Connecting system ground to mains earth

[johnmx]

Normally I use switching AC/DC power supplies (from XP Power, EOS, Traco Power, etc.) in my prototypes. Some power supplies have the outputs completely isolated from mains earth. Although they usually have the footprint of a capacitor in the PCB to connect the output DC ground to earth, they don’t have this component populated. The problem is that in most of the times appears an AC voltage between the DC ground and power supply’s case which is connected to earth. In some cases can be as high as 90Vac which is enough to give an electric shock (I tried  )

How do you recommend connecting the DC ground to the mains earth?

A direct connection eliminates the potential difference but may increase noise (EMI) in system ground.

A 1MOhm resistor in parallel with a 100nF/250V film capacitor may reduce the voltage but doesn’t eliminate it. Still 6Vac measured with a handheld multimeter.

If there is a ‘RF expert’ around here please tell me what you do in this cases, because the system ground is always connected to the equipment case.

[Kiriakos-GR]

You had just reply by your self , in your own question ...

In some cases can be as high as 90Vac which is enough to give an electric shock

And so anything lower , that it can not cause shock ( like the 6 Volts ) it does stand as solution.

[Zero999]

It shouldn't be a safety issue as the current is too low.

The only problem is it could damage some static sensitive components but as long as the case is connected to the same supply voltage as the circuit you're measuring (either +V, -V or 0V) via a low enough impedance, it shouldn't be a problem.

One thing you could try is powering the 'scope from a traditional mains transformer based power supply or using an isolation transformer with the SMPS.

[Feanor]

This 90VAC could be the difference between mains earth and mains neutral. This can occur when the mains voltage "floats". Can be caused by many thins such as poor load balancing on the 3 phase line supplying your area.

I have this problem at home with a poor earth connection. The voltage between neutral of my cellphone charger and my body for example gets up to around 80VAC when the weather has been dry for a few weeks. It can give a stinging tingle if you make contact with it.

The mains neutral and mains earth should be at the same potential and solidly connected to the ground by way of a ground spike. They should be connected together on the supply side of your home distribution board. Connecting earth to neutral on the load side of your home distribution board can lead to your RCD (residual current detection) circuit breaker tripping all the time.

My home is very old and there are no RCD circuit breakers at all, so I just buried a 2 meter piece of copper pipe in the ground and I connect it to the neutral of what ever I am using as the initial power supply (wall wart, cellphone charger, ATX power supply). Problem solved until I upgrade the wiring of the house, when this happens a proper ground spike will need to be put in and the neutral and earth will be connected to this on the supply side of the distribution board.

[alm]

This 90VAC could be the difference between mains earth and mains neutral. This can occur when the mains voltage "floats". Can be caused by many thins such as poor load balancing on the 3 phase line supplying your area.

This is easy to verify with a DMM. But for ungrounded switchers, I think it's more likely that it's due to a capacitive voltage divider formed by the input filter caps.

I have this problem at home with a poor earth connection. The voltage between neutral of my cellphone charger and my body for example gets up to around 80VAC when the weather has been dry for a few weeks. It can give a stinging tingle if you make contact with it.

Cell phone chargers (which are usually double isolated, so nothing to do with ground potential) should not allow any significant current to flow between people touching it and either the live or neutral. Both are considered identical from safety perspective in consumer equipment.

The mains neutral and mains earth should be at the same potential and solidly connected to the ground by way of a ground spike. They should be connected together on the supply side of your home distribution board. Connecting earth to neutral on the load side of your home distribution board can lead to your RCD (residual current detection) circuit breaker tripping all the time.

This depends on the local earthing system (eg. TN-C vs. TT). Some systems may even have ground in the middle between the two other conductors (eg. on ships). Connecting any other conductor to ground could cause some serious fireworks in that case. Even in TN-C systems, neutral and ground are usually not at the same potential, since the neutral wire carries current, and the ground wire doesn't, so the neutral wire has an IR drop (this is actually a neat way to measure wiring resistance). Allowing ground to carry current defeats the point, you'll just have to parallel neutral wires, and no additional safety compared to WO2-era appliances (chassis connected to neutral).

In my opinion, it's a bad idea to tell people to mess with safety features of electrical wiring, unless they know what they're doing, but in that case they wouldn't need anyone to tell them what to do. Especially if you don't know someones location, electrical system or regulations.

[Zero999]

It's caused by the Y capacitor(s) which divert high frequency noise on the DC output back to unregulated rectified mains.

The neutral shouldn't be at 90V unless you're on an IT earthing system and there's a fault and even then the neutral shouldn't be connected to the secondary.

[saturation]

90Vac between secondary negative terminal and chassis ground in a new switching power supply? That's a shock hazard. This sounds fairly odd and concur with hero999, check for a fault.

It's caused by the Y capacitor(s) which divert high frequency noise on the DC output back to unregulated rectified mains.

