TCs are sensitive instruments of measurement. TCs are abused mainly because they are misunderstood. First, go out and purchase a set of TCs with the proper rating for your process with the least amount of variability. Look at the thermocouples that you are using. Organize the TCs according to observed temperature values by plugging them into your profiler. For now, we are not going to enter into a discussion about electromotive force, thermometric sensitivity, linear vs. non-linear, TC conductor length, weld type, or even conductor material. What we are looking for here is that we are dealing with Type “K” thermocouples.
To characterize the thermocouples, you will need a few items. First, get a new set of thermocouples, where the welded bead has been pre-formed by the same procedures. The length of the conductors (the length of the TC) should be more or less the same for each TC, with the method of TC attachment remaining constant. My preferred method is Aluminum Tape (see TC attachment in Chapter 7 – Thermocouple Attachment Methodology and Materials).
Here are some suggested materials to use for characterization:
1. Three (or more) sets of Thermocouples (Type “K”, all from the same manufacturer).
2. One stainless steel plate (if the same plate is used for each run, it can be made of any material as long as it is capable of withstanding the designated highest temperature for extended periods of time). I recommend 1 or 2 mm. stainless steel, 12 inches long and 8 inches wide.
3. Squares of Aluminum tape to attach the TCs to the plate. Squares should be of equal size, with method of attachment consistent from TC to TC. Remember, we are looking for reproducibility.
Run the plate and profiler through the reflow oven. Be certain to document as much data as possible for the sake of the experiment and for reproducibility, especially with respect to oven set points, conveyor speed, TC attachment, distance of the profiler from the plate, length of the TCs, internal profiler temperature, and any other data point that varies, having an effect on the reflow process in the future.
Repeat the run several times. Like most data, the bigger the sample, the better the results. Take your time to set up this effort. If you are not getting similar results each time, then there is something wrong with the set-up of your experiment. Go back and check the repeatability of each step.
Here is an example of a run where I dumped my TC readings into a spreadsheet.
I had over 300 samples for seven TCs. The average TC reading for each TC is shown in the last row. The spread from the lowest value of 149.5°C to the highest value of 150.6 °C, shows a variation of 1.1°C, which is within the +/- 1.2°C rating for type K thermocouples. Since I went to all this effort, I decided to pick from my study: TC#2, #4 and #7 for profiling my next PCB, knowing that my variability is no longer 1.1°C, but now .3°C, which is the difference of 150°C and 149.7°C.
The point is this, it may seem like you are splitting hairs, but 1°C here and 2°C there has a cumulative impact. If you had a process that only allowed for, perhaps, a 5°C process window, which is not unusual today and you could cut out half of your variability due to your facility, equipment and TCs, wouldn’t it be worth it?