# 15V to 3.3V to power ESP12E

I need help designing a power conversion circuit from 15V DC to 3.3V DC to power an ESP12E module (which can handle max 350mA, but actual usage will be 80mA).

I have tested a prototype with Wemos D1 Mini powered by 5V via AMS1117-5 voltage regulator (like this one) from 15V. It worked quite well, but the regulator running is at about 30℃ (with ambient temperature of about 22℃).

In the final design instead of Wemos D1 Mini I want to use ESP12E which uses 3.3V (with the PCB ordered from JLCPCB with SMD components assembled).

I was thinking of adding AMS1117-3.3 to step down 5V to 3.3V.

Are there better alternatives for this situation?

UPDATE 1: The source voltage is actually 15V and not 18V as originally stated. So AMS1117 coped because it's maximum allowed input voltage is 15V.

UPDATE 2: I ended up searching for 3.3V voltage regulators on JPCPCB, going through datasheets and found a few options. The short list is:

Will these do the job? Any advantages of one over the other?

UPDATE 3: I've checked the actual current draw of my prototype and it's only 80mA.

• I am using cheapy LM2956 and also AMS117-5, AMS117-3.3 modules. Commented Jul 29, 2020 at 10:00
• Why don't you use a voltage regulator? Commented Jul 29, 2020 at 10:36
• @HelenaWells what do you think a AMS1117 is? Commented Jul 30, 2020 at 5:58

The AMS1117 is not designed for 15V input voltage. 15V is the Absolute Max voltage, while the accepted voltage range is 1.5V to 12V. Anything between the operating voltage to the absolute max is a danger zone, and anything at or above the absolute max, you are basically okay with blowing the IC up. If your voltage goes up by even a tiny bit, bad things will happen. See https://www.silabs.com/documents/public/application-notes/an1225-isodriver-amr-eos.pdf for why absolute max is not what you should be using.

Even at 15V, that's almost the limit for the output line regulation of the AMS1117-3.3 (if powered by 15V). See the data sheet, test condition for line regulation is 1.5 <= (VIN - VOUT) <= 12V.

That said, yes, you can run two voltage regulators in series. Using 15V in to AMS1117-5.0 to a AMS1117-3.3, ignoring the problem listed above, would be okay, but you are not reducing the heat problem. Linear regulators like the AMS1117 will turn regulated voltage into heat. First the AMS1117-5.0 will take (15V - 5V) * 0.35A = 3.5 Watts into heat, and the AMS1117-3.3 will take (5V - 3.3V) * 0.35A = 0.6 Watts into heat. And that's assuming it's exactly 350mA. The second one is fine, but the first one will need proper heat sinking, either on pcb copper traces, or a heatsink added to the top.

For an input voltage * current this large compared to the output, you want a switching regulator. They are much more efficient, 80 to 95%. Even at 80%, 3.3V * .35A = 1.15 Watts output means (1.15W * 120%) = 1.38W, 1.38W / 15V = 0.92A input, AND it only dissipates 1.38W - 1.15W = 0.23 Watts. You won't even feel it get warm.

There are a number of switching regulator modules you can use for cheap, or get the ics and make our own board. Some even come in simple 3 pin form factors with minimal external parts needed. Or if you just need a one off and don't mind hacking, cheap dollar store car USB adapters (the small inside the cigarette outlet ones, not the older fat ones) tend to have adjustable switching regulators you can change the resistors on to get 3.3V, and tend to accept up to 24V. (See https://hackaday.io/project/6200-quick-and-dirty-33v-power-from-5v-car-charger)

• Thanks for you valuable input. I have found some switching buck converters on Aliexpress which I can use. I have also checked the actual current drawn by my prototype device and it's only 80mA, so I assume using a voltage regulator to convert 15V to 3.3V (e.g. datasheet.lcsc.com/szlcsc/…) will be not too bad, since the heat loss will be about 0.94W. Which I assume is fine and won't need a heat sink?
– Muxa
Commented Jul 30, 2020 at 9:13
• BTW, 1W ends up costing approximately \$2.5 for electricity per year where I live.
– Muxa
Commented Jul 30, 2020 at 9:15
• @mux that ldo will work at 15V in, but the dissipation is still the same. Based in the junction to ambient without heatsink of 125 C/W for the medium sized sot223 you are running it basically at absolute max. And that's if ambient is 25 c or 77 f, open air with a breeze the regulator is at 150 C!. The larger to 252 pacakage would be fine at 100 C/W but it will be still be 125 c 257 f degrees at the case. Too hot to touch. If you add the required copper of 1 cm square to the pcb it will be 70 C/W to ambient but still much cooler. Add little fin heatsink on top and your be fine. Commented Jul 31, 2020 at 23:28
• @mux but again go with a switching regulator module. For the 2.50 youd spend on power, you can get one and forget about wasting power or heat issues. Commented Jul 31, 2020 at 23:29

There should be ample variants of step-down converters to meet the 18V to 3.3V step-down you require. Depends what you're looking for, like a complete eval board, or a particular device to do the job. Look at the LMR36006 part that will do this, but generally doing a search on most manufacturer websites like that of TI or Maxim etc. will get you many options, and mostly all of them will have ready-to-use eval boards you can purchase on-line.

Hope this helps out your search ... cheers, C

• Thanks for you reply. I was looking for an SMD part. I ended up searching for 3.3V voltage regulators on JPCPCB, going through datasheets and found a few options. The short list is: datasheet.lcsc.com/szlcsc/… datasheet.lcsc.com/szlcsc/…
– Muxa
Commented Jul 30, 2020 at 8:21