I'm trying to implement an inverting Cuk DC-DC converter on a 2 layers PCB, in order to generate a -3.3V rail from 3.3V, but the circuit affects greatly the input rail. The 2 rails are used to power 4 Op Amps. I'm using the LTC1617 from Linear Technology and I copied the circuit from the datasheet, modifying the components values to fit my needs. The schematic is the following: Converter schematic

I simulated the circuit in LTSPice and everything seems to work. I had to implement the design on a 2 layers board, and I followed the suggested PCB layout that can be found in this application note: suggested layout

my layout:

enter image description here

When I power the circuit, the output voltage is exaclty as in the simulations (stable -3.3V with 20mV ripple) but the 3.3V input gets greatly affected. The voltage supplied from a bench power supply becomes a large smoothed sawthoot wave, with a pk-pk value of 5V, mean value of 3.3V and frequency 10Hz.

Picture from the scope (sorry if it's not quite readable, but it's the only picture I took): enter image description here

It seems to me that somehow the switching voltage at the SW pin node pass through the inductor and present itself at the input pin. I also can't explain the very low frequency. From the simulations the voltage on the SW pin reaches exactly the same voltage values as the ones that I find at the input: Simulation

This is the first design I make with a DC-DC converter and I'm not really acquainted with DC-DC converter topologies so I'm not sure where to look for errors. I'd greatly appreciate suggestions on how to debug the circuit. Thanks a lot.

I'm also attaching a picture of the full schematic (the DAC is not soldered on the board): full schematic

  • \$\begingroup\$ What happens if you add a larger capacitor (eg. 470uF) across the input? Exactly which bench power supply are you using? \$\endgroup\$ Commented May 6, 2022 at 19:36
  • 1
    \$\begingroup\$ Why does C11 go down to -3V3 right after the inductor, instead of to GND? \$\endgroup\$ Commented May 6, 2022 at 19:58
  • \$\begingroup\$ L1 and L2 may be close enough to be magnetically coupled. You may want to test that and add that to your simulation and see what it does. K1 L1 L2 0.3 ? \$\endgroup\$ Commented May 6, 2022 at 20:02
  • \$\begingroup\$ How are you measuring it? \$\endgroup\$
    – winny
    Commented May 6, 2022 at 22:30
  • \$\begingroup\$ Is there a load on the output? I’m wondering (because it’s been too long) if when in DCM the cuk converter returns energy to the power supply. Many bench supplies will freak out if you try to put power into them - that might be what you’re seeing. \$\endgroup\$
    – Bryan
    Commented May 7, 2022 at 5:35

1 Answer 1


What is providing the 3.3V source?

I'm suspicious you are triggering a constant current limit in that supply. So when the -3V3 is operating it is drawing just a little too much current on the +3V3, hence the slow drop in voltage. When it drops to about 1V, the LT1617 shuts down and stop drawing current and voltage shoots quickly back up to 3.3V, the LT1617 powers up and process repeats.

If this is the case, adding capacitance wont stop the problem but it will slow the oscillation down. The current limit would need to be increased.

  • \$\begingroup\$ This is actually a very good point. I will try to repeat the test and check the power supply's current limit. Will make an update with the result. \$\endgroup\$ Commented May 7, 2022 at 8:54
  • \$\begingroup\$ Your intuition was right. The current limit was ok but the PS channel I was using turned to be not working properly. Changing the channel solved the issue \$\endgroup\$ Commented May 10, 2022 at 10:32

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.