It is currently Tue Sep 25, 2018 5:25 am



Forum locked This topic is locked, you cannot edit posts or make further replies.  [ 34 posts ]  Go to page 1, 2  Next
Author Message
 Post subject: @abcedarian on the web
PostPosted: Wed Jan 21, 2015 2:54 pm 
Offline
User avatar

Joined: Wed Jan 21, 2015 1:53 pm
Posts: 49
Location: WxSW U.S.
@abcedarian, who seems to use the same handle everywhere he posts, made a nice catch last month and posted about it on the rusefi.com forum:
http://rusefi.com/forum/viewtopic.php?f ... =40#p12886

@abcedarian even offered to give some away for free:
http://rusefi.com/forum/viewtopic.php?f ... =50#p13273

@kb1gtt begged off stating that he was too busy with life to make any use of it but eventually it dawned on @russian that he was offered something for free and didn't respond appropriately:
http://rusefi.com/forum/viewtopic.php?f ... 747#p13747

Had Freescale offered something like that for the MPC563x o5e might have turned out differently. Member @murpia on the old forum suggested making a dev board like that using either the PWR40 or MPC5644a but it wasn't well received. Meanwhile, @russian and a few others have been kicking around the idea of making their own dev board using the 176-pin version of the STM32F4 non-automotive, non-industrial grade MCU.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Wed Jan 21, 2015 9:01 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
Moved this to Backyard Chat, 'cause I think that's really more like where it belongs.

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Wed Jan 21, 2015 9:22 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
US56 wrote:
@abcedarian, who seems to use the same handle everywhere he posts, made a nice catch last month and posted about it on the rusefi.com forum:
http://rusefi.com/forum/viewtopic.php?f ... =40#p12886

@abcedarian even offered to give some away for free:
http://rusefi.com/forum/viewtopic.php?f ... =50#p13273

@kb1gtt begged off stating that he was too busy with life to make any use of it but eventually it dawned on @russian that he was offered something for free and didn't respond appropriately:
http://rusefi.com/forum/viewtopic.php?f ... 747#p13747

Yes, that's me. I suggested the TMS570 since it's ARM based, automotive qualified and has a peripheral similar to the eTPU, which could be leveraged to free up main processor cycles for other uses. Worth mentioning is that the N2HET has hardware angle generation functions so once set up, it could watch the crank position and automagically generate an event such as fire ignition coils or injectors autonomously, but I'm not going to go too far into how it does that, here.

The offer to provide two boards to them was passed on, initially, by both russian and Jared, until russian recently requested if the offer was still open. I understand his conundrum because switching boards from the F4 Discovery to the TI TMS570 would make the hardware they've developed mostly useless for anything other than wild-wire prototyping... not that an adapter board couldn't be designed, but would it be worth it?

US56 wrote:
Had Freescale offered something like that for the MPC563x o5e might have turned out differently. Member @murpia on the old forum suggested making a dev board like that using either the PWR40 or MPC5644a but it wasn't well received. Meanwhile, @russian and a few others have been kicking around the idea of making their own dev board using the 176-pin version of the STM32F4 non-automotive, non-industrial grade MCU.

Well, one could hope that with the success of the Kinetis processors Freescale has released, they might consider doing something low-cost MPC56xx based, but I wouldn't hold my breath, since even the TRK board doesn't have much out there on the Internet regarding projects, save for the O5E stuff, and it seems even within Freescale the fervor that once surrounded that chip has quelled substantially.


Either way, I'm not in any sort of hurry as I have other projects going on, one of which is getting my motorcycle back together so I keep my eyes peeled and when I see something going on, I jump in with my 2 cents, to see what might be a consideration for my bike.

*cheers*

*edit to fix "quote" tags.

