Newsgroups: comp.robotics
Path: brunix!cat.cis.Brown.EDU!noc.near.net!MathWorks.Com!panix!zip.eecs.umich.edu!newsxfer.itd.umich.edu!nntp.cs.ubc.ca!alberta!bigbrd!crs-sys!nextedm!spl4!cuugnet!hrynkiwd
From: hrynkiwd@cuug.ab.ca (Dave Hrynkiw )
Subject: Re: Solar Panels
Message-ID: <CoLEnL.484@cuug.ab.ca>
Date: Thu, 21 Apr 1994 04:47:44 GMT
References: <60.8180.4884.0N19A997@canrem.com> <2onaah$cha@handler.Eng.Sun.COM>
Organization: Calgary UNIX User's Group
X-Newsreader: TIN [version 1.2 PL0]
Lines: 178

Chuck McManis (cmcmanis@pepper.Eng.Sun.COM) wrote:
: In article <60.8180.4884.0N19A997@canrem.com> (Notre Dame) writes:
: >Hi, does anyone know how to get any significant current from a standard
: >Solar Panel(like the type found in most calculators) that might be
: >capable of powering a small motor.

: So to run our motor for 1 second we need to store 6 seconds of output from
: the solar cell. To schematic is something like :

: 	+---|>|---+----------------- Motor A
: 		  |         
: Solar		----- C1	  +- Motor B
: Cell		-----             |
: 		  |               |
: 	+---------+            |--+
: 		  | trigger  | |
: 		  |	+----| |<-+  Q1 Nchannel MOSFET
: 		  |	     | |  |
: 		  |	       |--+
: 		  |		  |
: 		  +---------------+

: capacitor 'C1' to the triggering of the FET. Some systems like some of the
: solar rollers I've seen simply wait a fixed amount of time and trigger
: the fet/transistor regardless. What I'd like to see it an extremely low
: power charge pump that puts out 15v from the 3.2v input and uses a zener
: or something to trigger the fet when the trigger cap is charged. So imagine


	Ahh, here's that article!  I meant to reply to this a week ago, but
was too busy.  Thanks for reminding me Keith...

	Anyway,  I personally haven't done much with the experimentation with
different ways to build a relaxation oscillator, but understand from a friend
who has that the MOSFETS do work.  Only caveat is, they still hate static.
He'd get one to work, but then it'd blow as soon as somebody said "Hmmm, what's
this do <picking it up>".  I'm still a great believer in Mark Tilden's design
using a 3904, 3906, and a bias resistor.  It'll cost you two diode drops in
voltage loss, but it is sturdy, and tunable thru the bias resistor.  I'll
tag on a msg Mark sent me once regarding the circuit at the end of my msg.

	In regards to a voltage-sensitive trigger mechanism, by all means,
YES - Figure one out!  We've been playing with various schemes to get the
RO circuit firing in the most efficient manner, and so far, the optimal way
is thru the time-based "Happy Birthday Singer" chip found in greeting cards.
It is very power efficient, drawing microamps at it's most efficient, and 
"maxes out" at about 2mA at 4 volts (or something near that - chk the msg
tagged on).  Ideally, one would want a voltage _and_ time dependant triggers
working in parallel in case if its too cloudy for it too hit the "release"
voltage.  At least some of the work would be accomplished that way.

	The following msg from Mark Tilden covers (in general) the HBS timer,
then the origional "BEAM Circuit", using a zener as a trigger.  The zener is
too "current hungry" to work consistantly, so use the output of one of the
speaker outputs thru a 0.1uF cap to where the zener connects to the 3906.

	BTW, I've got a 20 pg booklet & component kit explaining how to do
this in great detail with lots of pictures.  If interested, email me for more
info.

