Hey there,

I'm about to design a sensor circuit which is to be outside the whole year.
I took care that most of the ICs are able to deal with temperatures from -20 to 30 degrees. Unfortunately though, the photodiode/transimpedance amp IC (OPT101) is within spec only at temperatures > 0 degrees.

Now... I've got a couple ideas but I'm more or less lost at what you can do about that.

1) Have a small load resistance at the output, so that the increased current through the IC will heat it by itself...? (A voltage regulator for example would hardly need heating, would it?)

Problem: The output is logarithmic to the irradiance. High current will flow only when there is "a lot" of light. What about (cooold) mornings...?

2) Have some kind of heating. What would you guys recommend? Halogen lamp (hidden away form the light sensor obviously)? PTC heating?

I will most probably have a temperature sensor -near- the circuit as well, so the more local the heating, the better.

Any ideas?

Thanks in advance!

 

Roman black has a made a xtal oven that you could possibly adapt to the part.


A quote from his page:

How does it work?

Three small resistors are glued to the body of the xtal. These act as "heater elements" and get warm when current is passed through them and heat the entire xtal body. This current is controlled by a little darlington transistor. The temperature is sensed by a cheap NTC thermistor, which controls the current; increasing current if too cold, decreasing current if too hot. Simple negative feedback.

and the link:
http://romanblack.com/xoven.htm

 

Sounds like a ridiculous device, so many shortcomings. They also recommend that you bake it at 85 centigrade for 24 hours before soldering.

 

That's hilarious! I don't know if it's applicable though. The chip comes in quite a flat DIP8 casing and the top obviously needs to be free of anything blocking light.

 

Ummm... There has got to be a better way than using this particular IC. Is there a vehicle version or an outdoor version? Or one that doesn't want you to bake it at 85C for 24hrs?

 

Well, I guess it's easy enough to put a resistor on the board near the amplifier, and run current through it under the control of a temperature sensor. Apparently it doesn't need to be super accurate, just enough to keep the temperature above freezing.

That's assuming that the system has enough power available to throw away on keeping stuff warm!

Surely there's a substitute component around that isn't so fussy!

But edited to say: if you could control the output current of the sensor amp and thus cause self-heating, that would be a very elegant solution. (Power supply voltage minus output voltage, multiplied by current drawn from output, equals power dissipated in IC package.) Would there be any way to put a temperature sensor right up against the body of the chip?

 

But surely there's a substitute component around that isn't so fussy.

Thats exactly what im thinking.

Alos, the maximum ratings are -25C so if accuracy is not severely important in those temperatures, you should be good.

But, chances are, a package is made for other environments. Cars or outer space.

But there are probably other sensors that do what you want with less hassle.

 

You might find some of these sensors interesting - http://www.taosinc.com/Productfamily.aspx?id=1&SD=ltv - The TSL260RD is good for the range -25 to 85 C.

 

The reason for baking at 85°C for 24 hours prior to soldering is to eliminate any moisture that may have been absorbed by the IC's packaging. If you fail to do so, you may wind up with a ruptured IC package due to the absorbed moisture boiling when heated during soldering. You may need a microscope to see the damage. You might not be able to see the damage even with a high powered microscope.

Baking is not an unusual requirement. Most SMD/SMT IC's come in sealed ESD-protective packaging with desiccant nowadays. That keeps the IC from absorbing moisture, because there isn't any present. If the IC is out of the package for more than a couple of hours, then you'll need to bake it prior to soldering.

 

Thanks for clearing that up Wookie!

I checked out those TSL2xx chips - they're not responsive enough with their 1W/m^2 . The sensor's gonna be in the sun...

Alos, the maximum ratings are -25C so if accuracy is not severely important in those temperatures, you should be good.

Well, that's kind of what I was thinking too. If I lose like 5% of accuracy at deep temperatures... well, so be it.

Well, I guess it's easy enough to put a resistor on the board near the amplifier, and run current through it under the control of a temperature sensor. Apparently it doesn't need to be super accurate, just enough to keep the temperature above freezing.

I think I'll try that just for the fun of it .

But edited to say: if you could control the output current of the sensor amp and thus cause self-heating, that would be a very elegant solution.

The problem is that the output voltage is logarithmic to the irradiance. So there'd be alot (enough?) current through it when the sun shines, but next to none (relatively speaking) in the morning.

Hm, is there a non-linear device that draws more current (lower resistance) when there's less voltage...?

Would there be any way to put a temperature sensor right up against the body of the chip?

I dont have the chip yet, it looks quite flat though so dunno. I was planning to put it onto a socket though anyway, so maybe you could put a sensor right underneath it...?

 

Hm, is there a non-linear device that draws more current (lower resistance) when there's less voltage...?

Would you believe - an incandescent light bulb?
(It's true, check it out.)

 

Hah! A light bulb fixing the problem of my light sensor... !

...oh wait

 

Hm, is there a non-linear device that draws more current (lower resistance) when there's less voltage...?

Tunnel Diode.

http://www.eie.polyu.edu.hk/~ensurya/lect_notes/commun_cir/Ch7/Fig7-1.jpg

 

Interesting - thanks spook!

Now, my professor came up with another idea. One could use a voltage regulator (LM217) to heat the sensor. The whole circuit's gonna be wall mounted, so when you put the regulator below the sensor, a NTC resistor above it... that might just work, no?

The circuit I came up with seems a little complicated though... any comments? Any other circuit ideas?

http://img191.imageshack.us/img191/9430/lm217heater.png

The idea is basically to use a Schmitt Trigger for the temperature. The output drives a bipolar transistor so that enough current is delivered to the LM217.