Dimmer under potential but no load

An 82 ohm resistor would dissipate about 175 watts; indeed, a 15 watt bulb may dissipate that much power briefly. Something the size of a 15-watt bulb that dissipates 175 watts is going to get very hot, however, and when tungsten gets hot its resistance increases by an order of magnitude. You may have been expecting your 1K resistor to dissipate about 1.5 watts, but in practice it would have dissipated about 17.5 initially. Most resistors don't nicely increase their resistance with temperature the way a lightbulb filament does, however, so your resistor may have continued to dissipate that much power until it melted, whereupon its resistance might have actually decreased, causing it to try to dissipate even more power, until things went downhill very severely.

I would suggest that you might consider a Christmas decoration that's designed for a C7 bulb; such bulbs draw about 1 Watt. If you wanted to use a resistor that drew less power, two 33K resistors half-watt resistors in series would probably be reasonably safe if they were somewhat ventilated (the resistors should dissipate less than 1/8 watt each, but half-watt resistors should be used so that in case one fails shorted the other would be able to handle the 0.44-watt fault power) but might not let through enough power to operate the touch switch (if it does work, keep adding 33K resistors until it doesn't work reliably, and then remove one). The fact that incandescent bulbs have a much lower resistance when cold than when heated allows a much smaller energy-wasting element to be used for a given application than would otherwise be necessary.

The G3MB-202P 5V input (Zero crossing and built in snubber) is probably the problem. The data sheet says the operate time is '1/2 of load power source cycle + 1 ms max.'.

^{simulate this circuit – Schematic created using CircuitLab}

Figure 1. Simple SSR has zero-cross on/off switching - EDN.

This circuit demonstrates the action of a zero-cross detection circuit.

• If \$V_{tri1}\$ is low (above but close to zero) and Q1 is turned on by photo-action from D1 then SCR1 will be triggered. This in turn will pass enough current through R6 to bring TRI1 gate voltage high enough to trigger.
• When voltage exceeds a certain level Q2 will be biased on. The collector voltage will fall and there won't be enough to turn on SCR1.

The effect is that TRI1 can't turn on unless it is triggered close to the zero cross.

Figure 2. Triac waveform.

When your master dimmer does this to the mains waveform the downstream SSRs go from zero (where there is no voltage available to trigger the triac) to 'on' where the voltage is high enough to disable triggering.

Your best option in this case is to replace with the G3MB-202PL.