Author Topic: Grounding Bench Power Supplies (connecting GND to mains PE)  (Read 5277 times)

0 Members and 1 Guest are viewing this topic.

Offline neo2001Topic starter

  • Contributor
  • Posts: 14
  • Country: de
Grounding Bench Power Supplies (connecting GND to mains PE)
« on: March 01, 2019, 06:14:03 pm »
Since I have four different power supplies by now, I was thinking of how to handle the "GND situation" of the different devices.

I have one old ATX power supply, which I modified to be a bench power supply a few years ago. Since it delivers all the commonly used voltages for digital devices, without the need of adjustment, I kept it. Like all(?) ATX power supplies, the protective earth wire of the mains connection is tied to the GND pole of the power supply's DC output.

The next power supply is a "classic" 30 Volt, 5 Ampere linear power supply with a big transformer an power transistors. On the front there is a small piece of metal, which connects the terminals of PE/earth/case and DC GND. Since it came this way, I kept it in place.

Additionally I have two DPS power supplies from Ruideng. Both of these are "floating". Both use a grounded power supply (36 and 60 Volt respectively). The PE wire is connected to the metal case of the power supply, but not to the GND pole of the DC output.

I was asking myself, if it would make sense, to tie all GND poles to the PE/earth wire, especially if I use two of the power supplies on the same device, in which case they would need a common ground connection anyway.

Does this make sense? What would or could be the downside of "grounding" the latter two devices by connecting the DC GND pole to PE of the mains supply? I also thought of using a switch to make the connection optional.

Somehow having all devices on the same ground potential, seems favourable to me. Especially with USB devices or digital stuff connected to a PC. But I wouldn't ask, if I already knew the answer. :-)

 

Offline janoc

  • Super Contributor
  • ***
  • Posts: 3925
  • Country: de
Re: Grounding Bench Power Supplies (connecting GND to mains PE)
« Reply #1 on: March 01, 2019, 07:38:47 pm »
Somehow having all devices on the same ground potential, seems favourable to me. Especially with USB devices or digital stuff connected to a PC. But I wouldn't ask, if I already knew the answer. :-)

Actually, you most certainly don't want that. Try to think what will happen if you make a mistake and connect a ground lead of an USB oscilloscope/logic analyzer/whatever to a place of your device under test that is not ground.

Yep, the magic smoke escapes and depending on how you had it connected and what the voltage (and amount of current available) was, you will most likely blow out the USB port of your PC if not the entire motherboard.

It is exactly the same problem like with earthed oscilloscopes (watch the Dave's video on how to not blow up your oscilloscope). Having the device you are working on earthed severely limits your options and greatly increases the risk of causing a short when connecting two devices together.

Another reason why you don't want this is for connecting multiple of them together (e.g. to get a dual supply for testing amplifiers, opamps, etc.). Once you earth them you can't do that anymore or you will short something out.

For development work it is much better to use a lab supply with floating outputs - that's the reason your linear supply has the grounding strap there instead of being hardwired (I would take it out, IMO) and why the Ruideng supplies are floating. It is the configuration you want in 99% of the cases. The exceptions are few and far between.
« Last Edit: March 01, 2019, 10:02:12 pm by janoc »
 

Online Benta

  • Super Contributor
  • ***
  • Posts: 6420
  • Country: de
Re: Grounding Bench Power Supplies (connecting GND to mains PE)
« Reply #2 on: March 01, 2019, 07:46:58 pm »
What a mess!
PE is exactly what it says: protective earth. It's there to prevent dangerous voltages on something you can touch: enclosure, screws etc. and has nothing at all to do with your test setups.
All lab power supply outputs should be floating. The ATX supply's output should be disconnected from PE, the bridge on your second supply should be removed as well.
If at some point you need a PE connection in a test setup, fine, do it consciously. Not through the power supply, that's looking for trouble.

 

Offline magic

  • Super Contributor
  • ***
  • Posts: 7453
  • Country: pl
Re: Grounding Bench Power Supplies (connecting GND to mains PE)
« Reply #3 on: March 01, 2019, 09:02:18 pm »
Is it a good idea to disconnect GND from PE in a PSU which was designed that way?
I sometimes wondered if it's possible that some PSUs may rely on noise being shunted to Earth for EMC? Could things be so bad that it wouldn't work at all without Earth?

And one remark regarding the original topic: if you happen to use multiple PSUs to provide different voltage rails to one circuit, either none or only one of them should be grounded. Otherwise some current flowing through PE installation in the building could find its way through one PSU, grounding cables, the circuit's ground traces, another PSU and back to PE, injecting noise into the circuit's ground that really isn't needed at all.
 

Online Benta

  • Super Contributor
  • ***
  • Posts: 6420
  • Country: de
Re: Grounding Bench Power Supplies (connecting GND to mains PE)
« Reply #4 on: March 01, 2019, 09:42:18 pm »
Is it a good idea to disconnect GND from PE in a PSU which was designed that way?