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Thu Jan 22, 2015 9:00 am 
Offline
User avatar

Joined: Sat May 11, 2013 9:45 am
Posts: 729
Location: PA, USA
US56 wrote:
Had Freescale offered something like that for the MPC563x o5e might have turned out differently. Member @murpia on the old forum suggested making a dev board like that using either the PWR40 or MPC5644a but it wasn't well received. Meanwhile, @russian and a few others have been kicking around the idea of making their own dev board using the 176-pin version of the STM32F4 non-automotive, non-industrial grade MCU.



o5e has a lot of problems, mostly me but also the fact that a diverse group has diverse priorities. Bouncing around between priorities is the very best way I know to never finish anything and that is the life of o5e. Once I finish up my engine project this summer I'll pick o5e back up and see, still with the 5634 because bouncing HW platforms is bad.

I'm not all that sure how relevant o5e is anymore though honestly. The addition of the x-gate to the S12 line has given that family new legs and as Russian has shown normal modern processors have the power to do timing close enough for engine use.

We'll see when I have more time available.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Wed Jan 28, 2015 4:13 pm 
Offline
User avatar

Joined: Wed Jan 21, 2015 1:53 pm
Posts: 49
Location: WxSW U.S.
The LAUNCHPADXL2 is probably the most bang for the buck for any dev board out there right now. The RM46x version (LAUNCHPADXL2-RM46) has 100base-TX Ethernet and 2x USB OHCI (host ports) so would be better for tinkering.

There is a more obscure TMS570LS series, the TMS570LS10106 thru 20216 which are more on the scale of rusEFI. Not so big a step up as the TMS570LS1224. Unfortunately, the only dev board I could find for those is the Olimex TMS570-CAN with a TMS570LS20216. It does not use Arduino style stacking headers but could be easily hacked to stack. The processor on this board is comparable in power to the e200z6 used on the MPC5554 (200+ CoreMark), more powerful than the STM32F4xx in every way and it's automotive grade.

The big dogs of the TI Hercules family are the TMS570LC4357 and RM57. These use the ARM Cortex R5F processor. The dual cores can be run independently (like the Freescale multicore MCU) which puts those in the same class, at least in terms of CPU power, as the MPC5676R and well positioned against future Freescale MPC57xx. Interestingly TI does not market any of their automotive MCU for the engine control application. I still prefer the Freescale ecosystem for engine control but have no doubt the TI processors can do the job.

mk e wrote:
The addition of the x-gate to the S12 line has given that family new legs and as Russian has shown normal modern processors have the power to do timing close enough for engine use.

The S12P is a great little processor. It can do a great job running a single or twin. I'd have a lot more interest in the Microsquirt if they'd just upgrade the MCU to an S12P. The S12XE is another very interesting product but the XGATE is an I/O coprocessor. It doesn't raise the S12XE to the capability of the MPC56xx (it's comparable to the TMS570LS10K/20K series mentioned above). Don't see great reviews for the MS3 Pro. Doing a quick google turned up some surprisingly negative comments. (However, if anyone out there has one and would be willing to post high res pic's of both sides of the PCB, please do). When I read that the only hardware timer used by rusEFI was the one used by ChibiOS it honestly made me wonder. Why do that when you have so much more power built into the MCU? Seems more like not wanting to deal with the details of the hardware although it does make it very portable.

mk e wrote:
I'm not all that sure how relevant o5e is anymore though honestly

Of course, that's the real question. The Freescale Monaco/MSQorivva fail was a big letdown for a lot of people and it's not clear that there is now a community of capable people who would be willing to invest the time and energy in developing pro-grade DIY engine control. Certainly, some consideration must be given to the so-called "free"/"open" development model.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Wed Jan 28, 2015 7:02 pm 
Offline
User avatar

Joined: Wed Jan 21, 2015 1:53 pm
Posts: 49
Location: WxSW U.S.
The LAUNCHPADXL2 is probably the most bang for the buck for any dev board out there right now. The RM46x version (LAUNCHPADXL2-RM46) has 100base-TX Ethernet and 2x USB OHCI (host ports) so would be better for tinkering. Same low price.

