Page 3 of 6

Re: Is stm32f4 the right chip?

Posted: Wed Apr 12, 2017 9:26 am
by kb1gtt
Is this chip supported by ChibiOS?

How different is the pin out from this board https://github.com/rusefi/rusefi/tree/master/hardware/brain_board_176-pin

Re: Is stm32f4 the right chip?

Posted: Wed Apr 12, 2017 11:44 am
by AndreyB
S32K is currently PRE-PRODUCTION, right?
evaluation board for S32K148 should be available in Q2, probably in May (it's still subject to change).
https://community.nxp.com/thread/446113

This would be an ideal chip for rusEfi!

But someone would need to develop ChibiOS drivers :)

Re: Is stm32f4 the right chip?

Posted: Wed Apr 12, 2017 10:18 pm
by AndreyB
kb1gtt wrote:Is this chip supported by ChibiOS?
No.

I do not understand differences between different implementations of Cortex-M4F precisely - maybe it would require zero changes in the kernel code?

But half of the ChibiOS fun is in the peripherals drivers - SPI, UART, ICU, USB. All this is completely different I assume.

Re: Is stm32f4 the right chip?

Posted: Thu Apr 13, 2017 3:10 am
by marko
Everyone licenses the core from ARM, the core is the same, uses the same machine instructions, etc.

All the peripherals taped to the core are home-grown and likely incompatible between manufacturers, so pretty substantial rewrite, since the core part is more of a gcc thing than a library thing, and the peripherals are all library stuff.

Re: Is stm32f4 the right chip?

Posted: Fri Apr 21, 2017 12:39 am
by AndreyB
S32K144EVB dev board is pretty availbale for $49

Actual chip PS32K144UAT0VLLA I do not see available

We probably want the more powerful sibling S32K148 which is not yet available in any form.

I am now considering a kickstarter to gather $750 needed to implement some basic drivers for ChibiOS

Re: Is stm32f4 the right chip?

Posted: Fri Apr 21, 2017 1:34 am
by AndreyB
russian wrote:Actual chip PS32K144UAT0VLLA I do not see available
Oh, this P in the beginning stands for PROTOTYPE. Seems a bit too early to jump on this wagon?

Re: Is stm32f4 the right chip?

Posted: Fri Apr 21, 2017 8:15 am
by Rhinoman
russian wrote:
russian wrote:Actual chip PS32K144UAT0VLLA I do not see available
Oh, this P in the beginning stands for PROTOTYPE. Seems a bit too early to jump on this wagon?
That's normal Freescale/NXP practice, they tend to keep the 'P' designation until they are in mass production. If it has had more than one revision its probably OK, an initial design I would steer clear of. There should be an errata available but I wouldn't rely on it being complete.

The issue will be availability and minimum order quantities, NXP still shows it as pre-production.

Re: Is stm32f4 the right chip?

Posted: Fri Apr 21, 2017 10:24 am
by kb1gtt
Is there risk that this will not make it to production? Does this chip offer features that will cause OEM's to use it? My fear is that this chip may be hard to source in the near-ish future.

The STM32 is general purpose and used in many many devices, so it is not likely to go away.

Re: Is stm32f4 the right chip?

Posted: Fri Apr 21, 2017 12:21 pm
by AndreyB
kb1gtt wrote:Is there risk that this will not make it to production? Does this chip offer features that will cause OEM's to use it? My fear is that this chip may be hard to source in the near-ish future.
I guess no one knows at this point? They call it "S32K: 32-bit Automotive General Purpose Microcontrollers"

I will wait for non-prototype revision to show up on digikey or mouser.

Re: Is stm32f4 the right chip?

Posted: Fri Apr 21, 2017 12:41 pm
by Rhinoman
kb1gtt wrote:Is there risk that this will not make it to production? Does this chip offer features that will cause OEM's to use it?
They don't make a chip unless there is a market for it, it will have been developed for some OEM purpose with a high enough volume to justify it. Whether it will become available in small quantities through the retail outlets is the issue, I doubt if they would answer that at this stage but they are pushing it widely so I would imagine so, we can only wait and see.

Re: Is stm32f4 the right chip?

Posted: Wed May 10, 2017 12:55 pm
by mobyfab
I really don't see the point of starting over with lack of drivers, community support, documentation etc just because it's 5v and "Automotive qualified".

If it had some kind of timing co-processor (or dual core) why not, but it's just another general purpose MCU.

Re: Is stm32f4 the right chip?

Posted: Fri Jan 26, 2018 10:02 pm
by AndreyB
S32K update: not really available yet.

