Isolation/ safety earthing conundrum

[Al (astral highway)]

Hi folks,

I am using the shell of a binned PC SMPS as a handy case for a project. For safety reasons, these are manufactured with a return to mains (safety) earth hard-wired into the case.

This poses me a difficulty, which may be more or less minor or major.

I had initially intended to follow suit and safety earth the case. I then had second thoughts: I am using an epoxy encapsulated transformer (240:18V) andif I take appropriate measures, I'm wondering if I can add another layer of insulation around it (lexan sheet or similar, to physically shield the mains input) and avoid some complex grounding headaches. Only I will use the equipment, but I obviously want to follow best practice.

The headache arises because if I safety earth the case, then two potentiometers that are to be mounted on it will have their cases mechanicaly connected to safety earth. That takes care of a low-impedance pathway for RF that is induced in them, but optimally, I would prefer to offer this path to circuit ground via suitable small capacitors.

The low voltage side is mounted on plastic stanchions so can float above mains safety earth.

The other advantage of having no safety earth to the case is that I won't have to worry about blowing up my oscilloscope through ground loops, If I am forced to return circuit ground to mains safety earth.

The circuit is a high-powered modulator (100W or so) at HF (180-330KHz) with PLL, for precision resonance testing of inductors to be used in another circuit, and testing of home-made current transformers.

I realise I can solve the potentiometer problem by mounting them on a non-conductive panel and then making safety earthing arrangements for the case, but it is a lot of additional effort if there is an alternative.

The 'scope issue is still relevant.

Thank you!

[Al (astral highway)]

I've answered my own Q!

Taking a step back, I need to return the case to safety earth to RF shield it, both ways around. So please close thread, mods, unless anyone has anything to add before you get to do so!

Cheers folks.

[Argus25]

Al,

I found something interesting with very high voltage high frequency generating equipment while doing work with ignition systems.

Although it seems on the face of it desirable to omit the earth for earth loop headaches and testing, when the voltages and frequencies are high, if any part of that circuitry develops a connection to near earth, or just coupling capacitance to it, it then raises other parts of the circuitry, including that of what are apparently isolated parts, like the low tension transformer secondaries, to a very high voltage above ground and arcing or corona discharges occur.

For example on the bench I had a 30kv pulsed ignition coil voltage generator, in a circuit powered by 12V, an isolated output supply. Just the output capacitances to ground lifted up the 12V supply's output enough to cause internal arcing in the power supply. So it required grounding of at least one side of the HV output and the low tension supply.

This is the basic issue, that if the output of a very HV supply is differential, one leg or the other of it could be unpredictably well above ground, depending on where the other leg is connected or coupled and capacitive effects can raise the driver circuitry above ground too.

So for any very high voltage generating equipment I always have one of HV outputs grounded, for testing and often the negative of the low voltage supply too, along with all metalwork/enclosures etc earthed to avoid these effects and have a definite (known) potential (zero) on all these points and exposed metalwork.

In vintage TV's, for pots in high voltage chains etc, the usual way was to put the pots on an insulating panel and use an insulated coupling shaft-plastic knob.

One trick that is used with very high voltage transformers, like those used in neon signs with a 15kV secondary, is they center tap the transformer and ground the tap to the transformer case/earth. That way the insulation requirements are reduced too, to 7.5kV with respect to earth (transformer housing) for each output terminal.

[Al (astral highway)]

Originally Posted by Argus25

...for any very high voltage generating equipment I always have one of HV outputs grounded, for testing and often the negative of the low voltage supply too, along with all metalwork/enclosures etc earthed to avoid these effects and have a definite (known) potential (zero) on all these points and exposed metalwork.

Hey Hugo,

Ahah, yes, I am starting to have similar experiences. I also note that the human body is a giant capacity and so I can get RF burns just by touching the heatsink of a power transistor that is referenced to ground via an RF capacitor, when I'm operating my power PLL oscillator. So I'm carrying a charge that passes to ground via corona from my finger: I was surprised, as at the moment there's only 15-20W out, from a low voltage (15-30V) power supply.

I'll take your advice and safety earth the case and also circuit ground. Measurement is going to be challenging. I don't want to blow up the 'scope.
HV is now okay as I have built your HV probe, but for example if I want to look up a gate drive signal or the output of another active device, I will now have ugly and potentially destructive ground loops to watch out for.

Clearly, mains (safety) earth locally isn't always at zero potential, counter-intuitive as this used to seem to me. Operating the device I'm building switches on my TV when it is in standby . It then cycles through the options screen by itself. This suggests that there is a complex waveform entering the TV via the mains, unless it is induced directly.

Originally Posted by Argus25

In vintage TV's, for pots in high voltage chains etc, the usual way was to put the pots on an insulating panel and use an insulated coupling shaft-plastic knob.

Yes, I have memories from investigations in the past where the pots associated with the line output stage and some others were mounted in this way. I did wonder what happened to the casings: if they were left floating like this, they could also pick up at least a static charge.

[TonyDuell]

'There is no such thing as ground' is one of Don Vonada's laws (from Digital Equipment Corporation). It has 2 meanings:

The first is the obvious one that voltmeters (etc) have 2 wires, you can take any point as your reference. There is no reason why that has to be (say) the chassis of the unit (in fact when working on the primary side of a switch mode power supply, the chassis is a particularly stupid thing to take as the reference).

The second is that even once you have defined the point you want to call 'ground' the connections to it have impedance. And that makes a big different at RF (including digital circuitry which is essentially RF in this respect). The network that links all the points you think of as 'grounded' is complex (in the non-technical, not mathematical, sense). You can get some odd effects as a result.

[Al (astral highway)]

Originally Posted by TonyDuell

'There is no such thing as ground' is one of Don Vonada's laws (from Digital Equipment Corporation).... The network that links all the points you think of as 'grounded' is complex (in the non-technical, not mathematical, sense). You can get some odd effects as a result.

Hey Tony, this is fascinating terrain and I'm deeply interested in it as I observe these laws at work right here, in front of me.

I might try to map out some of the 'network that links the points' that I took for granted were grounded. It may help me to visualise what, if anything, I can do to protect my 'scope, when I'm taking measurements - as well, by the look of things, other consumer electronics that I previously considered to be outside it!

[Skywave]

I'm sure that "Ground" has it roots (pun not intended ) in the early days of radio. When I'm making a drawing of a cct. diag., I avoid using 'ground' - and use 'common' (usually abbreviated to 'Com') instead. In a practical implementation of same, I always try to follow the golden rule of having separate 'COM' return leads to one point - often the 0v. output point of the PSU.

Al.

[Argus25]

Originally Posted by TonyDuell

'There is no such thing as ground' is one of Don Vonada's laws (from Digital Equipment Corporation).

When I referred to "ground" in my posts, it was simply "mains earth" and the potential to which all of the metalwork, signal connections, of mains operated equipment in the laboratory is connected. It is in fact close to earth due to the grounding stake in the dwelling, even if there are some ohms resistance to that, and as you point out an RF impedance that could be higher. When was the last time you experienced an electric shock from the body of a mains earthed appliance, through to your body to the surface you were standing on, when the connection to earth was good and the earth at the dwelling not faulty ?

So it is the reference which is most stable, as it is pretty hard to charge this connection to a high voltage with respect to a human body standing on it (unless the RF currents and frequencies are very high).

So while I agree there is no absolute earth, it is the better and safest option for the experimenter to connect to. As a result there are less likely to be human accidents and equipment damage. So it is a practical not a theoretical solution/connection.