The neutral shouldn't be at 90V unless you're on an IT earthing system and there's a fault and even then the neutral shouldn't be connected to the secondary.

[johnmx]

In Portugal we use TT earthing system and there is nothing wrong with my earth connection. Also, the 90Vac it’s not just from static, it actually flows some current (mA) through this connection.

I checked some RF lab. equipment and all have the ground connected to the earth with a wire. So, this is what I’m going to do.

[saturation]

Yes, I think this is so, but its unusual to have such a high AC riding on the DC of your SMPS.  It suggests their is an isolation issue between primary and secondary or is this a high power radio transmitter located in this area?  Often, reflections are absorbed back into the device and enters through the AC power lines/ground lines.  Is the sometime 90Vac line 50 Hz or in the radio frequency?

Good luck and be safe.


deleted, erroneous assumption

In Portugal we use TT earthing system and there is nothing wrong with my earth connection. Also, the 90Vac it’s not just from static, it actually flows some current (mA) through this connection.

I checked some RF lab. equipment and all have the ground connected to the earth with a wire. So, this is what I’m going to do.

[alm]

90Vac between secondary negative terminal and chassis ground in a new switching power supply? That's a shock hazard.  This sounds fairly odd and concur with hero999, check for a fault.

Depends on the output impedance. If you connect two small caps in series between live and neutral, the center of them will be at Vmains/2. As long as the impedance at 50/60Hz is high enough, the current won't be dangerous, although it will tingle. This is the case with connecting an SMPS designed for grounded connections to an ungrounded socket: the ground will be floating, and the two Y-caps are connected to live and neutral. Not usually dangerous for humans, but it may not be healthy for sensitive electronics. Here's a guy that did some tests on various wall warts.

[saturation]

Ouch, this is not good at all. Thanks alm, this is good information.  I had always presumed it was as isolated as old style linear transformer designs. While it won't hurt normal people, this leakage current can misfire pacemakers and implanted defibrillators.  I just tested some of my newer wall warts, at most the I'm getting 0-6Vac.  However, older ones are doing 40 Vac, both are 60 Hz.  Currents are small, 15uA or less.  Will look into this further.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC487186/

http://www.nejm.org/doi/full/10.1056/NEJMc0805900

We've never looked into this assuming all modern equipment are isolated from mains voltage, but it seems all SMPS wall warts have a bit of leakage.

90Vac between secondary negative terminal and chassis ground in a new switching power supply? That's a shock hazard.  This sounds fairly odd and concur with hero999, check for a fault.

Depends on the output impedance. If you connect two small caps in series between live and neutral, the center of them will be at Vmains/2. As long as the impedance at 50/60Hz is high enough, the current won't be dangerous, although it will tingle. This is the case with connecting an SMPS designed for grounded connections to an ungrounded socket: the ground will be floating, and the two Y-caps are connected to live and neutral. Not usually dangerous for humans, but it may not be healthy for sensitive electronics. Here's a guy that did some tests on various wall warts.

[jahonen]

If you take a look at for example the datasheet of Schaffner inlet filter, then it can be seen that specified leakage current for standard types is around 370 µA (@230 VAC). Those 2.2 nF Cy-capacitors form a capacitive voltage divider.  My personal experience is that it can really kill most consumer-grade interfaces, when plugging or unplugging them while connected to the mains. The fact that most usual consumer connectors stupidly connect the signal first(!), and ground last, makes it even easier. The leakage feels also extremely nasty, for example when connecting an grounded antenna to floating system. It is not a surprise that medical types don't have Y-capacitors.

Problem is also multiplied if there are several devices plugged in same extension cord. That can cause much higher leakage currents. I guess that it is far from safe to disconnect the safety ground wire at big building where there are hundreds if not thousands of such devices.

Regards,
Janne

[Zero999]

I don't see what all the fuss is about. It's pretty normally for switch mode power supplies and is not an issue unless it's a medical device, lots of the PSUs are connected in parallel or you're using it to measure stic sensitive components.

As I keep saying, it's the Y1 rated capacitor which is leaking a tiny current from the mains to the DC side. If you care to measure it, you'll probably find it's probably under 250µA which the acceptable leakage for class 2 domestic appliances.

[johnmx]

I’m concern because the prototype is a controller of several semiconductor optical amplifiers (SOA) which are static sensitive devices and very very expensive. Although these SOA’s have the case electrically isolated, I prefer to follow the safest solution and connect the system ground to earth at the output of the SMPS.
I need to learn more about the SMPS medical type. Maybe next time I will buy a SMPS medical version, the price difference is just a few euros and has a big advantage of no minimum load needed.

Is the sometime 90Vac line 50 Hz or in the radio frequency?

I connected a 10kOhm resistor between the ground and earth just to check this signal and the result can be seen in the attached images. The announced switching frequency is around 70kHz. The SMPS is a XP Power, model: ECM40UT31.

