grounding/earthing problem with leakage PSUs

[greengold]

hi all, newbie here and in electronics too.
now I want to design a circuit: MOSFET driver controlled by PWM from raspberry. MOSFET will have 220V load connected to it with about 1A current. so this is what I have:

a bit sketchy but, you'll get an idea.
first, I need to have common ground with MOSFETs source and LTV-3120 driver. otherwise it won't operate. PS2 is there only for driving it (I didn't have much luck with combination of Zenner, Caps and resistors) so there it is Whole PSU just for it. PS1 is purely for Raspberry (like) device with GPIO on it with good PWM signal and some i/o for sensors and buttons that I need.
now, both this MeanWell PSUs have 220VAC leakage to out. as per specification just with a very low current .25mA, but you can feel it, trust me. This is unacceptable. I worked around it by connecting GND1 to the Earth as you can see on the schema's top. Now this is sketchy a I know it can trip a fuste easily, but the current is small, so it runs and phase is out earthed so I can touch the pi and it's periferals.
now, I know this is all fishy, but I just don't have any idea on how to do this correctly, so I want to kindly ask you for recommendations on what should I do to minimize hazard of this circuit or straight up do it correctly? I am able to rework it still, just need some suggestions...

thank you!

 

[Diver300]

a bit sketchy but, you'll get an idea.

Well I can't understand what is going on in the circuit diagram.

Can you explain where the load is attached?

It really helps with making circuit diagrams readable is power flow is from top to bottom. Your diagram has five connectors with no indication what they connect to.

 

[greengold]

J7 is connection for load; MOSFET is on the low side, so drain goes into the J7 with the + from the rectifier directly. load is resistive

 

[Tony Stewart]

Don't start a design until you have an acceptance criteria for noise out to a target load with a target cable and EMI line noise since it is> 100W.

The load can be any peripheral or step load or reactive motor and EMI noise. Otherwise you are not ready to design.

 

[Tony Stewart]

another example you are not ready to design is you have chose a bridge rectifier rated for 35 Vrms on the grid which can produce explosive transients from nearby lightning and you have no protection.

Start by reviewing similar existing designs and identify every function of most of the parts and how it is tested and rated with specs.

 

[greengold]

Don't start a design until you have an acceptance criteria for noise out to a target load with a target cable and EMI line noise since it is> 100W.

The load can be any peripheral or step load or reactive motor and EMI noise. Otherwise you are not ready to design.

i just told you it's resistive load.
don't complicate it

 

[greengold]

another example you are not ready to design is you have chose a bridge rectifier rated for 35 Vrms on the grid which can produce explosive transients from nearby lightning and you have no protection.

Start by reviewing similar existing designs and identify every function of most of the parts and how it is tested and rated with specs.

okay, don't like that rectifier, you can spill some milk in here post a correct part. did it even crossed your mind that that's why Im here? like when that thing is 100% why should I post here? to wait for your agreeable nothing? just ******* go some place other when you dont want to add nothing constructive. don't need that debility comments which leads to nothing

 

[Diver300]

So the live connections, which are the rectifier, MOSFET and load, as well as both sides of PS2, are all connected to the live and neutral so they should not have any connection to the chassis. Any connection to chassis would be dangerous or would trip an RCD on the supply.

The low voltage side, so the output side of PS1, the processor, and the input to the opto-isolator, should have no connection to the live connections.

It's likely that the 0.25 mA leakage of PS1 will be most of the leakage. That amount of leakage is safe, but I agree that you can feel it. There is probably a mains-rated capacitor between the supply and output of PS1 to reduce noise.

If you have a supply with a ground connection as well as live and neutral, you can connect the -ve of the PSU output to ground. If not, and the leakage is unacceptable, you need a power supply with less leakage for PS1

 

[Diver300]

J7 is connection for load; MOSFET is on the low side, so drain goes into the J7 with the + from the rectifier directly. load is resistive

It would have helped to have shown that on the circuit diagram.

