Watch Those 1% Resistors

Decades ago, electronic components were not as easy to acquire as they are today. Sure, you could get some things at Radio Shack. But you might not have many choices, and the price would be on the high side. TV repair components were another option, but, again, big bucks. Some places sold surplus parts, which could be cheap. These often came from manufacturing runs where a company bought 10,000 components and made 8,000 products. But today, you can order parts inexpensively and get them on your doorstep in a day or, sometimes, even less. Are these inexpensive parts really any good? [Denki Otaku] likes to find out. In a recent video, he checks out some Amazon-supplied 1% resistors to find out how good they are. You can watch his results below.

He starts with two resistor kits and examines them quite closely. For example, a magnet revealed that one brand of resistor had iron leads which are not as conductive as copper leads. They look the same because they are both coated with tin, but the magnet tells the difference.

The Elegoo kit measured the expected resistance value within the tolerance band. The sample size was a little small to draw a solid conclusion, but it does look like the resistor were well within their advertised tolerance.  The Longrunner resistors, however, did not fall in range. Some of the resistors were within 1% of the marked value, but others were as far out as 1.4%. That doesn’t sound like much, but if you bought them for their 1% tolerance, a 1.4% error is way off the mark.

We liked the test jig he built to do the four-wire resistance measurements more easily. Simple but effective. He did some experiments with hot air to determine the temperature coefficient and the power handling capability of the resistors. That led to an accidental smoke release. Later, he had more of a planned smoke release.

If you need general-purpose resistors that are within a few percent of their marked values, these kits are probably fine. But if you are relying on precise values, you might be disappointed. On the other hand, you would probably hand-match anything important, anyway, but it is still good to be aware that the markings on the resistors might not reflect reality.

We recently saw a case where being out of spec with cheap parts was much more dangerous. If you want to know more about 4-wire resistance measurements, you are in the right place.

54 thoughts on “Watch Those 1% Resistors

  1. “Decades ago, electronic components were not as easy to acquire as they are today. Sure, you could get some things at Radio Shack. But you might not have many choices, and the price would be on the high side.”

    Yup. And that was good. Because of this, hobbyist were being forced to use properly-sized resisters.

    Our local electronics shop had 1/8 Watts (abnormal, for miniature circuits) , 1/4 Watts (normal), 1/2 Watts (normal), 1 Watts (for PSUs etc), 2 Watts (for PAs, lamp controls), 5 Watts (ceramic, less used; used in TVs for tube heating)

    1. About accuracy.. There were at least two basic resistor types available, coal-layer and metal-film.
      The accuracy they had was higher than what a layman’s (hobbyist) measuring instruments would usually display. ;)

        1. It’s apparently not quite that bad anymore. It seems they now often make a value, then add the tolerance afterwards. This leads to an interesting effect. If you buy a bunch of 5% resistors, you’ll find it’s a nice bell curve. However it’s will often be chopped at 5%, rather than tailing off. It also has a lump missing, in the middle around the width of 1%/2% error.

          Basically a 5% resistor will often never be closer than 1-2%. A 10% closer than 5% etc.

          1. Cynar, That’s both fascinating and informative, I got an “Ah ha” moment reading your comment similar to ones I’ve gotten reading Richard Feynman.

          2. OR, what I see a lot in smaller resistors (0402 and smaller) is that the process is pretty much tailored to about .5%.

            Because the process practically guarantees the resistors to be .5% after manufacturing, 5% and 1% resistors are only sample tested. .5% resistors are fully tested.

            This is why in those resistors there’s not really any cost difference between 5% and 1%, and only a small increase in price to go to .5%. All of them are less than $.002 in moderate quantity.

          3. Patrick mentions the current method of manufacturing… build to a fairly small tolerance spec, and everything above is practically guaranteed (sampling is still done to ensure that’s the case).

            The bimodal Bell curve mentioned by Cynar was born out of how the components were binned. In that case, the tolerance on the components were quite a bit more loose (say, 5%), and binning had the effect of removing components from the middle of the curve. Every tolerance (except the tightest offered by the manufacturer) had their own bimodal curve as they were scavanged for the next highest tolerance. Just like with CPU binning, once they filled the bin for the tightest tolerance, they would continue binning for the next highest, and so on until they had the allotted number of units for sale in each bin. Definitely reduces cost for the manufacturer, and increases cost for the buyer as the manufacturer is now doing all of the testing.

