SELV 24VDC Power Supply - Measuring 90V AC!

[Soupercollider]

Hi Everyone,

I have been arguing with a supplier of mine about a 24V DC power supply which is supposedly SELV compliant to EN 60950. When doing my usual testing of SELV supplies I measure the voltage across any two conductors on the DC side to ensure they do not exceed ELV levels (as one should expect from an SELV supply). The +ve to -ve terminals I measured as 24VDC no problem, but when I measured from either +ve or -ve to earth I got 90V AC with 50Hz ripple!

Am I going mad, or does this not comply with the requirements of SELV?

The power supply in question is a Carlo Gavazzi SPM1-241 (datasheet)
I believe this 90V AC is due to the capacitive coupling through the transformer as it is not screened and there is no earth terminal on the power supply itself. They are only relying on double insulation, but this doesn't stop the capacitance.

My supplier is arguing that this is OK as when you test to the SELV clauses in EN 60950 you only test for AC or DC depending on the supply type. They reckon that if the power supply is DC, you only need to test DC voltages. If the supply is AC you only need to test AC voltages.

I completely disagree as by this argument you can have a full exposure to 240V AC on the ELV side so long as the power supply is DC.

Does this not scream potential for electric shock to anyone else and complete void the purpose of SELV in the first place??

 

[FreddyNurk]

Its been ages since I last recall any in depth discussion of such things, but from memory its likely to be as you suspect, its capacitive coupling. The last discussion didn't include 60950, but the implication was that its a capacitive potential that only shows up with a decent high impedance DMM, and if you use an analogue meter the measurement disappears as there's no grunt behind it.

Whether such arrangements are acceptable in terms of compliance with 60950, I've got no idea.

EDMS Australia

 

[Soupercollider]

Whether such arrangements are acceptable in terms of compliance with 60950, I've got no idea.

I haven't got the copy of the standard it in front of me right now, but I don't recall seeing anything in there as far as allowing capacitive potential. I figured it was pretty cut and dry when it said (paraphrasing):

The potential between any two conductors, or any one conductor and earth shall not exceed ELV

It pretty much says the same thing in AS/NZS 3000 for SELV..

 

[FreddyNurk]

The previous conversation came about because someone did pretty much exactly as you did, measured the output to ground (90V AC with 50Hz ripple implies that you're using an oscilloscope rather than a meter), and then asked how come such a dangerous voltage was present.

For what its worth, I just went and checked a bunch of double insulated power supply devices here, and most of them indicated around 100V output to earth too, with my trusty high impedance Fluke. If I had an analogue meter I'd go check it again, but I don't.

I agree though, the argument that only AC or DC is measured is spurious and doesn't inspire confidence.

This document, whilst not dealing with the same sort of equipment, tends to state the same thing, its a potential that only shows up with a really high impedance measurement device, rather than being any sort of safety issue.

EDMS Australia

 

[Soupercollider]

Thanks FreddyNurk, I hadn't actually thought about the impedance of my DMM (it was a DMM, not a scope that I was using). Mine doesn't have a lowZ option for measuring, but one of my collageaues might. I'll have to check it out when I got back in to work tomorrow.

Now that you mention it, I do recall something in 60950 about using a 5kOhm resistor across the voltage measuring instrument for measuring between a conductor and earth. I should give this a try, do you think this would still be valid with my highZ DMM?

This could be the way out, though not the way my supplier is thinking. If I measure ELV with this resistor in place then it should be compliant according to the standard.

I'll report back...

 

[VEBill]

Measuring open circuit voltages using instruments with very high input impedance often leads to meaningless results.

In this situation, the immediate instinctive reactive should be to consider if there's supposed to be a load of some sort applied.

edit: I also recall one similar case where, after I'd made sure it was safe to do so, I switched to measuring the AC *current* leakage to ground. Obviously one doesn't just switch the meter to AC Current and measure. One must first make sure that the current is just uA leakage and that it's safe to ground that terminal via the current meter.

 

[IRstuff]

The root rationale for such as test is typically tied to safety of people or equipment. While a voltage, by itself, could potentially ;-) cause an ESD-like damage, it's the current or charge capacity that typically makes the voltage dangerous.

In your case, that current or charge needs to be miniscule to protect electronics from getting zapped. A problem often can occur if one circuit is connected to another where the two are on different AC systems. Most of the time, little or no harm occurs, but if one AC circuit gets discharged through the other, the circuitry that carries this discharge current could get damaged, either outright or latently. The latter would be a big problem for a production environment, if the outgoing product is weakened by this configuration. In this case, imagine that your circuit is connected to this power supply and you probe the circuit with a properly grounded oscilloscope. That 90 Vac difference could be discharged through and internal node with the board and partially damage the internal circuitry,

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[OperaHouse]

It is more than transformer capacitance. In switching power supplies they typically ti the B+ of the switcher is tied to the output common through a low value Y capacitor to reduce EMI. This turns the secondary winding into a pseudo shield. Even a resistor of 100K will drop this leakage voltage to a very small value.

 

[Soupercollider]

Thanks for your replies, everyone.

I've been reading up on highZ measuring equipment and why it's bad for what I'm trying to do.
I've tested the PSU again this time with a 10k pot between +ve and earth and varied the resistance. Straight away the voltage dropped to less than 2V and kept dropping down to the mV range when I reduced the resistance.

I also went back into EN 60950 to find the clause about the resistance:

... When measuring a voltage between earth and a conductor in a circuit that will not be earthed in the intended application of the equipment, a non-inductive resistor of 5 000 Ω ± 10 % shall be connected across the voltage measuring instrument.

I'm satisfied now that this PSU is SELV compliant (and now I need to go test some of the other supplies I knocked back!)