Voltage on heatsink

[fatcat]

I have this old receiver with an atsc tuner that I am using for OTA channels. It's always been a bit finicky. I was checking out the power supply and accidentally touched a heatsink that has a rectifier or something attached, and I got zapped. I checked with a meter and the heatsink has 117vac on it. Could this be normal or is it weird?

[eccerr0r]

Re: Voltage on heatsink

Well, good thing you were able to walk away from it...touching any live SMPS can be deadly. Anyway, if it's normally covered by some plastic so that you can't touch it, then it's fairly common - one of the heatsinks tend to be hooked to the line input, after being rectified and filtered. Not all of them are like this but it's fairly common.

[budm]

Re: Voltage on heatsink

That is why I always suggested to at the very least plug the TV into GFCI outlet if you do not have Isolation transformer.

[stj]

Re: Voltage on heatsink

but would a gfci trip if the leak was after a full-bridge rectifier??
the fantom drain is sourced from both poles!

[Per Hansson]

Re: Voltage on heatsink

It's a good question stj.
I've been shocked exactly like described:
By touching the metal chassis that was grounded and simultaneously the primary side heatsink that was live.
I later measured 160VDC on the heatsink.
It's by far the worst shock I've ever had. (Meaning it carried allot of current & wasn't just static buildup or anything like that).
The outlet was however not GFCI protected due to running from a UPS.

So I had to do some tests now: finding PSU's that have live heatsinks was not a problem!
One had only AC present of about 30v, I did not test it.
Another had 160VDC and 130VAC, it tripped the GFCI instantly when I tried starting it up with the primary heatsink connected to ground through a 10Ω resistor.
A third one also had 160VDC and about 130VAC, in any case it seems pulsed DC, just like would be expected.

However on the last PSU I did not connect up the resistor first, but did a more "real world" test where the PSU was running and then I connected the resistor.
It tripped the GFCI instantly but the sparks from the resistor where really impressive!
This tells me one thing: even if the GFCI can trip we still have large bulk capacitors that store allot of energy:
Of course the GFCI can't protect us from that since there is still a return path to ground available...

[eccerr0r]

Re: Voltage on heatsink

I think it still should trip, current is still leaking towards ground instead of the other pole.

Worth an experiment? Not too hard to concoct one... (No, shouldn't do it ElectroBoom style. )

Ah looks like Per Hansson did do the experiment. That 10 ohm resistor is a bit too low however, it should trip with much less current; 10 ohms is getting close to where it would trip a regular circuit breaker (not quite, but getting there...)

GFCI should trip with milliamps of current, a 10K ohm resistor should trip it I think....

[Sparkey55]

Re: Voltage on heatsink

I think it still should trip, current is still leaking towards ground instead of the other pole.

Worth an experiment? Not too hard to concoct one... (No, shouldn't do it ElectroBoom style. )

Ah looks like Per Hansson did do the experiment. That 10 ohm resistor is a bit too low however, it should trip with much less current; 10 ohms is getting close to where it would trip a regular circuit breaker (not quite, but getting there...)

GFCI should trip with milliamps of current, a 10K ohm resistor should trip it I think....

A properly working GFCI outlet should trip at less than 2 milliamps.

[Per Hansson]

Re: Voltage on heatsink

I think it still should trip, current is still leaking towards ground instead of the other pole.

Worth an experiment? Not too hard to concoct one... (No, shouldn't do it ElectroBoom style. )

Ah looks like Per Hansson did do the experiment. That 10 ohm resistor is a bit too low however, it should trip with much less current; 10 ohms is getting close to where it would trip a regular circuit breaker (not quite, but getting there...)

GFCI should trip with milliamps of current, a 10K ohm resistor should trip it I think....

Yea, looking through my junk bin i found a 17k resistor, it does not trip it. (~10mA)
Around 7.5k does trip it though (~20mA) but then I got tired of wiring resistors in series

Anyway the experiment was to see how much power I'd see, sparks flying signify that we are dealing with allot of energy. (10Ω is afterall 16A at 160VDC).
The higher current I pull through the GFCI the quicker it should trip.
My theory was as I said that the bulk filtering capacitor provides that energy even after the GFCI has tripped.
However I just tested that: the bulk filtering capacitors are charged to 330VDC, they quickly dropped to 160VDC when tripping but then stayed there.
So I next tried connecting the 10Ω resistor to the live heatsink but the voltage did not drop any faster at all.
So it seems indeed that a GFCI is a very good protection even after a bridge rectifier
(On second thought this might be due to the APFC circuit, a PSU without APFC might not behave the same way, i.e. it might shock you...)

A properly working GFCI outlet should trip at less than 2 milliamps.

Most common GFCI are rated for 30mA, the one I used was too.