S32K144EVB evaluation board mouser Estimated Ship Date 22-Mar-18
https://www.nxp.com/products/processors-and-microcontrollers/arm-based-processors-and-mcus/s32-automotive-platform/32-bit-automotive-general-purpose-microcontrollers:S32K?tab=Buy_Parametric_Tab S32K family currently lists only two active part numbers FS32K144HFT0VLLT and FS32K144HNT0CLLT but only FS32K144HFT0VLLT is actually in stock today.

Re: Is stm32f4 the right chip?

Posted: Tue Jan 01, 2019 5:59 pm
by AndreyB

Re: Is stm32f4 the right chip?

Posted: Tue Jan 01, 2019 11:24 pm
by AndreyB
@ has pointed to the limited license "Free TriCore™ Entry Tool Chain" https://free-entry-toolchain.hightec-rt.com/ as long as "Useable for TriCore evaluation boards with on-board wiggler only, PCP assembler only, Debugger: No visualization functions at runtime, no MCDS support, no Script support."

wiggler?! What is wiggler?

Re: Is stm32f4 the right chip?

Posted: Wed Jan 02, 2019 12:17 am
by kb1gtt
I recall wiggler being a JTAG programming cable. From some old memory, I understood it to be a very basic cable for your parallel port, with software bit banging. No IRQ's and almost no special hardware. Perhaps this is that same wiggler concept.

Re: Is stm32f4 the right chip?

Posted: Wed Jan 02, 2019 11:41 pm
by AndreyB
russian wrote:
Tue Jan 01, 2019 11:24 pm
@ has pointed to the limited license "Free TriCore™ Entry Tool Chain" as long as "Useable for TriCore evaluation boards with on-board wiggler only, PCP assembler only, Debugger: No visualization functions at runtime, no MCDS support, no Script support."

wiggler?! What is wiggler?
Related to this - Hitex ShieldBuddyTC275 dev boards based on https://en.wikipedia.org/wiki/Infineon_AURIX
processor is running at 200MHz and hidden inside the LQFP package there are in fact three of them, along with 4MB of FLASH, 128kb of data flash and 500k of RAM.

Board is about $100, seems to only be in stock in Europe. They have arduino IDE so them seem to have some HAL library? But no free compiler anywhere in sight.

Re: Is stm32f4 the right chip?

Posted: Thu Jan 03, 2019 6:46 am
by andreika
russian wrote:
Wed Jan 02, 2019 11:41 pm
But no free compiler anywhere in sight.
Maybe all is not lost with that Sh*tBuddy? :)
https://github.com/TriDis/llvm-tricore

Re: Is stm32f4 the right chip?

Posted: Thu Jan 03, 2019 10:24 am
by AndreyB
andreika wrote:
Thu Jan 03, 2019 6:46 am
https://github.com/TriDis/llvm-tricore
but Updated on Feb 4, 2017 :(

Re: Is stm32f4 the right chip?

Posted: Thu Jan 03, 2019 11:12 am
by AndreyB
How comes we've never mentioned NXP Kinetis on this forum?

KE1xF is their strongest chip - it's same Cortex-M4 Core for Toolchain and Performance, and ChibiOS community (not "official") support seem to having some life at http://www.chibios.com/forum/viewforum.php?f=26

168 MHz Cortex-M4 w/ FPU
256 – 512 KB Flash
32 – 64 KB SRAM

Only thing I do not see so far is 144 or 176 pin package and well (only 64-LQFP, 100-LQFP), I have not really looked into it yet :)

There is also this weirdly-looking $100 TWR-KE18F dev board - https://www.digikey.com/product-detail/en/TWR-KE18F/TWR-KE18F-ND/6204874

Re: Is stm32f4 the right chip?

Posted: Thu Jan 03, 2019 11:31 am
by kb1gtt
I recall a discussion about the FreeScale (now NXP) kinetics. I guess there is a chance that was a discussion before my days at rusEFI. I recall it was expensive, you likely had to use there compilers and their tools, as well it's significantly large. The stack is quite large. It's mostly designed for robotics, not really designed for automotive. Also take note they don't use tiny little header pins. They use much larger edge connector.

Re: Is stm32f4 the right chip?

Posted: Thu Jan 03, 2019 11:38 am
by AndreyB
kb1gtt wrote:
Thu Jan 03, 2019 11:31 am
I recall a discussion about the FreeScale (now NXP) kinetics. I guess there is a chance that was a discussion before my days at rusEFI.
I guess there is a chance that things have changed? :) On this forum "kinetics" search returns nothing (only this thread now)

Re: Is stm32f4 the right chip?

