that you'd probably have to buy 10 or so of the carbon comp to get one that's as close to its labeled value as any old 1% metal-film resistor? In other words, is there a difference in the "quality" of the component, or is it simply the tolerance and hence a little more of a nuisance in that you have to measure more (perhaps) to get close to your desired values?
I know there's arguably a tone difference between metal-film and carbon comp. but besides that, if we're just talking about resistance values...
Same deal for capacitors?
I also realize there's some places where the exact value's not nearly as critical. If the "quality" is the same (but just a larger range of possible values), couldn't you set up a system of equations (for a small pedal anyway) that would tell you how to adjust the other component values to compensate for one value being "slightly" off? So you put in values that you actually install, and then for the rest it says something like "lower" or "higher" than the posted value, and by about how much you should shoot for. Kind of like hand-matching components, only in this case you're matching them to compensate rather than matching two that are the same value.
I'm not building mammoth gear, just some small effects pedals and maybe a modular synth project in time. Maybe some gear clones as well.
I know there's arguably a tone difference between metal-film and carbon comp. but besides that, if we're just talking about resistance values...
Same deal for capacitors?
I also realize there's some places where the exact value's not nearly as critical. If the "quality" is the same (but just a larger range of possible values), couldn't you set up a system of equations (for a small pedal anyway) that would tell you how to adjust the other component values to compensate for one value being "slightly" off? So you put in values that you actually install, and then for the rest it says something like "lower" or "higher" than the posted value, and by about how much you should shoot for. Kind of like hand-matching components, only in this case you're matching them to compensate rather than matching two that are the same value.
I'm not building mammoth gear, just some small effects pedals and maybe a modular synth project in time. Maybe some gear clones as well.
Carbon resistors are noisier than metal film. Also, if you buy 10 10% resistors you're not guaranteed to get any within 1% of the value.
1/4watt 1% metal film resistors are so cheap now, why not use them? If you're just building a few things, the difference is pennies at worst.
In general, you want to design so that the component values don't have to be exact for the circuit to work as desired. If your circuit is well thought out, for the most part you shouldn't have to compensate for the actual values of the components you've got?you should take the tolerances into account when you're designing your circuit, so you don't have to worry about it when you're building it. Biasing a jfet or something in a critical position might be an exception.
1/4watt 1% metal film resistors are so cheap now, why not use them? If you're just building a few things, the difference is pennies at worst.
In general, you want to design so that the component values don't have to be exact for the circuit to work as desired. If your circuit is well thought out, for the most part you shouldn't have to compensate for the actual values of the components you've got?you should take the tolerances into account when you're designing your circuit, so you don't have to worry about it when you're building it. Biasing a jfet or something in a critical position might be an exception.
- Darlington Pair
- takin' a dinner break
- Posts: 197
- Joined: Tue Jun 03, 2008 12:27 pm
- Location: Staunton, VA
- Contact:
Everything he said. Most audio circuits aren't so picky that they need things to be exact.dfuruta wrote:Carbon resistors are noisier than metal film. Also, if you buy 10 10% resistors you're not guaranteed to get any within 1% of the value.
1/4watt 1% metal film resistors are so cheap now, why not use them? If you're just building a few things, the difference is pennies at worst.
In general, you want to design so that the component values don't have to be exact for the circuit to work as desired. If your circuit is well thought out, for the most part you shouldn't have to compensate for the actual values of the components you've got?you should take the tolerances into account when you're designing your circuit, so you don't have to worry about it when you're building it. Biasing a jfet or something in a critical position might be an exception.
That said, I buy metal film because it's less noisy and I don't build enough stuff to make a big enough difference in my wallet and also to rule out the chance of tolerances working against me. Even with biasing a jfet, I don't build enough stuff to mind popping a little more for trim pots. If I ran Roland or something I would care about these cost differences, but not me as a home builder.
There is a great argument for the use of carbon comp in high voltage applications (if you can live with the extra noise) using science in this article, and from my own dealings with RG, he isn't one prone to unprovable arguments or mojo so I highly recommend his articles. http://www.geofex.com/article_folders/c ... oncomp.htm
-
- pluggin' in mics
- Posts: 30
- Joined: Sat Jun 30, 2012 9:38 am
Ah, thanks for the info. I agree that the cost difference is definitely negligible. That's a really neat article I hadn't come across yet on distortion in carbon composition resistors! So it's basically the sort of thing where you have a "tube" pedal but that uses such low voltages that you're not getting the thing that you typically associate with tube sound. What a surprise...to get that "great vintage tube sound," you really ought to buy a "great vintage tube amp" .
