I am prototyping for LT8362 for a "SEPIC doubled boost" converter for 80 V 100 mA output from 5 to 8 V input. All seems to work fine, EXCEPT that above roughly 7 V input the switching waveform becomes very unstable and "noisy".
This is the basic circuit:
I first assumed I had wrong compensation components, tried several things. The response looked pretty stable as long as I was at low enough Vin.
Then I figured I probably had too small an inductance (at first it was 6.8 uH and 33 uH so equivalent to a 5.6uH boost converter) and this might be too small and give subharmonic oscillation problem. I changed to the larger inductors above. Still trouble.
To rule out the loop gain / compensation I installed a 1 uF capacitor (only) for Vc pin. Stable at low Vin, but poor transient response expected of course. The jump to instability when Vin > about 7V is completely unchanged.
Finally, I guessed maybe the SYNC/MODE pin which is right next to SW pins may be picking up too much signal from SW pins and causing something bad. (Spurious syncing? SSFM? I don't know....)
Grounding SYNC/MODE has fixed the instability, the circuit works fine for well over 8 V input.
But, I'd like to use pulse-skip not burst mode, and not spread-spectrum, in this application. Is there any connection I can make at SYNC/MODE pin that will result in pulse-skip mode but be more immune to this SW pickup problem? Has anyone else seen this sort of issue with LT8362 or LT8361?
Thanks in advance,
p.s. I should mention, this is a hand-built prototype on copperclad board with a few "traces" carved by hand. There *is* a solid ground plane, but for sure the distances and capacitances are different than would be on a proper PCB, but if this problem happens on this prototype I believe there is not sufficient margin against it happening in a real board.
Yes connecting the capacitor should ok to gnd the noise. It might also be better if you use MSOP package to add distance from the SW and SYNC pin. So far we have not experience this kind of issue with…
My post was delayed for moderation and so never made the top of the list of open questions, it seems. Anyways I hope some expert could please comment on the possibility of troubles in floating the mode pin of LT8362, and what to do about it.
In the meanwhile I have tentatively connected it to a 100 pF capacitor to ground, this does prevent the problem that I was seeing. DMM says the pin comes to 0.987 V. Is it a reasonable scheme to use a capacitor to ground to reliably select pulse-skip, non spread spectrum mode? (And I still have the question whether anyone else has had trouble with floating the mode pin on this part? If so, it would be nice if the datasheet explained a bit.)
Yes connecting the capacitor should ok to gnd the noise. It might also be better if you use MSOP package to add distance from the SW and SYNC pin. So far we have not experience this kind of issue with our boards.
Layout can present unexpected issues so make sure you followed the layout guidelines.
Hi Mike, Fil,
Thank you for your answers, sorry for delayed reply here. I will include an 0402 cap to ground on mode pin on the board layout, and maybe it will work fine with DNI (or with cutting the trace right at mode pin pad. I am using the MSOP package, by the way. My observations above were on a hand-crafted prototype and I'm sure it will be better on a good layout, which will follow the guidelines as much as possible (though of course the circuit here is a little different).