Part 1: Introduction and test bed setup

Part 2 (you are here!): Experimentation with a heating element and basic measurements



This project is about creating a reflow oven for surface mount work.


With the mind still active thinking about some useful avenues that will help this project take shape (a stainless steel chimney as an enclosure is top of the list so far, along with heating from both above and below) I decided to start taking some measurements to either rule in or to eliminate some heating methods.

One main method and one backup method was employed. The backup was a thermocouple as a very vague indicator just for some reassurance that the main measurements were in the same ballpark (in the photo below you can see the thermocouple wasn’t even taped to the board).



The main method was a thermal array sensor (available from a few manufacturers – I used a Melexis one which is quite overpriced for what it is – other manufacturers like Panasonic or Heimann may have something more usable) connected up to a data logger (click here for information on it) to record an array of temperature data every second.


The photo below shows another view of the same thing. The thermal array sensor was just attached using an elastic band to an L-shaped channel just to keep it aligned (the alignment was first tested with a cheap laser pointer module that was attached to the channel, and then the pointer was swapped out with the thermal array sensor).



The data was then uploaded into Matlab and a ‘heat map’ type image was generated every second:



The video below shows the result over a 10-minute period (the video is accelerated):


Knowing what area of the heat map corresponds to which parts of the heated area is tricky. From some very rough calculations, the horizontal span of the heat map corresponds to perhaps half an inch less than the width of the entire aluminium plate. The sensor is at an angle to the board, so pixels will be stretched in the y-direction of the plane of the aluminium of the plate.


At the center of the heat map, the x-width of each heat map value corresponds to about 7mm width in real life, so I think the heat map has captured the hottest points to be the large ICs. Basically it can be seen that over time the chips ended up hotter than the rest of the board.


Here is a best guess at doing an overlay of the target area stretched over the heat map.



The ceramic heating element that was used is clearly not useful for the application since it took 10 minutes to heat the target area to 200 degrees, but the exercise was still helpful in proving out the testbed for measurements. And it does seem to eliminate ceramic based heating elements for quick heating (unless a higher power element is used – mine was 600W).


Summary and Next Steps

With some more refinement it might be possible to get a reasonably useful measurement of the target area. It will be easier using a large board with just one IC on it, to make it clearer to identify from the heat map.


The results so far might explain why it is important to quickly heat the board during the actual reflow stage of the process, because otherwise the chips will store a lot of heat. At the end of the 10-minute test the integrated circuits were perhaps 35 degrees hotter than the rest of the board, whereas at 2 minutes into the test the ICs were only around 15-20 degrees hotter than the rest of the board. It will be interesting to see what happens with halogen heating elements and tungsten ones (I still need to buy those).


Some progress also needs to  be made on the controller. To keep it all simple, it is likely to use a microcontroller development board with a basic PCB containing anything custom such as the buttons, display, etc. More on that later.