Fixing A Ground Loop

[iamdarkyoshi]

So I have a project that involves powering a monitor and desktop motherboard from six lithium batteries. The problem I have is that the monitor needed 16vDC so I could not use one of the rails from the main PSU, an M2-ATX-HV. The monitor is therefore driven off of a cheap buck regulator from China. Here is an image of it:


The heatsink on the inductor and the bottom of the board are my mods, the board has been electrically insulated from the heatsink.

I noticed a possible problem though:
The ground rail appears to have a diode on it.

Anyway, the issue is that the monitor (VGA only) is displaying a lot of scrolling lines on it, and it only happens when the PC and monitor are driven off of their respective supplies connected to the same battery pack. If I connect the monitor through the buck regulator to my bench supply, the scrolling lines go away. I hate VGA.

What would be the solution here? The ATX supply and buck regulator share a power source, and I think we are getting power flowing through the VGA cable's shield, causing interference. I tried putting a common mode choke on the input of the monitor with a 470uf capacitor, and it did not make much of a difference.

Could it be that diode on the DC/DC letting power flow through the shield on the video cable instead of the diode? I haven't the slightest idea on where to begin.

[sangvikh]

From an electricians point of view, the way to eliminate ground loops is to bond all the ground planes of the components together.
That way you eliminate any difference in potential between components and you will have no current going through screens etc.

This is why turntables have a ground connection point, so that no noise is introduced on the low level signal going to the RIAA

[ajb]

Just looking at the picture, that diode doesn't look to be in the ground path at all.  Notice the couple of vias right near In-, that presumably connect it to a ground plane on the back of the PCB.  Also a diode there wouldn't really make sense for a number of reasons.  Rather it looks like that's just the low side catch diode, it plus a high-side transistor built into the buck controller IC provide the low- and high-side switches for the buck converter.  If you can read the part number on the IC you could look up its pinout and trace out the circuit to verify, or just use a meter and check for continuity from In- to Out- and from Out- to In-.

So if that's not the problem, what is?  It depends on how everything fits together.  If you can provide some pictures of the overall setup, that might help provide some clues.  Some extra filtering on the buck converter will definitely help--those electrolytics are likely going to be the cheapest thing they could find, and probably have terribly ESR, so replacing or augmenting them with something better will certainly make for an improvement.

[uncle_bob]

Hi

Best guess:

You don't have enough filtering on your supplies or you somehow have messed up (broken) the common ground connection.

Simple answers:

1) Star ground. Tie all of the grounds for everything back to one common point.
2) Proper filtering caps and inductors on both the inputs and outputs of all your regulators.


Bob

[iamdarkyoshi]

I have added commonmode chokes and capacitors to the input of the monitor, atx supply and buck regulator and the interfefence has gone WAYYYY down. I shall check the continuity of the input-output on the buck regulator (disconnected of couse) and see if ground goes straight through.

[uncle_bob]

I have added commonmode chokes and capacitors to the input of the monitor, atx supply and buck regulator and the interfefence has gone WAYYYY down. I shall check the continuity of the input-output on the buck regulator (disconnected of couse) and see if ground goes straight through.

Hi

Do you have a ground through your HDMI cables? If so does it follow your grounding rules?

Bob

[iamdarkyoshi]

Here is my setup:

Battery branches off onto the buck regulator and ATX DC/DC. Each have their own capacitor, then commonmode choke, then another cap. The ATX board goes to the motherboard through standard ATX cabling. The buck regulator goes into the monitor, right before the monitor is another capacitor and choke. The motherboard is connected to a graphics card through a cheap PCIe riser cable. The graphics card then goes to the monitor through a VGA cable. If this was digital, I probably wouldn't have this issue, but the monitor only has VGA. The issue is nearly gone with the addition of all of these chokes and caps, but only on specific backlight brightnesses on the monitor. Hmm....

Here is an image of the setup, keep in mind that I have absolutely no shielding or whatever, really the only common ground point is where the battery branches off onto the DC/DC converters.

