Precision Resistors (.01% or better) - What are they good for?

These resistors go at Digikey for $6.60 qty 1 down to $3.44 qty 1000. What are these used for in industry? It seems they would make pretty good precision voltage dividers. But then again, less precise resistors would make good voltage dividers too if they were measured before going into a product and adjusting somewhere for the actual resistance of the resistor on a per-resistor basis, assuming their stability is good.

Just as an example, which would be cheaper? A $3.00 resistor or the labor associated with testing and selecting a large batch of resistors and then ‘manually’ soldering that resistor to a board.

When making one?

When making a thousand?

When making a hundred thousand?

You really can’t used automated assembly methods with manually selected components…

wanderson:
Just as an example, which would be cheaper? A $3.00 resistor or the labor associated with testing and selecting a large batch of resistors and then 'manually' soldering that resistor to a board.

When making one?

When making a thousand?

When making a hundred thousand?

You really can't used automated assembly methods with manually selected components...

Here probably the part.

In China, probably the labor.

Just a gesss. I bet you can test the parts at a workbench and bin them prior to soldering, assuming soldering doesn't change the value. The saving would add up very quickly at $3 a part saved if you are doing thousands. I guess it depends on the scale and what kinds of products actually need this precision. If it's only 6.5 digit DMMs then I agree with you, the quantites will just never be that high.

JoeN:
I guess it depends on the scale and what kinds of products actually need this precision.

My first thought would be medical instruments.

JoeN:
Here probably the part.

In China, probably the labor.

Doubtful, either would be more cost effective. Lower labor costs would yield lower product/distribution costs for the resistors, so that $3 resistor in the US may be a small fraction of that in China. Evidence that this is the case, is that China still uses automated manufacturing for electronics...

JoeN:
Just a gesss. I bet you can test the parts at a workbench and bin them prior to soldering, assuming soldering doesn't change the value. The saving would add up very quickly at $3 a part saved if you are doing thousands.

I believe your assumptions are faulty. One, I don't think you appreciate how much human labor costs, even in places like China. Cheap electronics exist because they can be assembled mostly/completely in an automated fashion.

Two, you can almost certainly not manually bin the parts for automated assembly. Indeed that is one ways hobbyists get cheap smd parts, leftovers from those large spools used in automated assembly. If it was cost effective, they would be reused.

JoeN:
I guess it depends on the scale and what kinds of products actually need this precision. If it's only 6.5 digit DMMs then I agree with you, the quantites will just never be that high.

Well, the demands for those itens are almost certainly reasonably high, since they can be purchased via non-specialized dealers like Digikey. They wouldn't be getting manufactured unless there was sufficient demand. And most likely you need them even in devices with relatively high error rates. The issue arises from cumulative error, which is almost always higher then (and can be much higher) than the error margins on the individual components.

But then again, less precise resistors would make good voltage dividers too if they were measured before going into a product and adjusting somewhere for the actual resistance of the resistor on a per-resistor basis, assuming their stability is good.

You already have the answer in your question. The less precise resistors have no guaranteed stability. Usually the precision is guaranteed for some temperature range. This is the reason why the more precise resistors are more expensive. They are manufactured differently to produce the better precision.

wanderson:
Evidence that this is the case, is that China still uses automated manufacturing for electronics...
Cheap electronics exist because they can be assembled mostly/completely in an automated fashion.

FoxConn alone employs 1.2 million people, almost all of them in assembly. I guess they are paid to watch the machines.

JoeN:

wanderson:
Evidence that this is the case, is that China still uses automated manufacturing for electronics...
Cheap electronics exist because they can be assembled mostly/completely in an automated fashion.

FoxConn alone employs 1.2 million people, almost all of them in assembly. I guess they are paid to watch the machines.

No, I would guess that they are paid, among other things to assemble boards into products... and for testing, moving boxes, etc... Can you find a manufacturer of large quantities that produce electronics boards by hand?

wanderson:

JoeN:

wanderson:
Evidence that this is the case, is that China still uses automated manufacturing for electronics...
Cheap electronics exist because they can be assembled mostly/completely in an automated fashion.

FoxConn alone employs 1.2 million people, almost all of them in assembly. I guess they are paid to watch the machines.

No, I would guess that they are paid, among other things to assemble boards into products... and for testing, moving boxes, etc... Can you find a manufacturer of large quantities that produce electronics boards by hand?

Surface mount, no. But I remember this video:

7:20 employees hand-mounting the through-hole components. At least I think that is where it is, I don't have audio here.

It still seems to me that if you wanted to pinch pennies you could use lower cost resistors if the stability was guaranteed. Another reply seems to indicate that it is not. It certainly wouldn't be with carbon film resistors, but metal film ones I would think would be very stable.

Remember Rigol overclocked (maybe still does) ADCs on their lower-end scopes and it worked just fine. It's not beyond Chinese companies to push the specs.

Anyway, this was not the main point of my post. I was wondering what the application of .01% resistors is. So far I have medical devices, maybe.

There is certainly applications for high precision components in industry. In the oil and gas chemical plant industry we would usually see 250 ohm precision (.1%, .05%, etc) resistors used in the instrumentation measurement equipment where the standard 4-20ma current loops from the field would be converted to 1-5vdc before they did a A/D conversion for the process control systems. Also some test equipment would use such high precision ratings.

