Homebrew PCBs

http://home.nc.rr.com/alan69/CNC/drive3k.brd
http://home.nc.rr.com/alan69/CNC/drive3k.sch

   Ok, here is a walkthrough for the schematic, pull it up in Eagle.

   First, turn on the tplace layer and take a look at the components.  Three 
complete 10 MOSFET drivers, 10 gate resistors are vertical at the left of the 
FETs, they're all pulled high so the upper ends are simply tied together and to 
the gate drive voltage or a resistor network.  At the right is two pin power, 6 
pin motor, and motor phase pull up resistor pack (to not blow gates without 
using a bunch of diodes, more on that later).  Also note the simple 7406 OC 
drivers for the gates.  With 12V gate supply it can run 7V motors with logic 
level FETs.  The FETs are arranged so 2 heatsink strips for 2 or 4 phase or one 
additional strip for 5 phase can give some heatsinking for all the FETs. 
Requires isolation tape and nylon screws and nuts since tabs are connected to 
drains, but easy enough.  Usually don't need heatsinking for <3 amp motors, 
these are 12 amp FETs for BUZ104SLs.  It's an all N channel driver too.

   Turn off the tplace, notice that the traces are decently wide.  Gate traces I 
made a bit narrower, so it's easier to follow, still wide enough that etch is 
not critical.  Drive traces could really be made a bit larger even, with only a 
few places remaining narrower for clearances.

   Hit Show and show the outer power pin.  This is ground, all three are tied 
together.  Lower FETs are low side, upper are high side.  Now start at the 
bottom motor connector pin, and work up.  Two vertical FETs are the phase drive, 
phase 1-5 from the inner FETs and that bottom pin to the outer FETs at the 5th 
pin up.  Note that phase 5 is tied to power.  Most motors are not 5 phase, so 
for normal use the outer FETs aren't stuffed.  The unused drive connection is 
tied to power, giving two power pins and 4 phase pins for a 6 wire motor, 
stuffing only the bottom 4 FETs for a 5 or 6 wire unipolar motor.  Also stuff 
the upper 4 and only use the 4 phase drive pins for 4 wire bipolar motors, use 
only 3 for 3 phase motors.  Cut the jumper from phase 5 to power, stuff the 
phase 5 FETs, and use the 5 pins to drive 5 phase motors.

   Show the middle driver + power.  Note the 5th phase when unused for 5 phase 
also takes power to the lower FET pin by way of the driver trace.  A short 
jumper from there to the bottom driver power will supply it.  A short jumper 
from middle + to the same upper driver FET pin will take power to the top 
driver.  Two jumpers and power only has to be supplied to the middle driver pins 
to run all.  Or keep seperate and have different voltage motors etc.  I should 
put another pin there to facilitate the jumper to the upper driver.

   Turn off the top layer.  Start at the bottom motor pin again and show.  Note 
that 1-3 have their entire drive trace on the bottom.  Only 4 and 5 have the 
connecting trace on top.  Phase 5 power jumper to cut if needed is also on the 
bottom.  Upper FETs power goes straight through the middle.  Lower FETs ground 
alternates.  Easy to follow almost the entire layout from the bottom to know 
what pin you're on, and troubleshooting the driver operation from the bottom.

   Turn the top layer back on.  Gates are paired, inner two are phase 3, then 2, 
1, 4, 5.  Not continuous but still easy to follow.

   Of course the drive pin numbers are relatively meaningless, you usually label 
motor pins 1 and 2 power, then phase 1 2 3 4 or 1 3 2 4 etc for a 6 pin motor. 
That is all logical remapping so it's easy.  This numbering is simply for layout 
and keeping phase 5 at the motor connector, leaving it out gives more room 
around the motor and power connectors when only using a normal lower phase 
motor.  I have some 36W per phase 5 phase motors to drive, and plenty of smaller 
5 phase motors as well.


   Now for why you need the resistors on the drive pins to power.  Turn on phase 
1 bottom side gate so pin 1 is low.  Motor coil connects through to pin 2, and 
this references phase 2 high side source to ground.  High side source at 0V 
through the motor.  Now hit that high side gate to supply the coil with power. 
12V on gate.  Source now connected to drain and rises to +V, say 7 Volts.  GS 
capacitance is relatively high, so gate stays at 12V above S for a bit.  19V 
total above ground.  Switch this too fast, and turn on the low side pulling 
source to ground, and you have your gate at 19V with a 20V absolute max. 
Inductance in the motor and switching means you end up blowing gates.

   Instead reference both sides of the coils to +V through a weak pullup.  Motor 
floats at +V of 7V.  Turn on the high side gates first to 12V.  Source never 
rises since it was already lightly referenced to drive voltage.  Nothing you do 
turning on the low side is ever likely to push the low side gate down all the 
way to -20V relative to the source to cause a similar failure through the bottom 
side.

   You don't need diodes from each gate to +V and ground like many people do, at 
least not for just 12V gate drive and 5V or 7V motors.  You just need to work 
out what the failure mode is and drive in a way that won't cause failures.

   That's pretty much all of the driver design considerations needed.  Missing 
about half of the parts most drivers have, yet still operates fine.  Had reverse 
diodes in an earlier version, took them out since there didn't seem to be much 
need with the motors and power levels I was running.  Easy enough to spread 
things a bit and put them in if you feel like it.

   Rest of the board was an in progress relayout of another version, but this 
had the fully advanced driver layout.


   I'm all open to someone doing it better, bring it on.  With 2000 of the FETs 
on hand at $.05 each, I could run 200 5 phase motors at not much over $.50 per 
driver.  Most people are limited to what some chip will let them do at best, and 
the high costs that go along with it.  This will run just about anything worth 
running for next to nothing.

Alan