There is a more obscure TMS570LS series, the TMS570LS10106 thru 20216 which are more on the scale of rusEFI. Not so big a step up as the TMS570LS1224. Unfortunately, the only dev board I could find is the Olimex TMS570-CAN w/TMS570LS20216. It doesn't use Arduino style stacking headers but could be easily hacked to stack. The processor on the board is comparable in power to the e200z6 used on the MPC5554 (200+ CoreMark), more powerful than the STM32F4xx in every way and it's automotive grade.

The big dogs of the TI Hercules family are the TMS570LC4357 and RM57. Those use the ARM Cortex R5F processor. The dual cores can be run independently like the Freescale multicore MCU which puts those in the same class, at least in terms of CPU power, as the MPC5676R and well positioned against future Freescale MPC57xx. Interestingly TI does not market any of their automotive MCU for the engine control application. I still prefer the Freescale ecosystem for engine control but have very little doubt the TI processors can do the job.

mk e wrote:
The addition of the x-gate to the S12 line has given that family new legs and as Russian has shown normal modern processors have the power to do timing close enough for engine use.

The S12P is a great little processor. It can do a great job running a single or twin. I'd have a lot more interest in the Microsquirt module if they'd just upgrade the MCU to an S12P. The XGATE is a switched fabric and I/O coprocessor in place of the plain old microprocessor bus on other S12 products. It doesn't raise the S12XE to the capability of the MPC56xx (it's comparable to the TMS570LS10K/20K series mentioned above). Don't see rave reviews for the MS3 Pro. Doing a quick searcj turned up some surprisingly derogatory comments. (However, if anyone has one and would willing to post high res pic's of both sides of the PCB, please do). When I read that the only hardware timer used by rusEFI was the one used by ChibiOS it honestly made me wonder. Why do that when you have so much more power built into the MCU? Seems more like not wanting to deal with the details of the hardware although it's much more portable.

mk e wrote:
I'm not all that sure how relevant o5e is anymore though honestly

That's a good question. The Freescale Monaco/MSQorivva fail was a big letdown for a lot of people and it's not clear that there's a community of capable people willing to invest the time and energy in developing pro-grade DIY engine control using the "free"/"open".development model only to see their work hijacked by Chinese cloners and random free-riders.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Sat Jan 31, 2015 1:59 am 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
Actually, the TMS570LC4357 cannot run its cores independently; to date, none of the "Hercules" TMS570 / RM4/5 processors can.

"The safety architecture includes the dual-core lockstep CPUs with ECC-protected caches and dual VIMs in lockstep."
"The TMS570LC4357 device integrates two ARM Cortex-R5F floating-point CPUs, operating in lockstep, which offer an efficient 1.66 DMIPS/MHz, and can run up to 300 MHz providing up to 498 DMIPS."

The key word there being "lockstep". But it is worth noting that DMIPS rating. ;)

How they work is:
Code:
/-- C1 -> 2 cycle delay --\                         /-- match, proceed
> = CODE                   >- core compare module -<
\-- 2 cycle delay -> C2 --/                         \-- do not match, throw an error, maybe re-execute the code in question


The cores run the same code, one core runs two cycles ahead of the other, and the cores are physically oriented differently so as to minimize affects. The core compare module compares the outputs of both cores, which given the 2 cycle delays does reduce throughput somewhat, but does provide some ability to detect processing errors and such.

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Sat Jan 31, 2015 3:21 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
US56 wrote:
The LAUNCHPADXL2 is probably the most bang for the buck for any dev board out there right now. The RM46x version (LAUNCHPADXL2-RM46) has 100base-TX Ethernet and 2x USB OHCI (host ports) so would be better for tinkering. Same low price.
I want to add to this that the RM46x LP board's MCU (RM46L852) supports Ethernet MAC and 2 USB host / 1 USB device, which are multiplexed with other functions, but not supported with Ethernet PHY or USB hardware on the board. It is otherwise pin / function compatible with the TMS570LS1224 used in the other LP board, albeit with a reduced temperature rating, and the two LP's actually use the same board.

