2 channels buck a boosted 5 to 3.3 ok?

Discussion in 'Analog & Mixed-Signal Design' started by Harley Burton, Feb 10, 2018.

  1. Harley Burton

    Thread Starter New Member

    Jan 1, 2018
    I'm working on a project where my Arduino Nano needs 5V, and I'm using a networking module (ENC28j60) that should run at 3.3V.

    My power source is an 18650 Lithium (3.7V nominal), so I'm currently boosting that to 5V with a boost converter module.

    The Arduino has a 3.3V supply, but it can't deliver the current that the ENC module needs, so I need to get that 5V supply down to 3.3V. My inexperienced knee-jerk solution was a voltage divider, but I've learned that, for obvious efficiency reasons, that is a bad idea for a battery powered project. My next idea is to use another converter, a buck converter this time, to bring the signal down.

    However, some part of me says that, since the boost is basically storing energy and "surging" it out to achieve boost, and the buck is basically going to build a PWM signal and smooth it with capacitance, they may not play well together. I'm not experienced enough to really know how that interaction will play out.

    Any thoughts?

  2. Harley Burton

    Thread Starter New Member

    Jan 1, 2018
    I'll add, for anyone who may find this looking for the same information, in case no more experienced members reply...

    In theory (theory being, in this case, what I think) it should be fine. At the end of the day, a good boost converter is delivering a steady DC signal at a given voltage. That being the case, there shouldn't be any significant interactions with the buck converter on the output of that signal. It's highly likely that this question is a ridiculous one, and this is the way things are done all the time.

    I have tested it, and in my setup (3.7V lithium cell -> Chinese boost converter (dialed to 5V) -> Chinese buck converter (dialed to 3.3v)) all signals look good under a 250mA-300mA load. I see no ripple at all. There may be some harmonics if I had a way to look at that, but I don't expect that would be something to care about.

    After I broke my test setup back down, I thought I should have put a higher load on everything to see if it breaks down, but I suspect as long as the 3.3V rail + the 5V rail doesn't pull more than the rated limit of the Boost converter, it should be fine.
  3. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    If the ENC supply voltage isn't too critical, wouldn't just a simple Schottky diode drop from the 18650 do the job?
  4. ebp

    Well-Known Member

    Feb 8, 2018
    Depending on the current, a simple linear regulator for the 3.3 V supply might be acceptable. A well-designed buck converter using synchronous rectification for very high efficiency would be "better", but with such a low input-output differential one using a conventional diode would not be a great deal more efficient that a linear (a linear will be 3.3/5 or 66% efficient, ignoring power for the reg itself; a buck could exceed 90%).

    You might use a low-dropout linear regulator straight from the lithium battery. There are some that will perform well down to a few tens of millivolts of input-output differential, particularly if the current is low. Pay careful attention to the requirements for capacitors at the input and output of LDO regulators. Some are very fussy. New designs work ok with ceramic capacitors on the output. Older designs had both upper and lower limits for ESR of the output cap.

    A potential (pun intended) problem with running different parts of the circuit from different regulators is power sequencing e.g. the 3.3 V circuit gets power before the 5 V circuit does, causing excessive current in I/O connections between. This isn't likely to be a problem unless you derive the 3.3 V straight from the battery and somehow connect it before you connect the boost converter (the boost convert will take a little time to get the voltage up to 5V, but the output will very rapidly reach battery voltage less a diode drop) or leave it connected when the boost is off.

    A switch mode regulator looks like a negative (in sign of the slope, not absolute value) resistance - if the input voltage goes up, the current goes down. This can sometime cause a regulator supplying the input to it to misbehave, but it is rare. Boost converters have an issue with frequency compensation in the error amp (the dreaded "right half-plane zero" problem), so they are usually made with pretty poor dynamic response. The slow response is to your advantage for what you contemplate. Occasionally you can get some bad behavior due to switching transients in two converters connected to each other.

    Don't use really fine wire for connecting the supplies. Keep connections as short as practical and twist the supply and return conductors together. Twisting the conductors together reduces the inductance and the "loop area" which is a player in radiating RFI.
  5. Harley Burton

    Thread Starter New Member

    Jan 1, 2018
    I'm going to have to look about half of this up, but thank you.

    I can't really simplify the 3.3V directly from the battery. The enc module is very finicky about its power, and the low end of the cell is 2.8V. The options are buck/boost to regulate over the entire discharge cycle, or buck the 5v output that will be boosted over the entire discharge cycle. A simple 3.3V regulator is something that I considered, but I had the same "something in the back of my head" about it as the buck converter. The efficiency difference should be negligible with such a small load.

    The power sequencing problem (from what I just looked up) is only a factor when running, as you say, 2 converters from a single source directly, so it shouldn't be a factor, unless I misunderstand it, when running the buck off of the boost output.