I printed a board meant to supply a variable 5–250VDC from this open-source schematic:

source schematic

This was my board

board in EAGLE

with this schematic (excuse my silly device symbols)

shematic in EAGLE

but I can't get the output voltage to vary. With the 5k potentiometer (the box at the bottom center of the board), it measured ~320V, invariant of the pot's setting. For a 10k pot, it measured the same voltage, again invariant.

Here's the soldered print (excuse the poor quality, someone else has them currently):

top soldered bottom soldered

Is something shorting? Is it somehow "skipping" the potentiometer? Is there anything obviously incorrect or any likely errors I could test for?

  • 3
    \$\begingroup\$ Looks this board routing was designed by Picasso... \$\endgroup\$ – Eugene Sh. Jul 10 '17 at 17:50
  • 2
    \$\begingroup\$ The routing on your PCB looks like an abstract art project ... \$\endgroup\$ – brhans Jul 10 '17 at 17:52
  • 1
    \$\begingroup\$ Without knowing what is to the right of your schematic it is difficult to understand your circuit. You have connected POT1 as a variable resistor rather than a potential divider. When you turn it to zero it will short that line to GND. Is that what you intended? \$\endgroup\$ – Transistor Jul 10 '17 at 18:05
  • 2
    \$\begingroup\$ At what load? You may be in deep discontinuous mode and your PWM circuit might have limited minimum duty cycle. \$\endgroup\$ – winny Jul 10 '17 at 18:07
  • 2
    \$\begingroup\$ Even if it were working correctly, C4 and C6 are underrated. There's a note in the original schematic that says the output varies from 65 to 260VDC - which exceeds the ratings that are also right next to the parts. \$\endgroup\$ – JRE Jul 10 '17 at 18:55

Unfortunately, this PCB is unlikely to ever work. You will need to redesign it.

Switching power supply designs are very sensitive to layout. In particular, it's critical to minimize the area and inductance of the current loop. Unfortunately, your design places the inductor rather arbitrarily, which leaves the current loop all over the place. For more information on these aspects of design, you may want to read some application notes on switching PSU design, such as:

I also see a couple of generic issues with your PCB design, including:

  • Your layout is not making use of its ground pour. A good ground is critical for all circuits, but is particularly important for a power supply design.

  • One of the terminals of C1 appears to be unrouted.

  • As other users have mentioned, it looks like you may have chosen some parts which are not rated for the voltages you're trying to produce.

  • You are using rather arbitrary packages and footprints for a lot of parts. For instance, you're using a Kelvin resistor footprint for RSENSE1, but ignoring the kelvin terminals, and it almost looks like you've soldered an electrolytic capacitor to a SMD capacitor footprint on the back side. Switching power supplies are sensitive to lead inductance; these choices matter!

  • \$\begingroup\$ Ah, layout. I've read that layout was critical to HV supplies but had no idea why. Though the ground pours didn't render, C1's other terminal is actually routed to the ground pour on that side of the board. But it makes sense that our other layout decisions would screw up the functionality. I will certainly read what you linked and be much more assiduous when re-designing. Thanks! \$\endgroup\$ – Jack Lynch Jul 10 '17 at 19:34
  • 1
    \$\begingroup\$ It's not specific to HV power supplies. Layout is critical to all high-frequency analog designs -- which includes switching power supplies, regardless of their output voltage. \$\endgroup\$ – duskwuff -inactive- Jul 10 '17 at 19:57
  • \$\begingroup\$ I'll definitely keep that in mind for the future. The PDFs you linked were super informative, too—thanks! \$\endgroup\$ – Jack Lynch Jul 10 '17 at 22:25

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.