Noob question about TI Stellaris Launchpad EK-LM4F120XL

[BravoV]

I realized upon unboxing this board, it has two identical LX4F120H ARM mcus, so the left one's purpose is solely for ICDI function only ?

Just curious why need such powerful auxiliary chip for just the in circuit debug feature ? Again, this is a noob question here. 

[chickenHeadKnob]

First I will make the assumption that this processor (the target processor) is an ARM cortex m4 that  is using the standard ARM JTAG/SWD circuitry and not some proprietary interface (I haven't read the data sheet). It does not require anything like  another cortex m4 to run the JTAG, all you need is a USB to serial like FTDI 2232, Dangerous prototype's Bus Blaster for example. Although I am almost certain the FTDI chip also has a common core processor (could be an 8051 derivative - don't quote me) onboard. All of the semi manufactures use one of their own  chips to provide jtag for their development boards . NXP uses an LPC1343 in their  LPCExpresso series, ST uses a STM32F103 in their  ST-Link2 STM32 Discovery boards ect. Why wouldn't they use their own products?

[Bored@Work]

Of course, when a company does a demo board they want to use their own chips on it for as much stuff as possible. It is a prestige thing to not have competitor products on a demo board.

Why they used the same chip? Maybe to reduce production cost. It is one feeder less for the pick and place machine. It makes logistics, procuring and storing the different parts for production a bit simpler.

The hardware team, when designing the board, only has to deal with one kind of MCU. One datasheet less to study. The same basic connection for things like power and clock. Same footprint. Same for the software team doing the firmware and demo software. It only has to deal with one kind of MCU.

We can also safely assume that TI's internal cost for the chip is fare less than it's official list price. In the end using the totally oversized MCU was maybe even cheaper than using something else.

[mrflibble]

Just curious why need such powerful auxiliary chip for just the in circuit debug feature ?

Reflected sound as of underground spirits.

[andersm]

JTAG and SWD are not very well suited to run over USB. By putting more smarts into the adapter, you can use better protocols over the (relatively) high latency bus, and translate into JTAG or SWD commands locally. Why TI chose an LX4F120H in particular is anyone's guess. Could have been a case of dogfooding on part of the Launchpad team, or making a point of the chips being so cheap you can put two of them on a single board or whatever.

[BravoV]

Thanks for the responses and insights, yeah, I've thought about dogfooding or the prestige too.

But I think the most reasonable is this Arm M4F is still new for TI, and putting both will be the best situation for their internal team and also "us" the early adopters to iron out the bugs.

Cause after digging deeper at TI web site, also into the datasheet + the so called errata documentation, lo and behold, TI's Arm Cortex M4F families basically is not even at the production stage yet.

Well, personally I can't complain too much since the price was dirt cheap at $4.99 during the promotional period, no wonder.

I guess the new batch with the $13.99 price tag doesn't have the e"X"perimental chip and much-much less erratas and hopefully with workarounds for fatal ones.

Just look what I've discovered, the chip shot + notes should be sufficiently explain what I mean. 

Edit : Btw, this board's batch code is on week 41, go figure. 

[BravoV]

Just curious why need such powerful auxiliary chip for just the in circuit debug feature ?

Reflected sound as of underground spirits.

Huh ? 

[rolycat]

Just curious why need such powerful auxiliary chip for just the in circuit debug feature ?

Reflected sound as of underground spirits.

Huh ? 

Echo-nomics.

See The Colour of Magic by Terry Pratchett for further enlightenment.

[TheDirty]

First I will make the assumption that this processor (the target processor) is an ARM cortex m4 that  is using the standard ARM JTAG/SWD circuitry and not some proprietary interface (I haven't read the data sheet). It does not require anything like  another cortex m4 to run the JTAG, all you need is a USB to serial like FTDI 2232, Dangerous prototype's Bus Blaster for example.

FT2232 JTAG is relatively slow and FT2232 adapted to SWD is even slower.

It is kinda annoying that we have all of these different proprietary standards.  LPC wont even release the documentation on LPC-Link.  At least Crossworks supports both ST-Link and ICDI.

Edit : Btw, this board's batch code is on week 41, go figure. 

41st week of the year, not 41st week of production.  That wasn't that long ago.  They are putting them on boards pretty quickly.

[rsjsouza]

Usually the parts that are eXperimental did not go through the full characterization process (mostly electrical specs)  due to the length this takes even after perfectly functioning parts are already available internally. Therefore the idea is to put parts out so people interested in them can start software development while the complete characterized parts for production become available.

This in general has little to do with the silicon Rev and its functional bugs usually listed in an errata - these take a lot longer to fix.

