Clarifying ppm/deg C for resistor tempco

[John B]

I'm gathering some tempco data on the no-name chinese resistors I have laying around to see just how bad (or OK) they are.

I have already gathered some test data points on a particular resistor, which shows a fairly linear, negative gradient of resistance vs temperature, for the range of ~25 to 120 deg C.

I understand ppm to represent 1/10^6 just as % represents 1/100.

However, when quoting a resistors tempco as ppm/deg C, that would imply a deviation from a standard starting temperature?

My ppm calculations so far have been verified by ppm calculators, such as vishay's one: http://www.vishay.com/resistors/change-resistance-due-to-rtc-calculator/

However they have the field to enter a reference starting temperature. So depending on what you set as your reference temperature, the ppm/deg C will vary with the same linear data points.

Is there an accepted standard temperature? For example in Dave's Wekomm resistance standard video, there is a handy chart showing 23deg C as the reference measurement. But in datasheets Ive read, the only measurement is ppm/deg C.

So, if there is a standard temp, I will need to linearly interpolate the resistance for that standard temp. Keep in mind I don't have fine control over the temperature, and I can only heat, not cool.

Further more, the weather where I am has varied by +25deg C/ 77deg F over the past week....Crazy. 

[neil t]

To my understanding it's rather arbitrary, but if you check data sheets there seems to be a 25 deg c measurement point , this temperature seems to be consistent across most data sheets. I'm guessing 25 deg c is considered "room temperature" climate controlled.

regards Neil.

[Vgkid]

For standards it is 23C, was also 20C in UK. As said earlier, it is arbitrary.

[JS]

  Usually ppm/ºC is used as a maximum value the resistance would drift with a variation of 1ºC in the operating (or specified) range.

  For example, it could be 5ppm/ºC (15ºC to 50ºC) and 20ppm/ºC (-40ºC to 125ºC)
In that case you know that from 25ºC to 26ºC it won't change more than 5ppm of whatever value it had at 25ºC with and from 90ºC to 91ºC won't change more than 20ppm of whatever value it had at 90ºC

  So you see it doesn't really matter the reference temperature.

  There are other cases (PTC, NTC, etc) where ppm/ºC is the spec you are after and you do know the sign and the tolerance it has. This is not your case, as you wan't to know the error you resistors would introduce, which is normally not linear and somehow lower at certain point, compensated components have "zero" TC at a certain point and would go both up and down independently if the temp raises or falls.

  A reasonable way to estimate this parameter in the home shop is to measure the resistor and heat it 20ºC, measure again, temps around the desired operating point of course, then if you got a 0.3% (3000ppm) between both measurements you know your resistor is around 150ppm/ºC. You are in the ballpark, it could misbehave and not change at all in some particular range and change twice as much at some other point but usually is a good estimation.

  I hope this clarifies your understanding on that particular topic, is a very important thing understand datasheets and to do so is a good thing to know how those parameters are measured (obtained) and how could you obtain them at home if you don't have them, or even check them if you do have them.

JS

[John B]

I think that clears up what you would be given on a datasheet. When you say: 5ppm/ºC (15ºC to 50ºC) and 20ppm/ºC (-40ºC to 125ºC), that would mean +/-5ppm/ºC of the resistors specified value, eg 1k ohms, between 15ºC to 50ºC.

When it comes to measuring individual resistors, I guess you have to measure it at SOME temperature, therefore you cannot just list the value on the bands.

For example, I measured a "10k" resistor as 9.827k at 25ºC (its a "1%" resistor - lol)

The data points are quite linear up to 114ºC where it measures 9.6065k

By my calculations, the change in R is:

(9.6065k-9.827k)/(114-25ºC)= -2.48 ohms/ºC approx

This value as a ppm of the resistance at T=25ºC is:

-2.48/9827*10^6 = -252ppm/ºC approx

So to accurately characterise this single resistor, I would specify it as 9827 ohms @ 25ºC  -252 ppm/ºC but I don't know if that's an accepted way.

Ill have to measure a few resistors from the same batch to see if the tempco is roughly the same even if the initial values are different. If you think of it as a linear equation, the different resistors would have the same gradient but different constants.

[wraper]

For example, I measured a "10k" resistor as 9.827k at 25ºC (its a "1%" resistor - lol)

Either faulty, counterfeit or not 1%
...Or you were holding both leads with your fingers while measuring.

[John B]

It's all on a test jig, so no hands. As per my first sentence, I am trying to see how bad they are. Low expectations, low disappointment.

I've got a few batches of "1% metal film" resistors from ebay china (they're different shades of blue on the body), plus some locally sourced 1% metal film resistors which are showing much better specs.

I may have some data to release in a few weeks. The question arises sometimes whether its worth getting cheap ebay kits of resistors. Hopefully this will help to decide when they are ok to use in a project, and when they're not.

[wraper]

The question arises sometimes whether its worth getting cheap ebay kits of resistors. Hopefully this will help to decide when they are ok to use in a project, and when they're not.

Chinese 1% metal film resistors may easily actually be carbon film crap.

[tszaboo]

What you should first understand, is the so called "box method". It is not guaranteed, that for a 100ppm/K tempco resistor, that it will change less than 1000ppm for a 10K change. It is guaranteed, that between -40 /+125 degrees, it will stay in a box of 16500ppm. It can reach it's maximum at 50 and drop down till 70 and go up again, as long as it stays in the box, that is fine (well, according to marketing anyway).
Now, resistors are simple devices they usually dont do this, unlike voltage references for example. Some high accuracy resistors will specify better tempco for reduced temperature range.
And then there are the lying datasheets, like VPG (or whatever they are re-named now) which will say 0.003ppm/degrees typical tempco on the first page in huge text, and then say that it is actually +/-3ppm.