21

I work at an electronics store and the other day a customer came in who was rebuilding a circuit board. I sold him some resistors, but later he came back in wanting to return some of them because they were only 1% tolerance, and he needed 5%. I'm not an engineer but I have a hobbyist's understanding of electronics, and I was under the impression that there are situations that would call for a more exact value, but never a less exact one. Am I mistaken?

Justin Pederson
  • 211
  • 2
  • 6
  • you are not mistaken ... the person may have other reasons ... best to ask that person ... it could also be that the circuit design is sloppy, and does not work with tighter tolerance components – jsotola Feb 20 '23 at 01:34
  • 2
    I'm interested to find out how he "knew" that what he got were 1% ? It's not implicitly written on the components, so was it from the invoice? Maybe he didn't know how to read the extended colour codes for 1% ? – Kripacharys Feb 20 '23 at 05:16
  • What type of resistors were they? – Bruce Abbott Feb 20 '23 at 06:55
  • 36
    It's highly possible they simply don't know themselves. They just know the repair schematic/BOM says 5%, and they got 1%, and don't know what "percent tolerance" means. I don't think there's a technical reason here, just a simple silly error. – Shredder Feb 20 '23 at 09:13
  • 21
    Well 5% is greater than 1% so it means 4% more tolerance, more is better, right? – floppydisk Feb 20 '23 at 11:38
  • 4
    @Kripacharya I assumed these were through-hole resistors, where the tolerance is the last band of the color code https://en.wikipedia.org/wiki/Electronic_color_code#Color_band_system – Vorbis Feb 20 '23 at 12:54
  • 1
    I can totally see the following example happening: Customer comes into store asking for (eg) 220 ohm resistors, he buys them. He goes home and checks his resistor colour chart and realises the last band doesn't match the ones he has/what he's replacing with. He comes back "these don't match the ones I want, I want the ones with the gold bands at the end". – stanri Feb 20 '23 at 15:01
  • 1
    For the question title, cost is certainly a reason to prefer 5% to 1% resistors, but that doesn't seem to apply to the exact situation here. – Jon Custer Feb 20 '23 at 17:19
  • 9
    I know of one major computer manufacturer that designed their own bus, in the spec was to use a %5 tolerance resister, this was to help randomize the retry delay in the event of a bus collision, using tighter tolerance resistors would result in more bus collisions. – Glen Yates Feb 20 '23 at 17:23
  • 6
    @Glen Yates - But that by itself doesn't guarantee randomness. Resistors, especially if they're from the same batch (most likely), all are probably very close in value to one another. – SteveSh Feb 20 '23 at 19:38
  • 4
    @SteveSh Nobody said it was a *good* design... – Michael Feb 21 '23 at 00:01
  • The following may be relevant: https://electronics.stackexchange.com/q/157788/11869 – copper.hat Feb 21 '23 at 05:09
  • 2
    @Michael or not *still* a good design - it was probably fine when the senior engineer was a youngster – Chris H Feb 21 '23 at 11:33
  • @SteveSh The assumption was that you would probably not have 2 of the same devices from the same manufacturer on the bus, not 2 mice, not 2 keyboards, not two trackballs, or whatever else you might have on a desktop bus. And actually, the recommended tolerance may have been 10%. – Glen Yates Feb 21 '23 at 16:11
  • @GlenYates that's the sort of assumption that tends to cause later trouble. I've seen the same with USB: plug a 2nd identical webcam (certain Philips models) in to the same port number on a different different root hub, and the PC would reboot as soon as it enumerated - and keep rebooting until you unplugged one. – Chris H Feb 21 '23 at 16:18

10 Answers10

27

If everything else is the same then there is no technical reason to prefer a looser tolerance, so I think you are correct.

Of course there are situations where a particular manufacturer and part number may be required for regulatory or other similar reasons and no substitutes are permissible. And there are many resistor characteristics that are important in some situations but are not specified by the nominal resistance value + tolerance + power rating (or package). For example, a composition resistor may be specified because of superior power pulse handling, but they cannot be made in tight tolerances due to the way they are manufactured. A 1% film resistor of the same value, package size and power dissipation rating might fail in short order.

