Potentially Dangerous USB Hub Power supply!?

[eveares]

Just teared down a cheap USB power supply from a USB hub, and to no surprise the isolation between the mains primary side and the SELV secondary side was suspicious at best.

Not only did my non contact voltage detection stick react to what should have been the separated isolated 5V output , but when measured with a MM between mains earth and the 5v output jack, it displayed 33V AC. (Shown Below)


Non-Contact stick reacting:

[GALLERY=media, 96231]1 by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]


33V AC between Mains Earth (COM jack on MM) and the 5V output (VΩHz jack on MM):

[GALLERY=media, 96232]2 by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]



So following the result I got from my MM, I decided investigate the workings of the switch mode power supply.


Moulded Casing:

[GALLERY=media, 96233]10 by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]


Back view of PCB: (Notice the dry solder joints!)

[GALLERY=media, 96234]3 by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]


Front view of PCB:

[GALLERY=media, 96235]4 by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]


Circuit Layout:

[GALLERY=media, 96230]Power Supply Circuit by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]


Transformer:

[GALLERY=media, 96236]5 by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]



Here I performed a 500V DC Insulation Resistance test on the PCB mounted transformer (between primary and secondary side) as well as on the PCB between where the transformer normally goes, but got a reading of >2000MΩ both times! Where's the mains isolation issue?

500V IR test on Transformer:

[GALLERY=media, 96237]6 by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]


500V IR test on PCB:

[GALLERY=media, 96238]7 by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]


Poor quality separation material between windings:

[GALLERY=media, 96239]8 by eveares posted 18 Oct 2015 at 2:54 AM[/GALLERY]


The Aftermath:

[GALLERY=media, 96240]9 by eveares posted 18 Oct 2015 at 3:01 AM[/GALLERY]



Summary:

One has to question that should it even be legal to supply such a poor quality power supply when supplied with a USB hub bought from a reputable on-line retailer like this one did.

In a nutshell, I can not work out if there truly was a issue between the mains input and the SELV output; the >2000MΩ reading on the 500V IR tests seems to be contradicting with the 33V AC that I got with my MM.

Also for those who have 6 minutes to kill, the video version of this thread is also available here on YouTube.

Regards: Elliott

 

[PBC_1966]

In a nutshell, I can not work out if there truly was a issue between the mains input and the SELV output; the >2000MΩ reading on the 500V IR tests seems to be contradicting with the 33V AC that I got with my MM.

Capacitive coupling between the windings. You're using a DMM with a high impedance input - probably in the order of 10 megohms or so - so it doesn't take much capacitance to leave you with an appreciable voltage into such a high impedance.

That's not to say that the transformer insulation isn't poor and likely to break down with disastrous consequences, as no doubt it is given how poor the construction looks. And there's the common issue here of no internal fusing, so potentially an upstream protective device of up to 32A.

 

[bernardgreen]

Because there is no Earth connection the unit's potential on the mains side winding of the transformer will drift to somewhere around the mid point voltage between Live and Neutral.

The potential on the output (*) will also be driven towards the mid point by capacitive coupling between windings. Connecting the output to anything that has a path to ground will reduce that potential. If the item being powered by this unit has no earth connection then it too will be pulled to the mid point potential.

(*) this is the potential that the 0 volt side of the 5 volt ouput will be at. The output is still 5 volts )

One would hope that R1 ( the 1 ohm resistor between mains and the capacitor ) is a fusible resistor with a suitable breaking current rating but I doubt it is.

 

[winston1]

So you don't understand what you are doing and have destroyed what was probably a perfectly safe power supply as a result! It's been stated many times on these forums that non contact voltage detectors are unreliable. You also don't understand the concept of capacitive coupling and high input impedance multimeters.
Your insulation tests show that there was nothing wrong with it and there seems to be a one ohm fusible resistor in the 240volt side.

I suggest you learn a bit more about the subject before condemning things.

 

[ericmark]

I do remember the argument many years ago when I bought a new TV. Good well known make well before China flooded the market. I got a shock from the aerial and complained to manufacturers. I was told it was permissible to leak a set figure to earth. I personally considered having a voltage at the aerial which could cause an involuntary action on the top of a ladder was undesirable it would not be the electric shock which kills but the shock when you reached the ground.

However as time went on I realised the whole point about voltage gradient and why you should not use metal ladders the TV was safe it was my method of aiming the aerial which was suspect.

It seems there was nothing wrong with the unit you dismantle and to give a 100 ~ 240 volt input it was clearly regulated and you don't show any components to regulate the supply some where there will have been a zener diode or other device which would control the mark/space ratio that the transistor was switching.

So as others have said you have destroyed a good power supply for no good reason. Not only that you have tried to scare people by publishing misleading information.

 

[JohnW2]

....and there seems to be a one ohm fusible resistor in the 240volt side...

