1
\$\begingroup\$

I'd like to save space on a circuit that uses two buck LED drivers. The circuit is designed for a low EMI solution and there are two separate PI filters on each Vin pin (recommanded on the datasheet). I was wondering if it would be possible to use a single PI filter for both buck converters. I should point out that both buck converters operate at the same switching frequency.

Are there any specific risks or performance problems that might result from sharing a PI filter between two buck converters?

The diagram shows one of the two buck converters. All components are the same for both. The PI filter is framed in red

enter image description here

\$\endgroup\$
6
  • 1
    \$\begingroup\$ I should point out that both buck converters operate at the same switching frequency. are they sync'd? \$\endgroup\$
    – Rohat Kılıç
    Commented Apr 8 at 10:35
  • \$\begingroup\$ I don't see why Vin would be that much of a concern, rather than Vout? That's where all the noise from the buck converter is. \$\endgroup\$
    – Lundin
    Commented Apr 8 at 10:40
  • \$\begingroup\$ @RohatKılıç No, the switching frequency is not synchronized. \$\endgroup\$
    – msch
    Commented Apr 8 at 11:25
  • \$\begingroup\$ @Lundin I guess Vin filtering helps stabilize the input power, preventing noise from affecting other devices on the same power line. I don't know if it's really necessary, I'm following the datasheet recommendations to reduce EMI and Vout is also filtered. \$\endgroup\$
    – msch
    Commented Apr 8 at 11:34
  • \$\begingroup\$ Ah you meant Vin on the LED driver. A schematic would mean fewer misunderstandings. That driver seems designed for relatively high voltages though. \$\endgroup\$
    – Lundin
    Commented Apr 8 at 12:29

1 Answer 1

Reset to default
1
\$\begingroup\$

Yes, it is possible, and generally preferred. The filter is linear and obeys superposition: as long as the attenuation is sufficient for the total amount of ripple, you are set.

Mind that this does nothing about common mode noise, which depends critically on layout, and cable length. You may also want other tweaks, such as dampening (an R+C or electrolytic somewhere; C36 already present is promising, though the EEE-HA1J470UP doesn't specify ESR and looks like it may be pretty high), TVS (to clamp inrush overshoot or other transients), polarity protection, or etc., depending on the application.

Note that the local bypasses can also be at least partially shared; how much, depends on the distance between ICs, and whether they are synchronized. Probably, if more than a few cm apart, the full set of capacitors should be provided for each IC; if adjacent, use, say, two 0.47us and two 4u7s each (i.e. add one and one).

As long as I'm mentioning capacitors by value, note that 0.47u doesn't do much here: you get the same ESL from any value of same chip size capacitor. I would use all 4u7s and be done with it; besides, staggered values tend to invite more problems than they solve.

If synchronization is an option, forcing out-of-phase or interleaved operation minimizes input ripple, and a single set of capacitors will suffice for both (again, assuming the ICs are adjacent). Maybe even less, but it's safe to assume given as a minimum (reducing parts count would be more of a production optimization option, I'd say).

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.