-
- moves faders with mind
- Posts: 2746
- Joined: Thu Jul 03, 2003 11:26 pm
- Location: Denver, CO
- Contact:
It's often rare to find low-tolerance parts that are near the nominal value. Here's why:
http://www.edn.com/electronics-blogs/si ... 7-solution
For many analog circuits, the absolute values aren't as critical as the relative values. It's the ratio of the resistors to each other that makes things go. Granted, there are some practical considerations: reistors that are really low in value tend to consume a lot of current, and many circuits don't want to drive low-impedance loads. Really large resistors are more noisy. 10K-ish seems to be a reasonable middle ground.
You can also calculate the effect of component tolerance...though the combinatorics tend to add up to too many cases to comfortably handle. You could do node-voltage analysis yourself, but some SPICE packages can do this for you.
If you do the analysis or simulation, you'll probably learn that with 10% resistors, you'll usually get within a couple dB of the desired goal. With 1%, it's a fraction of a dB. Sometimes things are more critical than that...and frequently, there are parts (like pots or caps) that you can't find in anything better than 10% anyways.
(and since you mention it, tuning related circuits in a modular synth is one of the places where you'll want to best tolerances you can get. 10% out of tune on a synth, is a lot less palatable than 10% too quiet or too loud in a preamp.)
http://www.edn.com/electronics-blogs/si ... 7-solution
For many analog circuits, the absolute values aren't as critical as the relative values. It's the ratio of the resistors to each other that makes things go. Granted, there are some practical considerations: reistors that are really low in value tend to consume a lot of current, and many circuits don't want to drive low-impedance loads. Really large resistors are more noisy. 10K-ish seems to be a reasonable middle ground.
You can also calculate the effect of component tolerance...though the combinatorics tend to add up to too many cases to comfortably handle. You could do node-voltage analysis yourself, but some SPICE packages can do this for you.
If you do the analysis or simulation, you'll probably learn that with 10% resistors, you'll usually get within a couple dB of the desired goal. With 1%, it's a fraction of a dB. Sometimes things are more critical than that...and frequently, there are parts (like pots or caps) that you can't find in anything better than 10% anyways.
(and since you mention it, tuning related circuits in a modular synth is one of the places where you'll want to best tolerances you can get. 10% out of tune on a synth, is a lot less palatable than 10% too quiet or too loud in a preamp.)
"What fer?"
"Cat fur, to make kitten britches."
"Cat fur, to make kitten britches."
-
- pluggin' in mics
- Posts: 30
- Joined: Sat Jun 30, 2012 9:38 am
I can't believe this never crossed my mind. And of course that explains why 10% resistors aren't 10x cheaper than 1% ones. Totally went over my head that...hey...these companies have to make money!!!The Scum wrote:It's often rare to find low-tolerance parts that are near the nominal value. Here's why:
http://www.edn.com/electronics-blogs/si ... 7-solution
Good point. My friend would probably come over and be like, "Hey, why'd you put the detune knob on backwards? It says it's all the way off."The Scum wrote:(and since you mention it, tuning related circuits in a modular synth is one of the places where you'll want to best tolerances you can get. 10% out of tune on a synth, is a lot less palatable than 10% too quiet or too loud in a preamp.)
Thanks again for taking the time to explain things. I think I'm just going to have to start building some stuff on breadboard and working out just what is audible or not in a practical situation. I have a decent amount of "book-learnin" on the subject, but as with a lot of things, it's probably only half or so of the equation.
- Marc Alan Goodman
- george martin
- Posts: 1399
- Joined: Tue Oct 28, 2003 7:57 pm
- Location: NYC
- Contact:
I'd always kind of assumed that but never tested it. I love the internet. Somebody's done everything.The Scum wrote:It's often rare to find low-tolerance parts that are near the nominal value. Here's why:
http://www.edn.com/electronics-blogs/si ... 7-solution
-
- re-cappin' neve
- Posts: 602
- Joined: Mon Mar 12, 2007 8:45 pm
- Location: Sunny Tucson
Re: Component Tolerance
In addition to the initial-value tolerance, you have to consider that the higher-quality resistors are more stable over time. In other words, the resistance doesn't change as much with a 1% as it could with a 10% resistor.ithoughticouldrelate wrote:Quick question about things like percentage tolerance. Is the only difference between say a 10% carbon comp. and a 1% metal-film resistor the fact that you'd probably have to buy 10 or so of the carbon comp to get one that's as close to its labeled value as any old 1% metal-film resistor? In other words, is there a difference in the "quality" of the component, or is it simply the tolerance and hence a little more of a nuisance in that you have to measure more (perhaps) to get close to your desired values?.
-a
"On the internet, nobody can hear you mix a band."
