• Locked Locked
  • Replies 7 replies
  • Subscribers 31 subscribers
  • Views 361 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

What's up with the M3?

mike fontes
Expert 1385 points
Other Parts Discussed in Thread: LM3S811

I just got the letter about the TIVA parts.  It looks to me like the M4 parts were renamed, but that is about it.  I am very disappointed for several reasons:

1. There seems to be no plans to handle M3 parts in a reasonable fashion.  I have seen many posts complaining about the drop the M3 strategy, with no replacements. I have noticed it has gotten harder to get the documents for the M3 chips.  I have also noticed the TI E2E forums turning into a place where engineers are recommending competition chips to replace those M3 parts that people were using! That was actually kind of funny!

2. For engineers like me who have been using TI parts for well over 30 years the biggest hit was that TI has made no effort to fill in the new TIVA or M4 lines with components over the whole product spectrum.  I was using low end, low pin count parts to handle some mundane, but necessary tasks.  With the new line the smallest package size is 64 pins, when I could actually do with a dual UART chips with 14 pins if it were available.  Not everyone needs every interface.

3. I find it amazing to see that with each new version of the chips that TI is keeping the performance limited to 80 Mhz max.  That may have been good 3 years ago, but it is not a good omen for most of us.  We usually try to do more, in less time, in less space, and these new designs look more like anchors when the competition is continually improving performance with each new model.

I would like to get this to someone at TI who actually makes some decisions about what is coming up next in the TIVA world.  At this point I am pretty disgusted with TI's lack of leadership in a market they are capable of dominating with a few logical product placement decisions.  I would like to see a more complete top to bottom offering of choices in peripherals, with top speeds in the 150 - 250 Mhz range, in packages from 24 pin to whatever you think you need for every peripheral known to man.

Sincerely,

Mike Fontes

  • TI__Expert 7165 points

    Mike,

    Thank you for your feedback on LM3S. We are working hard to fill in any gaps left by the NRND of the LM3S devices - including smaller packages.

    We will keep this forum updated on any public plans for new Cortex MCUs from TI.

  • Intellectual 550 points

    Well put, Mike, on all points.  I couldn't agree more.  The things I could do with a 24 pin Cortex M3...

  • in reply to Dustin Carr
    Expert 1385 points

    Dustin,

    I had planned on a global communications coprocessor on one of my new products where I had a single piece of code in an M3 on each of my external systems on a long bus. The M3 would monitor bus conversations, provide any non- computational bus replies, and only send information to each card's main processor that it was required to act upon. That way I could have one piece of code, that I could use repeatedly, and my daughter cards could use the processor of choice depending on their functional needs,  I could get away with a very small footprint on my daughter cards for this function. I was expecting the size of a large postage stamp plus a little for the required circuitry.

    The loss of the small footprint parts, and simple functionality that is caused by the change to the M4 has left me not able to afford the increased board space, so I now have to come up with a different solution.  The worst part is that I had already written a bunch of the code using an LM3S811 kit, so most of my work was done.

    Maybe if enough of us express our functional needs, or complain, to the Gods of Processor Marketing at TI they will understand that they are losing very good product placements by going with 50 versions of basically the same large footprint and not considering the small stuff.  In my case each system had an average of about 10 - 12 communications coprocessors in it so there would have been a pretty reasonable volume.

    Sincerely,

    Mike Fontes

  • in reply to miguel2667
    Expert 1830 points

    miguel2667 said:

    Mike,

    Thank you for your feedback on LM3S. We are working hard to fill in any gaps left by the NRND of the LM3S devices - including smaller packages.

    We will keep this forum updated on any public plans for new Cortex MCUs from TI.

    Would it not have been better to fill in the gaps and then NRNDed the LM3S devices?

  • in reply to Vito Casa
    Expert 1385 points

    Vito,

    It may have been easier to keep the LM3S low pin parts available for new designs until the TIVA or M4 equivalent was available.  I have seen indications that some of the parts can still be bought, but they are basically gone for new designs.  I can't even get a datasheet without some magic now because the site has pretty much purged the availability of the documentation.  It doesn't bother me that I have a couple of kits for parts that are going nowhere, but only because TI actually supplied them to me.  The part that irks me is that I have test boards designed using the parts, and a fair amount of manpower invested, and it sure feels like TI just dropped me. 

    I would like to see a commitment from TI to fill the holes from the top to bottom on the M4 TIVA line that includes performance improvements to at least 200 Mhz, and fill in the low end holes before worrying about every option on the bloated high end of the device offerings.  Since TI usually makes the top performer in the line work first I am sure that is happening with the TIVA chips.  I would think a lower pin count part with less peripherals would be a slam dunk. Keep in mind that a lot of us are trying to get rid of a bunch of lower end PIC type processors, and rally around single design tool chains.  Having a different development tool chain for each processor is very expensive! That is why the CCS looked so good.  It is one tool for a group of processor line, cutting down the costs of keeping multiple tools.

    This is the second time in two years that this has happened.  It also happened after I spent a lot of time trying to sell management on the use of 3530 and other Beagle related processors and was told they were no longer going to be available to me because I don't buy huge quantities.  I am forced to go the low end route with Concertos, and to be honest, I think I am going to run out of processing horsepower because they are so slow.  It is a major concern of mine.  The dual core paradigm fit very well for the systems I am working with, but I really need speeds more in tune with 300 - 500 Mhz with expansion to Ghz as software requirements increase.  It seems the Concerto, although it is nice for lower end applications, is going to force a processor based redesign sooner than I would like. 

    Sincerely,

    Mike Fontes

  • in reply to mike fontes
    Intellectual 550 points

    This is all very tragic.  Developing for the Stellaris was the best experience I had ever had with any TI product.  For my company, TI is now NRND.  That includes all analog and digital.  Of course, right now we don't have volume products, so I don't think anybody cares.

  • in reply to mike fontes
    Guru 117855 points

    Mike Fontes said:
    a lot of us are trying to get rid of a bunch of lower end xxx processors

    Our group believes that the ARM M0 devices (supported by others - rejected here - thus far) best satisfy this requirement.  Maintain consistency w/M3/M4 - run on the same IDE (if you've chosen well) and come closest to meeting LMI's promised 1USD price point.   And - lowered pin-count devices have already appeared - while others are, "in the wings."

    Believe your points to be well considered and effectively presented.  Only qualm - "locking one's firm" into one vendor (a la CCS) prevents the selection of "best part - targeted for specific Ap" - which is unlikely to result from one vendor - at all times!  IAR (our choice) and Keil provide this "freedom of choice" - and have certain clear operational advantages over CCS.  (i.e. SWD - and great robustness...)