Evolution of MI power supply designs from older Elements to more recent Warps
  • Hi pichenettes,

    I’ve been cribbing extensively from your designs these past few months, mostly using Elements as a primary reference as it maps well onto the sort of functions I need for my design. Today I was looking at your more recent Warps design, and I noticed some changes.

    Rather than blindly crib from your design, thought I’d ask you about those changes and hopefully can some greater insight into these circuits.

    What motivated these changes (component numbers mentioned below are from the current Warps schematic)?

    - Addition of ferrite beads between VCC, D1 and VEE, D2. I understand the purpose of a ferrite bead, but curious why you decided to add them now – Repositioning of ferrite bead in the 3.3v supply circuit to output of the LM1117-3.3 – Addition of R1 adjacent to VCC in 3.3v supply circuit – increase 22u panasonic electrolytic caps to 47u (both at C1, C3 positions as well C2 and C4 positions)

    Frankly, I don’t have a super deep understanding of the prior Elements design either (other than the basic functions of each circuit, motivation for the diodes), so any general comments about the new and old designs would also be appreciated :). Here are a few questions I have some guesses/vague understanding about, but would love to hear the reasons directly from you:

    - Why the use of panasonic electrolytic caps and not 0603 like elsewhere in your supply circuits? – Beyond just supply circuits, it seems that you have a preference for being liberal with decoupling capacitors – How critical are the ferrite beads? I’ve seen other designs that lack these.

    Thank you!

    • Ferrite beads: they prevent HF noise (generated by the module) from contaminating the supply rails – and being rectified by another module with poor power supply noise rejection (eg: discrete analog module using the rails as reference voltages) – I haven’t met a situation in which this actually happened, but it’s cheap and it’s a “defensive” move against dreaded noise problems.
    • R1: additional PSU HF filtering (at the cost of making the voltage entering the LM1117 vary with module current consumption). That’s the most debatable choice.
    • C2/C4: Before working on Elements I did a few modules in which I couldn’t accommodate bulky 47uF caps (Tides, Streams), so I used smaller 22uF caps. Trying to use as many common parts between modules is a good thing (the manufacturer can keep a large stock of them bought in advance), so Elements used 22uF too and I planned to get rid of the 47uF value across all modules. This didn’t happen and the next modules used the larger value for which there was enough room on the PCB.
    • Aluminium capacitor. Is there such a thing as a 0603 capacitors providing 22uF or 47uF of capacitance at 11.5V DC? To give you an idea, Murata GRM31CR61E226ME15L is already 1206, and provides only 6uF of capacitance at 11.5V. Check Murata’s capacitor specs tool. You’ll see how the capacitance drops when the capacitor handles a constant DC voltage.
    • Those 100nF decoupling capacitors contribute to about 30 cts in the cost of the module. There’s no reason trying to cut cost on those, even if half of them could probably be removed without any impact on performance.
  • Thanks pichenettes! Helpful comments!