SMPS leakage currents present in the output pins

[akis]

Back in another thread I discussed laptop adapters and other power supplies (LED supplies, various wall plug adaptors etc) and outside of pure horror stories, there is a commonality: a Y1 capacitor between mains and output, supposedly to reduce EMI.

Looking inside some adaptors have a wire link between negative DC output and mains earth. Others have not. All however have the Y1 capacitor (blue) and it is this what I am discussing here.

This capacitor creates voltages in the 10s or 100s of volts at the output pins and is enough to give me mild jolts as well as light up a 5mm LED connected to the earth.

To avoid being accused of an incorrect testing method: I connect the oscilloscope probe to any of the outputs and the ground clip of the oscilloscope via a 100K resistor to the same output: you will see large voltage swings (10s or 100s of volts). And to qualify this further, neight output of the adaptor should have any connection to the earth. Many times the adaptors have no earth pin at all.

It seems that most adapters and power supplies suffer this. Am I right? Is it impossible to find a truly portable SMPS that is truly floating with respect to everything : mains and earth?

[wraper]

It seems that most adapters and power supplies suffer this. Am I right? Is it impossible to find a truly portable SMPS that is truly floating with respect to everything : mains and earth?

Branded smartphone chargers usually do not have capacitor. Seems that they usually have shielding between primary and secondary windings to prevent output floating at switching frequency. Without it touchscreen will be unable to work normally when connected to the charger (what happens with crap China chargers). Higher power SMPS is, less likely to find one without capacitor. Look for medical graded.

[Pjotr]

Look for medical graded.

Medical graded ones have the capacitor too. The only difference is that there are 2 capacitors in series, because medical grade requires 2x MOPP (Means Of Patient Protection). And they are smaller to keep leakage currents within the limits of medical standards (usually < 100 uA).

The other option is to use a class 2 one with a shield in the transformer as you mentioned. But those supplies are not common.

[akis]

I noticed the two blue caps too! I even wrote to them about it.

I presume that a linear supply, transformer followed by rectifier and capacitors will be 100% floating and will have no leakages to earth or to anything else. I will build one today and measure it.

The linear supply is bulkier, heavier and does not cope with 90VAC-250VAC as do the SMPS ones. Other than that it is better, it has no RF or anything at all.

[Pjotr]

Be careful with LF transformers too.  I measured about 1 nF with some toroids. Small PCB block transformers are around 40 pF - 100 pF and much better. And the switching noise of rectifying diodes can disturb up to the FM band too. Not all makers of linear power supplies put HF suppression capacitors in parallel to de rectifier diodes (but that is not a  leakage current issue).

[akis]

Here is a simple test for leakage currents.

Take a normal SMPS. Assume it is "floating" with respect to the ground.
Take a 0.5mA-1mA/1.2V LED (very common on ebay).
Ground its cathode.
Now connect a wire from the power supply's DC output to the LED's anode. Either the positive or the negative wire, it makes no difference.
The LED lights up.
Now put your body in between DC output and LED, say left hand touches power supply's output (positive or negative, makes no difference), right hand touches LED.
LED also lights up (you may need to wet your fingers).
Yes the current will go through your heart. You can also try with one hand, the other behind your back. But you get the point.

Verdict: there are leakage currents enough to light up an LED, leaking from from positive and negative terminals of the SMPS.

Verdict 2: without a mains isolation transformer, how are we certain that some component will not give up and suddenly we get a bit more than "leakage" currents on either of the output terminals?

[digsys]

S/Mode supplies WITH an proper Earth pin are usually never a problem. The RF noise from the Y caps has a place to go.
In these types having a R+C (usually 1M+0.1uF 630V) from the DC- side to E is also found, and helps clean up any noise on the DC- side.
I often add R+C in units without them. The DC -ve is technically still floating, just has a bit more leakage.
It's the Isolated / no_E pin mains plugs that are the HUGE problem. The RF always manages to get onto DC- and build up quite a voltage.
I've had to modify many pipe inspection trucks to stop operators getting zapped, usually when they unplugged the last Earthed device in the chain.
In some industries, Isolated mains plugs cause no end of problems.

[Pjotr]

It is not a question if there are leakage currents, there are always leakage currents! The question is more how much can you tolerate. For safety as well as proper operation of the apparatus.

An easy way to look at leakage currents: Put a 1K resistor across the input of a grounded oscilloscope and simply connect one of the outputs of your PSU to the input of the scope. Now you see the leakage current flowing trough the 1K resistor flowing to ground. With a modern DSO, peak, average and RMS leakage currents are easily measured/calculated too. Do this only with a healthy PSU with safe leakage currents! You can put a 100 nF capacitor in parallel with the 1K resistor to filter off unwanted HF noise.

For really low leakage currents (< 10 uA) you'll need an isolation transformer with a grounded ground shield.

[SteveyG]

Look for medical graded.

Medical graded ones have the capacitor too. The only difference is that there are 2 capacitors in series, because medical grade requires 2x MOPP (Means Of Patient Protection). And they are smaller to keep leakage currents within the limits of medical standards (usually < 100 uA).

The other option is to use a class 2 one with a shield in the transformer as you mentioned. But those supplies are not common.

Not strictly true - You only need 2x MOPP if the ME has applied parts.

[Pjotr]

Not strictly true - You only need 2x MOPP if the ME has applied parts.

SMPS that are marketed as "Medical grade" are designed to be 2x MOPP. Whether they are Class-1 (with earth protection) or Class-2 (double isolation). And have lower leakage current than general domestic/industrial ones.