Fun With Temperature Data

16 October, 2007

There will be a real announcement of this on the main page later, but on November 16 during the West Racine Open House I will be presenting a short presentation a few times in the evening.

Discovering the Roast The best coffee in the world tastes like the worst coffee in the world if that coffee is not properly roasted. Learn about the coffee roasting process and try samples of various roasts at this event.

Presentation times will be decided later.

Recently we were having a little trouble with the older of our two espresso machines. The thermostat on the espresso boiler was going bad with the result being overheated water getting used for espresso. If the water for espresso is too hot, the coffee will have an unpleasant burnt taste. Fortunately, we have a great staff who noticed this problem first thing in the morning, shut off the machine, and did not serve the bad espresso. I made the appropriate temperature adjustments to keep the brewing temperature the same as our newer espresso machine, but it kept going out of adjustment within a few days. The thermostat had to go.

After considering my options, I decided that instead of replacing the thermostat, I would put in a PID controller. So far that is working very well. The promise of PID control is better tasting espresso through improved temperature stability. The potential for that is certainly there and we are taking advantage of it, but there are a couple points to be cautious of.

The controller we are using has an operating mode that attempts to figure out what settings are best. This automatic tuning does not work very well, so I would advise others attempting this modification to know their machine and understand the math involved. My settings give tighter control than either automatic tuning or the recommended settings for my machine. With this sort of controller starting to appear in new espresso machines, it will be helpful for a barista to know a little calculus and a little physics.

The other point of caution is a general concern with developing espresso blends with tighter control over the brewing variables. Laboratory style control over all of the variables is fine as long as you don't expect anybody in the real world to be able to get a decent cup from that blend.

The big gain with this is that it provides more information about what is going on in the machine and makes it very easy to change the temperature control. Now, given an espresso blend, I can figure out an ideal brewing temperature very quickly. In case you're wondering, for the latest Viking Blend Espresso recipe that's 194°F (90°C) which for this machine requires a boiler temperature of 200°F.

Chances are good that I will eventually make this modification with our other espresso machine, but it will be done on a day we are not open.

The other bit of hardware hacking I've done recently is adding a USB port to the roaster, allowing me to get temperature data on my computer. This is one step closer to my plan for reducing my paper use in the roasting area. Presently this is just at the proof of concept stage, but there has already been a benefit.

The temperature indicator on the roaster is not the one the roaster originally shipped with. That part is no longer produced and LED segments on the display started to go out. The new indicator gives me the option of 0.1°F display resolution, but there is some odd filtering going on that I do not like. My code does not do this, with the result that I am more consistantly getting measurements that match theoretical values for easily observable physical changes in the coffee.