      1. Because there’s no mentioning of the test conditions, or is it?
        I think there might be a small difference, if for example, the test was done outside in winter at -20°C or if it was done in summer at +40°C room temperature.

        1. The post talks about how the tester used hot air to determine the temp coefficient so it’s likely they weren’t testing in extreme and abnormal conditions. Or you know you could watch the video.

          Seriously, devil’s advocate is good and well but that’s a game of probability. And what’s more probable, that someone took the time to test 1% resistors for labeling accuracy, built a test jig to do so, did further tests to measure temp coefficient, filmed, edited, and published their findings, picked up by hackaday, and still missed an obvious confounding variable that the test environment was extreme conditions… Or that cheap mass produced products sold on Amazon are close but not 100% within spec of their labeling?

          1.4% labeled as 1% for a hobbyist product isn’t good but not end of the world and if someone is using hobbyist grade components in a mission critical application without doing their own internal QC, that’s on the hobbyist

    1. You have to remember that big A and E are just distributors. You need to pay attention to the actual source of the goods when you order from either. Both have reputable sources as well as ones that shouldn’t be selling anything to anyone.

      1. Digi-Key is “just a distributor” too but if a manufacturers were frequently selling out of spec resistors on DK, I’d expect them to do something about it. I would not expect the same from Amazon or eBay.

    2. There are no guarantees that the same sellers don’t operate on both platforms with different names, but for sure they sell stuff produced by just a bunch of manufacturers and sold with different names. This applies also to end products: their concept of brand is different from ours. About Amazon and Ebay, I’m not on Amazon and wouldn’t trust their mostly fake reviews anyway. Ebay’s feedback system is not perfect but makes life a lot harder for counterfeit sellers. But you have to choose carefully what to buy there as well; fakes and low quality parts are there too.

      Anyway, I’ve bought for over a decade on Ebay with mostly good results, but I avoided parts that are clearly fake because they simply don’t have them. Whatever the seller you buy from, here’s some advice regarding parts coming from the far east.

      Power transistors and mosfets: avoid; they’re mostly (sometimes like 99.9%) relabeled lower specs or reject parts. Sometimes, but you have to be really lucky, they are the real thing but salvaged from used equipment, with reconstructed leads.
      That is probably the case with some ready made boards and modules I’ve seen in which all parts appear new except some chips whose case and paint wear seems to date them a lot older than the board.

      Low power transistors and jfets: avoid. Same as the above, including those assortment boxes which most often contain the same 3 or 4 types relabeled in 10 different ways.
      Jfets will never be the ones advertised: you’ll never find a real BF245, 2N3819, BF256 or similarly famous types. If you’re lucky you’ll get a relabeled jfet, otherwise you could receive a bjt relabeled as such.

      Resistors packs: mostly ok; different power ratings and value steps, and they always have been within specs value-wise, but yes, they have thinner leads and probably shouldn’t be trusted in power applications.

      Electrolytic capacitor packs: avoid. Electrolytic caps are among the biggest single point of failure in modern electronics; get quality ones from reputable sellers and you won’t regret. Needless to say that Nippon Chemicon and similar ones sold from far east sellers are all relabeled fakes.

      Ceramic/poly capacitors: mostly ok. Values are consistent, never had a failure. As with resistors, they have thinner leads, and I couldn’t test them for noise or other parameters, therefore I’d say they’re ok for most breadboard saturday projects, but not for production or anything RF related or where very low noise is necessary.

      Chips: if they’re of Chinese production *and* design, they’re mostly safe; you’ll get the original because over there it costs nothing to them, and faking them would be bad PR for Chinese manufacturing.
      With US/EU/Japan chips it’s totally different however, and getting the real part is next to impossible. Almost all chips are either lower quality clones or fab rejects or even relabeled junk that wouldn’t even pretend to work once inserted in a circuit.
      Do not expect, for example, those 8038 function generators or those LM3886 amplifier boards to contain the real chip; they’re clones that work, but will never meet the original specs.

      Germanium transistors/diodes: stay away, they’re fakes. You’ll find much better deals and quality from eastern Europe sellers, and those are military grade ex soviet parts.

      Also, consider buying at HAM fests or in the surplus market; parts coming from there are usually of much better quality, and prices would be competitive.

  2. “If you need general-purpose resistors that are within a few percent of their marked values, these kits are probably fine. But if you are relying on precise values, you might be disappointed. On the other hand, you would probably hand-match anything important, anyway, but it is still good to be aware that the markings on the resistors might not reflect reality. ”

    There’s a German saying for this:”Perlen vor die Säue” (Perls in front of the pigs).
    It means that pigs are being fed with something more precious than the pigs themselves.