----------xxx----------xxx----------xxx----------xxx----------xxx----------
--------------------------------
 
The happy-birthday singer is a two bit wide by n words deep programable
sequencer with 4 commands: reset to start vector (00), decrease frequency (01),
increase frequency (10), and pause (11).  The device starts out with 
a 2 word vector jump at the start of memory which locates 16 positions 
in it's memory.  This jump vector is readdressed whenever a (11) command
is found in the execution table.  "Happy Birthday" is the first in this
chain and is at default (00-00).  There are at least 6 songs already in the
device which can be found by reprogramming the jump vectors.  The vector
locations seem to be at either 256 or 512 bit increments in a 4k or 8k
memory map.  These songs include "you light up my life", and xmas carols.
Reprogramming the singers is not simple in either process or technology,
and truth to tell I never completed my studies of the devices as it's
a bugger working with an already burned-in prom.  The only details 
solaroller builders need know are:
 
- The configuration is as follows:
 
                   --------------------------------
                   |  Vcc   |     switch          |
                   |        |                     |
                   |---------                     |
                   |    |      gnd                |
                   |   ***  __________            |
              clock|--**uP*-|                     |
            whisker|  *****                       |
                   |   ***                        |
                   |   / \          Battery Holder|
                   |  |   |         (Usually cut  |
                   | [ ] [ ]         off)         |
                   --------------------------------
                  Quad outputs
 
- Soldering a cap between the thin clock whisker and Vcc gives an
approximate ratio of 1sec period = .1uF.
 
- Device operates from .98v to 5v with a significant current drain happening
around 3.1v (approx 3ma nominal).  Device works optimally from 1.2v to 2.3v.
A short of the power pins will reset the device but the device has a power
down mode between .3 and .98v where it will keep it's internal count position
without resetting.  The input impedance of the clock whisker is very high,
and is obviously a direct link to the internal resistor-cap junction of the
internal oscillator.  This pin has the classic cap charge-discharge curve
when probed.  Resistors to this pin will also work, but at a significant
current increase for low power applications.  
 
- Operating current of the device at 1.5v is in the microamps with the
piezo removed.  
 
- The outputs are rail to rail drivers at about 2-3ma per transition.  The
outputs are quadrature encoded, so they are always opposite polarity to
each other.  They can be direct shorted for long periods without affecting
the device and seem to handle positive and negative current spikes well.
 
- The only way of destroying them under nominal conditions is to power them
up backwards.  They die very quietly and then only way to test is by
reattaching the piezo xtal to hear if the oscillator has stopped.  Another
fault is when the cap soldered to the whisker is not superglued down after
being attached.  There is no cure for this as the whisker is usually lifted
right from the pcb.  Replace the device with another one.
 
-----------------------------------------
Try this:
 
 
          |-------------------O-------------O-----------------|      
          |                   |             |                 |      
          |               |-------|         |                 |      
          |               <  mot. >-        |               + |      
          |               |-------|         |                ---     
          p      2.2k-10k     |           + |              -------   
           n|--O---/\/\/------O           -----  >2200uF     ---    >2v solar
          P    |              |           -----  cap       -------      cell.
          |    |              N           - |                ---     
          |----|------------|p              |              -------   
               |              n             |               - |      
              /-/             |             |                 |      
               ^ Zener        |             |                 |      
               | mc4742       |             |                 |      
               |--------------O-------------O-----------------|      
                                                               
 
Legend:
               |                      |
collector-     P                      N                      |
base-      -|n     PNP trans.     -|p      NPN trans.    ----O----  wire join.
emitter-       p                      n
               |                      |
 
Essentially it's a modified SCR design with supercritical feedback.  The 
zener starts clamping at vcc-.7v and will eventually trigger the PNP, 
triggering the NPN, and will lock so long as the motor is moving.  The 
beauty of this design is it's incredible high impedance until tripped.
For transistors, I've used 3904s and 3906s, but for higher power designs,
the NPN must be replaced with something like cascaded 2n2222s, or high 
gain power transistors, like TIP 31s.  The zener should be rated to trip 
at 2 volts below your solar-cell voltage rating to get maximum power.
 
This circuit is called the BEAM solarengine.  I developed it for the 
Robot Olympics, used to build a solaroller device which covers a 1 
meter distance in pure sunlight using only a cell 2-1/2" by 1/2".  I 
promote it in my BEAM robotics package because of it's simplicity 
and cheapness of parts.  What's your application?


----------xxx----------xxx----------xxx----------xxx----------xxx----------

	Hope this helps!

	-Dave Hrynkiw

-- 
Um, no - that's H,R,Y,N,K,I,W. No, not      \ hrynkiwd@cuug.ab.ca
K,I,U,U, K,I,_W_.  Yes, that's right.       \ Calgary, Alberta, Canada
Yes, I know it looks like "HOCKYRINK."      \ Home of the Western Canadian
Yup, only 2 vowels. Pronounciation? "SMITH" \ Robot Games - APRIL 9 1994 
