PID controller woes

I’ve been having trouble with my composites oven. I think it lies in two areas:

  1. it’s a little leaky – can be fixed with a bit of expanding foam and a more careful cut of the kingspan i’ve used to make it. I could build a wooden frame for it as well, but that seems a lot of faff
  2. The sestos PID controller I have is a bit hard to use – it’s well understood the chinglish instructions are abominable.

So, how to fettle this myself?

I’ve found this great page at mbed.org which gives a great description of what the variables mean and how to set up a PID. Now at least I am armed with the info I need to fix this.

I’m actually thinking of making my own and the plan is:

  • keep the triad switch for the oven, fan, etc.
  • write a C++ application to do the PID control from a PC
  • configure multiple ramping profiles that are saved and can be selected at will
  • the application can have multiple learning points, so the PID can be most efficient at each ramping step
  • use a pair of xbees to provide wireless sensing and control of the oven
  • have 3 temperature sensors – two in the oven and one for ambient temperature
  • have a set of fans in the oven as well as the initial fan so that hot-spots can be eliminated
  • logging so you can see what the temp was doing

Saying that, there’s also this quite sophisticated controller which does most of the above and has a serial interface and instructions in English. So, I could write the application to control the PID and just use it for logging – but I wouldn’t get the wireless control of the oven (mine is in the shed).

Your thoughts?

Choices, choices.

monitoring the engine temperature, pretty LEDs

Hmmm – when I build this engine, there are lots of things to sense (which go to the engine management system). However, there’s lots of data to actually display, which can be done on a piggy-back loom to the engine management.

For instance, there are multiple places to sense temperature, and other things, all grouped together as a set of systems. So, for fuel, the pump in the tank will warm it up, and the (to be fitted) fuel radiator will cool it back down again. So, I could put a sensor in just outside of the tank, after the regulator, after the fuel radiator, and in the return to the tank, so I can monitor fuel temperature and pressure. If I put some storage in the arduino I can over time, log most data.

Similarly, oil temperature before and after the cooler, as well as the head temperature inlet and exhaust are interesting, as is water in similar places can all be monitored, displayed and alerted. I could go the route Saab do, and have a black-panel button as well to keep the eye-clutter down at night.

So, rather than go for a gajillion gauges, I’m looking at going the micro-controller route, and the display shown here is available for $130; not bad for the eye-candy.