Repair: DIADEMA coffee machine
Getting familiar with the machine
A 'DIADEMA' by C.B.C Royal First. Made in Italy in 2006.
All polished up and with the cover on:
This is a small-size cafe coffee machine. A must have for any home barista. It takes a long time to heat, but once it's up to temperature it seems to never run out of steam or behave poorly.
With the cover off:
There is some discontinuity between these last two photos. The first one only shows one black box at the top of the machine (a pressure operated switch). The other shows the pressure switch with its cover off and another, smaller black box next to it. This second box is the main controlling circuitry (discussed later).
When the machine is turned on early every morning, it starts 'clicking' rapidly and does so for a few seconds to most of a minute. Once warmed up it behaves.
Investigations and repair
Bad water tank switch
Underneath the water tank at the back is a switch to detect if there is any water. From above:
Presumably the spring stops the button from being pressed if the bottle is empty. I didn't test with different bottle weights.
The switch had seized shut, so it always reported that the bottle was in the back. If this switch is released then all machine operations cease: power is cut to everything except for the red light on the front.
A bit of cleaning and relubrication fixed this switch. It's inevitable more water will get on to it so I expect this problem will resurface in the future. I don't like the design of putting a mains switch right underneath a waterbottle that gets regularly taken out to be refilled.
This wasn't the cause of the main fault.
Investigating the clicking
I tried for many hours to locate the source of the clicking. At first I thought it was a solenoid, then a different solenoid, then a third, and finally the pump. I couldn't reliably track down where the noise was coming from: the whole machine seemed to reverberate with it.
As it turns out this should have been a hint.
I disconnected power to all of the solenoids and the pump. When I turned the machine on I could still hear constant clicking, but it was coming from the control box. The relays inside of it that normally fed power to the pump and solenoids were oscillating on and off.
Cracking the bottom off the control box:
Here we have two very clean circuit boards in a mostly sealed box. They're sealed in because they're right next to a steam-releasing valve on the boiler.
Taking them out is easy, they are only linked by a single ribbon cable:
The right board handles switching on and off mains to the various components of the machine and provides a power supply for these two boards. The left is only logic. They both looked impeccable with no visual signs of component failure.
Checking the power supply
In the picture above is a large, red box on the right board. This is a transformer, not a relay, outputting around 20-ish volts AC. Right beneath it is a smoothing capacitor, bridge rectifier and voltage regulator.
The AC is rectified and smoothed to about 21-25V DC. The relays themselves required atleast 18V to turn on, so they were powered off this 25V rail, rather than what the regulator outputted (5V for the logic? I didn't check).
After making sure I wouldn't blow up my scope I probed the voltage across the cap:
Hint: the '0' volts position is marked on the right side of the scope screen with a ground symbol and a '1'. 5V per vertical division. Does something seem strange to you?
This cap is as dead as a doornail. The relays were clicking because their coil voltage was dropping too low every 1/2 cycle of the AC power.
Presumably the capacitor would start behaving a bit better once it heated up, which is why the problem would go away if you left the machine on. I didn't test this.
Replacing the only electrolytic in the whole machine
It looks perfectly fine from the outside. I tried to measure its capacitance but it was still too large for my multimeter.
I didn't have any 330uF capacitors at hand. Everything else I had that had a higher capacitance was physically too big to fit inside the case. Instead I made a daughter-board for 3 100uF capacitors:
Scope shot of the voltage rail after I replaced the cap:
1V of ripple is still too large for my liking, but perfectly fine to feed a regulator and relays. If I had designed this then I would have preferred to chuck some more faradicals in.
Given that the capacitors are:
- in a sealed box
- right next to a voltage regulator
- at a high point right above the boiler
- eating a volt of ripple (I didn't measure ripple current)
... I suspect that this problem will occur again in the future. I can't increase the ventilation to the control box without exposing it to steam, nor can I easily move it without having to rewire the whole machine, so I'm just going to have to live with it.
Other various notes
Previously the red neon in the power switch died:
I replaced it with four 1/4W resistors in series and a couple of red LEDs. The light is now not as well diffused as what the neon provided, but at least you can now tell when the machine is on.
The valves in the plumbing of this machine have a very clever design. Take this valve, which looks like a T-piece on a piece of copper line:
Inside this silver stub and the T-join is the actual valve. You slide a coil over the silver stub and this coil is what activates the internal valve to let water/steam through:
Not a drop of water can be spilled, nor can any water get onto the coils. Brilliant. Probably very common too, I have not worked on anything electro-waterical before.
Coffee-drinking parents are now feeling absolutely buzzed.