Re: More fun with VHDL

"Mike Treseler" <> wrote:
> Take sdata_in out of the sensitivity list and
> you have a synchronous process that will
> sim and synth the same way.
>
> Hmmm. Why are we making bytes in the first place?
> -_______--______----____-------_
>
> Don't we just want to preset a counter
> while the data is high and count down
> while it is low?

Well if we were doing one PWM channel perhaps. My goal is to have several
PWM channels.

Constraints on my design are that I want to be able to "clock in" the
various widths behind the running the circuit. Then, once I have the next
set of widths ready to go, to assert "load" and at the start of the next PWM
period, all PWM channels running at their new value.

As a roboticist, and a "hobby" one at that, one application of this
technique is to build a device for driving a number of R/C servos that can
be programmed with a serial bitstream, and then commanded to operate
synchronously. This level of control is essential in using R/C servos to
control articulated walking robots for example. (google on the term Robo-One
for examples)

I'd like to build it out of a CoolRunner CPLD. I am, perhap prematurely,
trying to be stingy on flip flops to save macrocells. If I create a single
"master counter" that is counting through its iterations, then I can create
a PWM output which consists of 'n' flip flops (equivalent to the counter
width) and some combinitorial logic that implements an n-bit compare. So if
my count width is 'n' then my macrocell count is (pwm_channels + 1) x 'n'.
This is because I can use the same macro cells for both the latch and the
comparator. The more channels I can get into the CPLD, the fewer CPLDs I'll
need. Many Robo-one robots have 27 degrees of freedom (that is a lot of PWM
channels .

The synchronous change facility is a "freebee" once I've isoloated the
serial shift register chain from the pwm generator channels (I end up
burning another n macro cells for each channel making it (2 * pwm_channels +
1) x 'n' macrocells.

There is some other magic with them being synchronous but not simultaneous
but I'm not close to needing that stuff yet.

--Chuck
http://www.mcmanis.com/chuck/robotics/

Chuck McManis