It turns out you can get these apart without cutting things (like Big Clive did in the video I linked to in post #29). You can de-solder the wires from the screw terminals.
With it all out in the open, the 56 ohm resistor in series with the capacitor dropper runs at about 140 degrees when fed from the MK dimmer. It's quite cool down here in the workshop, so it'll be a lot hotter than that when back together and installed in a product.
But the AC voltage across the resistor is much lower than expected for the temperature of the resistor - which shows the difficult nature of these sorts of measurements. As a reality check, placing a DC voltage that is the same as the measured AC voltage results in far less heating, which shows that the AC measurement is a gross underestimate - even though the Fluke I'm using has an AC bandwidth of 100kHz.
With the addition of a mains isolation transformer, I'm able to observe the voltage waveform across the resistor on the 'scope, and see just how very spiky it is (and those spikes are very narrow indeed). A measurement with my Fluke 8920A (which is earth-referenced, uses a thermal RMS converter and has a bandwidth of 20MHz) finally gives a voltage measurement that correlates with the heat this resistor is actually making.
Let's consider the current. While the
trend of the earlier measurements fits my expectations based on previous experience many moons ago with these sorts of circuits (highest current in the 150V region), I was aware that the currents measured were likely to be underestimates because of the very high crest factor of the current waveform (and the "hot" smell from inside the unit). But as we now have a reasonably accurate measure of the voltage across that 56 ohm resistor, we can determine the current much more accurately - it's about 110mA, believe it or not! Roughly double the earlier reading.
The final thing to try was using my Rigol 1054Z, as this has a much higher bandwidth than the 8920A. It gave an RMS voltage that correlated to a current of 135mA.
So anyway, I hope this helps to demonstrate that there's more to this subject than meets the eye. I remember being fascinated by all this when meeting it for the first time, and I hope it's of interest to some.
Just to reiterate, this is when the unit is being fed from a triac-based dimmer which, for my particular dimmer at least, is clearly not a good plan! When used with "standard" sinusoidal-ish mains, it'll behave much more as expected. But don't forget a fuse!
BTW, now that I have one here, I can see that my earlier suggestion of a dead PP9 case is just about perfect for these. Plenty of space for the unit and a panel-mount fuse holder
