Notes on Designing for Solar Energy Harvesting

by JP Norair on 04 Jan 2015

in Ingress

While designing the HayTag across a period of two years, I found myself spending a lot of time not only in design and test, but also consulting with the solar panel manufacturers, battery manufacturers, and semiconductor manufacturers that make the charging ICs.  I want to write down some of the lessons I learned.

I’m still in the business of developing ultra-compact, solar-powered electronics, so I’m not going to go beyond the proverbial “tip of the iceberg,” but if you’re starting from zero even the tip of the iceberg could be a big help.  What follows are some recommendations based on some stories about my own successes and failures building the solar charging part of HayTag.

1. Greatly Over-spec the Solar Panel

Determine the power requirements of your device. Then look at the solar panel datasheet and determine how much energy the thing can supply. If your device is intended to be an outdoor device, then you can use a monocrystalline solar cell, and it will work nicely. If your device is an indoor device or an indoor/outdoor device, you might as well use an amorphous solar cell, since the indoor regime will dominate the charging behavior. Once you’ve done all the datasheet researching, multiply your energy requirement by at least 4, and then go back through the datasheets to find a suitable solar panel.

HayTag Solar Panel Here is the monocrystalline panel used for HayTag. It is one of the smallest available. In the future, I want to replace it with a larger, amorphous photovoltaic film.

2. Also Over-spec the Battery

Li-Poly is a great chemistry for small, solar powered devices because it can reliably source a lot of current respective to its energy capacity (and physical size).  If you are using some form of wireless/radio in the device, the radio will create relatively large peak currents with steep transients.  Li-Poly can deal with this use-case happily.  Rechargable Lithium coin cells are another popular option, but the chemistry they use is much worse for sourcing currents above 10 mA, so if in doubt use Li-Poly.  You can check for small li-poly packs.  I would spec a battery that has at least 2 times the energy rating (mAh) than what you think you need, because otherwise you need to really carefully consider the effects age, temperature, and peak current loading have on the output voltage of the battery.

Battery v. Quarter Here is one of the 10mAh Li-Poly batteries used in HayTag (very tiny!). It can source 20mA without much voltage drop.

3. Consider your options for charging ICs

The good thing is that several major semiconductor companies are serious about the market for energy harvesting in low-power electronics. So, there are a bunch of monolithic ASICs that provide the bridge between the solar and battery. Some also have integrated voltage regulation on the output side. I will briefly summarize a few of the devices I’m familiar with and have tested.

$7.87 @ 1ku
$2.10 @ 1ku
$3.20 @ 1ku
$1.15 @ 1ku
DFN 3×3 mm
QFN 3×3 mm
QFN 3×3 mm
QFN 3×3 mm
Die 1.2×1.2 mm
1x diode
1x inductor
1x resitor
3x capacitor
1x inductor
9x resistor
4x capacitor
2x inductor
7x resistor
5x capacitor
1x inductor
6x resistor
5x capacitor
Internal DC-DC
Internal LDO
(75 mA)
1x 1.8V (100 mA)
1x 3.3V (100 mA)
Min Solar Voltage
220 mV
100 mV
180 mV
150 mV
Max Charge Cur.
100 mA
110 mA
110 mA
70 mA
Charge Efficiency


Maxim MAX17710
Pros: integrated LDO with 3.3V/2.3V/1.8V output select , low start-up voltage, best-in-class quiescent current
Cons: Very High price, Requires a lot of external components

TI BQ25504
Right now, this is the most mature energy harvesting ASIC, and the one that all others are compared against.
Pros: lots of design support, low cold-start voltage, low quiescent current, highly configurable via resistor bridges.
Cons: no integrated output regulator

TI BQ25570
This is like the BQ25504, except that it also integrates a micropower buck converter on the output side. It is the most advanced and most efficient energy harvesting ASIC on the market today.
Pros: low cold-start voltage, low quiescent current, buck dc-dc converter on output makes it extremely efficent.
Cons: High price

ST SPV1050
It’s a new contender, and it is designed to be an extremely compact solution.
Pros: Low price, Low amount of external components, available in bare die or QFN, integrated dual LDOs (1.8V + 3.3V).
Cons: Not mature, cold-start voltage is slightly worse than the others

Note on “Cold Start” Voltage: this is the input voltage required to startup the device for the first time.  For example, if the battery is dead and the device is in the dark, the solar panel needs to reach the cold-start voltage before charging can begin to occur again.  In all these devices, the cold-start voltage is close enough (330-500 mV) that selection of the solar panel will have more impact on cold-start lighting conditions than the charging IC will have.

Final advice: test the SPV1050 and make sure it works for your app – energy harvesting is a finicky thing. If it doesn’t, try the TI solutions, which have more history and thus more design support. SPV1050 has a great price and a small total solution size, which means you can use more of your devices size for solar paneling.

4. If There is a Radio in Your Design, Design for Low Noise

The energy harvesting IC will kick-out a lot of noise if you are not extremely careful about component selection, placement, and shielding. In HayTag, I was able to get the RF noise floor from -90 dBM (crap) to -116 dBm (pretty good) by improving the components and placement. Having a poor noise floor affects the ability of the device to receive data via radio. Shielding can help, too, but for cost reasons it is not in HayTag.

5. Exhaustively Test Solar Charging

Especially, test the bejeezus out of the solar panel itself. We had a horribly bad time with the particular cell in the HayTag (I will call it the XXXX solar cell). The manufacturing yields are poor, and XXXX isn’t honest about it. They aren’t the only ones: most solar panels are designed with extremely specific use-cases in mind, even though the marketing materials hail them as great, general-purpose devices. XXXX has now solved the yield issue, but if we had not done extensive testing — and alerted them to it — surely they wouldn’t have fixed anything. So, regardless of who supplies your panel and how the datasheet is written, make sure to do lots of testing in different lighting conditions to validate that: (A) the panel will work adequately for your application, (B) the yields are good.

{ 1 trackback }

Previous post:

Next post: