Go Down

Topic: How to measure chip (or other component) temperature? (Read 1 time) previous topic - next topic

JimboZA

This thread about an overheating 293 made me wonder how to measure the temp of a chip.

My Christmas Fluke 17B has a temperature scale (measures up to 400C) and came with a Type-K thermocouple. To measure the temperature of a chip, would it be enough to hold the thermocouple probe against the surface or does it need a better contact, perhaps with some gel or paste or other gloop to get a better contact?
Repeal Ohm's Law

No PMs for help please

bobthebanana

I guess regular old thermal grease would work. Thermal exopy if you want a permenent contact. But either way, thet will be package temp, not die temp. Maybe with a bit of math and the thermal conductivity figures in the datasheet it would be possiblr to get an accurate measurement

jonisonvespaa

hi bob
whats die temp, is that the core of the ic?
cheers

CrossRoads

We have a Ryobi Non-Contact Infrared Thermometer
http://www.homedepot.com/buy/electrical/electrical-test-meters/ryobi-non-contact-infrared-thermometer-ir001.html
Put its red laser dot on the middle of a chip, or a resistor, or transistor, get a digital reading back.

No more burned finger tips!
Designing & building electrical circuits for over 25 years.  Screw Shield for Mega/Due/Uno,  Bobuino with ATMega1284P, & other '328P & '1284P creations & offerings at  my website.

Msquare


afremont

#5
Mar 08, 2013, 07:36 pm Last Edit: Mar 08, 2013, 07:50 pm by afremont Reason: 1
Thermal imaging is probably the best way I can think of, but it really costs allot of money for an imager.  Guess that's probably why I don't have one.  Next is my your finger. ;)  I have a non-contact thingy, but mine includes too much surrounding area for parts like resistors.  Even held really close, I don't really trust it on small parts.  Gotta watch out for shiney parts too.  IR sensors seem to have trouble with really glossy surfaces like chrome or gloss black harley paint on the oil tank.
Experience, it's what you get when you were expecting something else.

Docedison

#6
Mar 08, 2013, 07:43 pm Last Edit: Mar 08, 2013, 07:53 pm by Docedison Reason: 1
Jim, A little dab of "Elephant Snot" as it is sometimes "fondly" called in Southern California is your best bet. Inexpensive heatsink compound. The thread linked was someone complaining that he didn't do his homework right... or didn't have the tools to do it right... It's all calculable but like writing code requires some training... first. When, without planning or at least writing down the basic requirements first... and starts "throwing parts" at an idea... one frequently ends up complaining about abused parts...
It's Too Hot To Touch, What Do I Do.. ? How many times have you read that here in this group alone?
If you are desiring of making some kind of thermal monitor... either write up your thoughts on preferred method here or PM me... either way I'd be happy to help.
The comment about the math being able to calculate the die temp was correct but I also know that the cooler a device is in operation the longer it will live, so anything uncomfortable to the fingers should be investigated... Probably the best thing is the aforementioned non-contact IR thermometers they can be had (Good quality ones) for less than $25.. on Ebay I didn't read all of it carefully, Edit RKJ

Bob
--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

JimboZA

Quote
Thermal imaging is probably the best way I can think of, but it really costs money for an imager.  Next is my finger


.... and in between is my trusty Fluke.
Repeal Ohm's Law

No PMs for help please

JimboZA

Quote
If you are desiring of making some kind of thermal monitor...I'd be happy to help


Thanks Bob, but no it's really just curiosity, wondering how useful my Fluke's thermocouple would be in monitoring temp of a component.

Repeal Ohm's Law

No PMs for help please

oric_dan

Heat kills. There are some guys who think whatever it says, up to thermal shutdown
[eg, 175 degC] is ok, but I don't like my electronics running too hot, or failing prematurely.
So I use the Little Pinkie Test = if it's too hot to hold my little finger on the chip, then it's
too hot. And some redesign should be done. I also have a non-contact IR-Thermometer like Bob.

As Erdin indicated on that other thread, the L293s are bipolar technology, and not MOSFETs,
so there tends to be a collector-emitter drop of upwards to a couple of volts in those parts
[because the output stages have darlington pullup and pulldown ckts], and that times the load
current gives the power dissipation, Pd = Vdrop * Iload. Newer MOSFET h-bridges are much
better.

http://idmax.free.fr/Aide/Stepper/l293.pdf

Also, you might look at the TI 754410 as a drop-in replacement for the L293D. 754410 can go
to 1-Amp load vs 600 mA for the 293.

BTW, one way to tell if an L293 is too hot is, the chip will literally blow up and catch fire. This
actually happened to me once. Not just magic smoke, but the plastic end blew off, and flames
actually came out. So much for thermal shutdown. A mini Boeing 787.

Docedison

With the tip about the grease, quite well... Mine does or did until the probe was "Lost".. I own a Fluke 179.

Bob
--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

afremont

Seen a 7805 open up once, guess the shutdown had an issue.  I also learned it's also a good idea to pay attention to voltage and polarity with tantalum caps.  Anybody got any really good incendiary stories?
Experience, it's what you get when you were expecting something else.

Docedison

--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

retrolefty

Just because I haven't seen it mentioned here but many of the AVR chips, including the 328P, have an internal die temperature sensor impeded into the chip die. By selecting the proper analog mux selector bits one can read this temperature value. It's an un-calibrated temperature so not too useful and any reading taken is more a reflection of how much current you are sourcing or sinking via the output pins rather then the chip itself is generating, but some have found it amusing in the past to display the value. No one to date as best I can recall has done anything really useful with that sensor.

Lefty

Docedison

Yeah it involves some 25 100000uF 3V electrolytics and a 100 foot or 33 meter extension cord... and the police too. I was 14 and found these parts in a surplus electronics store scrap-bin. Most made a hell of a noise... almost had a dud but my mother called me to lunch.. It explodes disconnected from 110Vac 15 minutes later...
Thats when the neighbor called the cops... Seems he tried to kick it off his sidewalk, dern  near blew his shoe laces off... <BG>

Bob
--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

Go Up