Scope ground issues...

[Al (astral highway)]

I'd connected the +ve lead from a bench power supply and without the -ve connected, the power-on LED lit up, showing a completed circuit via the 'scope ground lead, through the mains earth of the scope and through the mains earth of the power supply.

This is infuriating. I haven't opened up the power supply but I can't see why it couldn't be double insulated and with no need to safety earth the case.


(I am aware of the completely different scenario with some power pulse circuits, which requires differential measurements without grounding the scope probes to circuit ground - OT for this thread)

I'm aware that I could also lift the 'scope mains safety earth, given that the case is double insulated. However I haven't done this yet as I can't yet quantify how hazardous this could be.

Have people experienced this problem and what have you done about it?

[Electronpusher0]

None of my bench psus are grounded, they all have floating outputs and a separate ground terminal should you wish to ground one of the output terminals.

Peter

[Al (astral highway)]

Originally Posted by Electronpusher0

None of my bench psus are grounded...

Do you mean, none of them has ground connected to mains safety earth?

We need to distinguish between ground and safety earth.

This is half the story as the 'scope having ground and safety earth internally connected is the other half...

[ionburn]

I tend to design all of my power supply circuits for avoiding an earth link through them (double insulated). That way I can use either positive or negative as ground and either to scope earth. It has the added bonus that I can use several supplies to the same circuit with care.

Yes, all the bench power supplies I have come across have the output(s) floating, this makes it easy to do things like +/-15V power rails. They also have a ground terminal which, at ones whim, you can connect to anything you like. I usually have a resistor of a few 10's of ohms twixt the ground and the circuits notional 0V. This means that a) the circuit is at signal ground and b) if you use a 'scope on it, the lack of earth connexion doesn't really matter at low frequencies.

[stuarth]

Most power supplies have a mains earth (is that what you refer to as the safety earth?) for its case and exposed metalwork. This earth will in many cases have its own terminal, but should not be permanently connected to any of the power supply output terminals. The power supply will have a mains (safety) earth terminal and output terminals, but none of its output terminals will at this stage be defined as "ground".

You can connect the outputs of one or more such power supplies to the circuit you are working on, but so far, your circuit will have no connection to mains earth. There will generally be a node in your circuit which you consider the ground node. This node may one of the power supply output terminals, or in may be an internal node in your circuit. Optionally, you can connect this node to mains earth at the power supply earth terminal, or, more generally through a scope probe's ground lead (which will usually be mains earth).

From your initial post, it seem that your power supply has a permanent or semi permanent connection between mains earth and the negative output terminal - this is best avoided. On my bench power supplies, I have a 1Meg resistor between the mains earth and the negative output terminal, but this is only to avoid ESD problems and would be removed if it upset circuit operation.

Stuart

[cmjones01]

It would be extremely unusual for the negative side of a bench power supply to be connected to mains earth. That would be really annoying.

Is it possible that your power LED is lighting simply because of AC leakage currents? There is likely to be a class Y capacitor of about 4.7nF between mains earth and the bench PSU output, especially if it's a modern switch-mode one. Stray earth leakage currents from whatever else is connected to the circuit could light an LED well enough to see.

Chris

[JonSnell]

On a workbench it is good practice to have all test gear earthed, power supplies floating as will the power socket for the equipment on test.
So, to recap, the equipment under examination is running from an isolated mains supply, (transformer usually 500V/A), power supply is also floating and oscilloscope, audio test amplifier, signal generator to be earthed. That gives only one path to earth avoiding accidental short circuits.
It is not recommended to tie any supply to earth through resistors. 1M and half mains is enough current to destroy most ESD components.

It is not recommended to tie any supply to earth through resistors. 1M and half mains is enough current to destroy most ESD components.

I do hope my 10's of ohms wouldn't give 1/2 mains and be capable of zapping CMOS.

[emeritus]

All the commercial PSUs I have used have had floating outputs, essential when you need to cascade them to get higher voltages, or generate an ac signal with a dc offset. The coax connectors of most of the signal generators I have used have had grounded screens, so the dc has had to be floating.

[Electronpusher0]

Originally Posted by astral highway

Do you mean, none of them has ground connected to mains safety earth?

We need to distinguish between ground and safety earth.

This is half the story as the 'scope having ground and safety earth internally connected is the other half...

Clarification: All the bench supplies have mains safety earths connected. It is the outputs that are floating with respect to ground. As I said, the mains earth is brought to a green terminal on the front panel should you wish to ground one of the output terminals.

