Bent Genny
Part # 3 - Data Collection &
Analysis...
Jan 3rd 2007
Recapping
Section #2, Coils were wound, the stator poured,
the generator was assembled and the gearing was revised to achieve 1 : 6.5
increase on the rotor from the pedals.
You
can imagine my embarrassment, in error I posted 32V - 12V = 30V, it should have
read 20V... But that sort of transcription error is exactly the inspiration for
this section... Digital sampling, computer controlled recording and automatic
graphing of results.
PicLog
V1.2 PicAxe 08M based Data Logger
Follow
the link below to download the source code, schematics, build details & PCB
layout.
The
Pic Log is the brain-child of "Glenn" @ www.thebackshed.com
in Australia.
I
found Glenn's site and poured over the details of the project for a couple of
days in mid Dec of 2006 (a few weeks ago)...
Over
the Xmas/New year holiday I was able to find a few days to settle down to work.
The
PCB shown above has been modified to accommodate 6-pin Opto-Isolators... Beyond
that it is still Glenn's design (even if I re-branded it)
Another
shot of the tired Acid bath for etching...
As an
aside, I did try rejuvenating the "Ferric Chloride" with a couple of
Carbon rods scavenged from a pair of "D Cells" and a 12V PC Power
Supply...
After
a day of running, the solution was warm, and there was a pronounced build-up of
copper on the negative electrode (I'm pretty sure it was the Negative)...
Regardless
it didn't make much of a difference if any, as it still took an afternoon to
etch the two small brds.
The
traces look fine, and their ready to be punched and drilled.
As a
testament to the simple design of the system, I was able to scavenge every
single part from the brds (to the right) except the actual PicAxe 08M controller
chip...
Though
I did substitute the 4.7V Zener Diodes for 5.6V's, other wise things would have
ground to halt waiting for a couple of "fitty cent" parts.
The
assembled brd sports a dead mouse cable to interface to the PC's serial port.
The
screw down header block was scavenged as well.
After
considerable testing everything is mounted in a husked PC Power Supply box.
The
brd had to be re-soldered a couple of times & refitted with a different Opto-Isolator
(mounted under the PCB) before I found the errors (I made)...
Initially
I scrawled some marks on the cover with a black magic marker, but later thought
it deserved better than that.
Ok!
enough promo for thebackshed.com...
In
all seriousness, I find brief glimmers of hope for the human race when I find
projects like the PicLog available for free, no strings attached on the
internet.
Everything
else just makes me indifferent...
Pictured
below to the left is the RPM sense cable (from a dead drill) that wires directly
across any phase. The sense line reads every positive going pulse over a 1 sec
duration... The raw pulse count is forwarded to the PC where a "User
Defined" constant calibrates the count into RPM.
In my
case the Generator is a 16 Pole rotor, 8 North poles and 8 South poles, Each N
& S combined creates a complete sine wave. Since only the positive wave
forms are counted, the constant to enter is "8".
Above
to the right is the sense resistor 0.1 ohms +/- 5% tolerance wrapped in green
tape, I wanted to keep the lines as short as possible between the Genny, Shunt
and Battery.
The
flat gray wire with the Shunt, Ground, +ve & -ve lines is a line cord from
an old freezer.
It
took the better part of an afternoon to calibrate the ADC constants for the
Voltage reading and the Current reading.
Initially
I was trying to do it by sourcing the current via the generator itself. That was
a lost cause as the readings are jumping around so much on the meter it was
impossible to hold the pedal rpm constant due to the 1 sec delay between acquisition
on the PicAxe and the PC screen display.
Ultimately
I opted to drain a second battery below 11.5Vdc, and measured the V & I
parameters between the two.
This
took less than 1/2 hour.
As an
aside, the Calibration constants you enter on the PC side do not have to be
integers... Once I started to tune them in to the 1st decimal place I was able
to track Voltage and Current within 100mV and 100mA respectively.
I was
very impressed with how tightly the calibration tracked between the meters and
the PC once it was all done.
As
the PicAxe was monitoring the 2 (two) batteries the RPM value in the screen shot
above reads zero... Though in the course of later testing the generator against
my mechanical Starett tach the PC readout was dead on the money.
In
the course of several test runs I found that the secondary drive chain would be
thrown as the rotor approached 500 RPM, due to the whipping action of the slack
as the pedals were pumped harder.
To
the left is a quick & dirty tensioner fashioned out of an old derailleur.
Purely
by luck, the derailleur bolted directly to one of the stator support rods and
tracks the drive chain perfectly.
The
rotor has been pedaled over 700 RPM with out a moments grief since.
My
office system was used initially for the calibration, it runs Win2K.
Since
it is impractical to have all this gear near the desk, I recovered this
comparatively old tablet PC (233Mhz Pentium running win98) from the mundane life
of charging I-Pods for the kids and periodic nasty notes to each other.
Beyond
relocating a DLL & an OCX file per Glenn's instructions with the software it
ran very smooth.
With
the portable PC located beside me as I train for the summer cycling season, I
can watch in real-time the progress of my workout.
In
this case I was just testing...
While
watching the "WattHours" count slowly climb, I'm gaining a far greater
respect for the Grid, and just how much energy is required to create that intangible
unit of measure the Kilo Watt Hour.
Below
is a graph rendered from within the PC software that lays out a 20 minute stint
on the genny. If I was to do nothing else with this project and just use it as
is, I think that the ability to graphically see how you expended your energy is
amazing.
If I
was a real keener, I'd mount the laptop, extra batteries and the logger to the
LEV-1 electric trike, and run the 12Km route to the lake, plot the graph and try
to mirror the power curves to the distance over time including inclines and down
hill stretches.
This
has been the setup for the last 2 days... I will have to build a short cabinet
on castors to house and hold the PC & Piclog so the "Bent-Genny"
isn't stuck in the middle of the shop.
The
PC has since been tied into the network, and the PicLog software is running over
the network flawlessly, which saves having to copy logs to disk for later
analysis.
The
software also saves the logs in Excel and HTML format on the server, so any PC
on the network could be a test bed, with all data conveniently located in a
central location.
Initially
I ran a series of tests in a "Star" configuration of the stator as
pictured to the right.
Since
the log file was saved in Excel, I opted to sort the data by RPM and plotted the
graph below showing Wattage over RPM.
Otherwise
it would have been less useful as a time line of output.
The
some what non-linear graph is likely as a result of the erratic power generated
over a 10 minute span with increases and decreases in RPM and tired legs. But
the trend of the graph is clear. If the log was to span 30 to 45 minutes there
would be 100's of samples to create a more accurate general trend.
The
stator was re-wired to a delta configuration as shown to the right with it's
corresponding graph shown below.
At
present up to 50 watts is a good even pace, but to peak out the genny over
100watts is crazy and tends to leave a burning sensation in my knees...
Please
note from the graph below that for a brief instant I was half a world class
athlete as I cranked out just over 140watts.
In
closing, this particular section has been a blast, and I feel more confident
playing with the PicAxe controller. Again a sincere Thanks to Glenn at
www.thebackshed.com for developing the PicLog PCB, firmware and PC-Side software
and providing it to the public domain.
With
a slick test-bed completed I'll turn my attention to logging the performance
increase of the genny by adding a blank steel rotor on the back of the stator,
and then a complete magnetic rotor that matches the existing one.