Is it possible to trim 0,1% resistor to 0.05% value?

[001]

Is it possible to trim 0,1% resistor to 0.05% value with about 20ppm for a long time?
Grinders, shunts, statistical tricks, etc?

[mikeselectricstuff]

Is it possible to trim 0,1% resistor to 0.05% value with about 20ppm for a long time?
Grinders, shunts, statistical tricks, etc?

Trim with other 0.1% resistors in series and/or parallel

[try]

Hi 001,

Is it possible to trim 0,1% resistor to 0.05% value with about 20ppm for a long time?
Grinders, shunts, statistical tricks, etc?

it depends on the resistor value that you left out.
Try to trim 10M Ohm for instance. 

What is "with about 20ppm for a long time"?
Do you mean 20ppm drift over "a long time"?
Or do you rather mean trimming from 0,1% to 0,02% accuracy?

You might consider mentioning more quantitative terms. 

[Wolfgang]

You may try, probably by the following method:

- Buy a bunch of 0.01% resistors and screen them for initial tolerance (you have a 3458A, I suppose). Sort out the ones that do not have enough precision
- Then, put them into a climate chamber and temp cycle them from 0 to a bit below their max. temp for a year.
- Sort out the ones with too much tempco.
- Redo the screening. What is still within 0.005% is probably safe.

... OR  just buy them von Vishay right from the start. Probably a lot cheaper and you can start working with them tomorrow.

[iMo]

... OR  just buy them von Vishay right from the start. Probably a lot cheaper and you can start working with them tomorrow.

How the Vishay does it then? Such an resistor must cost you an arm and a leg..

[Wolfgang]

If you look at their datasheets they use laser trimming.

For the 10ppm grades, I am dead sure that they use screening and aging as well.

Measuring those is a challenge even for a 3458A. DC voltage precision with option 2 is 4ppm. Resistor precision is less than that, depending on value.

Link to datasheet:
http://www.vishaypg.com/docs/63146/vhp203.pdf

I did not find a price for them. A 50ppm one is a few 10€ per piece, I guess the range for a 10ppm resistor is a few 100€.

[Wolfgang]

Hi, I found a quote at Newark:

https://www.newark.com/vishay/y60721k00000s0l/vhp203-1k0000-0-001/dp/51X5889

a 10ppm 1K resistor costs about 100€ in single quantities. So its an arm and leg, but the rest stays intact.

The question of the day here is: Which meter can measure this thing with a precision of 1ppm or less in (4-wire, of course) resistance mode ?

[Vgkid]

If you want accuracy you will need a esi242d+ standard resistors(national metrology lab /primary standards lab certified) , or a dcc(direct current comparator)+ same standard resistors. I haven't looked into using a long scale dmm+ standard resistor method.

[splin]

Hi, I found a quote at Newark:

https://www.newark.com/vishay/y60721k00000s0l/vhp203-1k0000-0-001/dp/51X5889

a 10ppm 1K resistor costs about 100€ in single quantities. So its an arm and leg, but the rest stays intact.

The question of the day here is: Which meter can measure this thing with a precision of 1ppm or less in (4-wire, of course) resistance mode ?

Wrong question methinks - Newark aren't guaranteeing anything - when exactly was the resistor measured to 'within 10ppm' and how much may it have drifted by the time you receive it?

[Brak]

Applicable post by Edwin G. Pettis:

Precision resistor specifications and definitions.

https://www.eevblog.com/forum/metrology/precision-resistor-specifications-and-definitions/

Brak

[Wolfgang]

... they have a shelf life spec, and I suppose you will get a certificate when buying one of those.

[macboy]

Is it possible to trim 0,1% resistor to 0.05% value with about 20ppm for a long time?
Grinders, shunts, statistical tricks, etc?

Since you mentioned statistical tricks,
If you simply wire up N of them in a series-parallel configuration, you get sqrt(1/N) of the error, so four resistors of 0.1% gives you 0.05%. This assumes a normal distribution of error around the nominal value.

[mzzj]

The question of the day here is: Which meter can measure this thing with a precision of 1ppm or less in (4-wire, of course) resistance mode ?

Most of the "serious" temperature calibration labs have toys to measure resistance to around 1ppm or better within limited resistance range.
Fluke 1595, or various MI or ASL(isotech) bridges are the usual toys with accuracies in 0,2 to 0,02ppm range.

http://mintl.com/productcategories/metrology/resistance-measurement/resistance-bridges/
https://www.isotechna.com/Thermometry-Bridges-s/35.htm

We got two older Hart/Fluke 1590's that are good to about 1ppm. Reference resistor is almost another 1ppm, but if I really push the uncertainties I would probably end up at about 1ppm total.    (not for 0,1 ohm resistor tough!)

[001]

from the internet "It's well known that the cheapest way to get an accurate resistor is to put many together; ten standard-grade metal film 10kΩ 1% resistors in parallel gives a 1kΩ resistor with a 0.3% tolerance for less money than a 0.5% part would cost."

WHY? 

[feedback.loop]

from the internet "It's well known that the cheapest way to get an accurate resistor is to put many together; ten standard-grade metal film 10kΩ 1% resistors in parallel gives a 1kΩ resistor with a 0.3% tolerance for less money than a 0.5% part would cost."

WHY? 

As macboy explained above, the error is sqrt(1/N), which in this case of 10 resistors of 1% will be 1/sqrt(10) ~= 1/3%

This is a simplification of course. This is assuming that individual resistors have independent distribution of error, which is not really so if they come from the same batch or even the same manufacturer. There might be some systematic factor involved.

[GigaJoe]

it works , if the value normally dispersed , from my practice not at all,  mostly under.  means for a set of 20 10K resistors, only 2-3 will above 10k , the rest below ...
I'm usually building myself,  trimming surface by filers, heating for couple days at +160 for oxidation,  and many layers of conformal coating .....
some low tempco ( like 25 -50 ppm)  often has positive and negative tempco curve, in the same set (spin, roll, package, batch, etc)    possible to combine resistors, for a significant TC elimination.

[try]

Hello 001,

from the internet "It's well known that the cheapest way to get an accurate resistor is to put many together; ten standard-grade metal film 10kΩ 1% resistors in parallel gives a 1kΩ resistor with a 0.3% tolerance for less money than a 0.5% part would cost."

WHY? 

pricewise this is currently not true in my country.

Each time you sum something up the outcome is converging against a normal distribution.
This is obvious when looking at the formula for paralleling 10 resistors.

A simple way to understand it is the following:

Throw a dice once and write down the result.
Repeat this procedure a 100 times. How often do certain numbers appear?

Throw a dice two times, sum up the result.
Repeat this procedure a 100 times. How often do certain numbers appear?

Throw a three two times, sum up the result.
Repeat this procedure a 100 times. How often do certain numbers appear?

What happens with the sums?
Go and count which sum appears how many times and plot that on a graph.


By the way, if you ask questions in other threads you might consider answering people when they ask you.
It's disappointing for people that spend time on your questions when you don't react anymore.
Otherwise they might not take your questions seriously in the future and stop answering you.

Regards
try

[e61_phil]

... they have a shelf life spec, and I suppose you will get a certificate when buying one of those.

You don't get a certificate for this resistors (normally). I paid ~70€ for a 10ppm tolerance VHP101.