The TMS570LS0432 used on the lower end Hercules LaunchPad would likely run an engine well if its N2HET peripheral was leveraged properly. Given a few bits of data, it could toggle fuel injectors and ignition coils autonomously- it is actually a limited function microcontroller itself. The LS1224 and others have 2 N2HET modules which further increase the utility of the chip. It is a shame though that TI doesn't have full documentation for this peripheral, only having produced "draft" documentation for things over the last few years, and no real sample / example code.

Unfortunately, I'm still not at the properly level of experience to pull this off.
:oops:

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Sun Feb 01, 2015 10:35 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:45 am
Posts: 729
Location: PA, USA
US56 wrote:
Don't see rave reviews for the MS3 Pro. Doing a quick searcj turned up some surprisingly derogatory comments. (However, if anyone has one and would willing to post high res pic's of both sides of the PCB, please do).


I have never known anyone one who used MS stuff, but I've spent some time over on speedtalk recently and there are a few guys over there with nothing but praise for the Ms3 Pro and a good deal of love of MS2 &3 standard too.

Personally I suspect that since most US V8 race stuff is still carb anything seems better and they seem to curse tables larger than 12x12 as wasteful when I see 24x24 as a bare minimum.....but to be fair the MS stuff has come a long way in the past few years.

I don't know......I need to finish my engine then get back to ecu land


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Thu Feb 05, 2015 3:32 pm 
Offline
User avatar

Joined: Wed Jan 21, 2015 1:53 pm
Posts: 49
Location: WxSW U.S.
abecedarian wrote:
Actually, the TMS570LC4357 cannot run its cores independently; to date, none of the "Hercules" TMS570 / RM4/5 processors can.

Are you sure about that?


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Thu Feb 05, 2015 7:02 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
US56 wrote:
abecedarian wrote:
Actually, the TMS570LC4357 cannot run its cores independently; to date, none of the "Hercules" TMS570 / RM4/5 processors can.

Are you sure about that?
99.9% positive.
From the LC4357 device page / datasheet:
Quote:
The TMS570LC4357 device is a high-performance automotive-grade microcontroller family for safety systems. The safety architecture includes the dual-core lockstep CPUs with ECC-protected caches and dual VIMs in lockstep. The device has Built-In Self Test (BIST) logic for the CPU, the N2HET coprocessors, and for on-chip SRAMs. The device supports ECC protection on the L1 caches, L2 flash, and SRAM memories. The device also supports ECC or parity protection on peripheral memories and loopback capability on peripheral I/Os.

The TMS570LC4357 device integrates two ARM Cortex-R5F floating-point CPUs, operating in lockstep, which offer an efficient 1.66 DMIPS/MHz, and can run up to 300 MHz providing up to 498 DMIPS. The device supports the big-endian [BE32] format....


From the technical reference, section 2.1.2/table 2-1, page 112:
Quote:
Lockstep - This is the mode of operation of the dual ARM Cortex-R5F CPUs. The outputs of the two CPUs are compared on each CPU clock cycle. Any miscompare is flagged as an error of the highest severity level. In addition to the lockstep CPUs, the two Vector Interrupt Module (VIM) are also in lockstep.

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Mon Feb 09, 2015 10:18 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
It's always nice to see the chip vendors take notice of what's going on.

A TI employee happened to visit and comment on the project over on that other one, and even offering a bit of kit to help; won't link to it since that'd be disrespectful.

Hey, Freescale!!! HINT HINT HINT!!!

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Tue Feb 10, 2015 2:46 pm 
Offline
User avatar

Joined: Wed Jan 21, 2015 1:53 pm
Posts: 49
Location: WxSW U.S.
abecedarian wrote:
99.9% positive.

