Cheap Stellaris LM4F120 LaunchPad Evaluation Board from TI

[metal]

Do you mean that the controller put in this kit is buggy?

[sn00p]

Do you mean that the controller put in this kit is buggy?

Almost certainly!

For example, here's the errata for the silicon of the current shipping revision of the controller we used in our design:

http://www.ti.com/lit/er/spmz639a/spmz639a.pdf

This controller is now marked as "ACTIVE", it's no longer in preview and the list of faults and problems in it are just staggering!

From experience the one thing you can guarantee is that they'll fix those problems in a new revision and at the same time break stuff that was previously working fine!

For a hobbyist, I'd say you can't complain at the price of that board!  But don't be surprised to spend hours trying to figure out why something is not behaving properly only to find out that it's a bug in the silicon.

In a professional scenario, avoid!!!!

[metal]

you make me feel they just gave away these chips because they are buggy. Even as a hobbyist, I can complain, not because of the price, but because of wasting my time, when I receive them, I will simply throw them away.

[Magnox]

Hmm... I'm regretting spotting this now. I thought I was helping people  

My order with TI is still showing 'processing' and payment has not been taken.

There is no 'cancel' option that I can find (that would be too easy) but it might be worth anyone emailing TI to cancel their order if they don't want to risk the dodgy chips.

I'm thinking of doing so myself... not decided yet. They are still cheap, but my time is valuable to me.

Another thought though - it's still a cheap programmer if it can be used as such...

[metal]

I wish I knew this sooner, I could have cancelled farnell order at least.. any way, what do you want to program with it?

[Magnox]

Hopefully other stellaris boards, using the debug interface built into the launchpad. If the chips are still as bad as sn00p says though, that might not be worth the effort.

[Gallymimu]

I dunno, it doesn't look like any longer list of errata for a new processor than anything I've used in the past.

Just as an example here's a PIC32 errata from one of their early parts in the MIPS architecture.

http://ww1.microchip.com/downloads/en/DeviceDoc/80440J.pdf (to this day only a handful of them have even been fixed after 3 years!)  GEEZ.

What processor lines do you guys like that have a very low introduction errata rate?  My experience has been anything new from Microchip has a nice chunk of silicon and documentation errors.

At this point I always worry about switching to new architectures due to the "devil you know" vs "devil you don't know" gotcha.

[sn00p]

True, the Sam7 was pretty good on the errata front from day one, nothing particularly bad.

Energy micro probably takes the award for best bug with a non functional trace port.

If it wasn't for the fact that the atmel line of processors (arm) became stagnant then I think we'd still be using those.

[Gallymimu]

True, the Sam7 was pretty good on the errata front from day one, nothing particularly bad.

Energy micro probably takes the award for best bug with a non functional trace port.

If it wasn't for the fact that the atmel line of processors (arm) became stagnant then I think we'd still be using those.

Interesting.  I've considered moving toward some ARM architechtures but I understand that opens a whole set of challenges in itself.  You are unhappy with the Atmel ARM offerings?  I'm actually interested in the TI Hercules series, but only because of medical device development.

[alexisnik]

sn00p I believe that you were quite unlucky, we work only with TI processors (MSP430, Stellaris) and we are able to work with them with no problems, at a professional level. The only real problem we encountered was with a buggy RTC on an MSP430 model. USB are quite problematic from reading the errata also.

For most hobbyists, they won't even notice the problems mentioned in the errata... That is just my opinion of course! I am just saying that you make it sound as if TI processors are crap, and I think you are scaring people much more than they should be scared. You had a bad experience and I can understand your frustration, but I think that what the Launchpad is used for (and at 5$), you get 10 times more than it is worth, without running into any problems.

BTW, I don't work for TI or anything...

[sn00p]

sn00p I believe that you were quite unlucky, we work only with TI processors (MSP430, Stellaris) and we are able to work with them with no problems, at a professional level. The only real problem we encountered was with a buggy RTC on an MSP430 model. USB are quite problematic from reading the errata also.

For most hobbyists, they won't even notice the problems mentioned in the errata... That is just my opinion of course! I am just saying that you make it sound as if TI processors are crap, and I think you are scaring people much more than they should be scared. You had a bad experience and I can understand your frustration, but I think that what the Launchpad is used for (and at 5$), you get 10 times more than it is worth, without running into any problems.

BTW, I don't work for TI or anything...

I do agree, hence why I did say that I don't think you can complain at the price of the board! 