Picking up a few such high tolerance resistors and caps (say from e-bay surplus sellers) can be very useful for hobbyist as 'lab standards' for checking the accuracy of your test equipment.

Lefty

retrolefty:
Picking up a few such high tolerance resistors and caps (say from e-bay surplus sellers) can be very useful for hobbyist as 'lab standards' for checking the accuracy of your test equipment.

Lefty

I came to that conclusion too. The whole thing that started this line of thought in my head was seeing a few eBay listings with these precision resistors. I picked up a few. I want to see if they are really within spec or not.

[/quote]

I came to that conclusion too. The whole thing that started this line of thought in my head was seeing a few eBay listings with these precision resistors. I picked up a few. I want to see if they are really within spec or not.
[/quote]

And how do you propose to do that - by measuring something supposed to be within 0.05% tolerance with a pice of test kit that MIGHT (at best) be within 1% - a difference of 20:1

Two examples come to mind.

DC to DC converters. Most of the circuits I have seen (which I will admit aren’t many) specced high precision (<1%) components for portions of the circuit.

Shunt resistors. Typically very small values, which imply higher precision. But futher, the voltage across the resistor will usually get amplified. So a 0.05% shunt resistor that gets an amplification of 20 for the voltage differential will produce a +/- 1% precision (0.05% * 20)… consider what a 1% or 5% resistor would do if they were even available in the values needed.

And higher precision components, due to their higher costs, tend to be manufactured to better quality.

author=jackrae link=topic=119797.msg901821#msg901821 date=1345743610]

I came to that conclusion too. The whole thing that started this line of thought in my head was seeing a few eBay listings with these precision resistors. I picked up a few. I want to see if they are really within spec or not.

And how do you propose to do that - by measuring something supposed to be within 0.05% tolerance with a pice of test kit that MIGHT (at best) be within 1% - a difference of 20:1

I was going to bring up the same point. The rule of thumb in instrumentation standards, calibration, validations, and check out is that your test equipment should have X10 the tolerance specifications of the things you wish to measure. So validating a resistor with a .01% tolerance would take a measurement meter capable of .001% accuracy. What these components offer to a hobbyist is a way to tell if their test equipment is operating up to specs, not the ability to validate the specification of the component.

Lefty

jackrae:
And how do you propose to do that - by measuring something supposed to be within 0.05% tolerance with a pice of test kit that MIGHT (at best) be within 1% - a difference of 20:1

I am going to measure them with a 34410A. Maybe your test kit is 1%.

I specified a lot of .1 and .01% resistors for op-amp feedback networks, Very frequently and not because I had any use what so ever for the accuracy of the part which was really important but the real reason was temperature stability.
Thermocouple scaling amplifiers are one case that comes to mind. I once had a problem with a Max666 in an old design where there was a voltage divider for a battery reference comparator and both values were common E12 series resistors a 6.8 Mohm and a 3.3 Mohm resistor. Until I changed to 10 PPM/Deg C resistors none would pass a temperature test without showing a bad battery. The other part of my answer was done already... hard to make 1% accuracy cheaply with 5% parts and trimming... Believe it or not but thermistors were a frequent part of a design just to compensate for temperature changes.It's Really harder to manufacture .01% accuracy test equipment with 1% parts, It can be done and has for many years BUT not on a real production basis... like cell phones or MP3 players or television receivers, Even with digital calibration done automatically with EEprommed constants and Digipots... It's still a difficult process.

Doc

One thing that fails with selection of cheaper lower precision resistors when really you need high precision results, is both long term and thermal stability. The two manually selected components make look perfect on the day that you match them, but will differ slightly once you solder them into a PCB, and progressively deviate from the original 'perfect match' with both passing time, and with temperature changes. That is why the precision parts cost much more than the cheaper components. Stability, both thermal, and life time, are difficult to achieve, and so come with a high price!

Docedison:
I specified a lot of .1 and .01% resistors for op-amp feedback networks, Very frequently and not because I had any use what so ever for the accuracy of the part which was really important but the real reason was temperature stability.

ipadwarrior:
One thing that fails with selection of cheaper lower precision resistors when really you need high precision results, is both long term and thermal stability. The two manually selected components make look perfect on the day that you match them, but will differ slightly once you solder them into a PCB, and progressively deviate from the original 'perfect match' with both passing time, and with temperature changes. That is why the precision parts cost much more than the cheaper components. Stability, both thermal, and life time, are difficult to achieve, and so come with a high price!

Thank you both, that is the answer that I was looking for. It is stability. I had a feeling it probably was but I really didn't know the answer there.

jackrae:
And how do you propose to do that - by measuring something supposed to be within 0.05% tolerance with a pice of test kit that MIGHT (at best) be within 1% - a difference of 20:1


:stuck_out_tongue:

JoeN:
FoxConn alone employs 1.2 million people, almost all of them in assembly. I guess they are paid to watch the machines.

1.0 million employees are there just to prevent next-gen iPhones from leaking.
0.19 million employees are chasing Gizmodo writers

That leaves about 10K employees watching the automated machines, packing, doing administrative jobs, HR, Mkt, Accounting, security, etc.

(sorry, but I just couldn't resist)