The supplies are:
Rapid electronics 30V 3A supply
EST PD5 30/3B
Thurlby Thandar PL320
Farnell LT30-2
Amrel Programmable PSU

Peter

[duncanlowe]

I had an issue some years ago with a fairly large SMPSU (30v 100A). I was trying to measure current in the positive output using a shunt and a scope. I had expected that the output was isolated from mains ground but there was enough leakage current to make the probe ground sheath smoke somewhat. It was a Glassman PSU, that we still use somewhere in the lab 20 years later.

[Al (astral highway)]

Just an interim message - as I’m stepping away from online for a day or so - to say thanks all for your comments so far. I’ll establish as a priority what exactly are the earth connections inside my ‘scope and bench power supply.

Thanks to all who have contributed so far.

[trobbins]

I have a 1970's lab power supply with isolated floating output to +/-250V from ground, but it has a 1 Meg // 470nF from negative output to protective earth to restrain the output from floating too far from ground.

[Al (astral highway)]

Originally Posted by cmjones01

It would be extremely unusual for the negative side of a bench power supply to be connected to mains earth. That would be really annoying.

Is it possible that your power LED is lighting simply because of AC leakage currents?

I've looked at the issue again. Please see attached pix.

The first shows a current of almost 20mA flowing through an ultra-bright, 12V rated blue LED.

The second shows that the circuit is completed with the positive side of the DC power supply connected to the LED, whose negative side is connected to the 'scope ground lead only.

So power supply +ve - LED - scope ground lead - completes the circuit with full current flowing through the LED.

The disconnected lead is the DC power supply -ve. Both ends of the DC power supply lead are shown not connected.

So this is an unsatisfactory situation indeed and has resulted in several smoked circuits before I knew the cause.

The only remedy is to lift the 'scope earth at the plug.

The 'scope is a GW Instek GDS1022. The power supply is an off-brand thing but works well enough - I think it's Hulkon APS3005D , but it's hard to work out the stylised lettering.

I'm just posting this update in case it's of interest to anyone. I know that some members have superior bench set-ups and this may seem untenably unsatisfactory, but it's what I'm working with

[trobbins]

No pic attached

Originally Posted by astral highway

The only remedy is to lift the 'scope earth at the plug.

This is to be avoided ....

[stuarth]

This is to be avoided ....

Agreed!

Lifting the scope earth only hides the problem and has safety implications.

[Al (astral highway)]

Originally Posted by stuarth

Lifting the scope earth only hides the problem and has safety implications.

It doesn't hide it, it eliminates it!

I knew this would quickly go down a safety woe route.

Please can we not go too heavy on safety as over the years, I've noticed threads with this orientation this don't end up too well.

Please can we concentrate on

1) acknowledging the fact that this loop is tedious and an unacceptable artefact of the tech that I have and perhaps others have...

2) solving the problem

I know there are hazards, but if it's done methodically and working mindfully, lifting the scope earth is a risk but a manageable one. I have spoken to engineers off-forum who have done this for similar reasons.

[cmjones01]

I think that photo gives the game away. There's a strap between the black and yellow terminals on the front of the power supply. I'm betting that the yellow terminal is chassis/mains earth, and the strap is connecting the power supply's negative output to it. Unscrew the yellow and black terminals, remove the strap, and the problem will be solved.

There's a similar strap visible in this review video I found:
https://youtu.be/HmCSV5rOfs4?t=719

Many bench power supplies have a chassis earth terminal available (the two with me at the bench here, a Farnell L30-5 and Solartron AS1164.2, certainly do), but few have such a subtle strap connecting it to the negative output.

Chris

[Al (astral highway)]

Originally Posted by cmjones01

There's a strap between the black and yellow terminals on the front of the power supply. I'm betting that the yellow terminal is chassis/mains earth, and the strap is connecting the power supply's negative output to it.

Wow. I must have looked at that a thousand times without seeing the strap!

Originally Posted by cmjones01

...unscrew the yellow and black terminals, remove the strap, and the problem will be solved.

Thank you! Well sleuthed. Much appreciated as it was giving me a big headache.

Originally Posted by cmjones01

...few [bench supplies] have such a subtle strap connecting it to the negative output.

Reassuring to know it's one of the subtle ones. Pictured is the innocuous but in this case troublesome link. The aluminium looks bright against this wood but very easily overlooked in situ.

Thanks again, Chris.

Now I can concentrate on waveforms and measurements in peace and without worrying about frying things.