 

[greengold]

So the live connections, which are the rectifier, MOSFET and load, as well as both sides of PS2, are all connected to the live and neutral so they should not have any connection to the chassis. Any connection to chassis would be dangerous or would trip an RCD on the supply.

yes, I have no intention to expose this 'power' part, it will be well hidden.

The low voltage side, so the output side of PS1, the processor, and the input to the opto-isolator, should have no connection to the live connections.

yes, I think that's what I have, everything from this part connects to the 'power' part by opto-couplers..

It's likely that the 0.25 mA leakage of PS1 will be most of the leakage. That amount of leakage is safe, but I agree that you can feel it. There is probably a mains-rated capacitor between the supply and output of PS1 to reduce noise.

If you have a supply with a ground connection as well as live and neutral, you can connect the -ve of the PSU output to ground. If not, and the leakage is unacceptable, you need a power supply with less leakage for PS1

basically that is what I have, right? GND1 connected to Earth as you can see on the top of the schema.. GND and GND1 _never_ connects. this is what I wanted to validate here..if I can do any better..

 

[greengold]

It would have helped to have shown that on the circuit diagram.

sorry, thought that's pretty obvious as that MOSFET is there for powering the (resistive) load and it leads to J7.

 

[Diver300]

basically that is what I have, right? GND1 connected to Earth as you can see on the top of the schema.. GND and GND1 _never_ connects.

It certainly isn't clear from the circuit diagram.

You have three different earth symbols. I guess that GND is the -ve after the rectifier. GND_1 is the safety electrical earth and GNDPWR is what would normally be called neutral.

It would make things much easier for someone trying to understand what you are doing if the labelling was clearer.

Where are you measuring the 0.25 mA to?

 

[greengold]

that's correct.
GND1 being -Ve of the PS1 that I need to use to power the 'control' part of the circuit with chassis exposure.

which is connected to the earth (in my world) at the top:

not fighting over the symbols here, I'm really out of conventions in this regard, but I think you got that right.
0.25mA is the leakage current as per specification:
"
Class II equipment:

These equipment do not have a protective earth ground. Such equipment uses reinforced or double insulation to provide protection against the electric shocks. Maximum leakage current is 0.25 mA.
"

 

[Diver300]

Ok, so the specification says that the maximum leakage current is 0.25 mA

If the negative side of PS1 is grounded, and you are grounded, I can't see how you are getting any kind of electric shock, even at the low level that 0.25 mA would give.

Have you measured the voltage and short-circuit current of the point that you are getting a tingle from to ground?

 

[Diver300]

Just to add, I had a bad earth connection on a washing machine, and I felt a definite tingle when I touched it lightly and touched the radiator at the same time. I couldn't feel anything if I wasn't touching both.

I measured around 120 V ac between the two. I can't remember if I checked the short circuit current, but it obviously wasn't enough to trip the RCD. There was no measurable voltage when the earth was connected properly.

I suspect that if you are feeling a slight shock when touching the circuit, either it isn't earthed properly, or something else is putting a voltage on your body. There is always some pick-up on people. A standard way of testing if an oscilloscope is working is to touch the input pin while not touching ground. If there isn't a pick up of some mains voltage, the oscilloscope isn't working. Of course that relies on the person being close to a load of mains-powered items, but that's almost always the case in any house or lab or office.

 

[Nigel Goodwin]

Just to add, I had a bad earth connection on a washing machine, and I felt a definite tingle when I touched it lightly and touched the radiator at the same time. I couldn't feel anything if I wasn't touching both.

I measured around 120 V ac between the two. I can't remember if I checked the short circuit current, but it obviously wasn't enough to trip the RCD. There was no measurable voltage when the earth was connected properly.

