I've got a number of battery powered devices which require periodic recharging. I'm using notebook chargers to provide for the input voltage and they work fine… most of the time. Occasionally they fail (produce all kind of strange behaviors).
I've isolated the problem to a missing earth connection. Approximately half of the chargers has two pin mains inputs (no earth), while the other half has three pin mains inputs (including earth). The problem only shows when we use the two pin (unearthed) power supplies, or when we purposedly leave the earth pin unconnected on the three pin power supplies.
The three pin devices are Lenovo brand, the others are unbranded. All of the power supplies are switching and give 19V 3A output, since they are intended for notebook charging. Breaking them apart they look similar enough; we see the expected two stage switching (high voltage, transformer, low voltage), linked by a feedback optoisolator AND some capacitor, which is the culprit for leaking a small amount of the mains voltage into the low voltage section. The three pin models have earth connected directly to this capacitor in the low power section, in order -I think- to eliminate the common mode voltage. Which it does… provided earth is connected properly.
In short, the power supplies give the expected output voltage plus a common mode mains voltage (220V here) which is occasionally picked by different sections of our devices producing erratic behavior including internal charger circuitry burning up, li-ion protection FETs jumping in the air, etc. Unfortunately, my device has a rather big metallic case which is connected to circuit ground! My guess is that parts of my device (its case, internal cables, even ground planes) are picking up the other part of the 220V signal, creating potentials in areas that can not be prevented by the usual TVS I/O protection.
I need to deploy these devices to rural areas where there is little chance of having a proper earth connection, so I'd need some strategy to avoid the problem outright.
I'd like a simple hack that I can apply to all my devices to prevent this kind of weakness. That would be the perfect scenario. A less desirable scenario would be something I could do to the power supplies… for instance, why do they have a capacitor connecting the high and low voltage sections? Can I remove it? The least desired solution would be replacing the power supplies entirely for others with proper galvanic isolation: a lot of money! and no warranties that a new one would not have the same problem. Finally, there's no way I could completely cover my devices with shields, since they have RF antennas which could be in part responsible for the mains noise picking.
Thank you in advance.
Best Answer
They have this capacitor to make sure the low voltage and high voltage part are somewhat related to each other, even when no earth is present.
High frequency transients on either side could otherwise (slowly or quickly) induce a voltage across the transformer barrier of many kVs, which would potentially cause breakdowns of said barrier, which would be bad to a scale that's not even close to the scale of the graph on which your problem registers.
So... do not remove it.
What you can (try to) do in the device is just what a good filter would do: Keep everything above a few Herz out and conduct it away.
How you do that, exactly, depends on the application and what is tolerable. Officially you are not supposed to conduct incoming crap to an ungrounded metal case, on account of "tingly feelings". If you could ground the metal case externally, that would be the ideal solution where you don't really need to do very much else.
In either case, externally grounded or not, a filter may help:
simulate this circuit – Schematic created using CircuitLab
I am assuming your case-to-internal-negative cannot be reversed for whatever reason.