EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: rvalente on April 05, 2019, 05:26:29 pm
-
Hello Mates,
I`m trying to find a similar thermistor for my fridge (Bosch KDN43), from a similar service manual (which use the same sensor probe) I've found these values:
The B25/100 constant should be the same for B-10/40 or B0/30? This is not clear to me, I`ve found the below constants:
Well, I`m clearly missing something, can someone help me?
Points from service manual
T °C R
40 3000
35 3500
30 4100
25 5000
20 6000
15 7800
10 9300
5 10700
0 12800
-5 16000
-10 24000
-15 34000
-20 45000
Constants:
β25/-20 3681
β40/-10 3424
β25/0 3059
β25/10 3489
β25/20 3184
β25/30 3584
β25/35 3274
β25/40 3176
β40/-20 3574
-
3578
-
Can you please explain me how you got to that number?
-
Right, beta is an incomplete description of the real behavior of NTC thermistors. So it varies with what points you pick.
Anything with a value in that neighborhood will do, and you can check datasheets for actual curves/data to find one that matches best.
Tim
-
Right...
so in this case, I have only these points. Which of them should I pick to find a close replacement?
-
To put that in other words:
Your part's beta varies with points chosen.
Any part you would replace it with, its beta also varies with the points chosen.
The two most likely vary in different ways.
Because, again, there are more degrees of freedom here. Beta alone is an incomplete description. (It's actually the first element of a mathematical series approximating the value. The more parameters used, the closer it gets. Sadly, almost no one gives those parameters.)
In still other words, you might pick, say, B0/25, and look for only parts with the same e.g. 3059 +/- 1%, and select a dozen parts that match. On checking their data, you will likely find that they all have different curves, that only coincide with yours at exactly 0 and 25C. And the spread between curves, away from those two points, will be comparable to the spread seen between different B values (i.e., more like 10%).
So what you need to do is, select a range of B values (say 2900-3500?), read through each and every one, and find the one that happens to coincide most precisely with your data.
Yeah, tedious as hell. Welcome to shopping for imperfect components. :-//
Tim
-
Can you please explain me how you got to that number?
From your temperature and resistance table.
beta = ln(R_T1/R_T2)/(1/T1-1/T2)
T1,T2 temperature in K
R_T1 resistance at T1
R_T2 resistance at T2
-
Got it, I never though the simple LN equation was a math aproximation, but the math accurate description for the thermistor.
Your explanation was very helpful!
-
The variation between thermistors is crucial for those making thermometers or temperature chambers. Less so for refrigerators. Most people just turn the cooling control up or down a little to get the results they like without paying too much attention to actual temperature.
That said, in picking your replacement thermistor you obviously care more about how it fits from about 10 degrees down to about -20, those being roughly the usual temperatures for the refrigerator compartment and the freezer compartment. So I would look for a fit in that range. If you go ahead and plot the points and the beta curve you will find that beta is a poor metric in that temperature range. Fits at the two endpoints give bad errors in between.