Dirt-E Bike
Kawasaki KE-175 Conversion - Part 2 ...
(Brute-Force YK42-x tender)
The first
section covered the basic "Mechanical" and 'Logistics" of the
Dirt-E, while this section is completely Electronics related.
The objective of
this page is to promote the open sourcing of how to Hack a $35 Yi Yun YK42-x
controller into a 60V/72V 25A/40A/100A customizable controller. This is NOT a
"Programmable" Controller, but rather one that you configure to your
specs and assemble according to your application.
theworkshop.ca
nor any third parties associated or noted on this (or any pages) assumes
liability for any and all use or misuse that this information is put
to.
theworkshop.ca
owes a considerable debt of gratitude to Fechter, and Robert (RTL) of the
Endless Sphere EV Forums... Robert has broken the ground on the YK42-x upgrade
front with his postings on how to take the YK42 to over 120VDC. While Fechter is
a mechanical eng (Not Electrical but should be by experience) who has helped me
make appropriate components selection and proofed the final PCB layout. Many of
the issues I note below were discussed in depth with either or both of the
above.
YK42-x
Family of Controllers
YK42-2 24V/30A - I
haven't seen the inside of this model and can't comment on whether any info on
this page is applicable.
YK42-2 Reverse
24V/30A (with reverse function) - I have to get one of these and see what it's
all about!!!
YK42-3 36V/40A -
This is the model that this page is based on.
YK42-4 48V/100A -
All info on this page applicable to this model
YK42-6 60V/???A -
I know that there is a 60V YK42 in production but have not seen any for sale in
North America.
The schematic
below is saved at 200DPI and should print legibly on an 8 1/2" by 11'' page
(Right Click on image and select "Save As").
The
I've spent days
staring at this small board and have to salute the Yi Yun engineering staff as
it truly is a masterpiece of simplicity and compactness.
Essentially the
Brute-Force brd is duplicating the output stage of this board with heavier parts
and copper.
When done only
about 1/3rd of the parts will be remaining on this board.
The PWM (Pulse
Width Modulation) is accomplished via an LM339 Quad Comparator and a dozen or so
analog components.
When ever my head
would start to pound from concentrating on the schematic (like a monkey doing a
Math problem), I'd spend a few hours husking PC power supplies for some of the
parts that will go into the Brute Force X tender.
Don't throw
anything away!!!
The key parts that
actually made it to the production brd were the 200V 680uF capacitors and the
misc isolating hardware from the diodes and transistors shown.
The diode pairs in
switching power supplies are ideal for this application, but since I gutted
several different models of power supplies I had quite a variance in the reverse
recovery rating, between 40 and 35ns (Nano Seconds).
So opted to buy
new units that are all identical.
The PCB layout is transferred
to a ridged fiberglass copper clad blank board. (do not use thin or flexible fiberglass
brds)
Blank boards are
sold as 1/2oz, 1 oz or 2 oz spec, it would stand to reason that the heavier the
board the thicker the copper.
We want the
thickest we can legally procure and that can be acid etched.
The PCB is etched
for almost 45 minutes...
As this is a fresh
batch of Ferric Chloride, and it was heated to approx 80F, I have to conclude
that this is fairly heavy copper,
(Chemicals,
safety... re-read the disclaimer...)
The etched PCB was
drilled prior to scrubbing the toner transfer off the brd, similarly the
component side stencil was applied.
Have all your
parts, hardware and misc items ready to Rock before cleaning the toner transfer
off. Once the Copper is exposed to the air, it won't stay shiny for very long
and since this brd will require some substantial soldering, once you start it
should ALL be done in a single session.
Again prior to
cleaning the PCB I fabricated this monstrous heat sink for the driver
electronics.
The IRFB4110
MosFets are rated at 180Amps @ 100V with an Rds of 4milli ohms... Killer
hardware.
The heat sink
started out as a 1/4" thick plate that was slotted and drilled to increase
the surface to volume ratio for greater cooling.
The active cooling
fans are 12V Pentium class CPU fans.
The heat sink is
drilled and tapped to be screwed directly to the PCB very securely.
Each Flyback Diode
pair must be isolated from the heat sink but be able to dissipate significant
energy efficiently to avoid catastrophic failure.
Heat sink compound
is applied to both sides of the neoprene/rubber (?) isolation pads.
The IRFB4110's are
mounted identically on the other side.
Use an Ohm meter
to ensure that there are no parts shorted to the heat sink, if there is even
one, I can guaranty Sparks and Smoke.
I know, it looks
hideous, but the heavy slathering of solder is to offer a more conductive path
for the current to flow along.
