Whilst that would ordinarily be both advisable and sensible! It has however been noted that the sensor works better at 5v even when hooked up to a 3.3v IO, either (or both?) the emonPi or emonTx are wired 5v by design despite the MCU on both working at 3.3V.
On reflection, it might be worth hanging on for Robert to comment/confirm. He has done some testing of the output from these sensors, whilst I know what I’ve stated to be true of the emonTx/Pi etc, I do not know for sure that the RFu is as tolerant as the later devices (or even if it is a concern).
If it works ok, run with it? If you have issues I would put money on it being either the supply voltage or (sorry to say) the RJ45, they are renowned for issues with the temp sensors, I fully relate to not wanting to cut off the RJ45, but it’s certainly the place to look if you get odd readings, even if you get too many pulses counted.
There are details (as best we can extract from measurements plus one that was disassembled) here:
I’m using the optical sensor on an emonPi at the moment, and it’s certainly susceptible to a fluorescent light (with a glow-tube starter) nearby being switched on, to the extent of tens of pulses if the pulse repetition period is set to zero. A few days ago, it saw 18 kW in one 10 s interval - the c.t. saw 154 W.
Ah interesting, My meter is in the garage with a number of fluorescent tube lights. I had a look at how my pulse sensor wired to the Pi directly and using the PulseInterfacer behaved. Looks like nothing out of the ordinary is reported.