[Zero999]

I’m concern because the prototype is a controller of several semiconductor optical amplifiers (SOA) which are static sensitive devices and very very expensive. Although these SOA’s have the case electrically isolated, I prefer to follow the safest solution and connect the system ground to earth at the output of the SMPS.
I need to learn more about the SMPS medical type. Maybe next time I will buy a SMPS medical version, the price difference is just a few euros and has a big advantage of no minimum load needed.

Is the sometime 90Vac line 50 Hz or in the radio frequency?

I connected a 10kOhm resistor between the ground and earth just to check this signal and the result can be seen in the attached images. The announced switching frequency is around 70kHz. The SMPS is a XP Power, model: ECM40UT31.

As expected you've got both the mains frequency and the mains.

You could connect both the positive and 0V rail to earth via a 100nF ceramic capacitors which will get rid of most of the mains and RFI.

A medical grade SMPS is definitely the solution but you'll have to pay for it.

[FreeThinker]

Neutral conductors can ( and do! ) carry large currents and float well above Zero volts (Earth potential ).I started a thread about this and the problems it can cause. Truth is the mains supplyis dirty and getting dirtier by the week.The only answer is better filtering either at the incoming mains or at the polluting source, neither is likley to happen soon so if it is a problem you need to filter your equipment.Does your mains really float 90v to earth? seems very high to me.

[saturation]

Thanks jahonen and alm, I'll withdraw my prior statements as it was posted in ignorance, and I've striked them out in case anyone reviews them in the future to avoid misleading them.

I've always made assumptions about power supplies used in medical and consumer grade as fairly close together in terms of leakage current.  Never paid attention to it since we were happy with the small leakages akin to old style step down transformer based linear power supplies.

As jahonen linked, the medical grade Schaffner input filters have just 2uA leakage, without the Y cap, usually we get < 5uA leakage current on medical gear.

A theoretical problem is the increasingly large number of consumer grade office equipment making it into medical areas, and our internal labs have never tested leakage as acceptable just assume its adequate too; also there are so many personnel bringing their cellphone or other chargers [ e.g. laptop] into areas, and the eye opener is how leakage current can vary widely from model, make, and age.

Nothing has happened in regards to safety, but maybe we are simply unaware of the relationship, and its relatively simple to follow through. 

If you take a look at for example the datasheet of Schaffner inlet filter, then it can be seen that specified leakage current for standard types is around 370 µA (@230 VAC). Those 2.2 nF Cy-capacitors form a capacitive voltage divider.  My personal experience is that it can really kill most consumer-grade interfaces, when plugging or unplugging them while connected to the mains. The fact that most usual consumer connectors stupidly connect the signal first(!), and ground last, makes it even easier. The leakage feels also extremely nasty, for example when connecting an grounded antenna to floating system. It is not a surprise that medical types don't have Y-capacitors.

Problem is also multiplied if there are several devices plugged in same extension cord. That can cause much higher leakage currents. I guess that it is far from safe to disconnect the safety ground wire at big building where there are hundreds if not thousands of such devices.

Regards,
Janne

[saturation]

Thanks hero999 for the graphic and commentary.  Of concern in theory, are multiple consumer grade PSU in parallel in a medical setting all riding the same branch circuit in line with medical equipment.  Labs working in < 1V and mA/uA range projects need be aware of the potential for noise into their power supplies from sources within the line voltage introduced by their accessory gear or others using the branch circuit on the same floor.

So far testing hasn't shown leakage above the limit of 500uA, it typically less than 10% of that but survey continues.

I don't see what all the fuss is about. It's pretty normally for switch mode power supplies and is not an issue unless it's a medical device, lots of the PSUs are connected in parallel or you're using it to measure stic sensitive components.

As I keep saying, it's the Y1 rated capacitor which is leaking a tiny current from the mains to the DC side. If you care to measure it, you'll probably find it's probably under 250µA which the acceptable leakage for class 2 domestic appliances.

I don't see what all the fuss is about. It's pretty normally for switch mode power supplies and is not an issue unless it's a medical device, lots of the PSUs are connected in parallel or you're using it to measure stic sensitive components.

As I keep saying, it's the Y1 rated capacitor which is leaking a tiny current from the mains to the DC side. If you care to measure it, you'll probably find it's probably under 250µA which the acceptable leakage for class 2 domestic appliances.

[Zero999]

SMPSes typically leak more current than old laminated transformer PSUs.

A medical SMPS will probably be a class 1 device because the EMI will most likely be leaked to earth rather than to the mains. I would imagine they use an earthed screen between the primary and secondary windings of the transformer.

Isolation transformers are used to ensure leaky devices such as commercial grade SMPSes comply with medical requirements. I would advise you get an isolation transformer which can be used for testing mains powered devices safely using a 'scope.