[Sparkey55]

Re: Voltage on heatsink

Yea, looking through my junk bin i found a 17k resistor, it does not trip it. (~10mA)
Around 7.5k does trip it though (~20mA) but then I got tired of wiring resistors in series

Anyway the experiment was to see how much power I'd see, sparks flying signify that we are dealing with allot of energy. (10Ω is afterall 16A at 160VDC).
The higher current I pull through the GFCI the quicker it should trip.
My theory was as I said that the bulk filtering capacitor provides that energy even after the GFCI has tripped.
However I just tested that: the bulk filtering capacitors are charged to 330VDC, they quickly dropped to 160VDC when tripping but then stayed there.
So I next tried connecting the 10Ω resistor to the live heatsink but the voltage did not drop any faster at all.
So it seems indeed that a GFCI is a very good protection even after a bridge rectifier
(On second thought this might be due to the APFC circuit, a PSU without APFC might not behave the same way, i.e. it might shock you...)


Most common GFCI are rated for 30mA, the one I used was too.

It only takes 1 milliamp to stop the heart.

[eccerr0r]

Re: Voltage on heatsink

Well the energy stored in the capacitor is limited, once it's discharged, the person getting electrocuted should be "released". The main danger is the danger coming from the mains that will keep a continuous flow going.

30mA is huge, should be much less than that for home wall sockets. 10mA is enough to cause problems in humans as it is - enough for a person to lose control of muscles.

[stj]

Re: Voltage on heatsink

A properly working GFCI outlet should trip at less than 2 milliamps.

standard rcd's in the u.k. are 30ma

[eccerr0r]

Re: Voltage on heatsink

In the USA, the home wall GFCIs that were meant to protect against electrocution should trip with only 4 to 6mA according to . This does jive with the 10mA AC number that will disable a human from pulling oneself away from the shock source, but not enough to be a nuisance trip while driving non-resistive loads that may be coupled to ground.

[Per Hansson]

Re: Voltage on heatsink

A standard PC power supply leaks a couple mA through the Y safety capacitors.
It would be very annoying with GFCI outlets rated at 4 to 6mA, due to the following:

Leakage current in power supplies may occur due to the EMC filters, which utilizes Y capacitors between the live and neutral conductors. This causes some leakage current to flow from the neutral or the live conductor to the power supply casing which is normally connected to the earth ground.

Most power supply manufacturers specify this current which should always be lower than 3.5 mA as per the IEC-60950-1 regulations. This ensures that the current is very low and cannot harm the person who touches or comes into contact with the power supply case. A power supply with a good earth ground reduces the leakage current significantly by providing a low resistance path to the ground.

Source: https://www.sunpower-uk.com/glossary...akage-current/

So in effect a GFCI rated for just 4mA could only take one single PSU that is at the upper end of the leakage current spec.
I think the 30mA rating was chosen just to keep nuisance tripping as low as possible.
Because the worst would be if people start ripping out the GFCI devices because they trip needlessly.

The one I used for my tests is sold as 136-57-301 at Elfa.
That said they have a handful rated for 10mA, but over a hundred rated for 30mA, that should tell you what's normal in Europe at least
Actually I wanted one rated for 10mA because I use it when repairing power supplies to protect me if I do something stupid with my oscilloscope.
But they are not available here...

This is also interesting, from an ABB PDF on GFCI devices: https://cdn.badcaps-static.com/pdfs/...c4df9e9f52.pdf

How many fluorescent bulbs and computers can be
connected simultaneously if a 30mA APR type RCD is
installed?
APR RCDs are characterized by their high resistance to line
overvoltages from atmospheric origin, grid interference and
leakage currents. One of the main factors concerning the
origin of leakage currents is the connection of numerous
computers and fluorescent bulbs with electronic ballast at
the same time. APR RCDs, thanks to their anti-interference
characteristics, reduce the risks of unwanted tripping,
significantly improving operational continuity for numerous
applications (offices, supermarkets, datacentres, schools etc.).
The most critical aspect for dimensioning lines supplying
fluorescent lights or computers is undoubtedly the interference
these devices emit. Today it is impossible to give exact
indications on how many bulbs and computers may be
connected, as this is variable and depends on the connected
load. Despite this, we can try, as we have done in the table
below, to give a general indication which can help facilitate
dimensioning of these lines.

RCD type	N. of electronic ballasts	N. of workstations
						(computers / printers)
Standard	20				2
A-APR Type	50				5

[budm]

Re: Voltage on heatsink

I always use GFCI when working on the TV EVERY TIME, no fault tripping yet and if it does I will find out why because there is leakage current limit for monitor/TV. Look at service manual sometime.

BTW, here is the TV allowable leakage current.
I wonder how may repair shop bother to do safety test, it only take one law suit to wipe out your business. Most shops just do not care.

https://www.evaluationengineering.co...safety-testing

[eccerr0r]

Re: Voltage on heatsink

In the US there is also the 30mA spec GFCI that was not focused for end user personal protection but rather for reducing nuisance trips. However, 30mA is deadly.

I've only seen some nuisance trips on my GFCIs... only very rarely. A small handful of my outlets are GFCI protected despite not in a water area (i.e., outdoors, garage, kitchen, bathroom) and I've been using them with SMPS and have not seen trips. In fact the most often trip method I've seen is actually pressing the test button by mistake while plugging in a device ...

I still use my isolation transformer regardless if there's a GFCI there or not