1
\$\begingroup\$

I'm hoping to find some help here troubleshooting.

The circuit in question is a bipolar power supply that makes use of two LM2576-ADJ. The desired output voltage is +12v and -12v. I've breadboarded this circuit before ordering the PCB, and it seemed to work exactly as intended, where the 20k trimmer sets a positive and negative output voltage between ~1.23 and ~14V.

However, on the PCB, it doesn't function properly. The negative output voltage always reads as -13.4V and doesn't adjust with the trimmer. The feedback pin of the corresponding LM2576 reads instead as -11.5V.

On the positive voltage side, upon powering on, the output voltage slowly rises from about 2.5V to 12.4V. Here, the feedback trimmer also has no effect on the output voltage. The feedback pin on this LM2576 instead reads the same as the output voltage.

Here is a link to the datasheet. On page 22 you can find the schematic from which I based the voltage divider part, and on page 19 is the inverting buck-boost version.

Do you think the problem is related to the feedback resistors? Why does the voltage slowly rise on the positive side?

The input voltage used is 12.68V, from a 12V DC adapter.

Here is an image of the schematic, and also of the PCB layout.

Schematic PCB LayoutPCB Layout

\$\endgroup\$
5
  • \$\begingroup\$ I hope those inductors are screened from each other... were they further apart on the prototype? And they are far too far from the switchers... \$\endgroup\$ Oct 10 '20 at 11:44
  • \$\begingroup\$ Thanks so much for your help! On the breadboard they were further apart- looking at a photo I can gather they were about 8cm apart- is this something that could account for the kinds of issues I'm seeing? \$\endgroup\$
    – otivlasc
    Oct 10 '20 at 11:55
  • \$\begingroup\$ It won't be too hard to find out... \$\endgroup\$ Oct 10 '20 at 12:00
  • \$\begingroup\$ LM2576 is a switching step-down positive voltage regulator. I see no way to configure it to generate a negative voltage ! Shame about the time spent on the board which looks nice. \$\endgroup\$ Oct 10 '20 at 19:25
  • \$\begingroup\$ Weird ! It's on the datasheet, so I'm game I suppose. \$\endgroup\$ Oct 11 '20 at 7:17
1
\$\begingroup\$

What might be a showstopper is that your input supply voltage is only 12.86 volts and that is insufficient for this regulator if it's producing 12 volts on the positive output. You'll probably need another volt of headroom for this to work correctly. It's not an LDO switching regulator - the data sheet is probably a little cautious about this but it does suggest that 15 volts is the minimum.

Ignoring any other potential circuit faults, the layout is not-so-good in that you will have problems with noise and regulation when you solve the circuit issue. I will also mention that I can see that you have fixed the negative voltage regulator error shown in the data sheet - the ADJ pin does need to be set with a potential divider for that circuit to work and you have correctly done that it appears. Anyway, these are my layout recommendations: -

  • Keep the ground plane exclusive to one layer (underneath usually) and route connections only on the top copper where possible.
  • Where connections cannot be exclusively routed on the top copper, make a bypass connection on the underside but, keep it as short as possible so as not to disrupt the ground plane.
  • Use thicker connections for power feeds to the chips, inductors and decoupling capacitors.
  • Keep switching tracks as short as possible
  • Use component reference designators in the silk screen
  • Get the inductors much closer to the chips - at the moment those connections will form loop antennas (especially as the grounding method you currently use is poor).
\$\endgroup\$
5
  • \$\begingroup\$ The supply voltage was it! Thanks so much for taking the time to give me this information about PCB design- it's really helpful! \$\endgroup\$
    – otivlasc
    Oct 10 '20 at 12:35
  • \$\begingroup\$ What might be an even bigger showstopper is persuading the LM2576 to generate a negative voltage. \$\endgroup\$ Oct 10 '20 at 19:26
  • \$\begingroup\$ @GrahamStevenson - no, it's a standard way of making a negative voltage. Sure, everything looks screwed up but it works. \$\endgroup\$
    – Andy aka
    Oct 11 '20 at 7:08
  • \$\begingroup\$ Weird ! It's on the datasheet, so I'm game I suppose. \$\endgroup\$ Oct 11 '20 at 7:17
  • \$\begingroup\$ You just have to think about it for 30 minutes and it probably helps that you believe it actually works. 1st time I saw it I was sceptical. \$\endgroup\$
    – Andy aka
    Oct 11 '20 at 7:20

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.