However, with my professional cap on you'd still have to work hard to convince me to use another Stellaris in a product again, we just had too many issues with not just the devices, but the supply chain too (and I mean much more than the normal supply chain problems).

I will say that TIs support is for the most part excellent although they do have a habit of slipping into "politician" mode if you try asking detailed questions about specific issues.  Atmels support on the other hand was dire.

With regards to the Atmel product line, when we saw the SAM3 we were moderately impressed especially that they offered a migration path on some of the parts from the SAM7, they were pin compatible!  Fantastic!  Then we started reading the datasheets for them, the chips were far better in every respect apart from one.  They'd cut the amount of RAM from 64KB to 48KB.

We were left purplexed why you'd make a pin compatible chip to replace an older generation and then cut the amount of ram in it.  It was game over, we do a lot of signal processing and in one particular product had just a few hundred bytes of RAM remaining, the vast chunk of it taken up by our FFT requirements.  Incidentally, they do actually show some 64KB variants in their 3S family, but they appear to either very new or vapourware!

For those that have bought the launchpad, don't panic, I'm sure it'll do everything you want - just make sure you read the errata first to mitigate any potential issues you may come across.  Stellarisware is a very good set of libraries and is something I made great use of with the product we use the stellaris in.

[alexisnik]

I think you are one level further compared to us, I would say we are developing products/solutions <1000 pieces range. In that scale, TI has a lot of info / libraries / etc, compared to others. Maybe as you go higher in volume production designs, things turn around!

We agree on the Launchpad, the reason I replied was because I saw that 5-10 posts after your comment everyone wanted to either cancel the order or throw the Launchpad away! 

[visn]

Got my two boards in yesterday. Looking at the toolchain, would you suggest the CCS (1.3GB) download, or should I go IAR or Keil with sonsivri spice? Integration of the Stellarisware libraries are supported by TI for these alternatives.

[metal]

This page should be helpful: http://tinyurl.com/bj2k9wg

I received mein today. I am thinking about downloading CCS, but it is too big. StellarisWare appears to support the compilers on that page.

[sn00p]

I've used stellarisware with GNU compilers without a problem, although I generally build my own set of libraries, out of the box stellarisware is built for the lowest common denominator - it's possible to rebuild it so that it's more optimised for your target processor.

[metal]

I downloaded CCS and stellarisware, and I am also watching tutorials on youtube, Getting Started with the Stellaris® LaunchPad: Chapter 1, and there is the main page for all videos, they are good to start with, thu a bit boring and you can't clearly see what he is doing on the IDE while explaining, but you can pick the steps while he is talking. I also found interesting project: LED Frequency Display Demonstration. The real main reference that every owner should look at is TI's page, the wiki is nicely written!

[Langley]

Looks like the $5 promo is about over. Price on TI store and Digikey is now $12.99 and Mouser is $9.65. Only Newark still shows $5.49, but none have any stock.

Still not a bad deal if that's the part you want to use, but the $15 STM32F4DISCOVERY board with 3-axis accelerometer, MEMS digital mic, and DAC w/ speaker driver has a lot more bang/buck. As Metal and others have noted, though, the TI Launchpads are MUCH better documented and supported. STM makes some nifty stuff, but documentation and support are not their strong suit in my experience. I did get my Discovery board going under Keil running a modified ver of their USB host mode demo with little problem, though, and nobody comes close to that price.

[Parmin]

the $15 STM32F4DISCOVERY board with 3-axis accelerometer, MEMS digital mic, and DAC w/ speaker driver has a lot more bang/buck.

Well now, thats a nice board for nice price! I'll be ordering today.
The closest on this I think is C&K Mongoose (running arduino) for $99

[visn]

Hi, anyone experience any problems with the virtual uart? I installed CCS in a virtual machine and was able to download and run the demo examples, there was no problems programming the firmware. The Virtual uart shows up in device manager and I can open Hyperterminal or Putty. The program seems to be running but nothing shows up in in terminal.

I installed the drivers outside the virtual machine into the Win7 host and Putty communicates with it some of the time (fresh boot), if I fire up the virtual machine I can do nothing in the terminal, and it also fails thereafter in the host when I shut down the VM. (Programmer/ debugger always works fine)

It could be me though, but I never experienced any problems with VM's previously.

[dcsmrgun]

Looks like the $5 promo is about over. Price on TI store and Digikey is now $12.99 and Mouser is $9.65. Only Newark still shows $5.49, but none have any stock.