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • 4
    If, as it seems from the question, it was a walk-in at a brick and mortar store, your second paragraph seems very unlikely. – jaskij Feb 20 '23 at 20:00
  • 10
    @jaskij Yes, of course, it is included as a more general answer for completeness since answers here are expected to have longevity. – Spehro Pefhany Feb 20 '23 at 21:33
15

There might be a hidden specification in wanting 5% - the circuit might be relying on carbon film (typically 5%, "beige bones") vs metal film (typically 1%, "blue bones") resistors. Carbon film has different behaviour under temporary overloads (in a bad design, a metal film resistor might fail) and looks (which might be relevant if someone is restoring vintage equipment and wants their resistors as beige as they were).

rackandboneman
  • 3,085
  • 9
  • 13
13

If he truly did want 5% instead of 1% then he may have been trying to bin the resistors. That is, individually testing each one and selecting for a particular value.

Are you sure he did not say he needed 0.5%? Because no one says it is "only 1%" then proceeds to get 5%.

Otherwise you might prefer 5% if the resistor composition you needed simply did not come in 1%, but that's preferring one composition over another, not 5% over 1% and I would not expect anyone to phrase it as such.

DKNguyen
  • 54,733
  • 4
  • 67
  • 153
  • 7
    I once measured a number of Philips metal and carbon film 5% resistors, and found them all to be within 1 % of the marked value, so hoping to "bin" resistors to get values slightly off a standard value is a lost cause with modern components. – Peter Bennett Feb 20 '23 at 06:49
  • 2
    I remember seeing in more than one ancient circuit diagram instructions that imply what now is known as "binning" - e.g. "choose R4 and R5 to be as much equal as possible, ideally within 0.5%". The fact that one cannot get real 5% or 10% resistors does not stop some retro fans from looking for them. – fraxinus Feb 20 '23 at 10:42
  • 7
    Didn't lines of resistors exist where you were guaranteed NOT to be within 1% if you bought 5%, since those within 1% were binned at the factory and sold as 1%? – rackandboneman Feb 20 '23 at 15:14
  • @rackandboneman Those definitely existed in the past. Whether they still do, I couldn't say. – Hearth Feb 20 '23 at 16:12
  • 2
    @Hearth: There may have been a time where there would be no convenient way to order an 8.45K resistor, and so the only way to get a 8.45K resistor would be to order 8.2K and maybe 9.1K resistors and hope for a value close to 8.45K. Nowadays, however, if one wants an 8.45K resistor one can simply order one. – supercat Feb 20 '23 at 16:29
  • 5
    @rackandboneman not anymore, because the tolerance is expected to hold over the temperature range and the projected lifetime of the resistor. Resistors that are 5% are made with different technology and if you measure them new at room temp they may even be better than 1%. – fraxinus Feb 20 '23 at 17:05
  • I was led to believe years ago that 5% & 1% resistors all came off the same production line. The resistors that measured within 1% of the stated value were given the 1% designation, along with a higher cost. – SteveSh Feb 20 '23 at 19:40
  • @SteveSh I think it depends on manufacturer. – DKNguyen Feb 20 '23 at 19:47
  • Maybe, but I don't think so. I don't believe they have one line/process for 5% resistors, and another for 1%'ers. – SteveSh Feb 20 '23 at 22:29
9

There's no technical benefit. All resistors are ideally 0% tolerance. As they can't be, circuits should be designed for the part's worst value at tolerance. It's a design complication that leads to performance variance.

In the past, 5% resistors have been significantly cheaper than 1% resistors. That becomes important when buying in volume rather than handfuls. Nowadays, 1% resistors are the standard part and there's no cost benefit in 5% resistors.

My guesses here:

  • Your customer was buying replacement parts for an existing circuit and so wanted like-for-like, not understanding there were no downsides to a 1% part

  • Your customer was buying against a spec' or parts list and simply wanted the specified part

