3.3V simple vs zener

Hi, I need to drop the Arduino 5V to 3.3 to power a radio module and I need to do it in a power efficient way as the project is battery powered.

The radio module has a 3.0 to 3.6 supply voltage range and can absorb up to 150mA.

Using a 3.3V zener diode with a 10 Ohm resistor I get an almost precise 3.3V output, but I will be consuming 150mA regardless of the radio status (on/off/power down): am I correct?.

Using two normal diodes in series I will obtain roughly 3.6V (5 - 0.7 - 0.7) but with almost no consumption when the radio is turned off: is this correct? If I want to be more precise I can add a germanium diode to further lower the voltage to about 3.3V.

If my assumptions are right then zener diodes should never be used in battery powered projects .

Or you can use them but adding an NPN transistor between the zener and ground so that you can interrupt the ground connection when power is not needed: tricky and at an higher cost than using the series diodes.

Please correct my thinking...

No corrections needed I think you have it all worked out.
Zenners make poor voltage regulators at the best of times. Now with low dropout regulators there is no excuse for using a zenner like this.
A switching regulator would be even more efficient.

Rather than using 3 diodes in series, you could instead use a single zener rated at 1.7 volts. Simply place the zener in series with your 5 volt supply and you will "lose" 1.7 volts across the zener.

Can you convert the entire UNO to 3.3V or use a 3.3V derivative?
I have a 3.3V UNO converted from 5V using this tutorial.

An MCP1702 would be a better choice.. Data sheet attached..


MCP1702.pdf (428 KB)

Good to know I'm starting to get some stuff right XD

Looking at DigiKey a low price 3.3V regulator would be the LM1117MP-3.3 (0.91 a piece) while a normal junction diode would be the 1N4007 (0.09 a piece, one tenth!).

Why should I pay the difference? Voltage stability? If the voltage is already regulated at the source then I suppose the output will not be very rippled and a couple of capacitors would do the additional smoothing.

Power efficiency? The LM1117-3.3 datasheet reports a 10mA quiescent current (actually not that good at all IMHO) and the 1N4007 reports the exact same value or better as leakeage current (reported as 0.01A :D). And I picked just one of the many rectifier diodes out there, I bet there are ones with better figures than my very first pick...

In other words, apart from space consumption (which is not crucial to me at this stage) I see no counter indication for the rectifier diodes: they seem a cost and power effective solution.

I'm suprised as it looks a too simple and cheap solution...

Can you convert the entire UNO to 3.3V or use a 3.3V derivative?
I have a 3.3V UNO converted from 5V using this tutorial.

That's not viable as I have mixed power lines: 5V and 3.3V.

Using serial diodes is not regulation at all.

Lets say you use two 1N4007 at 25°C :

  • when the powered device consumes 150 mA, the voltage drop is 2×0.8V, and the output voltage is 3.4V ;
  • when it consumes 10 mA, the voltage drop is 2×0.7V, and the output voltage is 3.6V ;
  • when it consumes 1 mA, the voltage drop is 2×0.6V, and the output voltage is 3.8V ;
  • when it consumes 0.1 mA, the voltage drop is 2×0.5V, and the output voltage is 4.0V ;
  • when it consumes 0.01 mA, the voltage drop is 2×0.4V, and the output voltage is 4.2V ;
  • when it consumes no current, there's no voltage drop, and the output voltage is 5V !!!

Got it! So when the chip is in power down mode (less than 1mA consumed) the voltage will be way above the tolerance: definitely not a condition I want to operate within!

Ok, so the diodes would be an acceptable solution in case I have a stable and fixed current consumption and a decent wide voltage tolerance, not even mentioning operating temperature as an additional factor.

That makes a lot of sense now why the cheapest regulator is 5 times more expensive than the diodes solution.

First of all, that's not the same. Quiescent current is actually a loss of current, but the current thru the diodes is not.

Anyway :

  • a 10 mA quiescent current is better than the more-than-150mA total consumption of the zener solution ;

Correct, but that would not justify the 10 times price of the same "power loss" of the series diodes.


  • the quiescent current of the LP2985-33 I've given is only 65µA (at no load) to 0.85mA (at 150mA load) – moreover, this regulator can switch off the load ;

Yep. The max allowed current is just within limit (I must verify my current consumption) and I don't need the switching capability, but it is definitely cheap! Thanks!


  • and the MCP1702 given by Docedison is even better with 2µA (at no load), but without load switching.

This seems a perfect fit for my needs!

I want to thank you all for the suggestions and for the little lesson: I knew I should have been missing something obvious! On the other end I now know I haven't got much :~