Solar Phone Charger - from idea to design

[^_^]

Hey,

so I've had this idea in my mind to build a Solar Charger for my phone, that would not have any Li-Ion etc. as a buffer, but supercaps instead. For reasons, let's not get into why

I've came up with this data:
- 180mm x 80cm solar cell (my requirement is max. A5 size: 148mm x 210mm),
- 3.4 seconds phone charging time,
- 17 seconds idle time (supercaps charging/phone not charging), but I don't mind this time being longer if my assumptions were wrong.

Would it be OK for the phone to be charged with this kind of repetitive cycles?

My design procedure:

3 blocks: solar panel supercap charger, stepup DC/DC converter, controller

Solar panel supercap charger: takes voltage from solar panel as an input and charges the output supercapacitors.
Stepup DC/DC converter: takes supercapacitors as input and outputs USB 5V.
Controller: sets supercap charger registers and turns on/off the stepup DC/DC converter according to supercaps voltage.

Solar Cell (180.00mm x 80.00mm, 8 USD), 2W @MPPT,
https://www.digikey.com/product-detail/en/seeed-technology-co-ltd/313070003/1597-1416-ND/5488052

Capacitor 25F:
http://www.mouser.com/ProductDetail/PowerStor-Eaton/TV1625-3R0256-R/?qs=sGAEpiMZZMsCu9HefNWqpn%2fiL%252bA0OdUhubc6GYJ3YAmwlhG1SiyWEw%3d%3d

Energy in 50F cap:
https://www.wolframalpha.com/input/?i=1%2F2*(2.74V+-+2.0V)%5E2*(50Farad)

Supercap charger (from solar panel, with MPPT):
http://www.mouser.de/ProductDetail/IDT/ZSPM4523AA1W/?qs=sGAEpiMZZMv0keMCjoUGAbilUTDsaT%2fd

Output DC/DC:
something from TI/Linear, TBD.

Assume the phone takes 5V, 600mA (instead of 500mA): 3W.
Assume supercap to USB 5V stepup DC/DC efficiency 75%, that's 4W taken from supercap.

Charging the cap to 2.74V and discharging to 2.0V, at 25F and 4W takes: 1.7 seconds phone charging time.
At 2x 25F caps (50F): 3.4 seconds phone charging time.

Now, how long would it take to charge the cap from 2.0V to 2.74V. Let's omit the first cycle to charge the cap from 0V to 2.0V.
ZSPM4523 does not have any efficiency charts, so let's assume 0.8W of charging power (2W solar panel, but angle to the sun + the charger).
At 50F it's 17 seconds to charge the caps.
That gives (3.4/(17 + 3.4)) = 17% time of phone charging: first 17s cap charge then 3.4s cap discharge (phone charge).

Cheers!

[stmdude]

I've designed a few cellphones in my day, and while you won't _break_ your cellphone (if it's halfway decently designed), I doubt you'd get any charge into it that way.

When you plug in your phone to the charger (start of your 3.4 seconds), the entire system wakes up (uses power), and tries to figure out whats on the other side of the USB cable. This takes "a little while", and the phone won't pull more than 100mA over the USB port until it's done. This uses a bit more power. Also, the screen most likely turns on, which uses _a lot_ of power.

I think you'd end up draining your battery faster than you put power into it, unfortunately.

[^_^]

You're right - this is super annoying. But I consider it a phone problem, there might be a mod/hack to omit that.
So far I've found one for Samsung S5 and it didn't work on my Asus.

Anyway, the Power Supply measurement test was:
- V=4.98V I=0A, phone gets connected
- 4.98V 561mA (short time, phone wakes up: vibrates and screen turns on)
- 4.97V 446mA (phone says charging)

So it's not that bad, though would be better to not wake up.
Anyways, the battery chemistry AFAIK should be good with this.

[DavidMenting]

We weren't allowed to ask why, but wouldn't you be better of taking a 6V solar panel, adding 5.5V overvoltage protection and hooking that up directly to the phone?

[^_^]

Phone will stop charging at unsatisfiable input current (at least mine did, when I limited the current below 0.5A).
So the clouds and the charger movement will disturb everything.

Plus bigger solar panel to have room over 2.5W.

