Hi
I have a PCB that is close to a metal object (non painted). It's about 5mm away
I cannot connect diretly to it due to risk of condensation
I checked out this sensor:
But that is something like 1 USD per device. I need something cheap
Any ideas?
Cheers
Klaus
The cheap IR thermometers that you get from Horror Fright or eBay use analog-output thermopiles, and those must be pretty cheap (in a number of different senses of the word).
Dunno where to GET them, other than China.
You'd need the thermopile, a means of amplifying the snot out of it's output, a means of measuring it's temperature (a thermopile only measures the temperature _difference_ between an object and itself), and a means of turning that mess of stuff into the temperature of the far object -- taking into account the object's emissivity.
After all that, a dollar might be cheap.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com
Thermistors on the top and bottom of the board?
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
I only just read the "non painted" clause.
If the metal can start out shiny, and end up dull and tarnished over time, you're going to be in a world of hurt as your calibration changes with aging. Basically, the signal you get from the metal is inversely proportional to how shiny it is (look up "emissivity"). So you build your prototype, get it calibrated with the shiny piece of metal, and then when the metal gets tarnished your signal level goes up hugely.
If you can arrange for a nice black (or other _reliable_) colored spot on the metal dingus, you'll be much happier in the long run.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com
The cheapest close proximity non-contact method is probably the use of an IR photodiode:
-- Regards, Joerg http://www.analogconsultants.com/
Could you stick on a liquid-crystal thermometer, and read out its optical reflectance?
A torsion-spring thermometer can be outfitted with a capacitor vane instead of a pointer, for another kind of non-contact measurement. It'd sense vibration, though.
sensors-14-05278.pdf
So, first, that calls out an exotic photodiode with response down 2.6 microns (silicon photodiodes are good down to 1.1um or so). Second, it's going to work well for typical cooking-heat temperatures, but it's not necessarily going to work well for lower temperatures -- check out the black-body radiation curves for the wavelengths you mean to watch for and the temperature you mean to monitor (thermopiles inherently work over any wavelength -- as long as it gets through the window and gets absorbed by the paint on the thermopile).
And, you still have the whole emissivity uncertainty to deal with.
I'm not saying it _can't_ be done -- just pointing out difficulties that must be worked around.
-- www.wescottdesign.com
What's the temperature range?
Cheers
Phil Hobbs
The temperature range is 0 to 100 if possible.
And I have found a way to measure only on a black body object
Cheers
Klaus
That's fairly hard to do well. If you really really can't paint the metal, you have to optimize the conductive coupling while minimizing radiative. On e approach would be a thin flex circuit with no cover layer (ENIG plated), mounted on foam sticky tape and held close to the metal without touching.
With 1/3 oz Cu (common in flex) and a serpentine pattern of 8-mil lines and spaces, you can make a Cu RTD of a couple of hundred ohms in about a 20-mm square. (IIRC--I did this a year or so back.) put the shiny side toward th e metal--that'll reduce the radiative coupling, but also its variability.
Cheers
Phil Hobbs
, you have to optimize the conductive coupling while minimizing radiative. One approach would be a thin flex circuit with no cover layer (ENIG plated) , mounted on foam sticky tape and held close to the metal without touching.
nd spaces, you can make a Cu RTD of a couple of hundred ohms in about a 20- mm square. (IIRC--I did this a year or so back.) put the shiny side toward the metal--that'll reduce the radiative coupling, but also its variability.
It is possible to select the object, so we have a tight tolerance on the em issivity
Would it be possible with this one:
If I keep it in total dark environment
(outside my confort zone on this one)
Cheers
Klaus
You can't measure temperature in that range using any photodiode whatsoever, for $1. (Well, you could forward bias it and look at its V_F, but that's no use in this case.)
Something like cryogenically-cooled HgCdTe ("mercadtell") would see the object fine, but (*ahem*) wouldn't meet the other constraints.
InGaAs is pretty well useless for temperatures below 250C, and you couldn't afford it anyway.
You could mount a small, conformally-coated sensor board on the metal, and use LEDs and photodiodes for power and comms, but that wouldn't be $1 either.
So it's really some bolometric device or nothing. Or else figure out a way to relax the no-contact requirement.
Cheers
Phil Hobbs
What's the temperature range? (I like Joerg's IR photodiode, but it depends on the temperature.)
George H.
One way to avoid most emissivity issues is to create a small cavity with one wall being the surface to be sensed (best if painted black), the other walls being shiny (gold plating is best, but that blows the budget), and a bolometer inside the cavity.
This comes directly from how one implements a black body using a cavity in some material, like a metal block - so long as radiation has to bounce around many times before finding the exit aperture, the aperture will be black, regardless of the emissivity and reflectivity of the inside walls.
Joe Gwinn
0 to 100 C? GH
What is the reason for not being possible to use a photodiode, just so I know? Isn't that what the guy used in this one:
Cheers
Klaus
al, you have to optimize the conductive coupling while minimizing radiative . One approach would be a thin flex circuit with no cover layer (ENIG plate d), mounted on foam sticky tape and held close to the metal without touchin g.
and spaces, you can make a Cu RTD of a couple of hundred ohms in about a 2
0-mm square. (IIRC--I did this a year or so back.) put the shiny side towar d the metal--that'll reduce the radiative coupling, but also its variabilit y.emissivity
So can you say why you can't stick something on the metal thing? If the metal thing is going to get condensation then that will change the emmisivity.
George H.
Different temperature range, and expensive InGaAs diodes.
Blackbody radiation drops exponentially at shorter wavelength, and PDs have a fairly sharp long-wavelength cutoff that has a horribly large tempco.
Cheers
Phil Hobbs
Yeah, it's not going to be very precise. I was just thinking that if they start using them for cooktops they ought to be dirt cheap.
Another option would be to have a regular cheap NTC deal on the PCB, a simple oscillator where this NTC varies the frequency, a leaky inductor, another leaky inductor of same kind on the other side (after the 5mm air gap) and that goes into the comparator input of a uC or whatever else happens to be there. Measures frequency -> deduces temperature from that.
-- Regards, Joerg http://www.analogconsultants.com/
Phil is telling you everything I was trying to say, only better.
I'd endorse his flex-circuit RTD idea, except I've priced flex recently, and unless you're already using it, or making things in enormous quantities, you'll pay more than $1 just to go flex.
DigiKey lists thermopiles that start at around $7 -- I'd be tempted to see if there's something that you can get cheap out of China, or see if you can figure out how to mount a thermistor so that it's thermally well isolated from your board, and well thermally coupled to your mystery metal. You'll probably still need to measure board temperature and calibrate it out.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.