Good that you left yourself 0.1% wiggle room.

abecedarian wrote:
... won't link to it since that'd be disrespectful

Really? Why? It would be good if there was a forum for open, frank, yet civil discussion of this field of interest. Checked the forum to which you were presumably referring and saw nothing about a TI guy offering anything.

Funny that @russian seems to have given in to the attractions of the TMS570 but some diehards are tossing up tin-plated flak to derail the idea. Migrating to something like the TMS570LS1224 (or even the very nice LS1114) is a real stretch for their project but it seems like @russian now wants to make a real ECU. He also seems to believe for some reason that the tool chain for the MPC56xx isn't free. There are issues but those seem to be comparable to the situation with the TMS570.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Tue Feb 10, 2015 4:15 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
US56 wrote:
abecedarian wrote:
99.9% positive.

Good that you left yourself 0.1% wiggle room.

Everything I've read suggests I should be 100%, but that doesn't mean there isn't some way I'm unaware of to run the cores separately.

US56 wrote:
abecedarian wrote:
... won't link to it since that'd be disrespectful

Really? Why? It would be good if there was a forum for open, frank, yet civil discussion of this field of interest. Checked the forum to which you were presumably referring and saw nothing about a TI guy offering anything.

Okay, here's a link to the post.
"Hi, full disclosure - I'm a TI employee working on Hercules.
...
Obviously can't make this a standing offer everyone who reads this (I'd go broke if I wind up paying out of pocket ;) but for 2 boards you can send me a direct mail w. your address etc.
http://e2e.ti.com/members/33558 should link to my E2E profile...."
His credentials check out. ;)
The boards he's referring to are TI [/url=http://www.ti.com/tool/LAUNCHXL2-TMS57012]LAUNCHXL2-TMS57012[/url] kits.

US56 wrote:
Funny that @russian seems to have given in to the attractions of the TMS570 but some diehards are tossing up tin-plated flak to derail the idea. Migrating to something like the TMS570LS1224 (or even the very nice LS1114) is a real stretch for their project but it seems like @russian now wants to make a real ECU. He also seems to believe for some reason that the tool chain for the MPC56xx isn't free. There are issues but those seem to be comparable to the situation with the TMS570.

Yes, it is unfortunate so many are against the idea of using a more appropriate chip.
I think, however, the push back is because of the investment some people have put into what they have, which is understandable. Russian has mentioned he may be open to the idea current hardware could continue in development.

Re: LS1114- would probably work since the peripheral set and pinout are identical to the LS1224, but it is 256K less Flash and 64K less RAM.

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Wed Feb 11, 2015 3:11 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:45 am
Posts: 729
Location: PA, USA
US56 wrote:

abecedarian wrote:
... won't link to it since that'd be disrespectful

Really? Why? It would be good if there was a forum for open, frank, yet civil discussion of this field of interest.


link away, I agree discussion is fine.

now for chip swaps...its stuff like that that kills projects the fastest. That was the heart of the rift between Jon, Jarred and I, I simply have no tolerance for alternate HW and custom designs because it's a massive distraction and complication that derails everyone.

This continues to be the issue I have with this and all open projects....I want you're help designing something custom for my personal car.....f*** that! It has to be a take it or leave it deal but it turns out only peopel who want custom seem to care about open design to begin with so it turns into a cat herding exercise.


...end rant.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Thu Feb 12, 2015 3:25 pm 
Offline
User avatar

Joined: Wed Jan 21, 2015 1:53 pm
Posts: 49
Location: WxSW U.S.
abecedarian wrote:
Okay, here's a link to the post.
"Hi, full disclosure - I'm a TI employee working on Hercules ...

Thanks. I looked at that thread but didn't back up to the previous page. Thought it was breaking news. Getting two dev boards gratis isn't a big deal (of course, one should go to @abcedarian for his persistence in pursuit of the TMS570LS). I planned to get one of each flavor, LS1224 and RM46,just to see the difference. The big deal is having a potential friend on the inside who obviously knows the product. That's golden.

