Kindle Charger

[knc1]

Moved from the "Solar powered Kindle Touch" thread.
The only thing "Kindle" about this solar powered, USB charger is the physical size and layout of the pv cells on a Kindle sized bookcover.

* * * *

Put an end to hand drawn circuits and quick-guess design, get a copy of LTspice:
http://www.linear.com/designtools/software/

Don't be put off by it being a Windows program, the authors intend for it to be run under Wine on Linux and <whatever_name> Wine has on MacOSx.

A decent "getting started" set of tutorials can be found here:
http://www.simonbramble.co.uk/lt_spi...e_lt_spice.htm

Edits 1 .. 8 in the evolution of this project are now hidden in the following spoiler. The various links to external information are still present once the spoiler is opened:

Spoiler:

The thread that gave birth to this post used a Li-Ion battery charge controller with direct, internal, connections to the Kindle Touch. Since the Kindles have an internal battery management system, the version here will just be a regualated 5v charger with a USB connector.

One way to make the charger operated from a variable, solar cell source, would be to use an up/down, charge pump, regulated converter with 5v output:
http://www.linear.com/product/LTC1514

One package, 5v @ 50ma output, should match the USB dc input.
As this post grew over time, I have tried thinking out the project using other parts than the one above.
I leave all of the changes in plan here, someone might want to take one of the other branches in planning.

One useful reference on getting power out of a solar cell array:
http://etd.lib.fsu.edu/theses/availa...r_Tracking.pdf
The section(s) on Photovoltic (pv) considerations are well written.
There is a copy attached here also.

Includes SPICE modeling information.
Although the paper uses Pspice, LTspice can do all of the same modeling, from the same Pspice model.
Although a modification of that model might be more usefull to this project.

How big is our ballpark?
Maximum irradiance, equator, solar noon: 1000w/m^2
My Kindle-3 (less the rounded edges): 0.021 m^2
Record pv conversion: 0.43, current, common devices: 17 to 19 precent.
Which comes out to 4.2w per Kindle, under the very best of conditions.
(Think 840ma at 5volt - the Kindle's battery manager's highest rate draws 500ma charging a dead battery.)

A very high efficiency charge-pump converter can reach 90% to 93% conversion rate. (3.78w out, very best of conditions, I.E: 5v@756ma).
That should allow for those who do not live on the equator, or have clouds in their sky and still run the Kindle battery manager on its 500ma curve.

Add in some clouds in the sky, and maybe expect an average of 1/10 the maximum: 0.38w or 76ma at 5volt - - -
Which can still drive the low-rate (20 hour rate) charge curve of the Kindle's battery management.

I have corrected and/or refined the above numbers from my original post. Now bring on the design tools, this could work.

Example PV cells easily available:
http://www.bgmicro.com/PWR1241.aspx
60 mm square, 4.5v open circuit voltage, 90ma short circuit current, already a sealed, "weather proof" construction.
These are thin-film CIDS cells, not the mono-crystaline cells of the thesis paper. Another reason for diddling with the model in that paper.

Edit:
Little bit more "guesstimating"
Two of those cells in series would be 9v, open circuit (no load) under the best possible conditions. The maximum input of the above converter is 10v.

Three pairs in parallel could produce 270ma, short circuit (zero voltage), under the best possible conditions. Maximum output of the above converter is 50ma and the minimum input voltage is 2.7volts.
So the above "first guess" converter part is too small a capacity to make good use of this pv cell array.

Question 1: How bright does the sun have to be to pull 20% of the maximum short circuit current without pulling the array voltage below 2.7volts?

That sort of question is what they make <xx>SPICE programs for.

Question 2: Can you get six of these on a Kindle cover?
See image 1, six of the pv cells laying on top of my K3.
Yeah, that could be made to work if the (book)cover (leaf) was a bit wider than the Kindle.

Question 3: How to get the numbers correct in the model for these unspecified pv cells under normal conditions where I live?
See Image 2, one pv cell and one CdS photo-resistor mounted in a test jig.

Plus, not shown, a busy little micro-controller taking measurements.
Link: http://www.microchip.com/stellent/id...cName=en556208
Haven't decided which of the 4 micro-controllers to use yet, probably the 32-bit, MIPS micro-controller. My data-logger-to-be kit.

The test_jig is a photo frame, with hard covered, foam center, "construction board" or "sign board" or whatever your local name for it is in the local craft shop installed, in place of the usual cardboard.

The hard surfaced layer on one side is cut away from the foam center forming a pv cell fitting pocket. (and a hole for the CdS photoresistor).
Do not put the glass back in the photo frame!
It may look clear to your eyes, but it cuts down the output from these pv cells a drastic amount.

The CdS photoresistor provides a measurement of the incident radiance (as a resistance).
The micro-controller will be programmed to characterize the pv cell at various conditions.
OR - such was the plan when I made the test jig.

Edit 2:
A bit higher parts count, but maybe a better choice converter:
Data Sheet: http://cds.linear.com/docs/Datasheet/3533f.pdf
LTspice demo circuit: http://www.linear.com/docs/26438
(If you have installed LTspice, just click the link, it should open in LTspice)
The demo circuit has the feedback resistors scaled for a 3.3v output but the device can be set for 5v output by changing those values, see datasheet above.

This one should allow running all six pv cells in parallel, capable of powering the battery manager in the 500ma curve mode under best of solar conditions.

Edit 3:
If the six pv cells are run in parallel ...
The maximum oc (open circuit) voltage is less than the desired 5.0 .. 5.1 volt output. A simplier "boost only" converter would do the job.
Since the maximum sc (short circuit) current of the six cells is 540ma, a 0.5amp converter would be undersized. Probably a better choice would be a 1amp converter, running "de-rated" to account for our (expected) poor construction thermal paths.

