# 3.3V @ 10mA from 5V - 35V

Any ideas on how to get 3.3V from 5V - 35V at a low 10mA current level? 3.3V need not be precise. It can be 3.5V or 3V, it is powering a small microcontroller.

I'm looking for a small and cost effective solution. That pretty much rules out through hole stuff.

Use a linear regulator that can handle 5V to 35V input. It outputs 3.3V.

It is that simple. You can use a buck regulator, if efficiency is an issue. but 10mA at a 35V drop is still only a 350mW dissipation.

I suggest the linear, keep components/complexity/noise down.

• efficiency gain from a buck regulator at only 10mA would be questionable as well, especially at the 35V point, the duty cycle would be really low. – Mark Nov 10 '10 at 16:50
• 350mW is a lot for small surface mount components. – Thomas O Nov 10 '10 at 16:51
• 350mW absolutely worse case is not that bad. that is if you are at 38V in. If you get a good ground plane to conduct heat, or use through hole with a heat sink(which I can put watts into without a problem) you can handle this. The question of heat dissipation we have had other places. This will easily be handled with small surface-mount linear regulator(I have done this much power that way). – Kortuk Nov 10 '10 at 17:04
• Yeah, but I can't spare space for through hole. It's also a waste of space in that the micro is in a tiny SO20 package but the reg is in a TO-220... How would zener diodes work? they have higher dissipation ratings, maybe it would work better? – Thomas O Nov 10 '10 at 17:07
• something like a linear.com/pc/productDetail.jsp?navId=H0,C1,C1040,P86928 in a DFN package with the thermal pad tied to the ground plane with 3-4 vias would work fine. Based on the data sheet you would need something like 25cm^2 of copper to allow operation up to ~90C ambient. Which if you have a ground plane in the board and the board is bigger than 5cm x 5cm should be fine. If you can do a ground pour on the surface layer around the IC, even better. – Mark Nov 10 '10 at 17:27

You'd struggle to find a buck reg with 35v input range that didn't draw 10mA quiescent before you got any power out. 35V is a bit on the high side for many linear regs so you need to look carefully at specs.

• tcrosley disproved you without much effort :P. LTC3631 has a maximum Iq of 0.22 mA in active state. Iq is usually proportional to the full-load capability, so while you're probably right for a 5A supply, smaller ones exist. – Nick T Nov 10 '10 at 18:59
• The OP said "Cost effective" - LTC3631 costs maybe 3-4x what the "small micro" is likely to. – mikeselectricstuff Nov 10 '10 at 19:07
• Micro is a PIC16F690, about \$1.80 in quantity. – Thomas O Nov 11 '10 at 13:41

This LTC3631 handles 4.5 to 45V in, 3.3v out, and comes in either a 3x3 mm or MSOP-8 package. 81% efficiency at 36v in and 10 ma out (page 4). You should be able to get a suitable inductor in an 0805 (2 x 1.3 mm) or smaller package.

The MC34063 might also work but it is a larger package (SOIC-8), uses more quiescent current (2.5 ma) and would require more external components. But it is 1/4 the cost of the LTC3631.

At such a low current level I would be inclined to pre-regulate down with say a 12v zener and stick a cheap-as-chips LDO on there

• At such a low current it wouldn't be entirely out of the question to use a voltage divider instead of an LDO if they wanted to be really cheap, but yes, I think an LDO or standard 3.3 volt regulator would be better in terms of voltage stability. – QuickishFM Oct 3 at 18:24