Atmel ARMs

[richcj10]

I have a project requiring a lot more punch than a 8bit avr.
I would like to work with some ARM stuff.
I was wondering if anyone has dabbled in ARM and what thins I should look for and look out for.

Thanks

Rick

[mikeselectricstuff]

NXP has a good range, even some in DIP (allegedly - not sure if any have been seen in the wild yet)

[BravoV]

NXP has a good range, even some in DIP (allegedly - not sure if any have been seen in the wild yet)

Huh ? ARM in DIP ? WHERE ? ? ...<drooling>....

[baljemmett]

NXP has a good range, even some in DIP (allegedly - not sure if any have been seen in the wild yet)

Huh ? ARM in DIP ? WHERE ? ? ...<drooling>....

http://www.nxp.com/products/microcontrollers/cortex_m0/lpc1100_x_l/LPC1114FN28.html

"Status: Samples available", but no distributors listed with stock.  The 'request samples' buttons takes me to a page telling me samples are only available to customers with direct purchase agreements, and everybody else should go through a distributor, so I daren't click any further than that!

(Looking through the distributor list, only Future Electronics even list the product -- the quoted price is rather interesting, although the MOQ makes it a bit pointless for general dicking-around use.)

[sacherjj]

NXP has a good range, even some in DIP (allegedly - not sure if any have been seen in the wild yet)

Huh ? ARM in DIP ? WHERE ? ? ...<drooling>....

http://www.nxp.com/products/microcontrollers/cortex_m0/lpc1100_x_l/LPC1114FN28.html

"Status: Samples available", but no distributors listed with stock.  The 'request samples' buttons takes me to a page telling me samples are only available to customers with direct purchase agreements, and everybody else should go through a distributor, so I daren't click any further than that!

(Looking through the distributor list, only Future Electronics even list the product -- the quoted price is rather interesting, although the MOQ makes it a bit pointless for general dicking-around use.)

So customers most likely to wanting to use their chip in DIP form are least likely to be able to. BRILLIANT!

[baljemmett]

So customers most likely to wanting to use their chip in DIP form are least likely to be able to. BRILLIANT!

I'm guessing whoever made that decision is very proud of the MBA they have hanging on their wall

If they're serious about it, I would imagine the product is just not yet available in any volume; once it is maybe it'll be easier to get hold of!  Until then, I shall consider it vapourware -- would be pretty neat if it turns out not to be, though.

[alm]

NXP, ST and TI appear to be the popular ARM Cortex manufacturers for hobbyists. I'm not sure why you would consider Atmel ARM. For the AVR they are the only source, and the AVR has some advantages over other 8/16-bit platforms. They were late with their Cortex M3, and don't really appear to have any compelling unique selling points, so why bother? Unless price or availability is better than competitors in your part of the world.

[T4P]

Worst of all, Atmel has not been earning a single dollar from day 1 so that means they might suddenly discontinue their ARM cpu's when the time comes
Like the AT91 quoting directly from MCP forums

Ya mips  and cycles are important considerations but the most critical is availability and vendor stability. Remember EDN with Motorola ads competing with each other? 56000 vs Coldfire and so on? Now its all of the Arm variants. One of my past employers I won't mention will have to discontinue some Arm lines they picked up. I designed in the AT91rm9200 from Atmel. Guess what? It's gone but the original PIC is still available. 

So how valid is a comparison if the parts are history?(or one-batch marketing lots)

[Bored@Work]

Worst of all, Atmel has not been earning a single dollar from day 1

Just because Dave says so doesn't make it true.

I wouldn't touch Atmel ARMs, but not because Atmel isn't making money (they make money), but because

- how they treated their customers in the past and how they treat them currently, and their past availability issues
- their Cortex offerings are rather new
- their policy not to support anything but development on Windows and their attempt to force all customers to use Atmel Studio 6

For a hobby project I would go with NXP. ST coming only second, because of their laughable dev board license and their lack of Linux support, i.e. not thinking their discovery series through. TI coming third, because their stuff is too heavily geared at large commercial customers only.