Remember, this is
aiming for 72V @ 100Amps of power, that's 7200Watts, just over 9HP peak.
And if I find that
under heavy load conditions that traces of the PCB are melted off, I will embed
bare AWG#12 wire in the solder.
Given the heat
dissipating properties of the large copper traces, I had to file down an old
soldering tip to improve the heat transfer such that I could establish an even
flow of solder onto the brd.
Note the fine tip
in the tray will never be able to solder as required given that this is a 24Watt
soldering station.
I suspect that
using a butane torch would just delaminate the copper from the fiberglass.
Not shown on the
schematic or discussed elsewhere, R7 needs to be replaced.
It's function is
to drop the Battery Pack voltage to approx 20 to 25V as the Vin Value for an
LM7815 regulator that supplies the LM339 circuitry.
On the YK42-3
(36V) unit it ships with a 390 ohm resistor, the YK42-4 (48V) version has a 1200
ohm resistor.
For either the 60V
or 72V upgrade I've been using a 2000 ohm 5Watt resistor that has worked
flawlessly, a 2 Watt resistor sits at about 45C to 50C, since up grading to the
5watt, 30C (even at 72V) is the most I've recorded.
The 3 (three)
solid buss or shunts noted are rated at 40Amps for the current limiting on the
YK42-3 (the model being used on this page).
Each Shunt carries
approx 13 Amps, to lower the current limiting to 25Amps remove 1 (one) of the 3
shunts, to raise the current limiting either add heavier shunts or bridge 2
(two) of the existing shunts as the YK42-4 is. (I'll be revisiting this in more
detail if the BFX moves onto larger motors such as the Series
Wound GE on the electric UTE)
This section is
largely my own doing and feel quite proud of how it works.
As Robert noted
with his 120V hack of the YK42-x there can be some pretty spectacular sparks
when ever you connect the battery pack to the controller due to the large
capacitance of the modified controller.
Given what I'd
seen when connecting 36V and 48V controllers, I didn't want to think about what
it was going to be like at 72V.
So my solution was
to implement a "Manual" Pre-Charge across the mains switch. The idea
is that a momentary contact switch is engaged for 5 seconds to charge-up the
cap's through a 620ohm 2watt resistor. While holding the pre-charge down, engage
the main switch, no arcing, contact damage or damage to the brd.
All the switches,
and a handy digital temp sensor are mounted in a generic "Project Box"
to keep things neat.
Although the temp
sensor does have programmable limits that can trip the blower and Controller
fans, I like the idea that I can control them at start-up and keep them running
after the bike is shut down.
The original
YK42-3 was re-assembled into it's heat-sink box prior to removing parts and
tested to ensure that it was in good working order.
I don't think that
there would be anything more frustrating than to try and diagnose a problem on
the Brute-Force board, when I wasn't sure that the PWM circuitry was functional.
With
the Brute Force X tender assembled and the YK42 stripped naked, they are united.
The original V+
and grd lines are used from the YK42, while the "Shunt Grd" line is
upgraded to AWG #8 wire and all wiring between the Battery pack and the motor is
also AWG #8.
Close
visual inspection of ALL Soldering and part removal is essential prior to
powering-up!!!
The schematic
above hopefully clarifies some of the rat's nest pictured below.
And here it is 72V
at 40Amps being cranked through the Brute Force X tender...
No heat on the BXF
at all, I even ran the brd for 5 minutes with the fans turned off and there was
Zero heat gain on the heatsink, MosFETs or the diodes.
Though the motor
did rise to about 40C after 15min of "No-Load" testing.
(All temps are
measured with a touchless IR thermometer).
The motor's no
load RPM rose to 810 RPM on the output shaft, or approx 5,400 RPM on the actual
motor at 60V.
At 72V the output
shaft measured 920 RPM or approx 6,100 RPM on the motor.
The current set-up
has survived approx 30 minutes of total run-time at 60V and over an hour at 72V.
The next steps are
to reduce the current limit to 25Amps, and add the filtered air blower to the
motor. The battery framing has been re-welded to accommodate the full 72V pack,
and just needs to be assembled back onto the KE-175 frame.
The controller
will be mounted into a sturdy enclosure and wired onto the bike.
If this proves to
be successful and the original 750Watt motor survives on the Dirt-E Bike, I
think that I'll have the motor/controller issues worked out for the recumbent
trike that stalled construction earlier in the summer of 2007.
So in closing I
can't help but feel that the Brute Force X tender (BFX) is a success, though I want to be clear that it is still in the most preliminary level of
testing.