Bug in PicStep - bad FW

[Garfield2]

Fixed.

PICStep now has a V2.0 firmware available.

It has a new 1/16 mode (1/16, 1/8, 1/4 and 1/2, lost full), a 5 1/2 minute auto shutdown and the direction line bug fix.

The motor should sound exactly the same in either direction.

I'd like to thank RadekCX for pointing this bug out, but then again the goose chase and the headaches caused by deciphering his posts probably doesn't warrant any.

http://www.fromorbit.com/projects/picstep/

Alan.

[RadekCX]

Alan good job - source code is correct, but you can make this code faster.

RadekCX

[Garfield2]

Originally Posted by RadekCX Alan gut job - source code is correct, but you can make this code faster. RadekCX

At 20MHz the entire lookup and step decisions are done in ~65 instructions which equates to 0.0124 milliseconds per step (200ns per instruction). I could probably shave 10-15 instructions off, but I think it's fast enough.

Plus you said yourself, yours is written in Pascal and is 7.5K versus my 1.59K and mine has the timeout function too. In embedded processors smaller is usually better. That's 4.716x better!

Alan.

[RadekCX]

Originally Posted by Garfield2 At 20MHz the entire lookup and step decisions are done in ~65 instructions which equates to 0.0124 milliseconds per step (200ns per instruction). I could probably shave 10-15 instructions off, but I think it's fast enough.

Problem is bad regularity in your code, no velocity and quantity instructions. Quantity instructions affect mainly max input frequency. Bad regularity affect resonance and torgue.

Originally Posted by Garfield2 Plus you said yourself, yours is written in Pascal and is 7.5K versus my 1.59K and mine has the timeout function too. In embedded processors smaller is usually better. That's 4.716x better!

Uaaaa, problem is in words "usually better". :-))))

Good luck
RadekCX

[Garfield2]

Originally Posted by RadekCX Problem is bad regularity in your code, no velocity and quantity instructions. Quantity instructions affect mainly max input frequency. Bad regularity affect resonance and torgue. Uaaaa, problem is in words "usually better". :-))))

Whatever... if I was designing this for an orbital satelite I might be worried. It's for homebrew CNC machines and the like, it's really not *that* important, any decent CAM software should take care of the velocity and "regularity".

As for this pissing contest, I conceed you are the bigger pisser. Well done, be congratulated, here is your trophy...

Alan.

[RadekCX]

:-((((

Sorry, rather keep mum.

[jfong]

Stephen,

Please email me your email address so I can send updated software fix for the EAS microstep board.

regards,

jimf at embeddedtronics.com


PS
Alan Thanks for figuring out the LMD problem and posting the details.

[zoltan]

RadekCX,

Thank you, too.

Zoltan

[santiniuk]

Entertaining thread

Just wanted to say, updated to latest Picstep software and it's running fine.

Thanks for the quick turnaround Alan.

To be honest I never noticed any problems with the old version but any improvements are worthwhile.

Treat yourself to a Fosters....