3

I am trying to design a working principle scheme to step-down the input voltage from 24Vdc ---> different low voltages (1.8 V, 1 V, 1.5 V, 3.3 V with ripples ~ 1%).

Could you kindly explain me the the pro and cons of considering a single step-down converter (replicated for all the requested lines)?

...or differently, you suggest me to implement two different steps (e.g. 24V --> 5V and then 5V ---> other lines?)?

What are the pro and cons of this latter solution?

jusaca
  • 8,749
  • 4
  • 36
  • 58
g.mezzina23
  • 51
  • 5
  • Are you trying to create 24->1.8 and another 24->1.5 etc. or are all voltages going to be in one system? – Aaron Nov 18 '19 at 21:34
  • Hi @Aaron, the whole system must be used to provide supplies to a Zynq7000 microprocessor. This device asks for several power supply lines (in a certain chronological order). I was thinking that could be better passes from 24->5 by a single switcher and then, from 5 to all the voltage lines as suggested by TimWescott in the following answers. – g.mezzina23 Nov 19 '19 at 09:32

2 Answers2

4

As the step-down (or step-up, for a boost converter) ratio gets higher, the duty cycle of the switcher gets less. This means more switching losses, higher current than necessary in the switch, and general Bad Things all around.

The rule of thumb that was passed to me is that you don't want to exceed a 10:1 ratio before you either use a flyback converter or two stages of conversion.

A flyback requires a custom transformer, so you probably don't want that unless either your power levels are high enough or your manufacturing quantities are large enough to warrant that transformer. Even with a flyback, unless you're converting a lot of power you may just want to drop to 3.3V and derive the other voltages from that using ordinary buck converters.

For small volume production and small power running two stages of buck converter is probably best; you'll want to do some design studies to see if 5V is the best intermediate rail, though. (For that matter, you may want to do some design studies to verify that you don't want a flyback).

TimWescott
  • 44,867
  • 1
  • 41
  • 104
  • 2
    In addition, using an initial switching regulator followed by linear regulators for the final minor voltage drops improves the ripple voltage a lot without requiring the linear regulators to burn off a great deal of power. – Anders Petersson Nov 18 '19 at 16:44
  • @AndersPetersson: that's a good point. Depending on what tradeoff you wanted to make on power vs. design ease and noise it's a viable option. If I had a board that used a lot of 3.3V and not much of the lower voltages, I might just use linear regulators for those, with switcher(s) to 3.3V. – TimWescott Nov 18 '19 at 16:51
  • Ultimately, you are suggesting me that I should implement: switcher 24--->5 LDO 5 -> V<3.3 Switcher 5->3.3. Am I understanding? – g.mezzina23 Nov 18 '19 at 17:08
  • First, I don't understand your synopsis, and second, I'm suggesting that you **do some design studies**. 24 to 3.3 is a bit under a 10:1 ratio, so it's actually not a bad *starting point*, but it may turn out that 24 to 9 to 3.3 (with steps of around 2.7:1 each time) is better **you need to check**. If the bulk of your power is consumed at 3.3V, then going with LDOs from there makes sense, but if the bulk of your power is consumed at 1V, you want another switcher. **You need to check**. We can't think for you -- we can only give advise. – TimWescott Nov 18 '19 at 17:18
  • Of course. I'm simulating some solutions according to your advices ! In this respect, I would like to thank you ! – g.mezzina23 Nov 18 '19 at 18:00
  • A possible advantage of dropping to 5V first is that the device could be powered by 5V if needed. Also many low voltage regulators have a maximum input voltage not much higher than 5V. – Bruce Abbott Nov 18 '19 at 18:24
0

The drawback of having the regulators ‘chained’ like this: 24 -> 5 -> 3.3 and so on, is that the first regulator will have to carry all the current for all the supplies.

Jakob Halskov
  • 1,542
  • 12
  • 18
  • Yes, ofc. But, also ensuring it, there are some cons in using the step 24->1V rather than a double step? – g.mezzina23 Nov 18 '19 at 16:30
  • A 24:1 ratio with a typical buck topology would be very inefficient, because the duty cycle would be so small. A flyback would be more efficient, but would require a transformer that would almost certainly be custom, with all the hassles inherent in custom magnetics. – TimWescott Nov 18 '19 at 19:15