I am using an LT8390 DC to DC converter, I have a highly inductive load and it is causing dips. The dips are shown here (and about a 6V drop for 7ms):

enter image description here

I need to get rid of the dips, which I will increase the output capacitance from 100uF electrolytic to 1000uF.

The layout is such that I have moved the feedback line after the filter caps, thinking that this would keep the line more steady at the point I wanted it controlled at. I'm wondering if this could be part of my problem.

From a control stand point, where do I want to place the feedback line entry point in the circuit? Pictured below are the schematics, is A or B better?

Entry point A with feedback line before filter capacitors: enter image description here

Entry point B with feedback line after filter capacitors: enter image description here

Edit, here is the section of the board that has the feedback trace, I have ran it directly to the output of the board, instead of before the output cap. enter image description here

  • \$\begingroup\$ Two questions: What's the nature if the inductive load, is it transient, or dynamic (do the dips follow changes in its' state? What is the length difference between the two feedback points? from the schematic they appear to be electrically identical, but if there is significant length and current, then you could have some parasitic effects from the line length. \$\endgroup\$ – isdi Oct 18 '18 at 22:01
  • \$\begingroup\$ Also what node is channel 4 (green) measuring? \$\endgroup\$ – isdi Oct 18 '18 at 22:03
  • \$\begingroup\$ @isdi The green trace is the inductive load (motor) the yellow trace is Vout. I'll post a picture of the line \$\endgroup\$ – Voltage Spike Oct 18 '18 at 22:48
  • \$\begingroup\$ And the application note and PCB design example for LT8390 says what? \$\endgroup\$ – Ale..chenski Oct 19 '18 at 0:20
  • \$\begingroup\$ @Ale..chenski says nothing specifically about placement, why do you think I asked? On linear regs it is better to place the FB as close to the load as possible. \$\endgroup\$ – Voltage Spike Oct 19 '18 at 4:01

From their reference schematics

enter image description here

it is clearly seen that the feedback wire (R18) comes from the main filter cap C20, and not from the point of load. It is actually in between two caps, C19 and C20.

More, your entire layout has no resemblance with the suggested demo board:

enter image description here

All high-current loops in your layout have skinny traces instead of being solid rectangle of copper. I am afraid this is where your output ripple problem comes from.

  • \$\begingroup\$ And to be noted that reference design has 4 layers PCB. It could be also benefical to add an indutor cca. 800nH and an additional bank of capacitors, to isolate the load dynamics from voltage feedback sense \$\endgroup\$ – Marko Buršič Oct 19 '18 at 7:06
  • \$\begingroup\$ @MarkoBuršič, if a reference design has 4 layers, it very likely means that it is nearly impossible to make a good working converter in a two-layer PCB and meet advertised specifications. Otherwise the manufacturer's marketing would be more than happy to claim "low-cost 2-layer PCB" in the feature list for this product and expand its market. \$\endgroup\$ – Ale..chenski Oct 19 '18 at 7:24
  • 1
    \$\begingroup\$ My question is more along the lines of what is the difference between the two placements of the feedback line and how this affects the loop. What does this do to the feedback? It's a hard question, that's why I asked it here \$\endgroup\$ – Voltage Spike Oct 19 '18 at 15:01
  • \$\begingroup\$ The main body (inductor, fet, controller, bears a close resemblance to the reference diagram. I also have lower current. I would think it unwise to have to layout the reference board verbatim every time you did a new design. \$\endgroup\$ – Voltage Spike Oct 19 '18 at 15:11

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.