RE: [CAD_CAM_EDM_DRO] LMD18245, Gecko MULTIPLIER, and SLA7062 Sync Mode

Hi Alan,

> Hi John,
>
> You're talking about pin 14, the DAC reference?

Yes.

>
> I was hoping to add reduced current holding to the PIC on the
> EAS boards,
> but it looks like the PIC is out of pins!

Since I set the phase current D/A value, direction and brake lines with the
serial shift registers, I had two pins left over. I used one of them to
pull the Vref line to a lower voltage to implement the reduced current.

>
> As I mentioned in a previous post, I'm considering running
> the four EAS
> microstep boards from a single oscillator, so perhaps I
> should figure out a
> PIC to watch the step lines, and reduce the reference voltage to the 8
> LMD18245's when they are not stepping. I'll want to add
> control of a SSS
> module for spindle control; motor "standby" could even be
> controlled by a
> simple additional controlled circuit.

There are lots of 'slow' lines on any project like the one you've got.
Replace the slow ones with something like the 74HC595 (direction for X,Y,Z
and enable Z,Y,Z come to mind) and you have freed up 6 pins on the PIC.
Then use SPI bus signals, CLK, MOSI, SEL and add a Latch signal to drive the
HC595 and with 4 pins you can do so much more. Get really adventurous and
add a parallel to serial device (74HC573 I think) and you can monitor all
your limit switch inputs too. At 10Mhz the SPI takes almost no time.

> If the SLA7062 has a "weak" problem in stopping on
> non-halfstep positions
> (did I get that right?), then what would appear to be needed
> is a "STEP
> PULSE MULTIPLIER" as used on the Gecko G210 (A) drivers.

Won't make any difference. Check out the SLA7062 data sheet. They show
some nice curves. The most torque you get out of the motor comes at the
70.7% point when both coils have that percentage of maximum current. After
that the armature is moved with less current in one coil and more in the
other till it's at 100% for one and 0 for the other. Micro-stepping doesn't
produce as much current as full stepping.

The curves really show it all. Where the LMD18245 falls short is that it
has only a 4 bit D/A converter. That means even with only 8 steps per step,
there's a compromise between the best evenly spaced (or smoothest motion)
and the D/A voltage used for comparison against winding current.

An 8 bit SPI D/A still just working 8 steps per step would work better
because the sin wave could be approximated better. And values could be
adjusted (or different tables used) at various velocities to deal with the
resonance rough spots.

The ELS, or your board, could keep track of what micro-step you were at when
it was time to reduce current. It could then issue up to 4 steps to move
the armature in either direction to the point where one winding is off.
Then reduce the current into that winding.

Let's see. On my lathe, 16TPI leadscrew is 0.0625" per rev. With 8
uSteps/Step I have 1600 steps per rev or 4 steps worst case to the one
winding scenario which moves the carriage 0.00132978". Since I know which
direction I'm going and I know which direction has the backlash, the ideal
would be to move the motor in the opposite direction to current motion. Then
that 0.00133" movement won't actually move the carriage but will just unwrap
some of the backlash. So yes, I'd see the motor move but so little and the
carriage wouldn't move at all.

Regards,

John Dammeyer