Stray voltage/capacitance: different voltage when device is on and off

[GuidoK]

I have a question (or an observation) about stray voltage/stray capacitance that I can't quite answer/explain:

I have here a CD player (but it pretty much could be any other non earthed device) that has stray voltage on the rca out.
This is present when the device is turned on, but also when it's turned off (provided the plug is put in the mains socket in such a way that the live is connected to the transformer, not the on/off switch). (I'm in continental europe where you can plug in either way)
This stray voltage is (I presume) the consequence of parasidic/stray capacitance.
And this stray voltage can be measured anywhere in the device, so on every IC pin, capacitor, the motorleads of the cd tray etc.

However, when de device is turned on, that stray voltage is considerably lower than when it's turned off.
At first I measured it with a digital/LCD voltage tester (similar to the one BigClive tested on youtube: https://youtu.be/uT-OwmLkpx4), when the device is turned off, it registers almost 110v (between the 55v and 110v marker), when turned on, it registers about 36V), I did a µA current measurement (to earth pin), where when the device is turned off, a greater AC current runs (about 3,5µA) than when the device is turned on (about 1,5µA AC), and I did a normal AC voltage measurement (to earth pin) where the voltage is also considerably higher when the device is turned off (53Vac), than it's turned on(33Vac).

So I was wondering why.
Because this stray voltage is present everywhere in both on and off position, I can't imagine this is due to the internal circuitry of the device itself (say the DC part).

I understand that with different measurement methods you will measure different voltages because the resistance value to ground with that measurement varies depending on what measurement you're doing, but it seems to be very consistent that there is a significantly higher stray voltage is present when the device is off than when it's on, regardless of the type of measurement.

The only/main reason that I can think of why this is, is that the transformer forms a stray capacitor (the first one seen from the mains outlet), where the primary winding is one conductor of the stray capacitor, and the secondary winding is the other conductor.
When the device is turned off, the complete primary winding (which now is 1 long ending wire) has a 230V voltage (to ground) along the complete length.
But when the device is turned on, the other side of the primary winding is of course connected to neutral. So the voltage on that primary winding starts from 230V (on the live side) to 0V on the neutral side, so there's a complete voltage drop over that primary winding and therefore the net (or average) voltage on that wire is much less.

So that the difference in stray voltage in the device between On state and Off state is not the result of what the internal circuitry is doing differently when its on compared to when its off, but that the difference in stray voltage is a result of that the initial average offered voltage (over that primary coil) is lower in ON state than in OFF state.

Does that make any sense or is the theory behind my explanation and findings all wrong and something else causes the difference in stray voltage between on and off state?
I couldn't find anything about this phenomenon in my search.

[Ian.M]

There are two common cases of stray voltage from non-grounded device PSUs:

Switched mode PSUs typically have a capacitor (commonly known as the Y capacitor, because it requires a class Y safety rating) between the negative of the primary side DC bus and output 0V to provide a return path for HF leakage current (at the PSU's switching frequency) flowing from the switching node driving the transformer primary to the secondary through the transformer's interwinding capacitance.  Its purpose is to reduce emitted EMI from the secondary side and load circuit.  However it also provides a path for the half-wave rectified line frequency voltage (with respect to ground) on the negative side of the DC bus to leak to the output, with an open circuit  RMS voltage of approx. 54% of the line voltage.  (YMMV as your DMM on AC V does not have an infinite input impedance and may not read true RMS.)

Linear PSUs (i.e. with a line frequency transformer) only have the interwinding capacitance coupling the primary and secondary sides.  This depends on the transformer construction with toroidal transformers typically have a much higher interwinding capacitance than single bobbin E+I core transformers and in turn those typically have a much higher interwinding capacitance than split bobbin E+I core transformers.  It all comes down to winding separation and area of adjacent winding surfaces.   Also there is the matter of the voltage on the surface of the primary adjacent to the secondary.  If the   supply Neutral is connected to the end of the primary nearest to the secondary, you'll see a much lower open circuit leakage voltage than if it is connected to the supply Line.

In all cases, leakage current can be redirected to supply PE (ground) by a grounded interwinding screen, however this costs winding area on the transformer, requiring a larger core, + more copper for the screen so is only usually fond in PSUs for medical equipment or particularly sensitive instrumentation.

Assuming a switch in the supply neutral^*, when its off, the whole of the primary has Line voltage on it, as does the primary side of the Y capacitor in SMPSUs, so you'll get a higher open circuit leakage potential and thus more leakage current.

* a switched neutral is usually due you crazy Europeans with non-polarized wall sockets, but can also be due to imported products from manufacturers that don't adhere to the convention that applies in countries with polarized sockets: single pole switches and fuses shall always be in the Line conductor, not the Neutral.  If you see a product with a polarized mains plug or mains inlet with one wire going to the switch and the other going to a fuseholder you know they were too cheap to fit a double pole switch or add an extra wire nut or crimp to connect the neutral to the  internal wiring, so expect the product to be a steaming 💩 !

[GuidoK]

Assuming a switch in the supply neutral^*, when its off, the whole of the primary has Line voltage on it, as does the primary side of the Y capacitor in SMPSUs, so you'll get a higher open circuit leakage potential and thus more leakage current.

Thanks, this is indeed what I thought was happening too. I tested this indeed on a device with linear PSU, I hadn't even come to how this would be on a switched PSU, but it makes sense that it acts more or less the same (more stray voltage when the device is off than when it's on (and the switch is on the neutral side)). The Y capacitor and parasidic capacitor in a transformer are both...capacitors