It is one of these quiet Friday nights again, and I feel like writing
something semi-random from the past couple of weeks. In fact, I was invited
to a night out, but I happened to feel particularly dull and worn-out even
compared to my usual weekday state, if not quite physically ill.
I did have a moment of fun playing with a newly bought volume/expression pedal, a Boss FV-50L, which in my Triton Le also works as a general effects controller. Should be nice for those few surf-rock/blues tunes where I have thus far resorted to left-hand volume knob tweaking. In addition, it might be fun to build analogue or FPGA-based effects, now that I have the physical controller that is comparatively hard to build otherwise. Complete effect pedals are almost as cheap, but it is good to keep things modular a'la Unix.
Last week's undeniable highlight was a trip to the hydropower plant at Vaajakoski. Class trips are no less fun, even if they only last a couple of hours. There is always the special atmosphere of being away from the usual stuffy classroom setting. In my mind, there was the obvious comparison to the industrial excursions of technology students, and likewise the main point was not learning about the science, but networking. The students also got to partake in a quiz about the technical details, with a not-too-shabby prize for a lucky one. On the return trip, I yielded to the youth's request to stop at Panda for some sugary treat. Not that I would sneer at some cherry and strawberry liquorice myself.
This weekend's hack was a bypass surgery to my Tefal Quick & Hot water
heater. It had started acting up this summer, after nearly three years of
service. Usually it would require a mains reboot before each cup, but
recently the duty cycle was going further down, and it would often stop
The belly of the monster revealed a number of embarrassing, cheap design decisions. Even before this, I had noticed that a tiny amount of water is leaking from the bottom, no doubt having to do with the electrical issues. The pipeline comprised a number of separate components and silicone hoses with simple plug-in joints, so not surprisingly some water would leak through over time. There was indeed a little rust at the heating section. I would have preferred a solid metal build, but you would probably have to pay through the nose to get such industrial build quality.
At the heating section, there was also an obvious electrical culprit. It was connected to the control board with spring-loaded tabs, and there was clear oxidation and sparking residue on the surfaces. Cleaning these up gave a momentary relief, but the problems soon returned. It was time to show the French engineering marvel a little solder.
Prior to the operation, I did some measurements. I had already figured that the halting problem was due to control issues, instead of high voltage parts, because of the on-off nature of the symptoms. The temperature sensor on the heating section had the same spring-loaded connections, and it would be much more sensitive to contact glitches, compared to the high-current parts that could practically weld themselves. Playing with a multimeter indeed showed that its resistance would be way off with the springy hookups, compared to a direct connection.
Interestingly, the sensor showed a negative temperature coefficient. Thus, contact glitches that increase resistance, would show up as low temperature, leading to overheating. A sensible design would take this into account, though, instead of keeping the heat on when a sensor is detached. There might also be some galvanic effects from the water leaks on two distinct metals, but they should mostly be balanced by similar connections at the other end.
In the end, I soldered pieces of mains lead to parallel each of the heater connections, for both the sensor and the heater just in case. Since then, the Quick & Dirty has worked with zero glitches. One must wonder why the springy tabs were ever chosen instead of solder, particularly in an environment prone to oxidation. Assembly might be a little faster, because the only solder found by default is on the circuit board. But for a small price difference that causes such obvious problems, it is simply embarrassing.
Having studied the insides for quite a while, I am already thinking of possible improvements to the original design. For example, the temperature should be adjustable, or at least a little higher — I now get about 70 °C in the cup, though the marketed 80 may actually hold at the machine itself. This discrepancy may in part be due to the water left inside the pipeline, and the heating section actually holds quite a volume. When the pump starts up, this cool water is forced out, instead of letting it heat up first.
This could be fixed by having the heater and the pump under separate controls. Now there is only one relay on the circuit board, so they are probably under a single control. A completely new, custom control board would be nice, and some of my process-engineering experience could come in handy ;) Then again, the mechanical design has obvious problems, and simply improving the electronics may not be worth it. But at least it is fun to think about such issues, and be able to fix some of the simpler problems, instead of buying a new one like a docile consumer should :)