I am using LTspice to simulate a MC34063 boost circuit to bring a 3.7 V Lipo battery up to 5 V in order to power a Raspberry Pi Zero.

When I simulate the circuit it works just fine (shown in the first picture):


However, when I add a load resistor representing the Raspberry Pi Zero, the voltage drops back down to 3.7 V (shown in the second picture):


I have been trying to troubleshoot this circuit problem. Maybe there is a glaring mistake that someone sees that I don’t.

The math for the load resistor: V / I = R, (voltage with the load at ~3.74 V) / (~0.1 A) = ~37.4 Ω

I got ~0.1 A from this site.

I got ~3.74 V from measuring the voltage with the load.

Please tell me what you think; I would greatly appreciate feedback.

Edit: This where I got my model the circuit was originally for buck circuit but I changed it to a boost https://www.eevblog.com/forum/beginners/simulate-mc34063-on-ltspice/

  • \$\begingroup\$ Your feedback resistors are too small, comparable to the load. Add an order of magnitude or two (I'm referring mainly at R3). \$\endgroup\$ Commented Aug 3, 2022 at 20:38
  • \$\begingroup\$ Where did you get the model from? I found one from Onsemi, but it's riddled with problems, and I'm not really in the mood for fixing. \$\endgroup\$ Commented Aug 3, 2022 at 20:54
  • \$\begingroup\$ Hi a concern citizen, I got my model from the ti datasheet linked on MC34063. I tried changing R3 from 330ohm to 3.3k and 33k and it didn't make a difference. \$\endgroup\$
    – Ritz123
    Commented Aug 3, 2022 at 20:58
  • \$\begingroup\$ (when replying to someone, use @, otherwise the user don't get notifications, so they might miss the reply/-ies. Use @<TAB> to cycle between a list of available names, they should appear without any spaces. If <TAB> doesn't bring up any list then there's no need for @. The owner of the post is always notified.) \$\endgroup\$ Commented Aug 3, 2022 at 21:42
  • 1
    \$\begingroup\$ Yes. Just click on "Window" in the menu and choose "Tile Vertically". And use the "Space" bar to fill the window with the schematic. \$\endgroup\$
    – PStechPaul
    Commented Aug 5, 2022 at 20:21

1 Answer 1


For one, the feedback resistors (R3, R4) have low values that, when the load is added, they will influence the value of the load (it will be less).

The value for the capacitor C1 implies a ton of 3 μs, but the maximum frequency for the MC34063, as listed in the datasheet, is 100 kHz. A value of 1 nF gives a typical 33 kHz, so C1 will force the chip to switch at a much higher frequency. Building a switched-mode supply is really something that needs an oscilloscope so, if you have one, do check the oscillation.

The value for the current sense is quite low for a 0.1 A load. I don't know how you calculated it, but if you look on TI's datasheet, on page 14 you have a table with some design equations. Even the voltage divider seems off if you want a 5 V output. I imposed a 75 kHz switching frequency, a 3.3 V minimum input voltage, and what I get is this:

ton  = 10.5u
toff = 28.6u
C    = 420p
Ipk  = 0.357
Rsc  = 0.84
L    = 82u

Use 430 pF for the capacitor. And for the divider, with R1 = 11k => R2 = 33k, choose 33.2 kΩ. The initial model I found was from Onsemi, but it's riddled with problems, so I found another one made by (the unfortunately late) analogspiceman, here. I modified the Zener to have Roff = 550 because the reference was lower than 1.25 V. In rest, this is how it simulates for me:

34063 boost

  • \$\begingroup\$ Thank you for the explanation a concerned citizen! I will be redoing my math for my calculated values. \$\endgroup\$
    – Ritz123
    Commented Aug 4, 2022 at 15:33
  • 1
    \$\begingroup\$ The question was edited to link to the model they used which seems to be the Onsemi one, but modified by Helmut. Just a FYI, I can't get it to work either. The model you linked to seems to be much more stable. \$\endgroup\$
    – Ste Kulov
    Commented Aug 4, 2022 at 16:01
  • 1
    \$\begingroup\$ @SteKulov Looks like Helmut simply converted the behavioural expressions to avoid the ternary operator. LTspice can handle it as long as it's not nested. It's most probably parser thing. Analogspiceman was very good with modelling and control was his speciality. A pity he's gone. \$\endgroup\$ Commented Aug 4, 2022 at 20:09
  • \$\begingroup\$ @Ritz123 Those values should work for your case, since it was with your conditions that I calculated them. But, it never hurts to double-check. \$\endgroup\$ Commented Aug 4, 2022 at 20:10

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.