Boost 1.3V to 13.6V!?

[RogerThat]

Hi,

I have side project which is to charge a 12V AGM battery from three solar cells(3.6W each). The cells will have 1.3V at their MPP and I need to boost that to 13.6V which is the battery's float charging voltage.

I looking for some advice which type of boosting would work best in this case. Switched inductor, charge pump or maybe a combination?

[exmadscientist]

A Very Expensive blue-sky project with an even more harsh requirement uses digitally controlled boost converters with EPC GaN switches. That may or may not be helpful, but, y'know, it's possible.

[mariush]

I've used a LT1307 to boost 0.6v+  to around 6v  ... it starts at 1v

The datasheet says it can do 12v @ 30mA with 3v input so I don't know, maybe if you put those solar cells in series it would work.

[wraper]

You'd better use more of smaller solar cells in series. While you can boost 1.3V to 13.6V, it will be very inefficient.

[RogerThat]

Good I asked the questions here I thought maybe I was missing something.

I've tried a breadboard version of charge pump(1n4148 and 100uF...plenty) and switched inductor. These fast-and-dirty breadboard versions drew more power than they generated...

The three cells will, when connected in series, have a MPP at 1.3V.

My thinking right now is to use a TPS6120x for the first boost. It can only boost to 5.5v so I will need to have a second stage from there up to 13.6V.

[strawberry]

push pull transformer 1:12 selfoscilating
SO-8 dual 0.8V gate threshold MOSFETs

11W 1.3V ~8.5A

[David Hess]

Linear Technology has various "Energy Harvesting" ASICs intended for exactly this sort of application like the LTC3119 buck-boost converter.

[NiHaoMike]

Would be easy to make a boost converter using parts salvaged from a PC motherboard. Use a microcontroller to implement the MPPT.

[aeg]

There will always be at least 9V across the battery terminals, otherwise the battery is too far gone to charge. If you power the PWM controller from the output side, the low voltage coming out of the solar cells is less of a concern.

[EsPiFF]

when you like to avoid exotic parts, I would suggest at least 2 boost stages. As a rule of thumb, dont go for more then 1:4 boost ratio, to keep the efficiency in reasonable areas.

The otherwise suggested solution with a 1:12 transformer push-pull looks elegant, if you can source the part, and it fits also with your other parameter.

The 2 stage boost solution looks cheaper to me, because you can modify on all parameters, like switching frequency, coil size, ect. It would also be easier to manufacture, because of each part there will be a second source.   

[Terry Bites]

Have you considered a "Joule thief".

[Marco]

Usually you'd just use a small flyback/autotransformer, but if you really want to use a single inductor you can get the duty cycle and switch voltage down with a multiplier.

[Coordonnée_chromatique]

https://fr.aliexpress.com/item/32668340182.html

[RogerThat]

At lunch today I looked for 1:12 push-pull transformers....found zero. Maybe they exist but 1h searching rendered nothing useful.
Seams like most push-pull transformers are custom made for certain chips?

With the 1:4 recommendation in mind....use the TPS for 1.3->3.3V and then normal switched boost 3.3V -> 13.6V.

From the TPS datasheet It looks like around 70% efficiency could be had for the first stage @ 200mA. I guess paralleling a few TPSs to achieve higher current. Maybe use three with different UVLO to achieve three stage MPPT.....nah, probably better to regulate in the second stage with an MCU.

[Marco]

I'd get some low voltage boost converter just to get a decent MOSFET drive voltage, but only for the MOSFET and whatever circuitry is driving it. Boost the 1.3V directly go to 13.6V, intermediate steps are just a head ache.

A standard transformer will be hard to find without it being way more chunky than necessary, because of the low primary voltage. So maybe the multiplier solution (say 4x) ain't so bad.

[NiHaoMike]

Don't overthink it, CPU voltage regulators work the other way around (12V to 1.2V or so) with just a plain synchronous buck converter. It goes to reason that such a design would work in reverse just by using a different control circuit. To simplify the work needed, replace the high side MOSFET with a Schottky diode, the loss would be fairly small at a bit more than 12V output.

[rfclown]

Charge pump isn't efficient. Boost is the only way to do it. I've not done the 10x like your application, but I've done 3.7v Lithium battery to 28 V (7.6x) for RF power amplifiers (12W and 50W) with no transformer. Used a Linear Tech part. My application's output duty cycle was low, so the boost was charging a bank of capacitors for a periodic bang.

[Prime]

Build a joule thief and then couple that to a boost converter.

The joule thief is essentially a self oscillating flyback converter.

[Marco]

The primary currents are going to be in the 20A range, that's a large joule thief I guess a simple hard switched boost, just making sure all the components are ultrafast to deal with the extreme duty cycle, has some appeal. NihaoMike is right, motherboards do it, with more power/current than this.

So have a low voltage booster to get a driving voltage and some fast boost converter IC with an external MOSFET.

[wraper]

Build a joule thief and then couple that to a boost converter.

And it will be '"thieving joules", in other words efficiency will suck. If you'll get it working at all due to low gain of BJT at high currents.

[SiliconWizard]

If your input voltage is truly 1.3V *min*, you don't need to resort to "specialty" energy harvesting ICs. There's quite a few boost converters that can fit here. For instance, the MP3429GL-P at a reasonable price. 1V input min. Nice.

I'm assuming you already figured the approximate max. current you'd get out of the boost converter and thus how long it would take for the battery to charge though, taking the efficiency of conversion and the multiplying factor? Assuming a 85% efficiency, you're down to about 9W (for 3x3.6W cells), which is ~0.66A @13.6V. And those 3.6W/cell is probably the most favorable output you'd get out of them.

[RogerThat]

...CPU voltage regulators work the other way around (12V to 1.2V or so) with just a plain synchronous buck converter. It goes to reason that such a design would work in reverse just by using a different control circuit...

No, it doesn't? The inductor in a buck converter is only used as a smoothing filter. You could swap the inductor for a resistor if you wish but the efficiency would not be great. Boost converter uses the inductor's, in lack of better word, inertia to shoot the voltage up. If you swap the inductor for a resistor in a boost converter it would not work. Different things.

The MP3429GL-P looks promising, I need to read some more about it. The battery will always be connected so circuit doesn't need to rely on the solar cells voltage to run(it can be supplied 5v from the battery), it just needs be able to switch it. The circuit will work to just keep the battery topped up(or charge if it runs low) so output should be max 13.6V and current will never be too high. If I can reach 85% efficiency it would be great. The cell's power output is much higher than what is needed, if I can get around 200mA charge current(at max output) that is enough.

[bidrohini]

You can try Pololu Adjustable Boost Regulator 4-25V
 https://www.pololu.com/product/799

[Marco]

No, it doesn't? The inductor in a buck converter is only used as a smoothing filter.

It's still an extreme duty cycle making life hard for the switches.

The inductor has to deal with far faster current changes for boost, but is that really the limiting factor here? A uH range inductor has double digit MHz srf.