1
\$\begingroup\$

I'm using a really good 5V-35A switching power supply to power some mini PCs (Cubieboards if that matters) as below.

schematic

simulate this circuit – Schematic created using CircuitLab

The weird thing is that when I measure the voltage on the PCs input terminals, they are below the power supply's voltage. If I measure it directly at the PS it's always 5.2V, no matter how many PCs I use on the circuit. However, If I measure it on the terminal that connects all the PCs, it gets lower and lower as I increase the number of PCs. Even though they are connected in parallel!!!

They don't have any other connection between them, only the power source.

The terrible thing about this is that after about 5 PCs the voltage gets so low that I can't power any PCs anymore (and I need to power around 20 of them using this source). I've tried using a different power supply and different connectors and the results were exactly the same.

What could possibly be happening? How can I solve it and what should be the best way to power all of those devices using the supply I have?

Thank you in advance,

Theo

\$\endgroup\$
  • 2
    \$\begingroup\$ What size wire are you using between the power supply and the PCs? And how long is it? \$\endgroup\$ – Peter Bennett Apr 15 '14 at 23:51
  • \$\begingroup\$ Are you using a single cable run, or multiple? \$\endgroup\$ – Passerby Apr 15 '14 at 23:59
  • \$\begingroup\$ Do you get the same pressure at the end of a long hose-pipe as you do when using a short hose-pipe? \$\endgroup\$ – Andy aka Apr 16 '14 at 7:30
  • \$\begingroup\$ The cable is around 10cm. Not long enough to be a problem, I guess. The gauge of the cable is a little bit bigger than the cable that comes with the Cubieboard. It's a 2mm gauge, shouldn't be a problem either. \$\endgroup\$ – Theo Apr 16 '14 at 14:37
  • \$\begingroup\$ A link to the power-supply may be helpful. What type of connector/terminal blocks are used? How many pins for each output are there? Are you using them all? Are you using just one? Besides voltage drops in cables, a common location for a voltage drop is at the connectors. \$\endgroup\$ – Tut Apr 16 '14 at 14:50
4
\$\begingroup\$

As specified in comments, it is very likely that you are experiencing excessive loss in your cables. If you are using very long wire or wire of a too small gage, you are actually getting such an equivalent circuit:

schematic

simulate this circuit – Schematic created using CircuitLab

Where the vertical resistors are the "computers" modeled as loads and the horizontal resistor are losses in the wire.

The resistor values are arbitrary, but a too small wire may cause the assertion of having a negligible resistance to be false. You can see on AWG page on Wikipédia the resistance of a cooper wire per meter as a result of it gage (diameter). You have to consider that the higher the current, the higher the resistance.

In order to solve your problem it is you case, you have to: 1) Use larger wire (smaller gage); 2) Use shorter wire (if possible); 3) Avoid daisy chaining your boards, as the first wire will have a larger amount of current flowing that the last one, resulting in larger losses.

You should prefer a star configuration:

schematic

simulate this circuit

so that you have a different wire starting from your source to each of your board instead of a wire running from power supply to first board, then from first board to second one and so on. Same thing applies to your return path (ground).

You may or may not be able to apply all of the previous advices depending on your setup, though.

\$\endgroup\$
  • \$\begingroup\$ I agree with @Mishyoshi in principle. However, the resistances in that schematic (10 ohms for the wire) are quite unrealistic. Apparently a Cubieboard draws 0.5A, or if it has a SATA HD, then 2A. So even a 1 ohm resistance in the wire and connectors would cause a 0.5V drop. With five HD-equipped Cubieboards daisy-chained, one might see 10A, in which case even a 0.1 ohm resistance results in a 1V drop. \$\endgroup\$ – gwideman Apr 16 '14 at 9:25
  • \$\begingroup\$ "The resistor values are arbitrary, but a too small wire may cause the assertion of having a negligible resistance to be false." I just didn't both performing calculations as I removed a 0 to the default value in editor. I was only explaining the concept. Now that you provided more realistic values, I will edit my answer to improve it. Thank you. \$\endgroup\$ – Mishyoshi Apr 16 '14 at 14:34
  • \$\begingroup\$ @Mishyoshi, these are great tips. However, in my tests i'm using really short wires of around 10cm and with a gauge slightly larger than the cables provided as power cables for the PCs. I'll upload pictures of the tests on the original question. Also, If I measure the voltage drop on the cable, I get 0V. \$\endgroup\$ – Theo Apr 16 '14 at 14:38
  • \$\begingroup\$ @Mishyoshi, I believe you are correct. I still get no voltage drop on the cable, but when I used a higher gauge and shorter one, the voltage drop was reduced. I'll make some more tests here and let you know, but I'll mark your answer as accepted anyway, since it led me to a direction of thought! Thank you so much. \$\endgroup\$ – Theo Apr 16 '14 at 15:29
  • \$\begingroup\$ @Theo Also, don't overlook the connectors -- you may find a voltage drop either in the connection between mating connectors, or where a wire attaches to a connector. \$\endgroup\$ – gwideman Apr 16 '14 at 23:07

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.