I'm building a project I seen in an electronics mag that is using a C671 PIC,
that PWM's a hexfet to run a DC motor.  The original ASM code is generating
about 20ms wide pulse widths (50hz).  I am looking to use this to drive a fuel
pump at different speeds.  I don't know the exact specs of the pump yet, but I
assume that it can do something like 40+ pounds of pressure.

The question:  Is there an "optimal" PWM cycle size that I should be using for
this kind of DC motor?  Does it depend on what speed it is running at?  i.e.,
could/should I code it so that it has a different PWM frequency at different
motor speeds for optimal tourqe & minimal motor core loses?  Or would you
recommend just throwing a certain sized capacitor in there to even out the DC
pulses?

I could easily rewrite the code to do 5ms wide PWM cycles, or less, and wondered
how this would affect torque at different motor speeds.  I'm thinking that a
faster PWM cycle would mean a faster response time, when conditions that affect
the PWM duty cycle change.



On another note, but for the same project....

I've got a C671JW from Digikey to use on this project.  It has a plated
(brass/copper/gold, dunno) metal casing, and does not have the standard pin1
dimple that the plastic C671 OTP parts have on them.  Looked at/in data sheets
for it, and it doesn't show anything about that (that I can find any ways).  Can
I assume that Pin1 is by the MicroChip (M) logo on that chip?

Also, I noted in the data sheets for the JW part, that the internal frequency
calibration would be lost if I erase it (the current code uses that).  How do I
retrieve that information prior to programming it?  Just have my Epic read it,
look at the last memory location for that data?  I just did that on one of my
671 OTP's, and looked at the Epicwin "code" window.  All of the memory locations
are 3fff, except the last one, which is 3480.  So, do I put the x'80' value
there in my program at that location?  I assume that the x'34' is the "retlw"
opcode, or do I have it backwards?  I did read the data sheet, and just need
confirmation on this...

Can I also assume that I'll have to read that internal calibration on each OTP
chip I use before programming it too?  If so, does anyone have a somewhat
"automated" script, or method, for doing this; or does this process have to be
done manually for each OTP chip (read code, change calibration byte in code,
compile code, program chip)?

BTW, How accurate is that preprogrammed calibration?

Sorry about all the stupid questions, this is my first time working with all
these different parts and factors.  Any help on any of this is greatly
appreciated.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics