Earthing - potential ground loop problem?

I've currently got my eye on the earthing: the supplied ATX PSU ties the mains earth to the 0V on one edge of the Duet PCB and the USB lead can connect to another mains earth (via connected PC) creating a potentially large enclosed loop. In the middle of this ring is the ground plane upon which the microcontroller sits. Ormorod is too power-hungry to connect to my PC supply which is derived from a UPS so it's on a differnet mains outlet to the PC therefore I'm unable to bring the two earths together as close as I would like to.

I was wondering if potential grounding issues had been taken into consideration when I found that the USB shell had been left unsoldered (I have yet to solder mine to the PC. It might have been better practice to connect the USB shell directly to the 0V PSU connection using star grounding technique. But as supplied, any 'earth currents times cable resistance' Voltages superimposed on the USB 0v line appear as a direct input to the Duet's 3.3V regulator among other things.

Connecting the outer screen of the USB cable via the USB socket shell should reduce any volt-drops in the inner 0v wire but such currents will still be flowing along the full length of the PCB. That's why I'm considering isolating the USB shell and connecting it via a beefy wire directly to the PSU 0V input terminal

I've been trying to figure out if this might br what causes crashes when soldering irons and other mains items are switched off. It only takes -0.5V to cause a CMOS latch-up.

---

RS Components Reprap Ormerod No. 481

Hi Radian,

You will get lots of different opinions on this.

My thoughts.... the current DUET has the screen soldered on two or four pins, BUT it is still not grounded, i.e. the pads do not connect anywhere. Opinion is divided amongst the big names between and R/C to 0V or a short via a ferrite bead (plus some other views) but the DUET is currently open circuit on this screen. I pointed this out to Ian and he showed on the schematic it is grounded, but this is not the case on the PCB artwork or meter.

The best situation for your arrangement is a USB isolator, Analogue Devices do a good chip and these are used in most products that you purchase. This will isolate all the signals and 0V's. These cost money and need another power supply!

The floating screen could be an advantage for noise pick up as it should couple back to the PC and hence earth. A ferrite clamped on the USB lead just before entry into the DUET should also help, by putting a barrier to the noise.

Just as a note, I checked my laptop and the shell of the USB does not appear to have a resistive path to earth, so it possibly just has a capacitor (another option!).

This discusses noise and the variations in the possible arrangements, but ground loops are yet another issue/problem as these are also involving 0Vs and not just screens and earth. The usb isolator (if suitable for the speed used) would be the best solution.
The other option which someone else from a safety perspective may wish to comment is to have a lab earth point and run seperate earths from the PC and Ormerod supplies to that, but this arrangement is for fixed mains wiring to avoid the issues of equipment/cables being used.

---

Ormerod #007 (shaken but not stirred!)

Hi Treth, yes I know that there is no simple concensus answer on shield connection but I think this is simply because the most appropriate solution depends on many factors - for example, if the peripheral device has its own earth or not. Setting aside the full isolation solution (which might actually be a useful option for some if USB over CAT5 is used as it would also allow several tens of meters separation between printer and PC) I'd like to see what can be done without going to these expensive lengths and hope we can pool some more brain power to keep Ormorod functional and low cost!

Some USB recommendations are for indirect coupling of the screen to ground - either to provide a path for AC (capacitively coupled) or for LF/DC (inductively coupled). At the moment the USB shield apparently floats at the Duet end so it at least acts as a good Faraday cage around the 0V wire inside. So there should be no RF current getting in that way, but the big earth loop still provides plenty of opportunities elsewhere - especially seeing as how the PSU wiring and Duet board are exposed. I instinctively feel that a low-impedance path needs to be established between the printer PSU and the PC PSU so that ground potentials are equalised before being fed into respective 0V rails on the PCBs. The USB shield seems convenient for this particular role as, in a typical desktop PC, the socket shell is electrically bonded to the chassis (not so much on a laptop) so direct coupling the Duet USB socket body to the ATX 0V input terminal where it enters the Duet PCB ought to provide a bypass for any externally injected currents between 0V systems - oughtn't it?

---

RS Components Reprap Ormerod No. 481

I must admit to being slightly puzzled, the solution to ground loops and all the other problems on USB is Ethernet... and it is there on the Duet already. Ethernet is isolated, so there are no ground loop issues, the protocols are designed to be robust and recover from errors, and is probably faster than USB, depending on type.

You can even get a telnet serial interface so it looks like a COM port on the PC, so no changes are required to pronterface, although a plain TCP socket interface would be straightforward.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Quote
bobc
I must admit to being slightly puzzled, the solution to ground loops and all the other problems on USB is Ethernet... and it is there on the Duet already. Ethernet is isolated, so there are no ground loop issues, the protocols are designed to be robust and recover from errors, and is probably faster than USB, depending on type.

You can even get a telnet serial interface so it looks like a COM port on the PC, so no changes are required to pronterface, although a plain TCP socket interface would be straightforward.

Ethernet is my preferred solution too, but while the USB option is there I'd like to be able to use it without hesitation.

---

RS Components Reprap Ormerod No. 481