[DaJMasta]

It does sound like that wakeup period is the lost efficiency in the system, though, so even with the same cell, doubling the charge you can store and just making it run half as often should be a substantial increase to charging efficiency (and it sounds like more would be greater as that may only just be break even).

At least with my old Galaxy Nexus, it seems to want about 2A on a fully drained charge, with the requirement dropping to about 1A when it's finished, but it does seem to accept a charge when powered off (currently 500mA constant current limit is reading 4.2V on the bench supply).  While it may be a much less useful device, you could probably charge from a powered down state with a 5V panel and a comparatively small trickle charge of maybe only a few hundred mA.

[^_^]

I've found even better caps (5.5V max, 45F), so more energy stored, but as they are hybrid they have very low charge-discharge cycle limit (50k).
On the other caps (3V) used for calculations it said 500k cycles.

I've calculated before the number of phone charges this device could provide. With the 500k cycles caps it would be "only" 138 full phone charges:
https://www.wolframalpha.com/input/?i=500000%2F((11.5Wh%2F(1%2F2*(2.74V+-+2.0V)%5E2*(50Farad)))+%2B+20%25)

After 500k cycles the capacity can change up to 30%.
I hoped for "ever-lasting" device, that's why I'd settle on supercaps, but maybe it doesn't actually make sense. I didn't find any better caps; they are always cycle-limited.

Another option I had figured could be to regulate the charge current using USB protocol (by descriptor of max current the device can take from host) and skip supercaps.
Though I have no idea if the phone would be happy about that and would it be good for battery chemistry as it would mean way lower than typical charge currents, that would change often.

Anyways, I had replaced the charger IC to STM's because the other one became obsolete...
Schematic attached.

[xani]

At 50k cycles you might just go and use Li-poly cells. Just limit the charge cycle to something like 80%-20% and you should get 1k+ cycles out of it.

I've calculated before the number of phone charges this device could provide. With the 500k cycles caps it would be "only" 138 full phone charges:

That is way less than just putting a fat Li-po that just needs 1 cycle per phone cycle (altho it has a disadvantage of having to charge for a while till it is usable)

[ANTALIFE]

Is the reason for using super caps to have something that will last many charge/discharge cycles? If so why not use a well developed battery technology like NiCd, that's what I used with my solar phone charger here

[^_^]

Is the reason for using super caps to have something that will last many charge/discharge cycles? If so why not use a well developed battery technology like NiCd, that's what I used with my solar phone charger here

Yes, I aimed for something "infinite"/"ideal". Sun = charging, no sun = no charging, and no chemistry involved (ideally).
That "could" be done with just one MPPT DC/DC and a smart charger (MCU), that would communicate available USB current according to momentary solar power.
But, of course, not all devices would care about what the host has to say and assume 0.5A.

I really like your project. You have a holistic approach, man
Are you from Sydney btw?
The only thing that gets me is you didn't use real MPPT (digital core performing the algorithm).
Linear's MPPT just keeps the input at a set voltage.
Why not SPV1040?

[ANTALIFE]

You will never get rid of "chemistry" ;^) your storage will always have some sort of chemical behaviour you need to be aware of.

But, of course, not all devices would care about what the host has to say and assume 0.5A.

Oh wew is that even if you tie D+ & D- together with 200R?

I really like your project. You have a holistic approach, man
Are you from Sydney btw?
The only thing that gets me is you didn't use real MPPT (digital core performing the algorithm).
Linear's MPPT just keeps the input at a set voltage.
Why not SPV1040?

Thanks dude, am from Melbourne at the moment.
Yea I know that LTC3130 does not have proper MPPT but I just wanted to play around with LT chips since they can all be simulated in LTspice.
SPV1040 looks very interesting, will add that to my cool IC's folder, thanks for that.

[^_^]

You will never get rid of "chemistry" ;^) your storage will always have some sort of chemical behaviour you need to be aware of.

Maybe graphene supercaps will

Yea I know that LTC3130 does not have proper MPPT but I just wanted to play around with LT chips since they can all be simulated in LTspice.

I know what you mean. I used to do it often, too. TI chips on the other hand are so bad to simulate. TINA is though webench is great to choose the IC.

Schematic update in attachment.