Back to the datasheet library during KeK typing breaks.

Edit 4:
Presuming 90% conversion and scaling the numbers based on current (documented) CIGS arrays:
It looks like 5v @ 240ma is a reasonable minimum expectation under standard conditions.
It might be almost twice as much, will know more when the LTspice model is done.

Edit 5:
Have been playing with:
http://solarelectricityhandbook.com/...rradiance.html
At my location, at my utility power cost, this panel will save me $0.18/year.
I can save more on my energy cost by paying on-line rather than buying a postage stamp for a single, monthly check.
Will also take about 200 years to break even on this device at those rates.
Will my Kindle last 200 years? Will I last 200 years?
Duh....
Topic does not include the word: "Practical".

Edit 6:
The "boost only" converter for the revised plan of 6 (to 32) paralleled pv cells:
Description: http://www.linear.com/product/LTC3421
Data Sheet: http://cds.linear.com/docs/Datasheet/3421f.pdf
The demo circuit is included with the LTspice install; component -> search for ltc3421; load the demo circuit.
The input voltage range of this converter is 0.5v .. 4.5v with an adjustable output that can be set to a maximum of 5.25v.
Also handles a continous current of 1.5amp (3amp max) so if you have the "Large Print" Kindle edition with room for 18 pv cells...

These thin-film CIDS cells should be able to produce that minimum input voltage in a dark room. Well, almost a dark room.

Edit 7:
I had been thinking of using a simple, piecewise linear, approximation built out of parallel varisters. Each with "custom" parameters.
But it reads as if Mohamed Azab beat me to it, 13 years ago. (the pvSim.pdf attached)

Edit 8:
When inventing a custom model, it is nice to have data on actual devices to compare the model's behavior.
This site has a nice collection of commercial models, with datasheets:
http://www.solarelectricsupply.com/solar-panels.html
(No CIDS panels in that list though and manufacturer's are bogarting the thin-film technical information.)

Perlight Solar wins hands down, they have actual engineering grade characteristics curves in their consumer data sheets. (attached)

Also nice to know how the <xx>SPICE diode model works, since pv cells are just funny diodes. I have added a description of the SPICE diode model to the collection of information attachments.

Edit 9:
Building a model of a photovoltic cell in LTspice that would fit the needs of this project requires adding one or more custom components.

Edit 10:
The user's guide to custom components has been updated to include custom component Macros. This guide is written for the "new to LTspice" user and covers what might take a new user several days of googling to locate.
See: http://knetconnect.com/ltspice/CustomComponents.html

[twobob]

Running a little audio simulation

Spoiler:

Warning: The image is full screen width for visual clarity
Thanks to thomass for pointing out that a spoiler manages this quite effectively

Thanks.

Great package

[silver18]

Did you consider the tilt angle and azimuth? We can squeeze something more out of that PV!

[knc1]

Originally Posted by silver18

Did you consider the tilt angle and azimuth? We can squeeze something more out of that PV!

I may have edited that post since you last read it.

I corrected the major math errors in the "guesstimate ballparking" - this IS what consumers would find advertised as a "4.5 watt panel".

From now on, I will let the computer do the add, subtract, multiple, etc stuff for me.

[thomass]

Originally Posted by twobob

If someone can tell me how to control the size I happily will..

Great package

You can use the spoiler tag at least

like this one:

Spoiler:

[knc1]

I corrected the worst of the math errors in the top post.
twobob will shrink his image to fit this forum, sometime, RSN, when he gets a chance.

[geekmaster]

Originally Posted by knc1

I corrected the worst of the math errors in the top post.
twobob will shrink his image to fit this forum, sometime, RSN, when he gets a chance.

Shrinking the image may make the text less readable. The image may fit better if rotated 90 degrees. Or better, resize the window and recapture the image.

In the "select boot" thread I chopped out the (non-relevant) middle of my screen capture of MfgTool to get it to fit the forum column width. My hand-drawn "cut marks" are even visible on the copies on the Russian website. (Perhaps this is an example of why some sites put ugly watermarks on their images).

P.S. For those who do not know, the "RSN" acronym used above means "Real Soon Now" (i.e. sometime in the near or distant future).

[knc1]

I picked a new part for the "boost only" up converter. See: Edit #6, first post.

[knc1]

Edits #7 and #8 include some new info. links and attachments.
I think I have all of the background information collected to create the type of custom model I want, one that can be used with un-documented parts (such as the PWR1241 surplus pv cell).

[knc1]

Hid Edits 1 .. 8 in top post.
Posted a HowTo on adding custom components to LT spice:
http://knetconnect.com/ltspice/CustomComponents.html
Top post updated to match.

[knc1]

The new-to-LTspice user's guide to creating custom components has been extended to include creating custom component Macros.
http://knetconnect.com/ltspice/CustomComponents.html

That should complete that document. Well, other than a bit of editing as the need is discovered.

I will start another set of notes on creating the Mohamed Azab pv cell model.

Why is this OT stuff needed? I thought no one would ever ask

The pv cell parts that a DIY hobbyist is likely to find (surplus, seconds, used, taken out of something broken, etc) will most likely not come with engineering documentation.

The DIY person, with a cheap multimeter, can put their "find" in the bright sunlight and measure the open circuit voltage and the short circuit current.
Ah, but engineering models of a pv cell are based on many other factors than Voc and Isc.
The Azab model requires one other parameter that can be approximated from Voc and Isc - at least close enough for a hobby project.

Plus the DIY person will have to know the active area of the pv cell, which they can measure since it is easily visible. If it isn't visible, the cell will not work very well.

[knc1]

Located a paper that includes the characteristics of CIDS photovoltic cells.
Didn't find it earlier because it hadn't yet been presented when this thread was started.
Added it to the top post attachments.