Trojan Coffee Room Machine Returns

MKalus writes "It seems that when they turned it off it wasn't quite the end to the machine after all. The german magazine "Der Spiegel" bought it and got it repaired. And now it is online again, not in the Trojan room, but the same machine." You just can't keep a good coffee machine down.

Comments

Re:10 Thousand marks for what?

[Publicus]

Per the /. story about its retirement. It was the first ever webcam, and it was used so people knew when there was coffee and then there wasn't. Talk about a technology stemming from a developer's itch!

Ethernet port and built in webserver, eh? I don't see anyone stopping you, go to it man!

Re:10 Thousand marks for what?

[wik]

I actually did this to my coffee machine (Winter Break, 1998). At the time, it was a Mr. Coffee 12-cup commercial coffee machine (since disposed of, pictures at: http://www.rabidpenguin.org/images/more/pic00023.j pg . The machine had a bunch of sensors attached to it, including temperature of the pot (red tape on the front), air temperature, amount of water in the pot (see the ribbon cable). Everything was attached to a Basic Stamp II, which had a serial cable to a 486 Linux webserver (mrcoffee.res.cmu.edu at the time).

It turns out that sensing the amount of water in the pot is quite difficult. If you use a scale, it has to handle heat, humidity, and steam if you put it under the pot. If you put it under the machine, you will also have plenty of water screwing up measurements because it stays in the filter. I thought about bouncing a laser diode over the surface of the water, but that never materialized. I also tried measuring the capacitance of the coffee between two places (more coffee = more conductive dielectric). That didn't work. Coffee and tea are great conductors.

Finally, I took a plastic ruler, drilled holes in it and hooked wires from a ribbon cable up to it, at a regular spacing. The coffee would short between a pin at a certain height (each pin was attached to an R2 ladder) and the ground pin at the bottom. This actually worked reasonably well! (If you could stand a ruler in your coffee pot!)

Oh, I didn't want to figure out how to write a web hit counter CGI script, so I had the stamp store the number of hits in the stamp's EEPROM! Much easier! I still have the code and the hardware lying around, though the coffee machine is long gone (last attached to a DECstation 5000/260, actually).

Upgrades

[wackybrit]

Why don't more nerds in universities/companies do things like this? There's a whole heap of stuff you could do.

1) Hook up water and a supply of coffee direct to the machine. This way you could control the ENTIRE coffee creation process remotely so that you have coffee waiting for you.

2) Run Linux on it and have the coffee machine actually control the webcam.

3) Implement a thermostat so that you can go get coffee when it's at a nice drinkable temperature, rather than getting burnt straight away.

4) Hook a car battery and some wheels up to it, and make it mobile. Then program it with everyone's coffee preferences, and have it work out a route around the building delivering coffee. Y'know, just like those robots at the start of Short Circuit.

Re:10 Thousand marks for what?

[wik]

I like this idea. I'm just not sure where you'd mount the ball on a stick. If you mount it on the machine itself, you'll have problems getting the pot out. If you mount it on the pot, you're going to have to make it removable, so that you can clean the thing! (Yuck)

At any rate, I'm not quite sure how you would get a measurement out of this. Attach the stick to a loose potentiometer? Not to mention, the stick would have to have a useful range of movement 80 to 90 degrees in order to capture the information that you want. My guess is that it would be less accurate than the ruler, but if you just want a ballpark measurement, it might work fine.

Re:10 Thousand marks for what?

[wik]

Yes, this would work well for a binary IsFull or !IsFull reading. I actually wanted to know how many cups were in the pot (and especially when it was empty, as it is in the common case). A ping-pong ball might work better for the binary measure, though it is a little large. Calibration might be hard for anything more accurate with that setup.

The advantage of the ruler was that I only had to calibrate it once (the resistors were fixed and outside the coffee machine) and it was easy to remove for cleaning. It also had an obvious failure mode (sometimes it would read -13 cups, then you'd know there was short somewhere).

The point of this project was to have fun, of course, not to make a coffee machine that people would ever drink from. Overengineered? Probably. But it was cool to have an LED flash everytime someone hit the webserver. And hey, doesn't everyone want to know the temperature of their dorm room to within 10 degrees rankine? (yep, it reported kelvin and rankine)