TonyM
  • 21,742
  • 4
  • 39
  • 62
  • In some cases, devices have regulatory approvals which are contingent upon using particular specified parts. I recall a semiconductor vendor discontinuing a lower-grade part except on a special order basis, and suspect that orders for the lower-grade part would have been filled with parts that failed testing for the higher grade but passed the lower grade, if any were available, and otherwise filled with parts that passed the higher grade. If all parts that pass any testing pass the higher grade, there would be no need to sell any lower-grade parts *except*... – supercat Feb 20 '23 at 16:14
  • ...to customers who need a part which is labeled XYZ123-3 (the 3 representing some particular grade which is inferior to any parts that are still being sold). If the company's policy has always been to fulfill orders for XYZ123-3 chips using parts that pass more stringent tests if no inferior parts are available, the fact that a company starts fufilling *all* orders for such parts in such fashion would not represent a change requiring new design approval. – supercat Feb 20 '23 at 16:21
  • 1
    I'm not the downvoter. I thought your second scenario painted a picture of someone who naively wants a particular part, rather than ignoring the fact that in some industries even the most benign substitutions require lots of paperwork. If a board is assembled with 1% parts instead of 5%, all of the documented inspection criteria will need be updated to accommodate that, procedures for guarding against counterfeiting may need to be reevaluated, etc. Issues that can be dealt with, but may be expensive enough that special-ordering parts labeled with poorer tolerance may be cheaper. – supercat Feb 20 '23 at 18:52
  • 1
    @supercat, no problem but scenario 2 doesn't anywhere suggest the customer's naive, just someone quite reasonably following the specified process in the released documentation ('simply wanted the specified part'). May be technically knowledgeable, may not but doesn't matter in that context. Had you misread the meaning there? – TonyM Feb 20 '23 at 19:01
  • I know you didn't intend it to suggest the customer was naive, but the phraseology "simply wanted the specified part" suggests that the customer's demand was a result of personal psychological desire rather than a need to satisfy regulatory requirements. – supercat Feb 20 '23 at 19:08
  • 1
    Sorry @supercat, I'm afraid you're really reading something into it that just isn't there and it's generating many long comments. 'Simply' means there's no other factors, no personal psycho-whatsit stuff :-D 'I simply want a sandwich' etc. – TonyM Feb 20 '23 at 20:26
5

For a random signal noise source, you may want noisier resistors that tend to come with larger tolerances because of different technology. Admittedly that's extra-handwavy because the additional noisiness of, say, carbon resistors comes from shot noise while a good random signal noise source would rather tend to rely on Johnson noise (which is more dependable).

Another handwavy reason might be that more precise resistors tend to be trimmed, and bein suitable for trimming implies an underlying structure that is more susceptible to stray inductivity.

Note that either of those reasons is really grasping at straws as an intellectual game.

The straightforward answer is "no", really.

user107063
  • 2,507
  • 1
  • 13
4

Customers don't know everything, I once returned a suitable product as unsuitable just because I don't know any better.

However customers are always right (tm)

With beer 5% is better than 1%, perhaps your customer was thinking along those lines.

Jasen Слава Україні
  • 31,874
  • 1
  • 31
  • 65
  • 3
    I recently learned that expression in your second paragraph was actually corrupted from the original: "The customer is always right *in matters of taste*" which to be honest makes a *lot* more sense to me. – Michael Feb 21 '23 at 00:05
  • I'm accustomed to low precision resistors having higher tolerances to overvoltage. – Joshua Feb 21 '23 at 03:10
2

depending on the circuit you are making, you don't need the precision. For example, if it is a current limiting resistor it will do the work either at 1% or 5%.

Honestly the main reason to prefer 5% over higher precision like 1% is cost, the other is power, it is very hard to guarantee very high precision for higher power components.

Edit: now for your specific case, I can't see a case where you would want less in general, if everything else is equal then yeah 1% is within the 5% and should work in all conditons.

Juan
  • 1,088
  • 7
  • 18
2

Of course agreeing with all other answers that generally say "no", I see those four cases:

  1. In analog musical circuits not-tight values may produce "the tone", when all loose tolerances in filters and other elements sum up, making one's tone less reproducable. Thus, even when you build a clone of your sound processor and give it to a friend, you two sound slightly differently. Though this does not apply to Hi-Fi equipment - rather artistic ones (e.g. guitar effects, synthesizers etc.).

