DIY: UDS temp controller using Android

HogFan, when the time comes I may ask for your help in determining the best PID settings, since you seem to understand it better than me.

Update: I did my first smoke (heh) test last weekend. It was good and bad.

Good - The hardware worked and it was relatively easy to set up. The PID output (i.e. the blower speed) seemed reasonable. The graphing software works very nice and builds a pretty chart of the cook.

I have verified that the software runs on an original Droid using Android 3.7.

Bad - Program crashed a few times but worse it didn't keep any kind of a steady temperature. I started with the ball valve all the way open, which was too much, and it climbed past 300 (thermostat was set for 225). I kept closing down the valve until temperatures started to drop. When the temp got to around 225 the blower kicked in but temps kept on dropping. The blower is putting out plenty of air so I'm not sure what's going on. I fixed the crashes and I plan on trying again this weekend.

The bluetooth dongle I bought has a short range and that was annoying. I've ordered another one to try with a range (supposedly) of 50 meters.

For those that might be interested, the software can be downloaded from here: https://sourceforge.net/projects/udssupervisor/

An update and a request for some advice.

I updated the software and ran another test last weekend and still had problems keeping a constant temperature. The hardware/software worked great; I ran the controller for 5 hours and there were no crashes or disconnects. The PID algorithm seemed to be calculate good values. I ran this all on my old Droid and it used up only ~30% of the battery so I'm convinced that this is a viable platform for controlling the temperature once I figure out my problem. That's where I'm looking for advice.

I'm using the Minion method with lump. The smoker has a single 1/2" ball valve for intake. I got some lump pieces going and when the temperature inside the smoker got up to ~200 degrees I engaged the controller and set the ball value to about half-way open. Here's a graph of the result:



The red line is the pit temp, the blue (flat) line is the desired temperature, and the green line is the blower speed (0-255). (Ignore the magenta line for now.) As you can see the first hour or so was reasonable. Starting around 19:00 the temp begins to drop, even though the blower is going full speed. At about 19:30 I opened the ball valve a bit. You can see the temps rise for 45 minutes and then begin to fall again. At 21:00 I opened the valve a bit more, the temps rose then fell again. I shut off the controller at ~22:00.

I don't know why the temps keep falling even though the blower is going full blast. It looks like I still had plenty of lump in my basket, maybe around 2-3 inches deep. Any ideas?

As an aside, here's the initial screen:

How about here?

I'm using this 25 CFM blower: http://www.allelectronics.com/make-a-store/item/CF-339/12-VDC-BLOWER/1.html

I currently can't leave the ball valve all the way open because that lets in too much air with the blower off. What does the damper look like and how does it adjust?

Thanks for the pictures AustinKnight. It looks like the damper is a simple choke that restricts air-flow by making the intake opening smaller. I'll play around with my setup to see if I can restrict the flow from somewhere other than the ball valve.

The software is set up with a max blower speed. I'll try reducing the speed so it's not blowing in too much air. I had it on max because it was forcing air through the half-closed ball valve.

Thanks for the help/suggestions all, I appreciate it. I'll look into that Java PID library.

HogFan said:

I don't recall what pit you are using, a UDS?

It does appear the fan is way too big. I would get a smaller one but if you opt to throttle it down, do so on the inlet side of the fan, not the exit.

What tuning parameters are you using for the 3 PID parameters?

Click to expand...

The current PID settings I'm using are 5, 0.02, and 0. I stole those from an Arduino project (similar to the HeaterMeter). I'm using this on a 30-gallon UDS.

Weather permitting I plan on running some more tests this weekend with the blower turned down and a choke on the air intake. In the meantime I'm continuing to update the software. I've added a web server so the smoker can be monitored outside of bluetooth range. This will require that you have WiFi. My plan is to use my old Droid as a dedicated ATC for my smoker.

moontz has helped me out a little bit with the graphing library. If there's interest I could add some sort of interface to work with moontz's android app.

Yes, but this update doesn't have much good news.

I ran some more tests but the controller keeps overshooting the desired temp. I think the PID algorithm is throttling the fan too late so I will start playing with the PID settings. To make sure it's the settings and not the algorithm itself I swapped out my PID code for the WPI code you linked above (thanks for that!). That took a while because the WPI code had a lot of stuff that I didn't want/need. Even though the controller is overshooting the temp it's pretty cool to be able to chart a cook.

I did some re-wiring over the weekend so I can run the controller from a battery (I was getting tired of snaking an extension cord out of my second-story window). Like a complete moron I didn't pay attention to the polarity and fried my IOIO board when I attached the battery. *sigh* If I had paid an iota of attention to what I was doing it would have been averted but no, I'm an idiot. I can verify that an IOIO puts out a lot of smoke when you reverse the input polarity. Anyway, a new one is on the way and it'll take me some time to reassemble it.

While waiting for it to arrive I've been taking the time to work on the web server part and other additions. Once I get a basic web page working I'll move on to adding alarms.