There are several interesting tidbits revealed which I doubt are publicly documented. After reading about the issues with getting the HWAG to play nice with the NHET, I looked at the debug capabilities of the NHET and it's not bad. Very basic but probably good enough and it's integrated into CCS. Looks like it will work with the relatively cheap XDS100/110 so it's a much different situation than with the eTPU. For the eTPU, Freescale now has a separate Eclipse based IDE (6-month free trial available) which only works with an expensive Lauterbach debugger. If the HWAG at present only likes a 60-2 wheel that could be a problem on a smaller engine. Seems like it shouldn't be difficult to get a 30-1 to work.

In their responses, it was dismaying to read that the key s/w guy admits, in honesty, that he doesn't know what a JTAG debugger does. The supposed key h/w guy apparently doesn't understand what an FPU does. Could be that the capabilities of the TMS570LS are wasted on them as is much of the capability of the STM32F4.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Thu Feb 12, 2015 5:09 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
US56 wrote:
abecedarian wrote:
Okay, here's a link to the post.
"Hi, full disclosure - I'm a TI employee working on Hercules ...

Thanks. I looked at that thread but didn't back up to the previous page. Thought it was breaking news. Getting two dev boards gratis isn't a big deal (of course, one should go to @abcedarian for his persistence in pursuit of the TMS570LS). I planned to get one of each flavor, LS1224 and RM46,just to see the difference. The big deal is having a potential friend on the inside who obviously knows the product. That's golden.

The big news to me was (1) a person from TI, and (2) was willing to pay out of pocket if necessary to supply two boards to the project. And I agree having a contact on the inside could be a huge benefit. And thanks for thinking of me but I already have one each 1224 and RM46x. FWIW back in December I offered to supply two boards myself- one for the s/w guy, one for the h/w guy- and they were passed over.

The boards themselves are physically identical down to the silk: the schematics and (Eagle) board files provided by TI are for the RM46x with notes that the TMS570 is the same. Nice little touch TI put on them was the a footprint / pads for a barrel jack to accept 5v from. The processors are mostly the same with a reduced temperature range on the RM board and different functional safety qualifications, and IIRC the RM will run to 180-200MHz instead of 160 - likely due to the reduced temp range.

Quote:
There are several interesting tidbits revealed which I doubt are publicly documented. After reading about the issues with getting the HWAG to play nice with the NHET, I looked at the debug capabilities of the NHET and it's not bad. Very basic but probably good enough and it's integrated into CCS. Looks like it will work with the relatively cheap XDS100/110 so it's a much different situation than with the eTPU. For the eTPU, Freescale now has a separate Eclipse based IDE (6-month free trial available) which only works with an expensive Lauterbach debugger. If the HWAG at present only likes a 60-2 wheel that could be a problem on a smaller engine. Seems like it shouldn't be difficult to get a 30-1 to work.
While you can do N2HET work in CCS, the N2HET code is a struct passed to the N2HET, which consists of native code for the module so you'd need a good understanding of the module and its op-codes to do so. TI has HET IDE which provides a means to write and debug NHET/N2HET code before incorporating it into a larger project in CCS. A simulation engine and visualization tools are also included.

I think the 60-2 was chosen as an example since it is a common wheel. My interpretation of things is that any wheel with a missing tooth (or multiple adjacent missing) would work; non-adjacent missing teeth would be a problem they haven't really looked too closely at. This topic may be interesting to some in particular the PDF's in the first reply: http://e2e.ti.com/support/microcontroll ... 2/t/237688

Quote:
In their responses, it was dismaying to read that the key s/w guy admits, in honesty, that he doesn't know what a JTAG debugger does. The supposed key h/w guy apparently doesn't understand what an FPU does. Could be that the capabilities of the TMS570LS are wasted on them as is much of the capability of the STM32F4.