    In combination with this article, it becomes pretty obvious what I mean to adress.

    These guys use high-tech equipment and complain about accuracy, but the things they are finally create aren’t seldomly very inaccurate, imprecise or their soldering methods (I’m not using the word “skills” here) involve abuse of components. Like overheating components due to lack of knowledge, incompetence or due to using soldering baths and ovens. In short, situations in which an experienced electrician/technician would turn pale. And then, they have the guts to complain about less than half a percentage of accuracy. What a circus.

  3. In my 1972 Ladybird book ´Making a transistor radio´ (where I learnt the colour codes) the tolerance colours are Gold 5%, Silver 10%, Salmon 20%. I have never come across a Salmon 20% tolerance resistor but I guess they must have existed. In those days 5% was the normal tolerance.

    1. The Radio Amateur’s Handbook in that time listed 20% tolerance as no color band. Fortunately by then the technology had improved enough that even the sloppy carbon composition resistors were 10% or better.

  4. 1/4w 25 or 50ppm radial metal film rwsistors arw cheap enough that I just round up to the next price break when ordering from Mouser-key.

    I usually buy the common values for cheaper off the shelf in Akihabara though. 1% 50ppm metal film resistors in all the common sizes. Good made in Japan quality too, always test better than the rated 1%.

  5. I have been bitten by this more than 10 years ago.

    It turns out these “not quite 1%” resistors are actually carbon film instead of metal film. That gets really obvous when you heat them. Metal film has a PTC effect. Carbon NTC.

  6. I’m not so sure what this video is supposed to convey.
    Nobody in their right mind buys Ebay / Ali / Amazon resistors for “quality instruments”, but they are great for learning electronics, tinkering on breadboards and building all kind of “hobby” stuff.

    But it is good to give this topic some more attention.
    I had seen a similar video a few years ago, and when I bought a batch of 1k resistors from Ali, I measured a bunch of them upon receipt, and I would have given negative feedback if they were out of tolerance. Giving negative feedback is about all you can do for such “fake” parts.

    About the steel wires…
    This is quite common for resistors. Even resistors I bought in al local shop in the ’80-ies have steel wires. I actually prefer this. The resistance difference with copper is negligible (especially for any resistor > 1 Ohm)

    What is discomforting though is the remark from David Swanson on youtube. He claims that 4 out of 6 of the “Elegoo” resistors failed completely after use. But I don’t know how trustworthy that remark is.

    1. @ paulvdh I second that the steel wires are common. I just checked my inventory of resistors and about 80% have steel wires. My inventory is mostly 1990’s and early 2000’s american made quality products.

    2. I think the value here is showing that even the cheap components are more than serviceable for home projects. 5% and often even 10% resistors have been fine for most home projects. So it’s variable to know that you can trust the cheap white label resistors to be at least halfway accurate in labeling. If my circuit truly required 1% tolerances, I’d be doing QC on my own anyway. So it’s nice to know that instead of having to buy some “military grade” markup components , I can just pick up a lot from eBay of 1% and mentally panel them as 2% thanks to less than strict QC at the factory.

      That’s still way better than the 5% standard we got from radio shack

  7. heh this reminds me of two things.

    i have a bin of cheap BJTs. 2n2222 and 2n2907. i’m pretty sure i bought them irl from a reputable radio repair store at the turn of the century, a couple bucks for a blister pack with 50 of them, ‘jimpak’. i grew up digital so it was a new experience for me to make a circuit with 3 transistors in it (NPN and PNP in the same circuit: achievement unlocked), a circuit using a 555 to turn off a charger when a certain voltage is reached. it worked great and i was quite pleased but after i used it just a handful of times, it no longer worked. i don’t think i abused anything but maybe i did. anyways, it took me hours of poking around at it before i finally accepted the reality that one of the NPNs had simply stopped working (verified it by testing in isolation). replaced it with one of its brothers and it’s worked for almost a decade since. i’m convinced i was sold a transistor with a time bomb defect. you don’t have to use ebay or alibaba to get ahold of such a beast.

    the other is the dang blue pill. same for these cheap esp32 boards. people love these things. so fast! so cheap! they have a few flaws, but nothing you can’t work around with if you diligently peruse the aftermarket documentation. it redefines the hobbyist microcontroller market, effectively erasing the gap between the $2 pic and the $25 fully-featured stm32 “development kit” board. i disapprove…i want good documentation and a product with few flaws and a low amount of deviation from one unit to the next. i don’t think it’s a good deal that they’re so cheap you can buy extras and throw out the defectives. that’s why i love raspberry pico.

    but there’s no denying the value of cheap garbage, if you know what you’re getting into.

  8. The vast majority of resistors used today are logic level pullups. If I had to guess, I would say well over 99%, maybe even over 99.9%. They are as ubiquitous as decoupling capacitors. In both of those applications, the precision of the component just isn’t really a concern. If your 10K pullup is actually 9K or 11K, then the only real effect is a ~10% change in the current flowing through it and the resulting power consumed. In a 3.3V circuit, a 10% difference in resistance is about a 30uA difference in current, or about 120uW. The circuit will still function.

    I think it was Horowitz and Hill that made me realize that good circuit design minimizes the impact of component tolerances on functionality. Design such that as few as possible of the components’ values impact the signal, and put your money in the few that do. Design so that negative effects (such as aging degradation, or temperature) are common mode so they mostly cancel themselves out.

    We are spoiled today with 1% metal film resistors being considered standard. In the 1980s (the decade when the first edition of Art of Electronics was published), they were rather more rare (and expensive), with 5% being the common standard. Electrolytic capacitors were rather worse, and had significant degradation issues (they still do).

    Much analog circuitry has today been replaced by digital circuitry, proportionally reducing the need for accurate components. Telephony and communications for example. Digital signal processing means that signals that would have formerly been handled almost exclusively in the analog domain are now processed almost exclusively in the digital domain. Even in audio amplifiers, it is becoming more frequent to use Class D designs that minimize the number of components where precision matters. How many analog 555 timer circuits or 741-based measurement circuits have been replaced by Arduinos? (All of them, if past HaD articles are any indication…)

    Horses for courses. For most circuit prototyping or hobby use, availability rules, and an amazon kit is just fine – even if you do end up with iron leads in your resistors. For production you’ll go to a more reliable source anyway. If you’re designing high end oscilloscopes, or other precision circuitry, then you probably do need to worry a bit more during prototyping.

    That said, consumer fraud bothers me. I like to get what I pay for. If I make an informed decision about a purchase, I don’t want to be later surprised by problems caused by misinformation. This is a global problem though, not isolated to Amazon. Even reputable supply chains have problems with fake components, unfortunately.

    1. 100% all of this. As for the fraud, I don’t think these components fall under that category. They have bad QC, which is different from fraud that’s usually a malicious act. 1% labeled resistors sometimes measuring as 1.4% deviation shows that they are selling what they market, they just aren’t paying to ensure they’re QC is top notch. At worst they’re taking 2% batches and relabeling as 1%. That’s fraud, but given the price point and the uncommon use of 2% as a marketing segment (which goes back to us being spoiled these days for component quality), it’s not right, but it’s minimally wrong.

      These products are sold based on price, not performance and they come close enough to a spec that’s way tighter than it needs to be, especially at this price point. It’s like if you were sold flawless diamonds by a pawn shop for pennies on the dollar and got upset that, when tested, they were only excellent quality. Everything else was legit and excellent diamonds would sell for 90% of flawless, so they’re still pennies on the dollar good deals.

  9. I don’t know what to make of this video. As a PSA for “sometimes things aren’t what they claim to be”, it’s OK and has some value for people who might not have realized that.

    Beyond that, all his measurement proves nothing to me. It’s one kit from each vendor, precisely measured and all that, but so what? If he ordered the same things again, he might get the opposite results for both. The methodology is just not statistically convincing. It’s just a well-produced couple of anecdotes.

    1. I guess 3/4 of the video is an disguised advertisement for a EUR1800 Multimeter, and the resistor story is just used as a vehicle both to hide the advertisement, and to make it more “attractive” so it reaches a wider audience.

    1. Kovar: Alloy with about 50% Iron and with the intention to match the thermal expansion and bonding with glass:

      I’m not sure if this has advantages for resistors, but for glass diodes (such as 1n4148) this seems quite logical. I also had a long standing suspicion that it was not regular steel, because the attraction to magnets is weaker then I would have expected for normal steel.

  10. “Decades ago, electronic components were not as easy to acquire as they are today.”

    I was buying from Digi-Key in the early 1970s. Did you perhaps mean “centuries ago”?

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.