[BravoV]

Rafael, thanks, so this is common practice, especially for new product, guess I have to live with it.

For me as a beginner, pretty sure those unsolveable erratas should not be a deal breaker for me.

[westfw]

Note that the stellaris launchpad does better than the MSP430 launchpad, which used THREE TI chips to implement the debugger interface.

[vis1-0n]

The STM32F0 discovery had the superior ST32MF103 to target the STM32F100. The LPCXpresso shipped with the LPC3154 which is a wet dream compared to the LPC1114 (or LPC1769) target. The LPC3154 is a so much more desirable chip, wish they had broke that out and allowed firmware updating.

BTW anyone experiencing any serial comms issues with the Launchpads virtual serial port? I have not been able to get Hyperterminal/ Putty to work, although Putty works okay on fresh boot on the VM Host machine and fails (VM and host) when I attempt using the VM. The programmer/debugger has no issues in VM.

[BravoV]

The STM32F0 discovery had the superior ST32MF103 to target the STM32F100. The LPCXpresso shipped with the LPC3154 which is a wet dream compared to the LPC1114 (or LPC1769) target. The LPC3154 is a so much more desirable chip, wish they had broke that out and allowed firmware updating.

Never experienced with ST's or NXP's Arm though, any comment how these two compare with TI's "especially for beginner" like me ?

BTW anyone experiencing any serial comms issues with the Launchpads virtual serial port? I have not been able to get Hyperterminal/ Putty to work, although Putty works okay on fresh boot on the VM Host machine and fails (VM and host) when I attempt using the VM. The programmer/debugger has no issues in VM.

Weird, what host or client OS ?

Don't have any problem at all, everything run smoothly and it worked straight from the 1st time I hooked it up at the VM, host OS is Win 7-x64 with XP as the VM client OS.

Screen shot attached below with my browser pointing at this page, and the VM (XP) running Hyperterminal connected to the stellaris launchpad.

[andersm]

Never experienced with ST's or NXP's Arm though, any comment how these two compare with TI's "especially for beginner" like me ?

The Cortex-M's are pretty much interchangeable. They all use the same core, and all manufacturers have peripherals that offer more or less the same functionality. There's differences in the implementation details, but it doesn't make one make inherently more difficult to use than the others. I haven't really used the manufacturer's software libraries, but NXP's seemed the most rudimentary and TI's the most developed.

The LPC3154 that was mentioned is a low-end applications processor aimed at embedded Linux and the likes, and very different from the microcontroller lines.

[andersm]

In the same way the PICKit3(which is rubbish) uses a PIC32, so you're gonna call that overkill too right?

Not really relevant to the discussion, but the PICkit 3 uses a PIC24F.

[BravoV]

The Cortex-M's are pretty much interchangeable. They all use the same core, and all manufacturers have peripherals that offer more or less the same functionality. There's differences in the implementation details, but it doesn't make one make inherently more difficult to use than the others. I haven't really used the manufacturer's software libraries, but NXP's seemed the most rudimentary and TI's the most developed.

Thanks for confirming, so far my own experience at TI's tools, documentations, tutorials and also their training videos, I would say it is good enough for noob like me to start into ths arm cortex platform. No big ambition though, just aiming for 'enthusiast' grade projects in the future.

[vis1-0n]

Weird, what host or client OS ?

Don't have any problem at all, everything run smoothly and it worked straight from the 1st time I hooked it up at the VM, host OS is Win 7-x64 with XP as the VM client OS.

Screen shot attached below with my browser pointing at this page, and the VM (XP) running Hyperterminal connected to the stellaris launchpad.

WinXP sp2 on VMWare, host is Win7 32 bit. What VM software are you running? VirtualBox? This is pretty annoying as after many years I finally sat down to actually learn and I can't get this working. I guess I will have to install onto the host, I rather hate doing that as I don't connect to the internet with it at all. This will also be less portable than the VM, the last time I had a failing HD it was so easy to restore a backup of my applications.

[BravoV]

WinXP sp2 on VMWare, host is Win7 32 bit. What VM software are you running? VirtualBox? This is pretty annoying as after many years I finally sat down to actually learn and I can't get this working. I guess I will have to install onto the host, I rather hate doing that as I don't connect to the internet with it at all. This will also be less portable than the VM, the last time I had a failing HD it was so easy to restore a backup of my applications.

Same here using VMWare Workstation (see the screen shot) except my host is Win 7 64bit, and also it runs fine on VMPlayer as well.

No extra tweaking applied at the VM, everything run on stock setting.

Edit : Just verified it again by plug&unplugging the Launchpad board, no problem at all and running well at VMPlayer ver 4.0.4 build-744019.