Yeah, a little dismayed myself reading those posts. I'm not particularly well versed in hardware or software but I knew at least in concept what those were.

A lot of what's been done could be leveraged against for supporting the TMS. Working hardware designs would need new board layout; many of the firmware ideas could port over as well since both proc's are ARMv7 based. Design files and firmware for the JTAG emulator are available, so the board could incorporate those, and the emulators provide a USB serial back-channel so USB com would be built in.

I have a few thoughts on both hardware and firmware for this but until I see some serious motion with it, I'm not diving too deep.

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Thu Feb 12, 2015 8:58 pm 
Offline
User avatar

Joined: Mon Dec 23, 2013 9:55 pm
Posts: 3
Quote:
For the eTPU, Freescale now has a separate Eclipse based IDE (6-month free trial available) which only works with an expensive Lauterbach debugger.

Disclaimer: I am not a software person.

Is it the cost of the development tools what mainly drives people to look at other microcontrollers?

Since ST second-sources the 5634, has anybody taken a look at their free (I think) development tool stuff?

http://www.st.com/web/catalog/tools/FM1 ... 1/PF255702

Thanks


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Thu Feb 12, 2015 10:16 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
Hadn't considered ST had their own tools.

Anyhow, looks like ST is the first source now. Go to the FS page for the MPC5634M and click buy or sample and tell me what the part number they show you is.

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Fri Feb 13, 2015 12:32 am 
Offline
User avatar

Joined: Wed Jan 21, 2015 1:53 pm
Posts: 49
Location: WxSW U.S.
abecedarian wrote:
Hadn't considered ST had their own tools.

Have you forgotten user Nobody already? He chose the ST chip over FS for some reason which he explained at the time. Paid up for a pro toolchain and was disappointed.

abecedarian wrote:
Anyhow, looks like ST is the first source now. Go to the FS page for the MPC5634M and click buy or sample and tell me what the part number they show you is.

Don't panic. Didn't take the time to login to the FS site (would take too long to find my password) but the FS production part no. is SPC5634M followed by what is probably a revision code and then a package code. Check Octopart. Mouser has many versions. In any event, FS has a "Product Longevity Program" and most MPC56xx parts still have ten years or more guaranteed lifetime.

That reminds me though, I need to get a TRK board while those are still available. It's relatively nice-for-the-price.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Fri Feb 13, 2015 1:22 am 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
I was logged in. ;)
I have a few MPC33814, MM912JS812 and S9 Kinetis here if you want....

From what I gather with these PPC chips is...
MPC***** is Motorola / Freescale provided.
SPC***** is ST Micro provided.
- could be wrong.

...
And I have a TRK kit here....
Interested?

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Fri Feb 13, 2015 12:45 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:45 am
Posts: 729
Location: PA, USA
TonyS wrote:

Is it the cost of the development tools what mainly drives people to look at other microcontrollers?

Since ST second-sources the 5634, has anybody taken a look at their free (I think) development tool stuff?


Thanks


It seems mainly to be having to actually do the work that keeps people from the project honestly....seriously.

Then there is the "armchair quarterbacking" with you should have done this or picked that.....with not actually doing any work.

the 5634 is a great chip and the tools are free other than the eTPU, but working eTPU code is provided free and ready to go so I have no idea what the actual concern is with the 5634.....but I'm in no way married to other than that is the chip all the o5e code is setup for at this point.

Yes code can be ported from anything to anything but why? Projects will always fail when every decision is constantly revisited. There is no such thing as right and wrong on chip choice, its just what issues do you want to live with.

IF, switching to an ARM processor opens the door to a lot more people being interested in the project then I guess that makes sense, but I promise moving from a chip designed to be an ECU to one not designed to be an ECU will be new problems, I sure manageable but they will be there.

So what is it to be?


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Fri Feb 13, 2015 2:53 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
Well said Mark.