I suspect that if you are feeling a slight shock when touching the circuit, either it isn't earthed properly, or something else is putting a voltage on your body. There is always some pick-up on people. A standard way of testing if an oscilloscope is working is to touch the input pin while not touching ground. If there isn't a pick up of some mains voltage, the oscilloscope isn't working. Of course that relies on the person being close to a load of mains-powered items, but that's almost always the case in any house or lab or office.

Perfectly normal (as you probably know), it's simply leakage through the suppression components from live to chassis, and neutral to chassis, if the earth wire is disconnected then the chassis will float at half mains voltage, at a very low current.

Even with the earth wire connected the current is far too low to trigger an ELCB, and even less so when it's going through your body instead.

However, REALLY bad news to have the earth disconnected from a washing machine, water and electricity don't play well together

One little (TV engineer) tip, never touch a metal item (such as a washing machine - or even more so an old TV chassis) without rubbing it with the back of your finger (the back, so any possible spasm doesn't mean your hand grabs hold). If it's live, then it will feel rough like coarse sandpaper, if it doesn't feel rough then it's not live. Old TV chassis's used to be connected directly to one side of the mains, so if the mains plug was wired wrongly, or a two pin plug the wrong way, then the entire chassis was connected directly to the live wire.

 

[Diver300]

However, REALLY bad news to have the earth disconnected from a washing machine, water and electricity don't play well together

It got fixed very shortly after I realised what the problem was.

It was a worn 13 A socket, that no longer gripped the earth pin. I later went and changed all the sockets of that make in the house. They were quite old, and most were single sockets, so things got plugged in an unplugged a lot.

 

[Tony Stewart]

The safety leakage level is a well known tradeoff between suppressing harmonic noise to earth ground which is a nuisance and the tingle of a leaking current through an unearthed Switched Mode charger or floating supply powered device. Which is also a nuisance but is not unsafe. One can easily experience this with a laptop and charger and bare feet or socks and salty sweat leather shoes and worse on grass. The skin easily picks up these sensations and the wrist even more which might hurt in a small area with low current but high density, and feel like a burn.

Still harmless, so one way to avoid this is to connect to an earth bonded video monitor. But then (hypothetically) if you have a device with a wiring fault such as floating ground and hot chassis, then the concept of the floating power with leakage will be safer than an earth bonded 0V DC or simply and AC gnd to a DC gnd. But generally you don't use metal case appliances with ground faults and a short to chassis while using a laptop with an earth bonded connected. Then if you get in between these two conditions, there is no safe current limiting. that is the purpose of isolated chargers but then they have to divert some 1mA of harmonic noise to PE (protective earth) or AC gnd as you know it.

The proper bridge diode is a higher suffix letter for say 350Vrms and then surge suppression. like a polyfuse and MOV or a fuse. Again see commercial designs to see how it is done.

More important is always start with the purpose of this project. When dealing with grid power, it is wise to get hands on experienced guidance.

 

[greengold]

I suspect that if you are feeling a slight shock when touching the circuit, either it isn't earthed properly, or something else is putting a voltage on your body.

nahh... when earth is connected it's all good. no phase on 'control' (PS1 powered) circuit part. I really just need someone more oriented in electronics to validate that or propose some more top-notch solution b/c intention with this is to go commercial. so 'power' part will be isolated out of touch and 'control' part earthened. i have it in front of me, it works, so far so good, just really need to get this done properly.

 

[greengold]

The proper bridge diode is one hgiher suffix letter for say 350V and then surge suppression. like a polyfuse and MOV or a fuse.

in reality I'm using this GBP204. so far... which would be more suited?
fuse I have there, connected to mains phase but that might not be what you mean ;D

More important is always start with the purpose of this project.

purpose in higher overview is for this 'power' circuit to drive a heatpad. that is generating a heat, sensed by temp meter hooked on RPI on which user sets desired target value and software regulates the power to get there. with PWM. so really, driving the resistive load (200W and I don't have any information on this heatpad from the reseller) with feedback loop through sensors and algo to set a duty cycle.