I'm really terrible at circuit design, but I want to incorporate a dual output DC-DC buck converter on a board.
I am using a 36v/10A power supply, and since the two outputs would have to power an ESP32 and a Nextion screen, I think 5v/1.5A would be safe (ESP can drop 5v to 3.3v).
surepic:
If input is 36v 10A and output 5v 1.5A then why not to use ldo regulator?
Because of something called dropout voltage. And, even with an "LDO", the actual dropout voltage will be quite high, in this case. See, "LDO" is merely the minimum possible dropout voltage. If the input voltage to output voltage ratio is higher than the LDO voltage, then the dropout voltage will be higher than the LDO voltage.
For instance: Let's say we use a 5V LDO Regulator IC with a rated minimum dropout voltage of 300mV, but, we put a 12V voltage on it's input. The actual dropout voltage will be: 12V - 5V = 7V! So, what's the point of an LDO? -- you might ask. Here's another example that should clear that right up:
Suppose we want to run a device at 5V, using 4 AA cells. And we want to squeeze as much juice [thank you Ben] out of those cells as possible, like still get 5V out of our regulator even when the cells are down to 1.3V...well -- lets do the math!
1.3V * 4 = 5.2V
So, if we use an LDO regulator with a 200mV minimum dropout voltage, or better [like an LD2981CU50TR], then you can juice that battery to the max [nearly]!
So, in the OP's case, the input is 36V, and the desired output is 5V. That's a dropout voltage of 31V! At a current of 1.5A, that's 31V*1.5A = 47Watts!! Pretty much, that LDO Regulator IC will quickly be transformed into smoke!!
As for the OP's question -- there are others on this forum, far more qualified to answer it.
Start with the Ti datasheet. They have reference designs and even have online configurators. Switching supply’s are very challenging due to high frequency EMF/RF. It may take multiple tries to get it right. Good luck and have fun!
Mic293xx series ldo regulators are capable of working in that extremes +26v input voltage continuous and +60v with less than 100ms. They require heatsink for sure but its still possible to do without designing buck converter from scratch. In that case only 30v-4v must be dealt with to provide +26v to ldo. zener+npn bjt or some sort of simple voltage dropper.
With buck converter ics that i tried i wasnt able to get as clean no ripple supply as with ldos.
wolframore:
Start with the Ti datasheet. They have reference designs and even have online configurators. Switching supply’s are very challenging due to high frequency EMF/RF. It may take multiple tries to get it right. Good luck and have fun!
I've found some of TI's schematics, but they are all using the old single output version of the chip - I need one that uses the dual outputs so I have 2 separate 5v/1.5A lines. TI says this is a new chip, and they seem to have the old version's data sheet on there still
Input up to 45v second one and very affordable ics. Micrel`s datasheets are very easy to read and explanations are also very good. Am sure you will not have problems with those.
surepic:
With buck converter ics that i tried i wasnt able to get as clean no ripple supply as with ldos.
Dropping 26V over a linear regulator is crazy.
If you really have to get rid of that last bit of ripple set your buck converter to 6.5-7V, do the last with a linear regulator.
Indeed its crazy i never got into that much of voltage drop in ldos in my circuits( im staying within +1-2v). But tried to oppose ReverseEMF in that ldo will turn into smoke.
surepic:
Mic293xx series ldo regulators are capable of working in that extremes +26v input voltage continuous and +60v with less than 100ms. They require heatsink for sure but its still possible to do without designing buck converter from scratch. In that case only 30v-4v must be dealt with to provide +26v to ldo. zener+npn bjt or some sort of simple voltage dropper.
You do realize the OP specified a max current demand of 1.5Amp, right? What you are suggesting is a situation where 31V*1.5A = 47Watts!! would need to be dissipated with that heatsink -- water cooled, perhaps?
surepic:
With buck converter ics that i tried i wasnt able to get as clean no ripple supply as with ldos.
A way to deal with this is by placing an LDO Linear Regulator after the Switch Mode regulator, with a dropout voltage one LDO higher than the worst ripple voltage from the Switching Supply. Probably not more than a half volt, or so, of loss. If that makes a difference, then adjust the Switcher's output around a half volt higher, to compensate, for that loss.
@ReverseEMF suggestion using buck converter then ldo seems interesting! Will try that method for sure when will need to drop several volts from supply.
surepic: @ReverseEMF suggestion using buck converter then ldo seems interesting! Will try that method for sure when will need to drop several volts from supply.
Only needed when Switch Mode ripple noise is an issue.
I seem to remember reading a few decades ago that the main reason was the Pi filter of the two caps on the input and output of the linear regulator together with the internal resistance of the regulator. Not the correction speed of the regulator itself.
A simple LC filter could be as effective.
Leo..
Wawa:
I seem to remember reading a few decades ago that the main reason was the Pi filter of the two caps on the input and output of the linear regulator together with the internal resistance of the regulator. Not the correction speed of the regulator itself.
A simple LC filter could be as effective.
Leo..
You're probably right. One thing to consider, though: There appears to be a plethora of "noise characteristics" induced by various factors, making filtering a case specific thing. In other words, one SMPS's poison, may not be another's, which suggests the need to pick-your-antidote.
For instance, switching frequency will make a difference, as well as proficiency of layout [i.e. noise worsened by improper layout], poor component selection, etc.