1
\$\begingroup\$

I have a test set up that can more or less be broken down like below:

schematic

simulate this circuit – Schematic created using CircuitLab The splitter shown is custom made for testing, here is a drawing: splitter cable

PCB1A + PCB2A pass the 12V line through to the Mini-ITX CPU board, which can be powered by 12V directly via the 4 Pin ATX connector; it does not need the full 24 pin ATX connector. PCB1A and PCB2A each have a PMOS to allow cutting power to the Mini-ITX, however they do not condition the power at all, i.e no regulation, boosting, etc.

I notice that if I power cycle Mini-ITX-B using PCB1-B, Mini-ITX-A also power cycles as Mini-ITX-B comes back online. Probably due to a voltage spike or dip.

I do not notice this with any other combination of assemblies, i.e. A-C, B-C, A-D, B-D, etc.

Notice that A+B are both tapped off of 1 output of the splitter, and are the shortest cables. I'm guessing this has something to do with it.

Does anything obvious jump out as to why this is happening? And, hopefully, is there an easy fix (i.e. a big cap, or ferrite bead somewhere)?

\$\endgroup\$
1
\$\begingroup\$

It looks to me that your suspicions are right, as are your possible solutions. You can try adding a decoupling capacitor placed as close as possible to the load points, in that way, the disturbances in the load are absorbed by the capacitor and do not propagate through the cable.

It can be a "not big" capacitor, you can try a ceramic or plastic capacitor, if it is too small it will have no effect, if too big it will be a hard load for the supply, since capacitors behave like short circuits when they are discharged.

As you commented, you can also add a ferrite bead. As the capacitor, it should be as close as possible to the load causing the disturbance (but not closer than the capacitor). If it is difficult to put the bead because of the connectors, you can try hinged ferrites.

\$\endgroup\$
1
\$\begingroup\$

The cross-interference between units A and B happens because they share the same power segment of supply wire, wire "1A". Apparently your PCB assembly (that holds mini-ITX computers) has too much of inrush current, which creates voltage "droop" along the segment 1A-1B-1D, and the common segment 1A has a profound effect in assembly A. Other assemblies (C,D,E) don't share the segment 1A.

First, you need to reduce the length of segment 1A to minimum, like crimping two wires "1B" right at the "8-pin EPS" contacts. This might help - it looks like the main PSU holds its voltage at pins, and is not having any significant "droop".

It would be nice to collect scope pictures of the power-switch event and see the depth of voltage droop and its edge shape. Before advising on ferrites and caps, peak current consumption should be determined.

Alternatively, your boards (with M1 and M2 power-off FETs) should be designed to limit the inrush current when turning on, somehow to provide soft current limit. It could be done by slowing down the gate control circuit, a simple RC, if the FET can handle instant power during transient over its linear region. This might be not an option for the current design, but a lesson for other designs.

\$\endgroup\$
  • \$\begingroup\$ Thanks a lot Ali. Have a nice Rigol being delivered Friday at which points I can get some scope shots and really dive in. Hey, if nothing else, it was reason enough to get boss-man to buy a digital scope finally. \$\endgroup\$ – Jim Nov 16 '17 at 1:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.