gemellocattivo.com
http://gemellocattivo.com/forum/

TMS570 based ecu?
http://gemellocattivo.com/forum/viewtopic.php?f=22&t=301		 Page 1 of 5
Author:  mk e [ Thu Apr 13, 2017 1:04 pm ]
Post subject:  TMS570 based ecu?
I think we should do an open(ish) ecu based on this:
http://www.ti.com/tool/launchxl2-570lc43

Who's in! :)
Author:  essess [ Thu Apr 13, 2017 2:12 pm ]
Post subject:  Re: TMS570 based ecu?
I'll give it a shot!
Author:  cribbj [ Fri Apr 14, 2017 9:21 am ]
Post subject:  Re: TMS570 based ecu?
Are you thinking of dumping the AEM/Enginelab unit?
Author:  mk e [ Fri Apr 14, 2017 5:19 pm ]
Post subject:  Re: TMS570 based ecu?
cribbj wrote:
Are you thinking of dumping the AEM/Enginelab unit?


NO!!!!!!

I really like the enginelab stuff....its really perfect for someone like me who had ideas but is not really a programmer/electronics type.

I've been trying to get an open(ish) ECU working for years now, this is a continuation of that project. I was working with a freescale 56xx platform which is very good but the free tools are not very good of very complete....so the thought to switch to the TI processor.

My thought is that I will continue to develop with EL while people like Sean (essess) pull the tested control logic into the open project and someday a good low cost option will be working.
Author:  essess [ Fri Apr 14, 2017 11:03 pm ]
Post subject:  Re: TMS570 based ecu?
mk e wrote:
My thought is that I will continue to develop with EL while people like Sean (essess) pull the tested control logic into the open project and someday a good low cost option will be working.


Yeah, I torture myself for fun with things like this. I really like the ideas that float around here too -- so maybe I can capture this all and try to stuff it in a TMS570. My board is on its way.

The big attraction over the MPC stuff is that the HET compiler appears to be free and we'd not be limited to m-n crank or single edge cam --- so this now opens up possibilities for me to get this on my plain-jane oem platforms with little effort. I also like the RAM overlay abilities (POM). This controller is overall a little bit simpler too.

So why not try ? Any showstopper simply results in falling back to the TRK board siting on my desk here. Nothing lost.
Author:  darktype [ Tue May 09, 2017 11:21 pm ]
Post subject:  Re: TMS570 based ecu?
Will it be worth it?
Author:  essess [ Wed May 10, 2017 8:27 pm ]
Post subject:  Re: TMS570 based ecu?
darktype wrote:
Will it be worth it?


you're going to have to expand on 'worth' ...

When I need an ecu I just buy it and be done with it. Capable solutions exist now. The 'worth' of rolling my own is something you can't buy --- that's its worth to me. I also enjoy helping mark where/when I can - he's built a pretty neat community around here.

So far, I had code running within 5 minutes of opening the package. I have rudimentary HET code doing PWM stuff -- nothing special. I've got a friend hand coding marks el model when he has time. I haven't gotten further than that .... maybe I never get further than that. It is fun! well worth the 30$.

If anyone wants to try HET coding, the simulator works and you don't need a board to do it. You can generate stimulus waveforms and examine internals --- all for free.
Author:  darktype [ Wed May 10, 2017 11:52 pm ]
Post subject:  Re: TMS570 based ecu?
I see... that's pretty comprehensive thought you've shared. Thanks for the pointers though.
Author:  essess [ Wed May 17, 2017 10:09 pm ]
Post subject:  Re: TMS570 based ecu?
whelp - the HET is done for if we need to get down to <=.1 resolution and handle a reasonable velocity.

...

To keep it simple, I'll use Mark's setup as the example (10deg/tooth):

To do 6k rpm (conservative) and have a final angle resolution of .15 deg (the best we could possibly get with SWAG), we'd need to have our HET program 'loop around' at 44+MHz. That leaves time for slightly less than 2 instructions and I'll need 3 to do the interpolation (APCNT->SCNT->ACNT). This doesn't even begin to count the space we'd need for scheduling inj/spark and handling the gap.

So, then I worked backwards from 'instructions per loop'. I used the following values:

32 instructions per loop? 953 rpm
64 ? 476 rpm
128 (most realistic of all) ? 238 rpm

I'm not sure if HWAG gets us past these restrictions ... but if it's undocumented/unsupported by the mfgr I'm not even trying it based on these numbers.

If we 'settle' by going down to 1.25 deg resolution (still 10deg/tooth):
128 instructions ? gets us about 15k limit ... not bad.

so, lets go up to 6 deg/tooth and .75deg resolution:
128 ? 9155 rpm

so we might be able to get some reasonable rpm expectations in exchange for 1 to 2 deg resolution. This is probably acceptable for fuel, but not spark.



Is this the end? Could be.
Author:  mk e [ Thu May 18, 2017 8:39 am ]
Post subject:  Re: TMS570 based ecu?
YUCK!!!!!

Probably not OK for fuel either. Angle doesn't really matter but pusle time does, particularly at low rpm/power where the time error will be a much larger % of the total pulse. There probably is no practicle way do to this stuff without an interupt based system.

....this seemed like sucha good idea until didn't work :(

How about we take a step back and keep with the modular theme and do a family of simple modules. On decodes cranks and provides a common timing to a buss. Then you can plug in as may fuel/spark/whatever modules that are a simple timer and driver, maybe SPI to update the pulse width or spark angle? So no fancy custom massive number of timers chip...just plug together what you need?

Just thinking out loud a bit.....
Page 1 of 5 All times are UTC - 5 hours [ DST ]
Powered by phpBB® Forum Software © phpBB Group
https://www.phpbb.com/