LM335 reading value incorrect.. any help?

triple_x · Joined: 01 Dec 2005 Posts: 10

Hi all,
I try to implement this thread http://www.ccsinfo.com/forum/viewtopic.php?t=25665&highlight=lm335
as suggested by Thelmah, but I got invalid values when reading the temp sensor.

The LM335 pin connection as follows:

kender · Joined: 09 Aug 2004 Posts: 768 Location: Silicon Valley

My $0.02 guess is that you have a hardware problem. I have never used LM335 before, but I have skimmed through the datasheet and found that the output is 10mV/K. I assume that your sensor is at room temperature T~300K. Your voltmeter reading should be V(T)~3V, and your actual volmeter reading seems too high. Also V(T)=4.88V is suspiciously close to the Vcc=+5V, and it hints that you resistor value is too low. Try a higher value of, say, 10k.

By the way, thatnks for asking such a detailed question!

Ttelmah · Guest

One comment. Check the pinout.
The normal connection, is:

Vadj

V+

V-

Along the chip, requiring GND to pin 1, and reading the output from pin 2, so if you are connecting to the outer two pins (which your pin 1&3 comment suggests), it is no wonder it is not working.

Best Wishes

languer · Joined: 09 Jan 2004 Posts: 144 Location: USA

I agree with both Ttelmah and kender, check the pinout. The following link gives you a picture:
http://www.ontrak.net/LM335.htm

And the datasheet just in case:
http://www.st.com/stonline/books/pdf/docs/2158.pdf

triple_x · Joined: 01 Dec 2005 Posts: 10

thanks for the reply. After changing to 10K resistor and triple check the pins, value now is ~2.7V. unfortunately, no change at output

Les Scott · Joined: 18 Feb 2006 Posts: 2 Location: Cheyenne, WY

Hi,

I am pretty new to CCS C but am successfully using the LM355Z in my project. The 355 wants a minimum of 450uA to work so that means that the series resistor should be something less than 4k ohms (if your supply voltage is 5V). In your last post you said that you were reading something like 2.7V. That would be a temperature of about 0 degrees C. I think you may still have a wiring problem. Make the series resistor somewhere between 2.2 and 3.9k. At room temp you should have a sensor voltage of about 2.9V. If this is not what you are reading at the output pin of the 355 you either have it wired up wrong or the 355 is fried.

You do not need the 1uf cap on the output of the 355. It is a pretty low impedance device and the cap isn't going to do much for you.

This is an amazing source of information. I hope I haven't offended any of the experts here because there are many that post here that are way more knowledgeable than I.

Let me know what more I can do to help.

kel · Joined: 17 Oct 2005 Posts: 68 Location: Brisbane

He is right,you don't need the capacitor.I worked on lm335 too a few weeks ago and you should be able to get what i posted.I didn't connect any capacitor and i through many helps from this forum was able to get the right temperature readings if the Vcc is constant.However lm335 is voltage depended and the temperature droops with the dripping of the VCC.Note that very well!
I had trouble with back light LCD that cause a drift in the Vcc and my temperatures were very much affected.
Looking at your code, there is nothing wrong with it!!

triple_x · Joined: 01 Dec 2005 Posts: 10

Thanks for the info kel & Les Scott. resistor changed to 2.2k and 1uf cap is removed (already removed before read the posting :-) ), got value around 3.04V at the output. but what suspicious me is that result echoed to terminal is the same!

triple_x · Joined: 01 Dec 2005 Posts: 10

success!

Les Scott · Joined: 18 Feb 2006 Posts: 2 Location: Cheyenne, WY

Cool!

Glad it is working!

Take care,

Les

sonicdeejay · Joined: 20 Dec 2005 Posts: 112

sonicdeejay · Joined: 20 Dec 2005 Posts: 112

Guy...

my connection as followed..

Ttelmah · Guest

559L, is 559. The 'L', ensures the compiler treats a value as a 'long', and is a good way of making sure that 16bit arithmetic is used, rather than the default 8bit.

1KR, is a bit low for the resistor on the chip. Shouldn't matter for 'normal' temperatures, but implies 2.5mA is being drawn at 0C, which is closer to the upper end of the chip's range, than is perhaps 'ideal'. Keeping the current lower, also helps to reduce error from the 'self heating' of the chip.

Your voltage is rather lower than I'd expect. In a typical room, you should see close to 3v, and even uncalibrated, the 'worst' chip from the maker, should not give as low a value as 2.525v. The specification from the manufacturer, is 2.98v at 25C, when 1mA is drawn, +/- 0.03v. Now, to draw 1mA, at this voltage, would require a resistor of:

(5-2.98)/0.001 = 2K02

So I'd suggest a 2K, or 2.2K resistor, as putting you 'on' the manufacturers recommended operating current.
You really need to find out why your reading is this low?..

The point about '559', is that with a normal chip, using an ADC operating off 5v, this is the reading for 0C. 0C, corresponds to 2.73v from the chip. The PIC ADC, has a 'step' size, of:
Vref/1024 (see my previous posts for why this is 1024, not the 1023, of many other ADC designs).
With Vref=5v, this gives a step size of:
0.0048828125v

Then 2.73v/0.00488 = 559 counts on the ADC.
So, you need to solve why your hardware is giving 2.525v. This is blatantly a 'wrong' value, and implies something is not right.

Best Wishes