Newark's price just jumped up too. I had four in my basket at the $5 price but decided to back down to two to play with. They'll have stock on 11/19 again so here's hoping everything works out.

I have two more on order from TI that I put in just under the gun when they still had their $5 price, but they're not scheduled to ship until late December


As someone just getting into embedded development coming from traditional desktop programming, I found the Stellaris' documentation impeccable and StellarisWare is definitely making it easy to grasp concepts. I'm hoping to hit the ground running when my TI boards come in.

I also have an STM32F3Discovery which works really well, but I didn't find much that I liked about ST's documentation. It's not terrible or anything, but I think TI just blew everyone away.

[metal]

I also have an STM32F3Discovery which works really well, but I didn't find much that I liked about ST's documentation. It's not terrible or anything, but I think TI just blew everyone away.

I felt the same way!

[dcsmrgun]

I don't know if this is a fair comparison, but I attended an ST lecture/demo for the STM32F3Discovery board last week and it was a nightmare. Granted I got a free board and a copy of IAR on a USB stick, but the entire day was the lecturer reading his notes in a monotone while clicking buttons on his laptop to compile example code. The TI lectures that are available on their development wiki feel much more in-depth and were much more enjoyable to follow along with.

So I guess to contrast:

I walked away from the TI workshop (online) enlightened
I walked away from the ST workshop (somewhat) confused.

Maybe it was different for people who were already well versed in the materials. Who knows.

[sn00p]

Hi, anyone experience any problems with the virtual uart? I installed CCS in a virtual machine and was able to download and run the demo examples, there was no problems programming the firmware. The Virtual uart shows up in device manager and I can open Hyperterminal or Putty. The program seems to be running but nothing shows up in in terminal.

I installed the drivers outside the virtual machine into the Win7 host and Putty communicates with it some of the time (fresh boot), if I fire up the virtual machine I can do nothing in the terminal, and it also fails thereafter in the host when I shut down the VM. (Programmer/ debugger always works fine)

It could be me though, but I never experienced any problems with VM's previously.

Is the "virtual uart" a CDC device provided by the demo firmware?

If so, theres a few things you probably need to be aware of. 

Under Windows (All versions) there are a lot of bugs with the microsoft CDC driver (usbser.sys).

Namely, if the bulk-in endpoint gets transmission failures, the usbser driver will stop polling the bulk in endpoint.  End result is that although your device appears to receive data quite happily from windows, you can't send data back to windows.  Only fix for this is to re-enumerate the device.

Also unplugging and replugging a CDC device while the "port" is open results in a "dead" device, that is to say that the device  looks like it's working but you cannot communicate with it.

Not saying that these are your issues, but it's helpful to be aware of these.

And finally (not really related) the fact that to use the CLASS driver you have to supply an .inf file.  Nice little earner for Microsoft if you want to avoid the unsigned driver errors (although they'll always work because the actual sys file is signed).

All these points have resulted in us dropping CDC in favour of using a HID device, max rate is 64KB/s but it's driverless and doesn't have these horrible bugs.

[Gallymimu]

Is the "virtual uart" a CDC device provided by the demo firmware?

If so, theres a few things you probably need to be aware of. 

Under Windows (All versions) there are a lot of bugs with the microsoft CDC driver (usbser.sys).

Namely, if the bulk-in endpoint gets transmission failures, the usbser driver will stop polling the bulk in endpoint.  End result is that although your device appears to receive data quite happily from windows, you can't send data back to windows.  Only fix for this is to re-enumerate the device.

Also unplugging and replugging a CDC device while the "port" is open results in a "dead" device, that is to say that the device  looks like it's working but you cannot communicate with it.

Not saying that these are your issues, but it's helpful to be aware of these.

And finally (not really related) the fact that to use the CLASS driver you have to supply an .inf file.  Nice little earner for Microsoft if you want to avoid the unsigned driver errors (although they'll always work because the actual sys file is signed).

All these points have resulted in us dropping CDC in favour of using a HID device, max rate is 64KB/s but it's driverless and doesn't have these horrible bugs.

I've seen these issue too.  Took forever to figure out that the stupid drivers were the problem and not my UART!  Your driver problem is honestly why we go with FTDI 230XS chips.  I understand that not all designs can tolerate the added cost but for small to medium volume it's nice because the drivers are ubiquitous and the chip is pretty cost effective (less than $2 at qty 100)

[dcsmrgun]

Ugh, Newark just updated their expected in-stock dates and they've been pushed a month back. Is TI not making these in volume or what?