Microchip MCP25020/25050 CAN IO

[zappedagain]

Does anybody on this site have experience with these CAN IO devices? They are CAN controlled with digital IO, PWM outputs, and A/D inputs (25050 only). They look pretty slick.

The errata deals with one fairly severe error condition that the software needs to work fairly hard to support. We have these implemented on a system and the error rates with these devices on the bus goes from 1 out of several thousand messages to about 1 out of 40. This is even with talking directly to Microchip's evaluation board with their parts on it. We are pushing the data rate to 1 Msps over fairly short, well defined, cables. Almost every error that can occur (CAN defines 5 types) does occur when these parts are on the bus. They work, as long as you let CAN work through the errors, but the bus gets very 'messy' and busy.

If someone else has had good success with these parts we'll continue debugging, but it is looking like another solution may be more fruitful.

Thanks,

Z

 

[zappedagain]

I found a forum on the Microchip site regarding these parts:

http://www.microchip.com/forums/f161.aspx

It looks like people have been complaining about the same issues since 2004. Too bad I didn't see that earlier.

Z

 

[itsmoked]

Ouch.. I do mostly Microchip and am often disappointed/annoyed by MC's idiot errata that they frequently never bother to fix.

A lot of times it's even head-liner stuff on the data sheet front page that doesn't work at all.

Sorry I have no suggestions other than maybe replacing them with a processor that doesn't have the problems that you program to essentially cover the functions you were using with the 25020.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[MacGyverS2000]

I love MC stuff for what I would call "toy" projects... but when it comes to a project someone's life (or my job) may depend upon, I look elsewhere. As Keith said, they don't make it a priority to fix errata, just document it and expect you to find a way around it. I spend enough time looking for bugs, I don't need to question the silicon, too.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[MikeHalloran]

I was hoping they would eventually dig themselves out of that mode, and eventually ship chips that agreed with the original documents. Not yet, I guess.


Mike Halloran
Pembroke Pines, FL, USA

 

[zappedagain]

All is not lost! The chips work much better at 250 kbps. Don't trust those data sheets!

Z

 

[itsmoked]

That's a quarter of what you were running.. Still enough to get the job done?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[MikeHalloran]

Right, so all you gotta do is inverse mux four of them.

Mike Halloran
Pembroke Pines, FL, USA

 

[zappedagain]

It works for now. They are nice parts when they work; it is very nice to be on the CAN bus with no additional software. I'm still on the fence whether I'll accept this data rate or if I'll design them out. This design works as long as two features in the spec stay mutually exclusive like they are now, but I don't believe marketing will stick with that plan.

Z