That Tingly Feeling (SMPS)

[FauxFrench]

Thanks a lot for the explanation, Rod. You are the kind of expert we needed. I have never designed commercial equipment.

Have you spotted any safety risk with the three "bricks" we have been discussing?

 

[Rod Coleman]

The leakage current is in the specification, usually, and it is not considered to be a safety hazard.

Still I agree with DF that we should not connect these cheap-looking supplies to anything without a solid safety-earth, connected separately. The quality of insulation of some of them does not inspire confidence!

For audio use I would look for something with a sub-1mA leakage current specification, to have some chance of controlling the common-mode noise.

 

[Elvee]

So I decided to crack one open last night.

The earth connection from mains was not connected. There is an X2 cap across the live-neutral. Decent separation between primary and secondary.

I tidied up some of the big globs of solder and ran a heavy, double insulated single core from the earth to negative to dissipate that AC.

There is a simple way of getting rid of the tingling, without any additional earth wire: you can simply remove the cap Y2, it is the "culprit" (it may not be alone though).

Doing so will in no way compromise the safety, the opposite in fact since you remove a (remotely) possible letal conduction path, and the supply is obviously built as a class II device.
In theory, it could compromise its EMC compliance, but that's something entirely theoretical since the main role of Y caps today is just to comply with the letter of the regulations, with a total disregard of actual impact in a real environment.

In practice, there is a very small chance it makes matters worse in this respect, but there are large chances it actually brings improvements, esp. regarding "noisy grounds" in interconnected environments.
In such a context, earths are often synonymous with hell (which is not a valid reason for lifting them!), but Y caps often bring a different kind of inferno: choose your evil.

Having a true, class II device here and there generally brings some welcome relief (based on 50/60Hz transformers, not SMPS)

 

[29285]

Removing the y-cap in a class2 ac-adapter will rise conducted noise level on mains by approx 30..40dB. Whether that disturbs real world, might be debatable as EMC standards are based on bygone AM/FM analog radio requirements. A valid compromise might be connecting the y-cap between PE and secondary GND.

 

[PRR]

> EMC standards are based on bygone AM/FM analog radio requirements.

I still listen to analog FM and even AM radio. While my main QRM is 60+Hz trash from bad power lines (and weak signal), I don't need more digi-hash on top.

 

[avtech]

Would it be preferable to connect the Negative output to the Earth via a capacitor or by dead short as I have done?

 

[FauxFrench]

The dead short you have made is the most safe as it will switch-off the safety relay in case of a short between primary and secondary.
The 200nF I measure on my HP brick has a 16K impedance at 50Hz which leaves only 14mArms to ground with 230Vrms. I believe my safety relay is invoked at 30mA. 200nF seems not to be safe, it should be around 1uF.

 

[29285]

If you want absolute safety, do a hard wired grounding to PE.
If you want low noise, excluding hum loops, a 100nF cap is the preferable solution.
Together with 1nF y cap this gives a cap divider 1:100 - nothing you will feel with your fingers.

You may connect two back-to-back wired rectifier diodes parallel to that cap to bypass any harmful currents and activate the RCD in case of a fault. The residual voltage is restricted to -/+1V without creating a hum loop.

 

[FauxFrench]

If you want absolute safety, do a hard wired grounding to PE.
If you want low noise, excluding hum loops, a 100nF cap is the preferable solution.
Together with 1nF y cap this gives a cap divider 1:100 - nothing you will feel with your fingers.

You may connect two back-to-back wired rectifier diodes parallel to that cap to bypass any harmful currents and activate the RCD in case of a fault. The residual voltage is restricted to -/+1V without creating a hum loop.

You write "back-to-back". Shouldn't it be "anti-parallel"?
Else, a good solution.

 

[DF96]

Whether that disturbs real world, might be debatable as EMC standards are based on bygone AM/FM analog radio requirements.

EMC standards are based on not seriously polluting a limited natural resource - the RF spectrum.

EMC enforcement often seems to be based on the assumption that nobody uses the spectrum any more, yet from time to time we hear that emergency services radios don't work too well when handling an emergency.