Manufacturing Process of Resistor

[onemilimeter]

Resistors are graded with different tolerance, e.g. 10%, 5%, 1%, 0.1%, 0.01%, etc. Do you think resistors with different tolerances are manufactured with different processes? Or, all resistors are manufactured using the same process and they're sorted to be in different tolerance classes.

[Zero999]

Good question.

I know that carbon film resistors only come in 5% tolerance or worse because they're inherently exhibit poor drift and stability characteristics so there's no point in making them any better.

At the low end, i.e. 1% and 0.1% probably not, there's just better process control on the narrower tolerances but the really good resistors probably use different processes entirely and are binned. It also wouldn't surprise me if some manufactures sell 1% resistors as 5% so they can shift more stock.

[alm]

At the low end, i.e. 1% and 0.1% probably not, there's just better process control on the narrower tolerances but the really good resistors probably use different processes entirely and are binned.

More accurate resistors often also have a lower tempco and drift, so I doubt that it's just binning.

It also wouldn't surprise me if some manufactures sell 1% resistors as 5% so they can shift more stock.

Some chinese vendors also sell 5% resitors as 1% .

[onemilimeter]

Some chinese vendors also sell 5% resitors as 1% .

Digikey also did that... as shown by Dave in one of his videos.

[onemilimeter]

Is it really matter to have resistors with very low tolerance (e.g. 0.1% or 0.01%) in circuit design? For example, in designing a differential amplifier...

[Neilm]

Manufacturers of some of the lower tolerance resistors will laser trim the resistor. This is particularly noticable on precisiion high power thick film resistors.

Neil

[Simon]

Is it really matter to have resistors with very low tolerance (e.g. 0.1% or 0.01%) in circuit design? For example, in designing a differential amplifier...

it depends on what your doing and the acuracy you want. If your playing with audio you will be happy with 5% usually. If your making some test equipment or some other precision circuitry you will want as low as you can afford and unfortunately no money can buy 0%

[allanw]

In differential amplifiers you'll want both halves of the circuit to match up as best as possible to maximize the differential gain, so better tolerance resistors are obviously better. However at a certain point other errors will have a bigger effect. You could put your circuit in a SPICE simulator and play around with tolerances to see what effect it'll have.

[tecman]

Most resistors are graded after they are made.  Some manufacturers only do a sample check of each batch as a check on tolerance bands.

As stated, carbon film are generally restricted to 5%.  Better than that is the realm of metal film.  By keeping the deposition process under tight control, the tolerance band can be predicted to a tight value.  Thus the batch sample testing is statistically significant and is most often used.

Laser trimming is generally the realm of less than 0.1% and although a fully automated process, it adds significantly to the cost (relatively speaking).

paul

[Zero999]

it depends on what your doing and the acuracy you want. If your playing with audio you will be happy with 5% usually.

It depends on the part of the amplifier. For example if it's an op-amp pre-amplifier with a gain of around 20 it doesn't matter but if it's part of a bridged amplifier it might be more critical to minimise the DC in the speaker coil, which is bad.

Also be careful as lots of cheap and nasty 5% resistors are noisy carbon film which is bad, unless you're building a valve amplifier lol.

[Simon]

Yea, these days I aim for 1 or 2% resistor by default

[onemilimeter]

In differential amplifiers you'll want both halves of the circuit to match up as best as possible to maximize the differential gain, so better tolerance resistors are obviously better. However at a certain point other errors will have a bigger effect. You could put your circuit in a SPICE simulator and play around with tolerances to see what effect it'll have.

I read this as well somewhere. Is there a way to get around this design issue without the need of "tight tolerance" matching resistors?

[Simon]

not really, say you use 5% resistors, you have two gain feedback loops and usually they are matched, say on one feedback loop you have a resistor that is +5% and the other is -5%, that's a 10% drift in your desired value, now suppose it works out that one branch goes +10% and the other -10%, you now have a gain different in the circuit of 20% from the intended value. It's up to you wether that is acceptable or not.

[alm]

I read this as well somewhere. Is there a way to get around this design issue without the need of "tight tolerance" matching resistors?

Sure, just lower the CMRR requirements. The tolerances for matching are proportional to the CMRR.

[Zad]

I am designing some test gear at the moment, for the majority of components absolute tolerance isn't vital but I have been quite surprised how cheap 1% resistors are. If I go for 0603 (urgh, I hate anything below 0805) then they are generally between 1.1 and 2.3p (£0.011-£0.023) each for quite obscure E192 values. It is amazing how many inaccuracies you can calibrate out with some smart software.

1% seems to be the "new" 5%. Having said that, I did some spot checks on (new, big brand name) 5% and 10% components the other day, and the majority were better than 1% tolerance, which might imply that they weren't tolerance binned.

[onemilimeter]

Resistor sorting machine

http://www.electrosort.com/chip_resistor_sorter.htm

[AdShea]

If you go for the really high precision (and yes, sometimes you do need it) they laser trim each one and have fancy parallel paths on the film.

http://www.vishaypg.com/docs/63113/z201.pdf

We're using some of these in my lab because they're about the highest precision and absolutely the best stability you can get without building it yourself or buying crazy one-off equipment.

[onemilimeter]

If you go for the really high precision (and yes, sometimes you do need it) they laser trim each one and have fancy parallel paths on the film.

http://www.vishaypg.com/docs/63113/z201.pdf

We're using some of these in my lab because they're about the highest precision and absolutely the best stability you can get without building it yourself or buying crazy one-off equipment.

Thanks. By the way, how much does this high precision resistor cost you?

[AdShea]

It's about 20USD each on digikey.