Macho coffee machine makers seem to enjoy using a pump that can squeeze the heck out of your coffee, but that's often not what you want. I fiddled with the OPV and found a workable pressure point, but that made for a substantial bypass flow. My carefully PID controlled boiler was cooling off pretty fast as most of the water went straight back into the tank. And I still I found myself having to grind my coffee very finely and crush it with the tamper to slow down the flow rate, so I decided to control the pump flow rate electronically.
I'd seen some projects that use a lamp dimmer to slow the pump, but that seemed a bit primitive. There was a better way: All of these machines use a vibratory pump; how about we make it vibrate at a lower rate. The idea appealed, so I sketched out a schematic using a microcontroller to switch a solid state relay. The microcontroller detects the mains positive-going cycles and disables the pump for a fixed percentage of the cycles. For example, to get a 20% pump flow rate we only had to ignore four out of five positive-going cycles.
The unit can run freely in a manual mode, or it will step through a program in 5s intervals. There's a display and an incremental encoder so you could teach it a program by pressing in the button and spinning the knob to select a percentage flow rate. The unit will remember up to ten 5 second segments as a brew flow-rate profile. The profile is stored in non-volatile memory so next time you turn Silvia on it comes up with this profile. This has the advantage that it stops the pump after your programmed sequence so it lets you do other stuff while it's pulling a shot.
The circuit uses an Atmel 8-bit micro, the ATMega88P, and a quite small DIP solid state relay. The unit derives its power from the Miss Silvia coffee machine and clicks to the side of the machine using some strong rare-earth magnets. While it's connected to a Miss Silvia it could be used with pretty much any vibe pump machine.
Here's a video of the unit in action:
For those who are curious, here's a schematic:
And a double-sided PCB layout:
Development under way:
Construction near completion:
Testing:
Sooooo ... does it improve the coffee? Well, yes, because it gives me many more options for grind and tamp. But I'd have to say that the biggest improvement I made to my coffee was buying beans someplace other than Costco!
The project took (guessing) about 100 hours of my time and a few PCB versions. As usual the first one ends up getting hacked around as you 'edit' the circuit. If someone wants to make a copy of the unit, do remember that there's mains voltage on the board. It'd be pretty easy to hurt yourself with this thing, but I guess if you're happy to tackle a project of this complexity you're probably not a neophyte.
Chris