Re: gEDA-user: reflow oven

[Bill Sloman <bill.sloman@xxxxxxxxx>]

At 12:23 7-5-2006, you wrote:
> On Sat,  6 May 2006 16:25:24 -0500
> "Phil Taylor" <phil@xxxxxxxxxxxxxxxxxxxxxxxx> wrote:
>
> > Do you know if there is heating occurring during cool-down?  If 
> there isn't,
> > can you remove heat another way?
>
> Thanks for the reply. The original idea was to have a piece of PCB, 
> and measure the temperature of the PCB, because the target will be a 
> PCB too. So I attached the pt100 to the pcb with some heat 
> conducting paste. I think this was a bad idea. I think I'll mill out 
> the area of the PCB where the pt100 mounted, so the only heat couple 
> will be is the pins of the pt100. I'll see if it helps or not...

Your original idea strikes me as extremely sound. Isolating the Pt100 
sensor, on the other hand, strikes me as a really bad idea.

Your control loop consists of an infra-red heater irradiating your 
Pt100 sensor, and a feedback loop that regulates the voltage across 
your infra-red heater to maintain an infra-red flux that sustains the 
desired temperature at your Pt100 sensor.

What you actually want to control is the temperature of the board you 
are reflowing.

For a given infra-red flux, the equilibrium temperatures of your 
sensor and of the board being reflowed are unlikely to be the same.

The heat flowing into the sensor depends not only on the infra-red 
flux, but also on its absorbtion coefficient (1 - reflectivity) at 
the wavelengths involved. This doesn't have to be the same as the 
absorbtion coefficient of the board you want to reflow.

Heat flows out of the sensor and the board being reflowed by 
conduction, convection and radiation. Conduction won't be an issue 
for the relatively large board being reflowed, and probably won't 
matter much for the sensor. Radiation isn't too important at the 
temperatures in question - the board and sensor will be re-radiating 
at much longer wavelengths than they are absorbing, so that while the 
re-radiation coefficient would be the same as the absorbtion 
coefficient at the same wavelengths, they don't have to be the same 
in practice, not that the relatively small proportion of the heat 
being lost by radiation will allow this to be a real problem. 
Convection is the real problem.

The sensor is going to be much smaller than the board being reflowed, 
which makes a difference. In practice, the convention currents around 
the board being reflowed and the sensor are going to be determined by 
the board, which is going to be much bigger than the sensor, so that 
the sensor is likely to be more effectively cooled by forced 
convention acting on the sides as well as the top and bottom of the 
sensor than is the board, which hasn't got any significant side area.

Putting the sensor onto a significant lump of printed circuit board - 
something that is at least ten times longer and wider than it is 
thick - seems to be to be pretty much essential if you want your 
sensor to be sensing a temperature that is anywhere near the 
temperature of the board to be reflowed.

Such a sensor will necessarily react more slowly to changes in 
infra-red power input than a bare sensor, so you are going to need a 
proportional/integral/derivative controller to get a stable and quick 
control loop. You might also need to linearise the voltage output of 
the control amplifier to get a roughly linear relationship between 
error signal at the input and corrective temperature rise at the 
sensor, bearing in mind that power dissipation in the infra-red 
source is proportional to the square of the applied voltage, its 
temperature rise (if dominated by convection) will thus be roughly 
proportional to the applied voltage, while its radiated output 
increases as the fourth power of its absolute temperature.

Messy physics ...

--
Bill Sloman, Nijmegen