It looks like a bog-standard regular 1Ω resistor to me, not an explicit 'fusible' one.

At that size, it would probably self-destruct fairly quickly when dissipating a watt or three - so perhaps a current in the range 1 - 1.7A - but if its just a standard resistor, there is no total guarantee that it would end up O/C..

Kind Regards, John

 

[bernardgreen]

Winston

I said that the resistor R1 should be a fusible resistor capable of breaking at least 16 amps ( though 32 amp would be better when plugged into a UK ring final )

It not a fusible resistor, that is unless all the other resistors are also fusible as they all appear to be the same ( colour code excepted ) and that would add considerable cost to the manufacture.

And not really much point as the track separation at A is barely adequate should the "fuse" blow without the PCB being damaged.

There is a "necked track" fuse but its rupture rating will be several amps and the transformer will be very hot if not burning before that fuse blows.

The component most likely to advert catastrophic break down is the doide D1 as this will almost certainly fail before R1 or the necjed track.

 

[JohnW2]

... you don't show any components to regulate the supply some where there will have been a zener diode or other device which would control the mark/space ratio that the transistor was switching.

One of the two diodes labelled D2 and D3 on the circuit diagram is labelled "ZD1" on the PCB, and the other one looks extremely similar (but label is largely obscured by a capacitor). I would therefore assume that one, quite probably both, are zener diodes.

Kind Regards, John

 

[JohnW2]

And not really much point as the track separation at A is barely adequate should the "fuse" blow without the PCB being damaged.

I'm not clear as to what problem or hazard that would represent.

Kind Regards, John

 

[bernardgreen]

The "regulation" of the 5 volt output relies on the feedback from the transformer's 14 turn secondary winding. The voltage on this will be slightly higher than the winding supplying the output. ( this allows for the voltage drop ( 0.7 v ? ) of the diode rectifying the 5 volts. ) Hence the regulating control aims to achieve 5.7 volts on the feedback secondary.

 

[bernardgreen]

And not really much point as the track separation at A is barely adequate should the "fuse" blow without the PCB being damaged.

I'm not clear as to what problem or hazard that would represent.

Basically to stop 230 volts the gap needs to be a minimum of 3mm clean PCB. If the PCB was scorched by R 1 running hot then carbon deposits would reduce that gap and could bridge the gap and allow current to continue to flow. Good practis is a 6 mm gap between fused and unfused tracks. Not always possible in a compact unit.

 

[JohnW2]

And not really much point as the track separation at A is barely adequate should the "fuse" blow without the PCB being damaged.

I'm not clear as to what problem or hazard that would represent.

Basically to stop 230 volts the gap needs to be a minimum of 3mm clean PCB. If the PCB was scorched by R 1 running hot then carbon deposits would reduce that gap and could bridge the gap and allow current to continue to flow. Good practis is a 6 mm gap between fused and unfused tracks. Not always possible in a compact unit.

Yes, but if, as you said, the "fuse" (necked PCB track) had blown, no current would flow, since all that bridging that PCB gap would do would be to partially replace the blown 1 resistor - the other side of the supply would still be interrupted by the blown "fuse". ... and this is all on the LV side of the tranny and main PCB gap, so (unless the tranny were damaged) should not carry any hazard of allowing 230V to the ELV output. Am I missing something?

Kind Regards, John

 

[bernardgreen]

The necked track doe not protect the capacitor. If that explodes then aluminium foil and other debris will be all over the place in side the casing and over the PCB. The risk of this foil bridging from 230 volt side to 5 volt side is very high.

 

[eveares]

So you don't understand what you are doing and have destroyed what was probably a perfectly safe power supply as a result! It's been stated many times on these forums that non contact voltage detectors are unreliable. You also don't understand the concept of capacitive coupling and high input impedance multimeters.
Your insulation tests show that there was nothing wrong with it and there seems to be a one ohm fusible resistor in the 240volt side.

I suggest you learn a bit more about the subject before condemning things.

I was only going to through it in the bin anyway, lead was only 70cm long what was way to short. True that it was capacitive coupling that did not occur to me, but now In hindsight I realize my oversight.

Not only that you have tried to scare people by publishing misleading information.

That certainly was not my intention! A bit harsh may i say.

you don't show any components to regulate the supply some where there will have been a zener diode or other device which would control the mark/space ratio that the transistor was switching

All components are shown.

 

[JohnW2]

you don't show any components to regulate the supply some where there will have been a zener diode or other device which would control the mark/space ratio that the transistor was switching

All components are shown.

Indeed - but, as I said, at least one (quite possibly both) of the diodes shown as ordinary diodes (D2 & D3) in your circuit diagram are probably zener diodes (which would be consistent with the polarity of both of them as shown in the circuit diagram).

Kind Regards, John