A very good point and one that I didn't mention. ATX supplies have a precisely defined function.
Using a modified ATX supply in a lab environment is questionable at least, but defensible when you know what you're doing.
My feeling is that it's not the case here.

In general: confusing/mixing PE with ground is not a good idea.
 

Offline janoc

  • Super Contributor
  • ***
  • Posts: 3925
  • Country: de
Re: Grounding Bench Power Supplies (connecting GND to mains PE)
« Reply #5 on: March 01, 2019, 10:00:36 pm »
Is it a good idea to disconnect GND from PE in a PSU which was designed that way?

A very good point and one that I didn't mention. ATX supplies have a precisely defined function.
Using a modified ATX supply in a lab environment is questionable at least, but defensible when you know what you're doing.
My feeling is that it's not the case here.

In general: confusing/mixing PE with ground is not a good idea.

Yep, definitely no messing with the ATX supply, that's asking for a big problem. OTOH, removing the earthing strap on the output of the lab supply should be fine, that one is designed to be removed (or even moved to the other terminal, as required).
 

Online Doctorandus_P

  • Super Contributor
  • ***
  • Posts: 4004
  • Country: nl
Re: Grounding Bench Power Supplies (connecting GND to mains PE)
« Reply #6 on: March 01, 2019, 10:14:45 pm »
Same thoughts here. Do not modify the ATX by  modifying the PE connection.

Your DPS modules are another concern.
They may internally have a buck or flyback topology. and the output "GND" may well be at different levels (if both modules are set to different voltages). The inputs of each of these modules should only be taken from a floating power supply.
With cheap modules such as these the current shunt is often in the GND lead.
If you short the shunt by "random" wiring between all the GND leads of your modules, at best the current measurement does not work, and at worst: "If you connect the supply power with output, the module wil be burnt". (Taken directly from DPS5005 user manual).

Lab power supplies should always be floating. Then you can make connections on the front panel as needed. Use common GND by connecting GND's together, or use them in series, or even parallel.
If the power supplies themself are floating, then you can also mak a star ground on a convenient place and connect them individually to that.
 

Offline rdl

  • Super Contributor
  • ***
  • Posts: 3667
  • Country: us
Re: Grounding Bench Power Supplies (connecting GND to mains PE)
« Reply #7 on: March 01, 2019, 10:56:04 pm »
An ATX power supply is rarely suitable for use in an electronics lab. If you need to use one, you'll know, otherwise don't.
 

Offline neo2001Topic starter

  • Contributor
  • Posts: 14
  • Country: de
Re: Grounding Bench Power Supplies (connecting GND to mains PE)
« Reply #8 on: March 05, 2019, 05:56:55 pm »
Thank you all for your replies!

I realize that there isn't simple and easy right/wrong answer for this. It depends on the situation and intended application.

To be honest, I wanted to replace the ATX power supply by the other ones long ago. But I still keep it, since its provides precise and stable voltages, I often use or need. It's also nearly impossible to select the wrong voltage, since every output has it's own colored banana terminal.
 

Offline janoc

  • Super Contributor
  • ***
  • Posts: 3925
  • Country: de
Re: Grounding Bench Power Supplies (connecting GND to mains PE)
« Reply #9 on: March 06, 2019, 08:04:41 pm »
Thank you all for your replies!

I realize that there isn't simple and easy right/wrong answer for this. It depends on the situation and intended application.

To be honest, I wanted to replace the ATX power supply by the other ones long ago. But I still keep it, since its provides precise and stable voltages, I often use or need. It's also nearly impossible to select the wrong voltage, since every output has it's own colored banana terminal.

An ATX supply is pretty far from what one would consider "stable" and "precise". They are specified with about +-5% tolerance (that's 11.4 to 12.6V on the 12V rail and 4.75 to 5.25V on the 5V rail!) and the outputs tend to be full of switching noise - the ATX spec for max. ripple specifies 120mV peak-to-peak on 12V rail and 50mV on 5V rail. Many older supplies with poor quality capacitors or bottom of the barrel design are even worse than this.

120mV ripple is going to ruin your day if you are trying to power anything sensitive with it - e.g. various amplifiers or radio receivers.

Even the cheapest lab supply is going to be better than this, especially the linear ones. ATX supply is good to have on hand for things that require a lot of current - such as motors or large LEDs. However, if you need a stable, accurate and clean supply, stay away!

This is one of the reasons why almost everything in a PC has local voltage regulators - the "raw" power from the power supply is simply not good enough. **

Oh and the biggest issue of all - an ATX supply has no current limiting. It does have a short circuit protection but it will happily pump 20+A into your breadboard if you make a mistake, melting stuff or setting things on fire, without ever tripping that protection.



** The other reasons are need for different voltage rails and reducing losses from high current on the 5V and 3.3V rails - it is better to use 12V and down-regulate locally next to the load to minimize the length of the low voltage but high current connections.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf