2
\$\begingroup\$

I am designing an experimental PCB. I have included a split rail DC/DC converter LM27762 on it. Since I wasn't sure it will work, for testing purposes I have included a 2x3 pin header for bridging the outputs of the converter to the board power planes. This is so that I can test the output voltages first. In the event that they are good, I will use a set of shunt jumpers to enable the board power planes. In the event that they are bad, I will supply external power to the power planes.

Is this ever a bad idea other than waste of resources? Will it pick up more noise? Should I always remove them after confirming the board is good? There are more capacitors after the pins.

enter image description here

\$\endgroup\$
2
  • \$\begingroup\$ I've seen it used often enough, so it seems unlikely that it's a particularly bad idea. \$\endgroup\$
    – Jos Bergervoet
    Commented Jul 16 at 6:40
  • 2
    \$\begingroup\$ Not worthy of a comment on its own but you could run PCB traces between the pins of the header. During development you can manually cut the traces with a knife and attach the jumpers. When you're happy with the design you just no longer cut the traces or populate the header pins. \$\endgroup\$
    – scotty3785
    Commented Jul 16 at 8:50

4 Answers 4

Reset to default
3
\$\begingroup\$

Beats me -- depends what the circuit is, what's on the other side of it.

But regardless of that, I can express what possible concerns I would expect, and what alternatives I would consider:

  1. Ampacity. Is the header and shunt rated for enough current? (Typically 2-3A each, less when many are used together -- mutual heating is a problem.)
  2. Reliability. Connections come loose, become flaky, etc. over time. Would there be any consequence of failure here? (Including momentary connections causing hot-plugging transients → you need to know the PDN characteristics.)
  3. Stray inductance. Would the, say, 5-10nH of the jumper cause any issues for supply quality / PDN?

You'll have to decide which of these are applicable to your case.

For design test, even just putting in a visible (surface) trace might suffice, and I can cut that for test, then leave it alone in production, assuming that no board rev is required (i.e., none at all, or only component changes). Scraping off soldermask and jumpering it back together allows using the board for further prototype testing.

Alternately, a zero-ohm jumper can be used, a chip component that can be desoldered for test, and left in place otherwise. It might be a zero-ohm jumper resistor, particularly for low currents, or a solid metal jumper for high currents. This might even be a standard component, say a small-value inductor when you need additional filtering but you just remove it for testing the regulator, and don't care about ripple during the test.

\$\endgroup\$
3
\$\begingroup\$

It's fine if you are comfortable with adding 50 to 100milliOhms of contact resistance and a small amount of inductance to the rails. I'd leave the caps in.

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

Typically during first-round board design I include in schematic/PCB a discrete zero-ohm jumper between the power supply (just after the minimum required output filter caps) and the load net, and size appropriately for the supply current, which often means adding a 0805 or 1206 resistor footprint before going to power plane or distribution fills. During bringup and testing this resistor can be removed to allow debug by separating source from load as you mentioned and allows measurement of the average current drawn by load by inserting meter (assuming loads will tolerate the meter inductance). Note that a typical standard zero-ohm jumper will have datasheet specs with maximum resistance of 30-50 mOhms and should state maximum current too. If the standard jumper could result in too much voltage drop (for example on high-current low-voltage core power rail), there are high-precision, very-low (1 to 10 mOhm) resistor parts that are intended as series shunt for current sensing, but they are expensive. So likely would remove the part entirely before production PCB.

This is just different way to accomplish the same goal as your series header pins, and maybe is smaller and more reliable but less convenient. I doubt the header pin method would have any noise pickup or EMI emission issues, but I would remove to increase reliability when going to production.

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

You could even use several jumpers for different supply domains going to different parts of the circuit. That could help for problems that are not in the supply but in the circuit, to find out which part of the circuit is at fault. And it would also mitigate the extra milli-\$\Omega\$s you introduce.

Picking up interference is not likely a problem, since the supply coming from the DC/DC converter is polluted anyway and will require further filtering in sensitive parts of the circuit. Also the "waste of resources", which would be a concern for a production version, does not seem to apply here since you say it is an experimental board.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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