Posted: Thu Jan 03, 2019 11:52 am
by kb1gtt
I recall it was Kinetis Tower System. I'm thinking that was the o5e days, and was likely before my days at rusEFI. So in my head, but not in this forum. They have kits which include the edge connectors, and similar supporting equipment.
https://www.digikey.com/product-detail/en/nxp-usa-inc/TWR-P1025-KIT/TWR-P1025-KIT-ND/3594924

With a starting price around $500, it's a kind of pricey option. However from a development stand point, it gets you a prototype quickly.

Re: Is stm32f4 the right chip?

Posted: Thu Jan 03, 2019 4:48 pm
by mobyfab
There's no point in evaluating Kinetis chips for now, since they are pretty much the same thing minus driver/OS support.
Just like most other Cortex-M mcu.

For the next few years/decade the STM32 range is perfectly suited for this job.

just my $0.02

Re: Is stm32f4 the right chip?

Posted: Thu Jan 03, 2019 5:05 pm
by kb1gtt
The connector boards might be a handy reference for the entire Manhattan style theory.

Re: Is stm32f4 the right chip?

Posted: Sun Jan 13, 2019 7:25 pm
by Simon@FutureProof
I am inclined to agree with @mobyfab, the only downsides to the STM32 seem to be that it is not "automotive" and that more timers would be nice.

The lack of 5v input is only a small limitation really as we deal with that on the rusEFI PCB.

While the STM32 is not automotive grade are we actually seeing any issue because of that? has there been any failures of the STM32 boards since the project began in 2013?

My experience has been that most of the aftermarket ECU tuning equipment is not using automotive grade parts in some area of their system, I can only assume that a lot of these parts are perfectly up to the job but are just not certified as "automotive" due to cost and a lack of need for the original intended application.
The OEM ECU manufacturers need to have practically zero in service failure rates thanks to market expectations and they have the muscle to get the chip manufacturers to provide things in "automotive" grade to their specifications. (thinking of Bosch who have pretty common simple ICs custom branded with Bosch numbers and no other ID markings from Harris, ST, Phillips etc etc)

The STMF4 also allows us to easily move to a more powerful chip easily with few modifications to the hardware, given there are lots of things still in development a complex migration to a new chip would seem counter productive?

Re: Is stm32f4 the right chip?

Posted: Mon Apr 01, 2019 5:35 pm
by AndreyB
russian wrote:
Fri Apr 21, 2017 12:39 am
S32K144EVB dev board is pretty availbale for $49

Actual chip PS32K144UAT0VLLA I do not see available

We probably want the more powerful sibling S32K148 which is not yet available in any form.
S32K148 is now available on Mouser - see https://github.com/rusefi/rusefi/issues/590

Infineon Aurix tricore tc275

Posted: Tue Jun 25, 2019 1:08 pm
by juanj
Hi :D ,what do you thniks about to port to the Aurix tricore 200mhz with generic timer module GTM for crank angle and free tool chain?
https://www.hitex.com/microcontroller-support/aurix/shieldbuddy-tc275/

Is this possible?

Re: Is stm32f4 the right chip?

Posted: Tue Jun 25, 2019 1:13 pm
by AndreyB
Please start working on that port and we will help if you have any questions.

I am not aware of any software developers with time for this large commitment, but if you are a developer or if you can hire developers, please start.

Re: Is stm32f4 the right chip?

Posted: Tue Jun 25, 2019 1:46 pm
by kb1gtt
I understand two large obstacles with the tri-core is the difference in how DMA is handled, as well as a lack of support with ChibiOS. Currently ChibiOS handles the low level registers and similar stuff. To use the Tri-Core would require allot of low level driver development. I understand the ChibiOS folks can be hired to develop drivers like this.

Re: Is stm32f4 the right chip?

Posted: Sat Jul 20, 2019 12:53 am
by faisal
russian wrote:
Tue Jun 25, 2019 1:13 pm
Please start working on that port and we will help if you have any questions.

I am not aware of any software developers with time for this large commitment, but if you are a developer or if you can hire developers, please start.
If the vendor already has some peripheral driver code, you can take that as is, and write a thin wrapper to make the API conform with ChibiOS/HAL .

I've done this with a proprietary architecture before for which a "HAL" already existed. I wrote set of hal_<periph>_lld.c/.h files which called out directly to the existing proprietary HAL functions. It was relatively painless, and gave me further confidence in the elegance of the ChibiOS/HAL design.

Once you have things functional, you can take your time and iterate on the low level driver wrapper, and optimize for whatever axis is important to you.