Soalr SMPS Part 1: Overview
No. There's no typo in the title. I named project with a mistake and it stayed like that.
Background
My family likes outside lighting at night. And I hate light pollution ... and wasted power. So I convinced them to use solar powered lamps. They come in all sorts of shapes and sizes but fall into one of 2 main categories:
- Too small to light anything up. You can see LED directly but not much else is lit up
- the useful ones. They're big, bulky and most importantly, they're expensive, yet still don't deliver in my opinion.
I thought it's simply a matter of how bad solar is at powering stuff. But I did some calculations, run some experiments and arrived at a simple fact. These devices must have terrible efficiency. Surely, it's because of trees, a bad angle, etc. It has to be, right? They're well designed! They have to, these are commercial products!
Teardown
As a matter of fact I opened one of the commercial units. Unfortunately, I don't have any photos from a teardown (it was long time ago). And inside, I see 2 linear converters. Dropping around 6V of a solar panel (maximum power point) to 3-4.2V of li-ion cell and then to ~2.8V of LED. That gives overall system efficiency of 47%, not accounting for non-optimal operating point of solar panel (lack of MPPT tracking).
If we could increase its efficiency to 90% light up time would double and with proper light beam shaping we could light only what's needed without polluting sky too much.
A point needs to be given to designers though. The design is dead simple. Charging "circuit" is just series diode that ensures the panel doesn't discharge battery at night. Discharge circuit requires few more parts but still simple ones. LED is current limited by a resistor and turned on-off by a MOSFET. This MOSFET's gate is controlled by the inverted voltage on the solar panel. Simple and effective.
Resistor current limiting circuitry also shows one more downside of these lights. They dim when they're discharging. This stretches their effective on time but decreases their usefulness as they need to be oversized or you can use only small part of their power. This is especially true with white power LEDs which have forward voltage in a range of 2.8-3V. If we consider lamp "off" when light flux density drops to just 10%, we get another drop of efficiency as li-ion doesn't work anymore on it's full 3-4.2V range, but only on 3.15-4.2V. Roughly 2.5% capacity loss.
What's soalr SMPS?
It's new a circuitry for solar powered lights. There's just one goal. Highest possible efficiency while keeping reasonable cost and board size. How can we achieve that?
- We could better match voltages of all components. If the solar cell had 5V MPP, the efficiency would increase to 56%. Diode's forward voltage has to be below min battery voltage and panel's above battery max voltage. Assuming the best case we get efficiency of 71%. Much better then original design but that's only a theoretical value.
- We could push previous method even further by decreasing the working range of the battery. But this drives the cost and size up, and component matching is a nightmare. Decreasing working voltage to 3.5-4V, drops capacitance to just 33%
- Just for bonus points, we could swap output resistor for a linear current source, but low dropout current source is not a simple analog circuit anyway.
- Instead of using linear converters we could switch to a switching mode models. Theoretical efficiency is only limited by li-ion's cell efficiency which can be over 99%
 |
| system diagram |
I sketched a few features I want for a circuit to have:
- high efficiency (ofc)
- can drive 1W and 3W LEDs
- single cell li-ion as power storage
- 6-12V input range (min)
- compact design
- built-in battery compartment
- battery overvoltage, undervoltage protection
Comments
Post a Comment