Resistor heat calculation help please.

snipped-for-privacy@hotmail.com wrote in news:b8fa7d1a-5865-4b17-9b0c- snipped-for-privacy@w1g2000prm.googlegroups.com:

No. You'd need a very complex model indeed to predict thermal coupling for an SMT part in almost any context. This really IS one of those things best done empirically. If anyone posts saying different to this, ask them how they're going to quantify the thermal coupling when they don't know your board, your solder fillet profile, ambient conditions, etc...

What you can do is look for a manufacturers figure for thermal coupling, hopefully given in the °C/watt rating used for heatsinking. They might specify some conditions you can use as a guide. None of which is a substitute for the test you'll have to make anyway, by which time you'll know all you need to know.

tute

Hi, Thank you for your advice....I did think I was asking for the impossible....trial & error it is then...

Regards.

snipped-for-privacy@hotmail.com wrote in news:c1343ba3-d66e-4e0e-9c3b- snipped-for-privacy@r41g2000prr.googlegroups.com:

I rethought this a bit, there are some clues.. first, the 4.3W in that small a volume might make a temperature rise big enough to make the subtleties moot, it might melt the resistors off the board. second, a small change of voltage, like just half a volt or so, might make the difference between the parts staying put or falling off. If you're sure the 4 watts is enough to do what you need, and it couples well to the load to be heated, then you'll probably solve the parts-falling-off thing. If the contact is indirect, then it's better to use more resistors, maybe 4 x 120 ohms, or even a 2x3 matrix, i.e. parallel three series pairs with 200 ohms for each resistor.

I don't know what you're doing so I don't know if you need a controller, but if you do, maybe the simplest way is to use an LM317 to control them. Adding a thermistor to the control network can allow it to form a thermostat if mounted with the heater resistors.

I don't know of any formulas. It depends strongly on the details. How big are the pads? How thick are the traces? How far to the nearest via? How big are the vias?

You might get some ideas if you look at the right data sheet or app note for power semi-conductors. I think the ones I'm think of are for National regulators. The have numbers for various patterns of copper that the heat tab gets soldered to.

What are you trying to do? Warm up a large area, or get a small region very hot?

Is this a one-off project, or do you expect large production runs?

If I was trying to warm up a large area, I'd use more resistors. If you have the typical 4 layer board, I'd use big pads with several vias to the power/ground planes.

If I was trying to get something hot, I'd probably use an axial leaded ceramic power resistor.

Hi, Just to fill you in.....

I=92m using the resistors as a heating pad to warm a QFP 64 processor. I believe this chip is approx 14mm x 14mm. This chip controls the dashboard gauges & warning lights in my car. When the chip is cold the gauges don=92t work properly=85once the chip has warmed up a little everything works fine. If the ambient temperature is below approx 10 degrees C the chip has this fault condition, if the temperature is more than 10 everything is fine. The problem is definitely with the processor & not anything else. I=92m very limited by space as obviously the processor is inside the dashpod.

The resistor heating pad will be bonded to the processor & will have trailing leads to come out of the dashpod. The heater will only be in use for a few seconds (to warm the processor enough to work), I will control this via a switch initially so I can see how long the heater has to be on before the chip starts to work, obviously this will vary depending on ambient temperature. Eventually I could use a timing circuit to do this. I suppose if I knew what I was doing I could build some form of temperature control into the timer circuit so if the temperature was over 10 degrees the heater wouldn=92t operate.

Regards.

-40 ... +85´C C grade

-40 ... +105´C V

-40 ... +125´C M

MfG JRD

Hi, The processor is a Freescale MC68HC908AZ60 device. Apparently the processor develops faults when it is cold due to poor programming when manufactured. A lot of vehicles from different manufactures suffer this problem & obviously it shows up a lot more in the winter months! The solder joints have been checked & re flowed & are fine....honestly the problem is with the chip not liking to be cold....I know it seems unlikely but it is the cause.

Regards.

