New Microchip PIC processor.

[Bill the arcstarter]

"K. Toyofuku" <toyofuku at cats.ucsc.edu> wrote:

>>The important thing here is that uchip has finally done two things to 
>>their
>>flash parts:
>>1) Included enough space (8K) to use useful
>>2) Allowed the processor to read/write the program space at runtime.  This
>>would be very useful for fuel maps and the like.  Prior to this processor 
>>-
>>that sort of thing would have required addition components.
>>
>
>i thought that there was another chip out there that did all of that
>already...hmm.. i can't remember which one it was....

Its possible.  One of the main advantages with this new family is the speed 
and cheapness of doing in-circuit code development! :)  If you have enough 
money to buy the in-circuit emulators and whatnot for the other processors - 
most anything will do.  PICs are cheap to work with and fairly powerful.

Atmel (one of microchip's competitors) has had fully flash parts for years.  
Scenix is another candidate.

>i do.  i'm new to this diy dfi thing, so what do you recomemd to look for 
>in microcontrollers when building your own dfi.  i've used motorola's 
>68hc11's (i don't think that they could hack it..) and some strong arm 
>stuff to build dataloggers and other non-car related devices.  is there an 
>advantage to using PIC?

"Advantage" is hard to quantify.  I suppose it's mostly what you are 
familiar with.  This 16F874 part looks a LOT like an HC11 rip-off.  Even 
some of the port names are the same.

I'm sure there are 100s of processors up to the task.

>i figured a microcontroller that had lots of I/0 lines @ 10 to 15bits of 
>resolution, decent processor speed,  lots of RAM and EEPROM, and uart.  
>what else?

Yea, I/O lines are nice.  Timers to run injectors are nice.  A/D converters 
are very nice to read sensors, etc.  But the basic fuel tables are rather 
small.  I've heard that the common GM ecu's use a fuel table only (at most) 
16x16 points in size.  Thats a max of 512 bytes, even at two bytes per cell. 
  Not much memory needed for the fuel tables.  (Somebody correct me if my 
impression here is incorrect)


I suppose the basic routine would be:
1)Read the MAP, or MAF or whatever sensor. (A to D converter)
Measure engine RPM (timer functions or input capture pins on an HC11)
2)Read the air charge temp (A to D)
3)Do a 2-D table lookup, possibly with interpolation.
4)Compensate for air temp, cold start etc (gets slightly complicated here)
5)Send this injection duration to the injector drive hardware.
6) Do it again!

I think thats about it.  You will also need some sort of "background" 
process/program to let you mod the table on the fly, or whatever you are 
planning on doing for data acquisition, etc.

Not sure why you'd need a lot of ram either.  Granted - PICs are rather weak 
in the RAM department, and are not easily extendable (unlike an HC11 would 
be).  But for the above steps you don't need more than a few bytes of ram 
IMO.

One of these days I'll try rolling my own after I get familiar with the 
system I presently have.

My $0.02.
-Bill


_______________________________________________________________
Get Free Email and Do More On The Web. Visit http://www.msn.com