# Why are we still using resistors with %5 tolerance while they can even manufacture a 14.318182MHz crystal?

Year is 2012 and I can only find %5-tol resistors in the local market. They can make transistors at molecular scale, they can manufacture 14.318182MHz crystals, they can place trillions of flip-flops inside a memory chip.

Then why don't they start manufacturing %0.01-tol resistors? Is resistor manufacturing a more difficult job compared to the ones I mentioned above? What is the reason for still manufacturing %10-tol and %5-tol resistors?

(I'm asking this because I learned that the following circuit may not work because the resistor values may differ greatly from the rated ones.)

• They can make 0.01% resistors, but it is more expensive. Here is a 0.005% example – Oli Glaser Sep 2 '12 at 17:08
• The circuit should have positive feedback for hysteresis. – stevenvh Sep 2 '12 at 17:08
• @stevenvh The physical system will have a thermal capacity. Do I need an electronic setup for hysteresis? – hkBattousai Sep 2 '12 at 17:15
• @hkBattousai - Not if that opamp is a comparator with hysteresis built-in. Otherwise yes: noise may cause the output to oscillate around the threshold. – stevenvh Sep 2 '12 at 17:21
• The question looks like a wrong observation. There possibly is no manufacturing of new 5% resistors anywhere for last few decades, except current sense shunts for milliohm, microohm range. – user924 Sep 2 '12 at 20:51

One more point worth considering: Maybe there's a problem with the local market?

In my local market, I have no problems getting 1% resistors and sometimes there's a larger choice of 1% resistors compared to 5% resistors. It's not always the question of can it be made but will people buy it too. Maybe your merchants for some reason believe that not enough people will buy 1% resistors, so they don't bother having them in stock (Basically what's it worth to them to have a part in stock when others sell well enough?) or they may be just lazy*. Maybe very small amount of people actually expressed their desire to use such resistors. Maybe people are so used to 5% resistors that they don't feel the need buy more expensive resistors since they haven't actually had the chance to see them in action.

Perhaps there's a non-obvious way for those resistors to enter your local market? Here where I am, we have companies that specialize in obtaining components which nobody else has in stock in amounts low enough so that working directly with foreign distributor would be too expensive.

Since we know that 1% and better resistors are commonly available in some parts of the world, the reason could be something specific to your market.

*For the end a short story about human nature possibly related to this issue: I lived in another country for several years and found there a brand of printers that I like very much. When I returned to my homeland, I noticed that nobody even heard of that brand. It so happened that I stumbled upon the office of the distributor for that brand and talked to them for a while. I was basically told that they're not expanding since they already have enough customers to sustain their company and that they don't want to bother having more customers than it's necessary for them to continue existing.

0.1% resistors are widely available. Digikey lists 59,000 part numbers. But the price is higher, like @ $0.04 in reel quantities, instead of$0.001 each for 5% tolerance.

If your design needs high tolerance, and your market isn't sensitive to a few dollars of price difference, there's absolutely no reason not to design with tighter tolerance than 5%.

At my previous job, we used 1% resistor tolerances as a standard, figuring that it was cheaper to spend a tenth-cent or so extra per resistor, rather than do the extra engineering to be sure the extra tolerance is acceptable (or to skip it and end up shipping bad products).

But in other markets, a few cents per resistor (say you have 1000 resistors in a design, those pennies add up) does make a difference. Also remember that any cost difference in the BOM gets multiplied up by a few times by the time the product gets on the shelf at Best Buy.

The basic answer to your question is that for 99.9999% of the applications of resistors, the improved tolerance would have no value. Circuits are generally designed to work just fine with 5% and 10% resistors.

In the specific example you show, it really isn't the absolute tolerance of the resistors that's important, it's how well they're matched to each other. You can indeed purchase matched resistor arrays (and tight-tolerance resistors) for such applications. Needless to say, they're rather more expensive then the jellybean 10%, 5% or even 1% parts that are more commonly used.

That's also why monolithic instrumentation amplifiers are valuable in such applications. They have all of the matched resistors integrated into the chip, and they're constructed so that all of the thermal, process and geometry variations (mostly) cancel out, so they're very well-matched indeed.

• "Back in the day" (vacuum tube era) 20% was the most commonly used value. For many commercial circuits to work, all that was required was that a certain voltage be "large" compared to some other voltage. This is still essentially true in many cases. – mickeyf Sep 3 '12 at 0:54

You are mistaking the number of digits in the specification of the nominal frequency of the crystal for its tolerance.

An inexpensive 14.318182 MHz crystal is not accurate to single-digit Hz. Those in the Digikey catalog are rated at between 10 and 50 parts per million, which is to say they are specified to have an error of +/- 143 to 716 Hz depending on which one you pick.

Just as with resistors, the tighter tolerance you want, the more you will pay. And at a certain point, the specified tolerance can only be achieved if you use it in a temperature controlled environment matching the calibration temperature - this is fairly common with precise crystal oscillators, such that you can buy "oven" crystal oscillators which include a heating element and control circuit for it.

You would also need to match the load capacitance for which the crystal is designed: conversely, by varying this you can "pull" the frequency of the crystal a few KHz - non-linear, but enough to have been historically useful for giving some frequency choice in crystal-controlled narrowband morse-code amateur radio transmitters, without the more substantial calibration and stability problems of an LC variable frequency oscillator.

While 5 % resistors will be fine in a lot of circuits I tend to use 1 % resistors. The word is reproducibility.

Murphy's Law: tolerances will work hand-in-hand to bring the system as far as possible from its stable setpoint.

The reason for the 5% is that the resistor values in the series then overlap in value. For example, the E24 series resistors have a 5% error rate, so that one resistor with +5% slightly overlaps the next resistor value with a -5% value.

So is there also a E12 series with a 10% error and a E48 with a 2% error. This is a good article about that: http://www.logwell.com/tech/components/resistor_values.html

• I am pretty sure the values in the series were chosen based on the tolerance rather than the parts being designed with a tolerance to match up with the values in the series. In fact, if you buy a batch of 5% resistors, there's no guarantee they'll actually be spread over a 5% range. Quite likely they'll actually all be clumped within 1% or so of each other. – The Photon Aug 21 '14 at 22:21
• Of course no company is going to set out to make resistors that are intentionally 5% off the intended value. They'll make a bunch, and test them. If they are spot on, they go into the tightest tolerance pile. If they are within 1%, then that's the bin they go into. And if they are within 5%, they go into the 5% bin etc. As The Photon alluded to, all the resistors manufactured together in one batch will probably be very close together in value and end up in the same bin. – tcrosley Aug 22 '14 at 0:15