And I'm fairly much without immediate need for anything so can afford waiting for
something appropriate for my use, regardless the processor architecture. While waiting, I am not afraid to voice concerns regarding various things, mainly because I might see something that may not be suitable for my intentions, which Mark did mention. However, I hope those are constructive comments and willingly accept criticism in kind.

Mark wants to run 12 cylinders; I, two. Obviously there will be differences in opinion regarding hardware. If the firmware runs 12, two shouldn't be much work, but I can imagine the reverse wouldn't necessarily be true.


To add, the TMS570, though not specifically targeting ECU use, is marketed at automotive use with steering, braking, hybrid/electric and "off-road" listed as possible use cases. Obviously they won't get too specific, but often times a chip 'meant' for use as one thing finds a use in and grows from entirely unintended use... PIC being one example.

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Fri Feb 13, 2015 9:15 pm 
Offline
User avatar

Joined: Sat May 11, 2013 9:52 am
Posts: 304
Location: Over here, doing 'over here' things.
Seems Mr. Seely made good on his offer:
http://rusefi.com/forum/viewtopic.php?f ... 241#p14240

Also:
http://e2e.ti.com/support/microcontroll ... 07#1421507

_________________
/me goes off to the corner feeling like Jerry Springer with a mullet.

My O5E candidate: 1982 Honda CX500TC motorcycle.


Top
 Profile  
 
 Post subject: Re: @abcedarian on the web
PostPosted: Sat Feb 14, 2015 8:02 am 
Offline
User avatar

Joined: Sat May 11, 2013 9:45 am
Posts: 729
Location: PA, USA
I was exchanging a couple emails and thought I'd just post the gist of it here for all

For most people, engineers in particular, the hardest part of any project is picking a starting point. I can't even count the number of project I watch spin in circles trying to decide where to start. The chip is almost irrelevant once you cross a minimum performance threshold. I was on about 5554 then 5674.....but the cheap 5634 does everything most people need in a non-oem application and I'm sure the same is true for most of the ARM family. It just doesn't matter.....but you have to pick and the only way to finish is to....you guessed it, put your head down and FINISH.

The almost as hard step which actually SHOULD form first is deciding what it is you are trying to build. Nothing will ever be everything to everyone and you simply can't have it all.....so you need to be honest about what you need and how much time and money you have and set a realistic goal. You can always add more features later, but if you don't limit the scope to get started you will end up with NOTHING

These 2 things have plagued 05e. I'll very happily admit that I have been very strict about what the project is and what it is not and that has pissed off a lot of people....but there is simply no way to ever finish or for anyone to know what they should be helping with if there is no plan.

Designing your own custom this or that is not helpful to the project. You're certain always free to do it but then be definition you are not part of 05e (or whatever project) team unless you are working on common tasks. Paul had his own ideas but also wanted to be part of the team and balance the 2. Jon was able to balance the 2 in the beginning and contributed a LOT of help, but when we got to a point where his personal goals diverged from the team we just couldn't keep things in balance anymore. Sean has given us some good stuff but he too has other goals.

Then there are the vast majority of the people around the project who are simply waiting for the core system to be finished and tested...so they can use it to start their custom projects......which is fine, it's an open project but its going to take forever to the core to be finished if everyone believes it's ok to wait for someone else to build it......

The curent hold up the flash is limited to 32k. To get the optional outputs on-line the way they were planned that needs to be expanded to 64k. I'm told it's not hard to change but I don't know how to do it so here we sit. If come june/july when my car is done and I have time we're still sitting I'll kill the fancy optional output stuff and plug in a simple version that will work in the 32k limit and give it a try. My strong preference is that someone who understands embedded programming takes a look at the issue and helps...but one way or the other a frankenECU will be built and tested.


Top
 Profile  
 
Display posts from previous:  Sort by  
Forum locked This topic is locked, you cannot edit posts or make further replies.  [ 34 posts ]  Go to page 1, 2  Next

Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Theme designed by stylerbb.net © 2008
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
All times are UTC - 5 hours [ DST ]