Hybrid Car – More Fun with Less Gas

Low cost 1 kW photovoltaic generator ?

register ::  Login Password  :: Lost Password?
Posted by Jeff B. on July 27, 2008, 12:53 pm
Hi there,
Looking to a proposal made to us for a 2 kW PV generator (16.000 ), I
was wondering whether we could design a cheap 1 kW PV generator using
Fresnel lenses in combination with high efficiency multijunction PV
What if we put 4 triangular multijunction PV cells (or TASC, assembled
into a 3.1cm x 3.18 cm square) close to the focal point of a Fresnel
lens ?
Could a PV generator comprising 25 Fresnel lenses in a 5 x 5 square
(2m x 2m square) with their PV cells approx. 30 cm behind them, the
whole thing being mounted onto a passive solar tracker, do it ?
From the Spectrolab spec: 27 mW/cm (for 1 sun) are produced by the
TASC. The Fresnel lens focuses approx 155 times the solar irradiance
reaching the 4 TASC cells (3.1cm x 3.18cm) thus producing
27x155x(3.1x3.18)=41,25 W. If we consider 25 units of this kind
(Fresnel lens+4 TASC), this would produce approx. 1 kW for approx.
2250 $ or 1450  (plus framing structure to add). Either there is an
error in my rough calculation or this would be a cheap way to produce
electricity. Do you think it is feasible ?

25 large Fresnel lenses from 3dLens or BhLens (http://www.3dlens.com/
shop/largefresnellens.php)=25 x 22.68 $
1 spectrolab box of 100 triangular PV cells (http://www.spectrolab.com/
DataSheets/PV/PV_NM_TASC_ITJ.pdf)=250 $
1 Zomeworks UTRK40 passive solar tracker (http://zomeworks.com/files/
pv-trackers/Track%20Rack%20Matrix%202008.pdf) = 1450 $

Posted by Roderick on July 28, 2008, 7:25 am

I would be concerned with how you will be cooling the cells to
operating temperature at this concentration.  I have read about
experimental concentrators at 100x, but they seem to have massively
sophisticated cooling systems on them.

Also, that's an excellent price on the cells.  Are you certain it
isn't $50 for each cell ($5,000 for the box)?

Posted by jp838 on July 28, 2008, 1:49 pm
 Jeff B. wrote:

You only need one cell and one lens to test it, ???$0 (you could even
sell a kit, many world wide would like to try that :)

- I doubt it's that simple, some must have already tried. I read about
that kind of cells used with concentrators
- a problem must be that the light spot power distribution is not
homogeneous (*)
- the time life of the cell maybe reduced ?? (*)
- the cell must saturate at ?some light flux ?? (*)
- dissipating the energy is not a problem. With a 100 sun
concentrator, it would give at worse: 1 kW/m2 >> 0.1W/cm2 >>  * 100 =
10W/cm2. A 4C/W heat sink  used in electronic can easily be fit under
the cell ($.5) and will work without a fan. With this,
you keep your cell below 70C.  With that much power to dissipate, a
smart design, profiting of convection, would probably deeply boost
your heat sink efficiency.

Choose your heat sink at:    http://www.mouser.com/catalog/634/1876.pdf
Usually they are in aluminium.

If you use circulating water heated at max +20C, you'll need 0.43 L
of water per cm2 of cell per hour, 10 liters per cm2 of cell per day,
~1m3/day for 1 square meter of panel. Not a good idea if water is not
recycled :)
If these cells could work at 100C one could grab the heat use it for
heating ...

(*) I'd like to have your views about these points ???


Posted by jp838 on July 29, 2008, 12:34 am

for (*) http://www.spectrolab.com/prd/terres/FAQ_terrestrial.htm#section1
a very interesting FAQ written by a cell producer.

BTW, Spectrolab have real products you can buy, not like many others !

Posted by Jeff B. on July 31, 2008, 2:09 pm
There are two difficult points, the more concentrated:
- the more accurate your tracking system must be
- the more calories you have to dissipate

for the first point, I was in search of cheap/simple passive trackers.
liquid mass transfer (with shadowing apparatus) is attractive by its
simplicity but I have doubt about their accuracy and their reliability
with variable weather conditions. It may require a lot of
experimentations to get worked.
you have to collimate a centimeter radius focused light circle on a cm
square of multijunction PV cells a few centimeters before the focal
point (say at 30 cm from the Fresnel lens (sub degree tracking).
I can't see any Shape Metal Alloy (for solar tracking purposes)
commercially available ?
The simplest tracking system I can see is first to use a square based
pyramid of basic PV cells to ensure a rought differential tracking
(~degree accuracy), orientation of the optical frame by two linear
motors. This is to bring the sunlight closer to the axis of the
Fresnel lens & multijunction cells (this couple form a pyramid, with
mirrors covering the lateral inside faces).
The high quality PV cells are the Spectrolab ones designed to cope
with high concentrations.
There are 9 of them in a 3 x 3 square.
The first tracking system is to ensure that the focussed light spot
intercepts this square (the target is then logged on).
Knowing the amount of light reaching each of the nine cells, you can
finally track in a very accurate manner and ensure maximum electricity
There are simple circuitry available on the web. The first and second
tracking phases could form a vernier. This system could compensate
analog biases (components in the tracking circuit) increasing with
For the heat dissipation, I am also looking in the direction of high
performance cooling  systems available for processors (IBM has its own
There is a liquid metal system (with a magnetic pump) from danamics
that seems very reliable/efficient and with reasonnable appetite (1W),
I don't know its price but you are right, heat dissipation is very
important for the best performance and life time of the cells. If you
expect to get 500 suns foccused on 10 cm and given that your cells
are ~0.3-0.4 efficient, this is a lot of heat of a small solar furnace
to evacuate.

PS) it appears that spectrolab do not sell small quantities of
multijunction cells. no order under $000 (120 cells, enought to make
a 3 kW generator). no possibility to build a small 170W PV generator
(suitable for our roofs). looking to Emcore.

This Thread
Bookmark this thread:
  • Subject
  • Author
  • Date
please rate this thread