# Bipolar energy harvesting circuit

I'm planning an energy harvesting circuit using LTC3105 (400mA Step-Up DC/DC Converter with Maximum Power Point Control and 250mV Start-Up).

The problem is that my energy source is bipolar, although polarity switch is a matter of seconds (very low frequency). I very much like the 250 mV startup voltage of this chip and such low voltage is required by my application.

The obvious solution would be to use bridge rectifier, but it introduces voltage drop that is significant compared to 250 mV startup voltage. Is it possible to solve it in another way? It is allowed to use two LTC3105 chips.

EDIT: LTC3105 supports only positive voltage with regard to GND. The problem is that I'm using small motor with gearbox as a dynamo which will be turned by hand in both directions. If connected directly and turned in one direction it will produce positive voltage, if turned in opposite direction it will produce negative voltage. The slower you turn the motor the lower voltage you get and the goal is to produce energy from smallest turn speed as possible.

So maybe instead of rectifying the motor input and losing some voltage, there is a method to route each motor polarity to separate LTC3105.

• Well, since you're hinting in that general detection: Does the LTC3105 tolerate inversed input voltage, or might you be damaging something? What is the circuit around the IC you're planning so far? What are the relevant in- and output voltage ranges, and which currents are we talking about?? Jan 27, 2017 at 18:52
• 'The problem is that my energy source is bipolar' - the problem is we haven't go a clue what the the energy source is because you haven't told us. Jan 27, 2017 at 18:53
• @JImDearden I clarified the question
– PSz
Jan 28, 2017 at 13:26
• If it's bipolar, what's the frequency? Can you use a transformer to get higher voltage? What's your source impedance? Jan 28, 2017 at 13:32
• @winny Ok, my fault. I didn't explain this well enough, but now it should be clear. I doubt that transformer will help.
– PSz
Jan 28, 2017 at 13:41

I found an article about a "Zero Voltage Diode", which is actually using a 2N7000 mosfet to act much like a very-low-drop diode. Several of the links are dead, but archive.org has copies.

Searching on a few parts retailers, I found the TI SM74611 Smart Bypass Diode with a forward voltage drop of 26mv at 8A, which is not insignificant compared to the 250mv startup voltage, but is still extremely tiny for a diode.

I know that I've seen somewhere information about a type of diode that conducts with 0v bias in one direction (reverse, iirc) and blocks up to a small voltage (maybe 500mv to 1v) in the other, but I can't remember what that type was called or where I read about it.

• "Ideal Diode" is what you're likely thinking of. Jan 27, 2017 at 19:31
• No, it was an actual physical device. It might have been an academic paper about an only-in-the-lab situation, though. I think it was something like a zener diode with (near) 0v breakdown backwards and normal drop forwards. Jan 27, 2017 at 19:42
• A 2N3904 is not a MOSFET. Jan 28, 2017 at 13:34
• @OlinLathrop Thanks for pointing out my carelessness​. Jan 28, 2017 at 13:52
• This zero voltage diode circuit cannot work if the input voltage never rises above the MOSFET gate threshold voltage. If the source always stays below 1~2V, it will never start up. Also, to act as a bridge and do a full rectification, you need four times this circuit.
– dim
Nov 30, 2017 at 14:17