My multimeter is capable of AC current, which I connected directly between the negative on the PSU output and the earth pin of the adjacent socket. MM is saying 0.04mA - picture attached just to check it's on the correct setting.
I started on the 10A setting and worked down until there was a measurement.
JohnRob:
I've tested several old AC adapters and get at most a few volts.Are you still measuring voltages in the 90's from the adapter output to ground?
Where is the ground you are measuring to?
What country are you in ?
Also what else is connected to the Arduino?
Yes, I'm measuring ~90v from the negative of the PSU to an earth point. I'm un the UK.
I'm measuring ~90v from the negative of the PSU to an earth point. I'm un the UK.
From your working I'm assuming the PSU is not connected to anything. This being true I'm at a loss for further recommendations, except that you should ground the PSU negative output.
Just thinking of the problem, here in the US the plug in power supplies use only two wires so the third (ground) has no effect on the PSU output.
US = one "leg" of the mains is at or near ground potential
the other "leg" is at 120 VAC
UK = one "leg" of the mains is at or near ground potential
the other "leg" is at 230 VAC
I don't think leakage current would go from a few volts (in US) to the 90V you are measuring.
I would measure from the Ground you are using to measure the 90V from, to each socket of the mains connector. If possible I would measure:
From the ground, and both Mains sockets to a water pipe (assuming you don't have plastic plumbing). You might use a standard extension cord to extend the meter lead.
The voltage means nothing unless you know the resistance. In this case, 10M. 90V/10M = 9uA.
apeng:
My multimeter is capable of AC current, which I connected directly between the negative on the PSU output and the earth pin of the adjacent socket. MM is saying 0.04mA - picture attached just to check it's on the correct setting.
I started on the 10A setting and worked down until there was a measurement.
Correct procedure.
So you have a source of voltage which will provide either 90 V or 40 µA. Its effective resistance impedance is then 2¼ Megohms. This is consistent with the capacitance between primary and secondary of the power supply transformer (I see no point in calculating the approximate actual capacitance).
My conclusion is that this is a safe situation (given that the construction of the power supply is adequate to exclude sudden failure) and unlikely to be improved further. You are able to feel the 40 µA as a "tingle" which is perfectly reasonable, a thousand times as much would be quite dangerous.
This level is unlikely to affect the fish if indeed there was another ground in the tank to make a circuit, but you could simply ground the negative of your assembly.
Thank you confirming.
For future reference, I want to check I measured this correctly. There was no load in and the meter was connected directly between the negative PSU and earth pin in socket, with the +12v simply floating. Should I have added a resistor inline ?
I assume that because the current was low measuring like this is fine - however, on the other hand if the current was higher, then either the fuse in the meter would have blowed or the MCB in the house fuse-box would have tripped ?
Thanks
It would make you quite extraordinary if you can feel 40uA, especially on your fingertips.
The best I got was 300uA in between my fingers.
apeng:
strange tingling effect on the metal surface of the sensor
Hello,
When reading the other posts we can note that the power supply may be with the insulation compromised,
But even for the new power supply, it is recommended to use earthing.
I tested the sensor here, and the metal housing has no continuity with any of the wires. Then it is possible that water has seeped into your sensor, this compromises the insulation and can also promote other problems, such as leakage of current between the pins of the sensor, and oxidation of the terminals and wires.
I also use sensors of this type, with water, but I used a resin to protect the non-metallic part against infiltrations.
polymorph:
It would make you quite extraordinary if you can feel 40uA, especially on your fingertips.The best I got was 300uA in between my fingers.
In this case, I think an 8:1 discrepancy is easily explained by individual variations (in skin conductivity).
Sample size of 2? Hardly conclusive. 
Hi,
Can you post a picture showing where you have the sensor and if it is in a bracket?
Thanks.. Tom... 
I knew someone who was convinced his new digital watch (back when LCD wristwatches were a new thing) was shocking him. He gave as proof that if he touched the bezel, he could feel a shock.
In fact, he only felt it when rubbing. It was actually his skin catching and releasing like a rosined bow on a violin string causing a vibration that he interpreted as an electrical shock.