  2. It was the point to measure them and prepare some report/statistics - for example for students classes.

  3. Preparation for bigger order: as 5%s are cheaper, he searched for best under that price label.

  4. Thinking about next serviceman: seeing 1% would think "why it is so tight and do I really need the same for replacement?". About half a year ago I had to replace 1200 uF capacitor in somehow regular switching power supply. As a hobbyist I was wondering for a while if I really need this value (I had only 1000 uFs in my box and live in a place without electronic store by the corner). I had used 1000 uF I had and since then owner of the equipment hasn't complained - and we are still friends! ;D.

The customer may have not mentioned that being sure that you will provide him with 5%s as defaults.

Marjan
  • 21
  • 2
0

People forget economics

Why would you want a 5% tolerance resistor? Because it's cheaper than a 1% tolerance resistor and your circuit can stand the variation.

A good example of such a circuit is a voltage divider where both resistors are 5% tolerance. Generally (not perfectly, but generally), resistors of the same tolerance will vary in value in the same way under identical environmental conditions. Thus, it's not worth paying for 1% tolerance when all you care about is the unitless ratio of the two resistances.

In this case, if the rebuilt board already had a 5% tolerance resistor, another 5% tolerance resistor would be required or the ratio would vary with environmental conditions, which can be bad.

Note: This isn't perfect, obviously, but I've not only seen this done, I've done it myself, especially in microelectronic design where you know for a fact that the manufacturing process of the two resistances is identical. Heck, we'd use short strips of polysilicon rather than actual resistors that would take up more real estate so long as they had the same orientation to guarantee the same processing encroachment.

I can't help it... I'm reminded of an old joke from my college days. A mathematician, a physicist, and an engineer are asked to answer the question, "what is 2+2?" The mathematician proudly answers "4!" The physicist smirks and says "4.00000!" The engineer looks a bit bewildered by their answers and responds, "any damn thing that works."

JBH
  • 594
  • 5
  • 11
  • 1
    4! = 24. Also, a related joke that 2+2=5 for sufficiently large values of 2. – Michael Feb 21 '23 at 00:08
  • @Michael Hah! That's a twist on the mathematician's answer. In the case of the joke, the exclamation point is literally that and not a factorial operator. – JBH Feb 21 '23 at 00:14
  • Apparently a good answer but how could it be worth the gas to return and exchange? – Joshua Feb 21 '23 at 03:11
  • @Joshua Without an explanation of the user's circuit, that's next to impossible to answer. But if the resistors were used in a current or voltage regulator, then the fact that the regulator isn't stable would justify the gas and time. ... And that's assuming your customer actually knew what he was doing, which I hope was the case, but might not have been. – JBH Feb 21 '23 at 03:13
  • But 1% resistors may be more economical. Our automated PCB assembly area had reels of standard value resistors (4.7 K, 5.1 K, 10 K, ...) and capacitors (0.1 uF, 0.01 uF, ...) on the pick and place machine. The resistors for the most part were all 1% film resistors. It would cost more to not use those 1% parts, and replace them with 5% tolerance components (that might require a hand operation) than to use the better tolerance, slightly more expensive parts. – SteveSh Feb 21 '23 at 13:17
  • @SteveSh No argument there, but remember that we're talking about the OP's circumstances, not all circumstances. Bulk resistors for PCB assembly are a different breed than individual resistors off-the-shelf. I don't know enough about it (I design integrated circuits), but the 5% resistors may not be available, or may not be as common at that level of bulk. It's why I don't say "always" in my answer. I don't even know how old the OP's customer's circuit was. It could be 20 years or more old, in which case the 5% would be required to maintain ratios. – JBH Feb 21 '23 at 15:01
0

Yes. If you are building a run of (say) 100 RC oscillators and you want them to have a spread of frequencies from (say) 95-105kHz. So you buy a lot of 5% resistors and select to get that range.

Of course there are better ways to do this, but it isn't an impossible scenario.

In other words, the customer is always right.

(Just playing devil's avocado here.)

danmcb
  • 6,009
  • 14
  • 29
  • You can't guarantee that a batch of 5% resistors is going to have any sort of distribution (Gaussian, uniform) across the tolerance range. – SteveSh Feb 21 '23 at 13:19
  • no, that's why you'd have to select from a batch. In fact you might have to buy a lot, because 10 adjacent resistors on a reel might be 4% away from nominal, but very close to each other. As I say, I am just playing the game here. Yes, there just might be a reason - although not a great one. – danmcb Feb 21 '23 at 13:56