All the others, including Dave's new friends? Well, difficult to get for hobbyists.

[T4P]

For a hobby project I would go with NXP. ST coming only second, because of their laughable dev board license and their lack of Linux support, i.e. not thinking their discovery series through. TI coming third, because their stuff is too heavily geared at large commercial customers only.

All the others, including Dave's new friends? Well, difficult to get for hobbyists.

Ah well.

ST ... dev board license ... ? Well i did see people hacking ST to work in linux but that's a bugbear really
"Dave's new friends" ? Well i'm not going to pick TI Stellaris because foremost their stellaris ... Cortex-M4 of theirs only has 80MHz ...
I would pick NXP first but hah, i'll stick to ST

[AndreasF]

I've spent some time looking into Atmel's SAM3U series, and to me they seem quite nice, mostly because of the set of peripherals it comes with (USB device port, MMC interface). Their SAMBA ("SAM Boot Assist") also seems like a very nice feature. And Atmel provides several app notes on how to develop for their ARM devices using open source tools - another plus for hobbyists one would think. However, I didn't spend too much time looking at other manufacturers, so they may very well offer similar things.

[T4P]

I've spent some time looking into Atmel's SAM3U series, and to me they seem quite nice, mostly because of the set of peripherals it comes with (USB device port, MMC interface). Their SAMBA ("SAM Boot Assist") also seems like a very nice feature. And Atmel provides several app notes on how to develop for their ARM devices using open source tools - another plus for hobbyists one would think. However, I didn't spend too much time looking at other manufacturers, so they may very well offer similar things.

Vendor stability is horrible
You might be able to design it now, who knows whether it will magically *poof* dissapear the next day?

[AndreasF]

...

Vendor stability is horrible
You might be able to design it now, who knows whether it will magically *poof* dissapear the next day?

Agreed, if it's for a project that is expected to be manufactured in large quantities for commercial purposes - if it's for a one-off/small run hobby project, I don't think that's a big concern.

[codeboy2k]

Worst of all, Atmel has not been earning a single dollar from day 1 so that means they might suddenly discontinue their ARM cpu's when the time comes
Like the AT91 quoting directly from MCP forums

Oh you didn't hear? Atmel changed their logo... they gonna be rollin' in money soon....they smell fresher already
I'm guessing the employee breakroom will have a slide next year. Just you wait and see .... 

[pkrobot]

...

Vendor stability is horrible
You might be able to design it now, who knows whether it will magically *poof* dissapear the next day?

Agreed, if it's for a project that is expected to be manufactured in large quantities for commercial purposes - if it's for a one-off/small run hobby project, I don't think that's a big concern.

Even if the project is expected to be manufactured in large quantities, going with an ARM core is not a bad decision. For microcontrollers at least, design of digital electronics design is not difficult. If you have chosen an ARM core, at least you can port your software, and not have to change the software tools significantly. Changing the electronics design would mean just replacing the microcontroller.

Having said that, I'd not go with AVR myself. If you want to consider cost, I think ST is the best. My personal preference is TI. I used to prefer their DSP's, but recently, have started going with Stellaris. For consumer items, maybe ST and NXP is a better choice, but in my experience, TI fares much better in reliability.

One more option to consider is an ARM softcore in FPGA, but in my career, I have seen many managers/ designers bring it up, but haven't met anyone brave enough to go ahead and chose one in a real project. I've seen a lot of engineering prototypes, but perhaps complexity vs. flexibility tradeoff does not tip the balance towards going with a soft-core.

-pk

[Psi]

I wouldn't go Atmel branded ARM unless i had a valid reason to.
Instead I'd go with a more common ARM brand with more community support.

If you're afraid of 40+ pin smd packages and want DIP then there is only one solution.
Buy 40+ pin smd packages and teach yourself to solder them, it's not hard.