Open in a new tab to see details:

[uncle_bob]

Here is my setup:

Battery branches off onto the buck regulator and ATX DC/DC. Each have their own capacitor, then commonmode choke, then another cap. The ATX board goes to the motherboard through standard ATX cabling. The buck regulator goes into the monitor, right before the monitor is another capacitor and choke. The motherboard is connected to a graphics card through a cheap PCIe riser cable. The graphics card then goes to the monitor through a VGA cable. If this was digital, I probably wouldn't have this issue, but the monitor only has VGA. The issue is nearly gone with the addition of all of these chokes and caps, but only on specific backlight brightnesses on the monitor. Hmm....

Here is an image of the setup, keep in mind that I have absolutely no shielding or whatever, really the only common ground point is where the battery branches off onto the DC/DC converters.

Open in a new tab to see details:

Hi 

Simple solution:

All the signal cables go to or from the PC. The best thing is to make it your central ground point. Tie all the power grounds back to the PC case. Tie the  battery ground to the PC case. That way there are no "ground loops". Everything heads back to that single point.

Past that, switchers generate a lot of hash and EMI. That is why you see big metal cases on them rather than parts just hanging out in mid air. The volume of a typical switcher is about 50% filtering, 20% cooling, and 30% the sort of parts you see on those little PC boards. Doing that filtering right takes some care.

Bob

[iamdarkyoshi]

Here is my setup:

Battery branches off onto the buck regulator and ATX DC/DC. Each have their own capacitor, then commonmode choke, then another cap. The ATX board goes to the motherboard through standard ATX cabling. The buck regulator goes into the monitor, right before the monitor is another capacitor and choke. The motherboard is connected to a graphics card through a cheap PCIe riser cable. The graphics card then goes to the monitor through a VGA cable. If this was digital, I probably wouldn't have this issue, but the monitor only has VGA. The issue is nearly gone with the addition of all of these chokes and caps, but only on specific backlight brightnesses on the monitor. Hmm....

Here is an image of the setup, keep in mind that I have absolutely no shielding or whatever, really the only common ground point is where the battery branches off onto the DC/DC converters.

Open in a new tab to see details:

Hi 

Simple solution:

All the signal cables go to or from the PC. The best thing is to make it your central ground point. Tie all the power grounds back to the PC case. Tie the  battery ground to the PC case. That way there are no "ground loops". Everything heads back to that single point.

Past that, switchers generate a lot of hash and EMI. That is why you see big metal cases on them rather than parts just hanging out in mid air. The volume of a typical switcher is about 50% filtering, 20% cooling, and 30% the sort of parts you see on those little PC boards. Doing that filtering right takes some care.

Bob

I agree, after adding the chokes, the interference went WAY down. It is nearly gone. I shall try more after dinner to see what helps and what doesn't

[iamdarkyoshi]

It's a common positive converter, you can not use it with connected grounds. You need another module.
The reason people design common positive converters is because low side drivers are cheaper to implement than high side gate drivers.

not sure this is accurate. I checked with a multimeter and it seems to be common ground.

[iamdarkyoshi]

It's a common positive converter, you can not use it with connected grounds. You need another module.
The reason people design common positive converters is because low side drivers are cheaper to implement than high side gate drivers.

not sure this is accurate. I checked with a multimeter and it seems to be common ground.

Can you post a photo of the back of your module?

I would need to clean off the thermal paste... I have it running right now with my scope across it and the only thing I am picking up is noise from the "antenna earth lead"

[iamdarkyoshi]

I would need to clean off the thermal paste... I have it running right now with my scope across it and the only thing I am picking up is noise from the "antenna earth lead"

So it is a high side converter, and the trace the diode is connected to is actually the switching node?

If so, try pass power in+ and power in- together for a few turns in a ferrite ring? Keyword: TDK clamp filter.

You know those commonmode chokes they use in standard switchmode supplies for the mains input filtering?

These ones?


I have one of those on the input of each switchmode supply, and one right before the input on the monitor. Each noise choke also has a capacitor on the input and output of the coil. Using my scope I was able to see the noise go WAY down on the voltage rails and the noise on the analog screen is a lot less noticeable, but it IS still there. And yes, I saw your comment on the suggestion topic