You'll be pleased to hear that there are no CPC 3.3V hardware (that I am aware of) that has this issue. The C4CPC for example connects to the CPC via a Xilinx XC95144XL which has 5V tolerant inputs and a 74LS245 Buffer which is also a 5V part.
So you can sleep well knowing that your Amstrad isn't about to die (at least not due to some 3.3V hardware being connected to it.
Bryce.
C4CPC is a 3.3V hardware that interface nicely with the 5V CPC
As Bryce said the pld used is 5V tolerant, and all signal from the CPC are going to it.
The output (data and acid output) are converted to 5V. This conversion is only needed because the 3.3V output where not high enough to be interpreted as 1 by the CPC.
Now a bit of warning regarding the 5V. While the PLD inputs are 5V tolerant and connected to IO and not straight power, the upper margin is small.
Using a power supply that is higher than the 5V + 10% may damage the PLD input, hence the C4CPC.
So be sure to use proper power supply, mainly when using switched ones. While the voltage reported by a multimeter may looks OK, you may have ripples that will be more than the max 5.5V with light load.
BTW, why is that thread in the German section