I have a SEPIC converter for 12 to 48V in 36V 3A out, based on a LM5020 controller running at 100KHz.
It's prototyped using separate 80uH inductors, a 10uF coupling cap, 680uF on input and output, a 0.025Ohm sense resistor with a 300 Ohm / 1000 pF filter. Compensation is arbitrary 100pF with 3.3K - 1000pF. This is my first SEPIC design.
At low input voltages and reasonable load (about 400mA) I see a sinusoidal envelope on the Q1 voltage (it is still switching at 100kHz), as follows - with 12V input the calculated Q1 voltage should be about 48V but I see a 4KHz 15V P-P sine wave around the nominal value.
If I increase the coupling cap from 10uF to 20uF the sine wave goes away unless I increase the load again. I have tinkered with the compensation and current sense filter without observable effect.
I'm at a loss. 10uF is adequate according to Power Stage Designer. Can anyone explain what is going on?
Thanks
Paul,
Can you send me your schematic, I'll look it over.
Terry
It's a scan of a hand drawn schematic. It's a little bit light but it looks readable on my machine. First time attaching a file....
4370.08091200.PDF
Let me know if anything isn't clear.
Thanks Terry.
Paul
I'm not a SEPIC expert, so I'm not sure why your seeing the low frequency oscillation, other than a loop problem or resonance. My recommendation is to use a LM5118 which is a Buck-Boost controller, it will do the same thing, and you will find that it is much easier and less expensive to implement. I have attached the data sheet.
Tried that first. Blew up two devices (they are hard to replace) and probably some other parts on the board so I gave up on it.
As a separate subject - I tried using the 5118 evaluation board and found it extremely difficult to modify which seems sad because it didn't need to be. Had the designer allowed a little more space, access to heat up pads for soldering, and maybe options to install some different sized parts it would have been a lot more useful.
I do like the SEPIC topology, the numbers look good, the stress is low, it seems to perform ok. Is there anyone there who is a SEPIC expert?
Thanks, Paul