- Marc Alan Goodman
- george martin
- Posts: 1399
- Joined: Tue Oct 28, 2003 7:57 pm
- Location: NYC
- Contact:
Re: Component Tolerance
Now that I find hard to understand. As far as I know (and has been covered in this thread), all the tolerances are made in one batch together. Then they simply test the values to see which ones are within 1%, 10%, etc and separate them that way. The tolerance value doesn't mean a higher quality resistor, just one that happened to land closer to the approximated value. So why would they drift less? They should drift exactly the same amount. I can imagine different brands and different series resistors would drift differently, but even that seems like it's probably not a case of more expensive = less drift.Andy Peters wrote: In addition to the initial-value tolerance, you have to consider that the higher-quality resistors are more stable over time. In other words, the resistance doesn't change as much with a 1% as it could with a 10% resistor.
-a
A 1% resistor isn't a better made resistor, it's just one that happens to be within 1% of the expected value.
- Darlington Pair
- takin' a dinner break
- Posts: 197
- Joined: Tue Jun 03, 2008 12:27 pm
- Location: Staunton, VA
- Contact:
Re: Component Tolerance
Except when it's a Mil Spec part, they have a wider operating temperature range with less drift over it's expected life. I'm pretty fond of using Dale 1% Mil spec resistors when I can, like I said the price difference just isn't that big for a guy like me that doesn't sell commercially. And except Carbon vs. Metal Film, they have drift differences.Marc Alan Goodman wrote:Now that I find hard to understand. As far as I know (and has been covered in this thread), all the tolerances are made in one batch together. Then they simply test the values to see which ones are within 1%, 10%, etc and separate them that way. The tolerance value doesn't mean a higher quality resistor, just one that happened to land closer to the approximated value. So why would they drift less? They should drift exactly the same amount. I can imagine different brands and different series resistors would drift differently, but even that seems like it's probably not a case of more expensive = less drift.Andy Peters wrote: In addition to the initial-value tolerance, you have to consider that the higher-quality resistors are more stable over time. In other words, the resistance doesn't change as much with a 1% as it could with a 10% resistor.
-a
A 1% resistor isn't a better made resistor, it's just one that happens to be within 1% of the expected value.
Honestly, in all my years in electronics I haven't seen too many resistors that have drifted so much that it caused a problem, sure it happens every now and then but there are much more common problems to worry about, audio stuff is VERY forgiving for the most part, now the microwave cryogenic amps I build for radio astronomy as my day job... that's another story.
- Marc Alan Goodman
- george martin
- Posts: 1399
- Joined: Tue Oct 28, 2003 7:57 pm
- Location: NYC
- Contact:
Re: Component Tolerance
Absolutely correct. We're saying the same thing, though I would call a Mil Spec part a different series, rather than just a different tolerance.Darlington Pair wrote:Except when it's a Mil Spec part, they have a wider operating temperature range with less drift over it's expected life. I'm pretty fond of using Dale 1% Mil spec resistors when I can, like I said the price difference just isn't that big for a guy like me that doesn't sell commercially. And except Carbon vs. Metal Film, they have drift differences.
Honestly, in all my years in electronics I haven't seen too many resistors that have drifted so much that it caused a problem, sure it happens every now and then but there are much more common problems to worry about, audio stuff is VERY forgiving for the most part, now the microwave cryogenic amps I build for radio astronomy as my day job... that's another story.
I don't have that many years experience, but I've yet to find a single drifting resistor in all the vintage gear I've waded through. I'm sure they happen, but it's one of the last places I check when troubleshooting.
-
- re-cappin' neve
- Posts: 602
- Joined: Mon Mar 12, 2007 8:45 pm
- Location: Sunny Tucson
Re: Component Tolerance
Late reply, I know, but:Marc Alan Goodman wrote:Now that I find hard to understand. As far as I know (and has been covered in this thread), all the tolerances are made in one batch together. Then they simply test the values to see which ones are within 1%, 10%, etc and separate them that way. The tolerance value doesn't mean a higher quality resistor, just one that happened to land closer to the approximated value. So why would they drift less? They should drift exactly the same amount. I can imagine different brands and different series resistors would drift differently, but even that seems like it's probably not a case of more expensive = less drift.Andy Peters wrote: In addition to the initial-value tolerance, you have to consider that the higher-quality resistors are more stable over time. In other words, the resistance doesn't change as much with a 1% as it could with a 10% resistor.
-a
A 1% resistor isn't a better made resistor, it's just one that happens to be within 1% of the expected value.
You're not correct. 10% and 5% etc resistors are not made on the same production line as 1% resistors which are not the same as the 0.1% resistors. The materials are different, the construction is different. They have to be. The tighter-tolerance resistor drift and aging characteristics are better than the wide-tolerance part. You can't test for something that will happen in the future. (Well, you can do the accelerated aging tests, but then you've destroyed the parts ...)
And when your production line is spitting out a million of the things a day, binning for those wide tolerance bands is not possible.
-a
"On the internet, nobody can hear you mix a band."
Who is online
Users browsing this forum: No registered users and 88 guests