Converting a NTC voltage into temperature in Celcius

Use a lookup table in software, major points, then interpolate between them, to required accuracy.

Chris

I have thought of this but I thought it was on the market !

I don't know of any. The Steinhart-Hart relationship has been around since 1968

Getting the parameters in the first place needs work, so the "cheap converter" wouldn't be cheap to use.

The look-up table approach with linear interpolation between a limited number of points is easier, and thermistor suppliers do published tables of resistance versus temperature, but the user would have to get the tables into the "cheap converter" which takes skilled labour, which isn't cheap.

The usual method of approximate linearization is to put a carefully chosen resistor in parallel, which may or may not be good enough depending on what you're doing. (See e.g.

A positive resistance works with an NTC, because its resistanance vs temperature curve is concave upward. A metal RTD's characteristic is concave downward, so you need a negative resistance for the job. Because its nonlinearity is smooth and gentle, you can do an amazingly good job that way--theoretically under 1K error from -100 to +150C iirc, and much closer over narrower ranges. (I designed a couple of those in my misspent youth.)

You wouldn't bother nowadays, since it's going into an ADC anyway, and code is much cheaper than op amps.

Cheers

Phil Hobbs

I tried to do that in the 1970s to control the gain of an APD with temperature. I ended up using a diode as a temperature sensor as I couldn't get it linear enough.

Later on I wanted a non-linear control for another application.We tried OP amps with piece-wise shaping circuits. We concluded however the best way to do it was a PIC with a built in ADC and DAC and a lookup table as you suggest. That might be the cheapest way for the OP

I've also used AD590s which are pretty good.

Brian

Linear-mode APDs are much tougher, I agree. For one thing, the nonlinearity is much stronger.

For another, the high bias voltage leads to a lot of dissipation which makes the die temperature fail to track the board temperature. And of course currents are less convenient to compensate than resistances.

Cheers

Phil Hobbs

Interesting. Thank you for sharing. When operated in Zener mode (as opposed to avalanche mode) a Zener diode's NTC apparently remains approximatly the same, regardless of orientation (forward or reverse biased).

Danke,

Whether a reverse biased diode breaks down in Zener mode or avalanche mode depends entirely on the diode you have selected.

if it breaks down at less than 5V reverse voltage it is breaking down by the Zener mechanism (with a negative temperature coefficient) , and if it the breaks down at an 8V or a higher reverse voltage the avalanche mechanism is dominant (with a positive temperature coefficient.

Am 11.04.25 um 08:13 schrieb Bill Sloman:

Noisewise, the mode is easy to be seen. The avalanche effect is MUCH worse. Already small avalanche contributions spoil the pot.

in

I have measured the NXP BZX84 family. The 2V7 or 5V6 in the type number mean exactly that. At 4V7 the low noise property is already gone.

The pics to the left & right show some LEDs and regulators.

BTW: the leftmost pic of the peaceful Blau river valley is where the Gauss family runs a hotel. Yes, "our" Mr. Gauss. They have an oil painting of him in the dining room.

??? of him / of his: which one is correct?????

BTW2: For the sake of scientific honesty, Prof. Zener sued the semi industry that they should not use his name for avalanche diodes. They settled to call them Z-diodes and to pretend it was for the form of the U/I curve.

cheers, Gerhard

If it's a painting of Carl Friedrich Gauss, "of" is correct. If it were a painting by Carl Friedrich Gauss, it would be one of his paintings.

Professor Zener was being optimistic. Language is defined by the words people use, and if the majority makes a habit of using words incorrectly that usage becomes part of the language.

My statement

When operated in Zener mode (as opposed to avalanche mode)

pertains Zener diodes in Zener mode. Why do you want to talk about the positive temperature coefficient of Zeners in avalanche mode?

Danke,

The problem is that you choose the mode when you choose the diode. Your formulation suggests that you think somebody could to operate one diode in one mode or the other. Only a hopeless newbie could think that, but some of the lurkers aren't all that sophisticated, and a few of our posters aren't much better.

Gerhard Hoffmann has pointed out that even a 4.7V zener diode has enough avalanche component in it's breakdown mechanism to introduce lots of noise though the temperature coefficient of the reverse voltage is still marginally negative.

Thank you for taking the time to clear things up. It's time to move on to a another interesting (from my own perspective) topic. At least one person refers to this Shockley pn junction equation as "famous":

VBE = VT*ln(IE/IS)

It makes me feel like a hopeless newbie for such a well known equation to remain unknown to me until relatively recently. My curiosity moved me to peek at the 12.5 THE THEORY OF A p-n JUNCTION section of ELECTRONS AND HOLES IN SEMICONDUCTORS by Shockley. So far, nothing recognizably resembling the above equation jumps out at me from the Shockley. Fortunately, "success has many fathers." So this paper (credits Widlar's work) will suffice:

Danke,

Nice. Thanks for posting that.

If the picture shows what Mr. Gauss looked like, it is a painting of him. If Mr. Gauss painted a picture it is a painting of his. It gets even more complicated if Mr. Gauss painted the image of himself - a self portrait. Then it would be his picture of him. Groan.

Reminds me of a Japanese lady who learned English and had a hard time with prepositions: do you ride in a bus or on a bus; why do you sit on a stool but in a chair (sometines on a chair) but you sit on a sofa, not in a sofa etc etc I helped her with that, but when she asked me to explain PIN diodes to her I couldn't.

Ed

Like incorrectly using mantissa for the significand of a floating point number. Note that the word mantissa does not appear anywhere in the IEEE spec.

I don’t remember the last time I used a log table, so that one hardly matters anymore.

Cheers

Phil Hobbs