On Mon, 12 Jan 2009 06:30:35 -0800 (PST), snipped-for-privacy@hotmail.com put finger to keyboard and composed:

IIRC, you wrote that the bidirectional reset pin was cycling low when an internal watchdog timer timed out. Maybe you could set up an external timer for your heater which runs for 3 seconds, say, and use the reset pulse to retrigger this timer. Once the uP is up and running, the reset output will be inactive, and the heater timer will time out.

- Franc Zabkar

as

d text -

Hi, Your idea sounds excellent, but as I'm learning as I go with electronics a manual switch will have to do until I figure out how to build a simple timer circuit.

Thanks for the suggestion though.

Regards.

snipped-for-privacy@hotmail.com wrote in news:c5429276-1410-4a45-990c- snipped-for-privacy@x16g2000prn.googlegroups.com:

Something's wrong here. Very wrong. This is to heat a CPU for critical warning lights and gauges in a car? And in a situation that normally demands a naojr manufacturer's recall at the behest of national government if they don't do it voluntarilty, a person with limited knowledge of electronics, below that needed to buld the simplest of timers is expected to do what for once, really ought to be handled by formal authorites?

If this thing fails while you're driving, and the car crashes or hurts another person, you're in for a vastly bigger and nastier shock than you're going to suffer if you pay for it to be replaced properly. You specifically mentioned warnign lights and gauges, so I don't think I'm over-reacting here. This is critical stuff. Last time I heard of some guy who tinkered with his vehicle and this resulted in deaths of people (in this case by breaking down in frozen inaccessible conditions rather than a crash) the driver and owner got seriously jailed.

If you have isolated as much of the problem as you have, good on you, take that info to a places registeres ad qualified to do it under terms that won't invalidate your vehicle insurance, etc. Use the info to make sure you get the job done fast, as cheap as possible, as it saves diagnostic time. This isn't one for DIY.

Yes I agree with your point that the manufactures should be the fixing this problem however like most car manufactures they seem to ba a law unto themselves & if they say no well quite often that's the final word. Public pressure of the same (well very similar) fault on the Audi TT did bring about free replacement dash pods regardless of the age of the vehicle, however a recall wasn't the case...probably because that would be admitting liability!

I do take your point, however I would still be in the same boat if I had the dash pod fixed anywhere other than the manufacturer, who won't fix the defective part but will fit a new dash pod for a exorbitant amount of money.

The resistor pad heater I intend to fit is merely to save me having to sit & wait until the processor has warmed itself up enough to spring into life, approx 5 minutes in frosty conditions. I don't drive the vechicle until the gauges have started to operate...not because of not knowing haw fast I'm going or anything silly like that but because if the car is moved when the gauges arn't working when they do come on the mileage goes silly & clocks on a big chunk of miles!

Regards.

The answer is in your question. This is a heater - therefore you would logically be controlling the target temperature by removing power when the required 'device' temperature is achieved.

You can reduce power consumption and improve thermal gradients between heater and target by thermally isolating the bonded target and heater from the surrounding ambient.

Limiting rise is only an issue if the power must be constant and is determined by thermal impedance to the physical environment from the source.

If the source of power is isothermal within a volume that has a specific surface area, the surface temperature rise can be calculated as a ball-park figure with better than 20% accuracy - but that's not what you're asking for.

RL

legg wrote in news: snipped-for-privacy@4ax.com:

Of course it's not what he's asking for! Did you reply late just to play the smartarse with lectures on thermodynamics? It's a small processor on a board in a car, how the hell is he supposed to thermally isolate it? You sound like a guy with a head full of blurry ideas who didn't even read any of the posts that were done before you came. By trying to make others look like fools, answering the first post as if all the rest are stupid and beneath your exalted threshold, you've made yourself look like a total fool yourself.

Hi, just to let you all know that I made the heater & it works fine to warm the processor to get it working. I ended up using 6 x 33 ohm resistors for better surface area coverage. A 5 second blast of the heater is all that is required to get the processor up & running fine.

Regards.