0
\$\begingroup\$

I use a constant current SEPIC boost regulator ([TI LM3410X) to run a 8.5V 400mA OLED panel from a single Cell Lithium polymer battery. I have noticed that the input voltage ripple is too much (~1V peak) when using a digital PWM signal on the DIM pin of the regulator that in-effect gates the regulator's 1.6 MHZ OSC on and off. NO issues what so ever when the OLED is at its brightest (i.e. DIM pin is HI, or always on). The design is validated with TI's Workbench tools.

I have input low ESR MLCC caps and played with the value (up to 80 uF) and the regulator's output cap but not much improvement in input voltage ripple when the PWM signal does a low to high transition. Layout is clean and tight with a solid GND plane.

To me this seems to be a fundamental issue with digital PWM dimming methods where the inrush requirements can be large every time the PWM signal makes a rising edge and enables the regulator's OSC, isn't it?

What are the remedies besides playing with input inrush storage cap and output cap of the regulator?

SEPIC BOOST REgulator using LM3410X

\$\endgroup\$
8
  • \$\begingroup\$ Please add a diagram that shows us what caps you have been experimenting with. \$\endgroup\$ – jippie Jul 15 '14 at 5:07
  • \$\begingroup\$ "Inrush" is a term that describes initial power-on situations. \$\endgroup\$ – Andy aka Jul 15 '14 at 7:51
  • \$\begingroup\$ What is your PWM frequency? \$\endgroup\$ – Dave Tweed Jul 15 '14 at 11:24
  • \$\begingroup\$ What is the reasoning as to using a SEPIC regulator over a different boost regulator? SEPICs are generally meant for systems that won't vary the power draw too often. \$\endgroup\$ – Funkyguy Jul 15 '14 at 14:05
  • \$\begingroup\$ 1)Added schematic, 2)I have been playing with C5,C6,C12 , 3)Inrush is relevant here because the regulator 1.6MHz PWM itself is gated with the DIM pin of the device so it is in a sense starting everytime the DIM pin is asserted HI. 4)PWM frequency is 1KHz although I have tried 500Hz too. I need to make sure the inductor does not hum so I have been playing with that a bit and it changes the signature of the input ripple but it is fundamentally there at every rising edge of the PWM-DIM signal \$\endgroup\$ – ezrcer Jul 15 '14 at 14:12
0
\$\begingroup\$

Try adding another inductor identical to L1 on the input, and use as much capacitance as you can fit in between the two inductors.

You could also reduce the output capacitance C12 to reduce the amount of energy that needs to be supplied every new PWM cycle. This will increase your output ripple, and eventually become unstable, so take care there.

\$\endgroup\$
4
  • \$\begingroup\$ Hi Daniel, I have played with C12 and that definitely impacts (reduces) the magnitude of negative ripple on VSYS and increases output ripple. I have tried a combination of lower output capacitance (C12) and a cap across the load and its return point at sense resistor and that helps with output ripple. Unfortunately, I do not have the luxury of another L in the path since the board is small. I was wondering if there is a way to trick the regulator but summing in some current at sense resistor and not gating the regulator's OSC completely? \$\endgroup\$ – ezrcer Jul 15 '14 at 21:35
  • \$\begingroup\$ Yeah, I dunno.. I've seen that done with voltage regulators before but that could be tricky in this case. I'm curious to see what you come up with. I guess just a large value resistor to a microcontroller pin, switched from HI-Z to HI might do the job? \$\endgroup\$ – Daniel Jul 16 '14 at 5:23
  • \$\begingroup\$ I was thinking of using the PWM output of a PIC controller into a RC filter and then feed current into the sense resistor via a series R. Or perhaps the DAC output could be used. (I currently use a PIC Chip to feed the DIM pin of the regulator). Then I would tie the DIM pin HI and control things via the summing point. This way the regulator is not shutting down and again up. At least that is the concept :) \$\endgroup\$ – ezrcer Jul 16 '14 at 20:00
  • \$\begingroup\$ It sounds like you're getting pretty hacky with this chip... if possible, you could flip over to a Linear Technology LED driver. Many of them have both analog and digital dim pins, so you have your pick. Up to you to keep trying to make what you have work or pick up and try again heh. Good luck \$\endgroup\$ – Daniel Jul 17 '14 at 21:17

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.