If you want something 'DIP-like' buy TQFP breakout boards, once you've soldered your TQFP micro to them you have a nice easy to use board.
Some even come in DIP form and give you a DIP package out of a TQFP that you can put into a DIP socket.

Soldering a TQFP can actually be quite easy and you don't have to solder all pins individually.
-Choosing a micro or device with a large TQFP pin pitch can help when starting out. The small pin pitch packages are more tricky.
-Apply some solder to all pads on the pcb (so you can see they are no longer flat and a little rounded with solder)
-Hold the TQFP component with some tweezers (or blue-tak on the end of something, thanks Mike) and solder one corner pin to hold it in place.
-Once you're happy that the pins are all sitting where they should solder a few more corners so it doesn't move.
-Put a huge chisel tip on the iron and give the tip a good clean. (you don't want solder blobs on the tip).
-Now use the chisel tip to press down on groups of pins at a time. This will heat them up, melt the solder on the pcb and join them together.
-There shouldn't be any solder bridges using this method but if there are they can be removed with some solder braid


It doesn't yield as good a result as soldering them one at a time or with hot air reflow, as you're bending the pins down to meet the pcb, but it does have the advantage of being really easy to do.

[T4P]

Or one of those round 45 degree tips with a well inside, i know one can only get it in Asia with a hakko but i am getting mine tomorrow ( Not very expensive at 10SGD it's only a dollar more then the normal tips )

[pkrobot]

I wouldn't go Atmel branded ARM unless i had a valid reason to.
Instead I'd go with a more common ARM brand with more community support.

If you're afraid of 40+ pin smd packages and want DIP then there is only one solution.
Buy 40+ pin smd packages and teach yourself to solder them, it's not hard.

If you want something 'DIP-like' buy TQFP breakout boards, once you've soldered your TQFP micro to them you have a nice easy to use board.
Some even come in DIP form and give you a DIP package out of a TQFP that you can put into a DIP socket.

Soldering a TQFP can actually be quite easy and you don't have to solder all pins individually.
-Choosing a micro or device with a large TQFP pin pitch can help when starting out. The small pin pitch packages are more tricky.
-Apply some solder to all pads on the pcb (so you can see they are no longer flat and a little rounded with solder)
-Hold the TQFP component with some tweezers (or blue-tak on the end of something, thanks Mike) and solder one corner pin to hold it in place.
-Once you're happy that the pins are all sitting where they should solder a few more corners so it doesn't move.
-Put a huge chisel tip on the iron and give the tip a good clean. (you don't want solder blobs on the tip).
-Now use the chisel tip to press down on groups of pins at a time. This will heat them up, melt the solder on the pcb and join them together.
-There shouldn't be any solder bridges using this method but if there are they can be removed with some solder braid


It doesn't yield as good a result as soldering them one at a time or with hot air reflow, as you're bending the pins down to meet the pcb, but it does have the advantage of being really easy to do.

If soldering SMD's is an issue with you, consider Schmartboards:

http://www.schmartboard.com/index.asp?page=products_qfp

I have used these boards, and can't see how you can go wrong with them. However, in my opinion, these days, designing and having prototypes of your own boards has become so inexpensive that you should considering soldring your own SMD's. Its not as hard as it seems if you haven't done it before. Get some scrap boards from www.sparkfun.com, and practice a bit. It doesn't take long to learn it and overcome any inhibition that you might have in soldering your own SMD's.

Using a good soldering iron and the right tip is a key to it.

-pk

[richcj10]

I forgot about this post! wow! I have been busy lately...

I know how to solder. I hand soldered thousands of surface mount packages (8 pin msop to 144 TQFP's) all the time. My clients have boards with 500 different components that are all hand soldered and inspected. soldering is not an issue...

I looked for code support. Atmel has nice support for ARM 3 and 4 with there ASF system.

SAM-BA is really cool as well. It is like there FLIP (software to program there ATU series) software for the ARM's