Can someone (perhaps resident Atmel guru Ulf Samuelsson RSVP) arguably explain how AVR32 is posioned against competition in terms of market share and future "road map". I guess I'm asking for a kind of marketing pitch to convince me that investing money in development tools and time in learning them is a "good" investment in a long run, in contrast with the same time/money dilema pertaining ARMs (therefore ***no*** PIC32 dilema here).
I'm not asking for a ARM vs AVR32 technical dissection(s), although I wouldn't mind, just some courage to jump into deep AVR32 waters. I still have a kind of fobia regarding infamous Hitachi's "6 months notice before EOL".
One indicator is usage, and you can get one measure of that, from Digikey :
If an item is currently in stock, but also showing 'non-stock', then that means they will not re-stock (sales too slow?) when stocks do run out, and so that could be a good package/device combination to avoid.
Not really. It could be that as the AVR32 is so popular with the big users Atmel does not have time for small players like Digikey. Most reps for Silicon distis don't get out of bed for deals at less than 50K USD p/a.
I doubt Digi sells the AVR32 in volume. It is not as though they are quoting for reels/trays with a MOQ of 250, their highest price break is
Any one doing serious development will be able to get 10-20 parts as free samples so no need of Digikey.
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/
Ha. I work for a company with around US$33BN in market capitalization. We use HUNDREDS OF THOUSANDS of Atmel devices a year.
Per policy, it is extremely difficult to get more than 5 samples of any part out of Atmel. Direct or through disti, it is just not possible. You have to get your rep to split the orders into multiple fake orders to fake customers just to get 20 parts.
The big AVR32 devices, AP7xxx, seem to have died without much warning. The planned new generation devices never appeared, and then the existing devices were suddenly re-labelled "mature" and re-filed in the attic in Atmel's website. It's a pity, since they had a lot of value for money, a lot of processing power per milliamp, and good support on the software side (solid gcc port from Atmel, buildroot embedded Linux environment, evaluation cards, etc.).
Last I heard, the UC3 AVR32 microcontroller line was solid and with new parts in development. The tools are not expensive - you can use the same jtag debugger for the AVR32 and the 8-bit AVR micros, and the avr32 studio is a free download along with a complete gcc toolchain. For those who prefer to pay money, I gather there is also an IAR toolchain - I don't know how it compares in code generation quality.
I use the AVR32 toolset from IAR. There has recently been a spurt of activity on their development with 2 major update releases in the last year. We went with this toolset primarily because we have been using the IAR tools on MSP430 and as a result we had a minimal learning curve when starting the AVR32 development. The quality of the code generated seems to be very good. There have not been any mis- compilations that we have found.
When comparing with the ARM Cortex-M3 about a year ago the things that made us select the AVR were:
1 - The low power modes with RAM retention. These were much more comprehensive than those on the ARM.
2 - The dedicated I/O functions on the AVR32B were a better match to our requirements than any ARM version that we could find.
3 - The unit price was competitive with the equivalent ARM devices.
If you look at it, your investment in tools is not that high. You get the S/W from Atmel for free so your investment will be for a development board and an emulator which will set you back some 100$s. If you get JTAGICE mk II and STK600, you can use them also for 8 bit AVRs. The JTAGICE Mk II can be picked up at eBay for ~$100.
Each package will need a socket adapter for the STK600. If you go for the Evaluation Kits. the chip is soldered to the device, so that will be cheaper but chip specific.
There will soon be a major new release of tools, merging 8 and 32 bit AVRs toolsets.
UC3A0/1 - Networking/communication/Audio UC3A3 - Focus on Audio UC3B/D - General purpose Micro UC3C - Motor control with Event System and Floating Point. UC3L - Real low power
The peripherals are the same as on the Atmel ARM7, ARM9 and Cortex-M3, so learning the peripherals will allow you to move between these chips with comparatively ease.
With UC3L going into mobile phones as an accessory processor and the UC3B going into netbooks etc, the market share of these devices will be significant. The UC3A Audio solutions seems to rule the Apple Iphone adapter market
Saw a cool new device for guitar tuning the other day. Previous devices allows you to tune one string at a time, The UC3 device in this unit, allows you to strike a chord of all 6 strings and analeze them together. Possible with UC3, but not with CM3.
They are not that deep. Get a local Atmel contact, and request some training material, and it should not be that hard.
I guess the second generation chips I have at my desk running Linux, are just imagination then ;-) Needed a respin, but people decided to spend the money elsewhere. When you have mobile phone companies prepared to commit for x Mu/year, the bean counters have their say.
Yes, talked to the guys at the OpenWRT project, and they told me that in that priceclass, the AP7000 was unbeatable.
All the internal benchmarking is done with the gcc compiler so you get decent performance out of the free toolchain. IAR has some advantages, like early support for floating point.
As far as us mere mortals are concerned, the chips never appeared - no web pages, no press releases (as far as I saw), or anything but rumours.
Actually, it's nice to know that you had the chip in a working sample - it shows that Atmel had a working prototype device before activating the marketing machine. That implies that when they /do/ market and publicise a new device, they probably have working hardware.
Yes, I know how it goes - it's the big customers that have a big effect, and us little guys just tag along.
I've always thought the AVR32 would be hard to sell against the various ARM devices, especially as Atmel makes ARMs themselves. I know there are differences between the cores - we picked the AP7000 rather than an ARM for a project. But for many customers, ARM is the standard and that trumps things like power consumption and even price to some extent.
Sometimes "samples" are for prototyping. Other times "samples" are just to save the hassle of ordering through channels for a pre- production run (delays, small quantity costs, etc). I once asked Harris for samples on a 7400 type part and was refused because they figured I should know how it worked. I asked for bloody three pieces of a 10 cent part!!! Needless to say I didn't design in the Harris part, but then I don't think the company shut down because of it either.
Unless the part is hot off the press and hard to get, samples are more a courtesy than anything. Like a sales person buying lunch. I can afford to buy my own lunch, but it shows they have some value for your business. I can also afford to buy my own samples, but getting them without having to generate the paper trails of procurement can be a cool breeze on a sweltering hot work day.
On the other hand, engineers shouldn't abuse samples and ask for more than needed or for parts you aren't really thinking hard about using. Personally I have no use for parts sitting on my shelf. I much prefer to put them on boards and ship them out to customers... ;^)
If you had actually read and comprehended them, instead of merely looking at them through the "I hate OSS users" glasses that are perpetually glued to the bridge of your nose, you would realize this isn't true. TI and Microchip for instance are very responsive (unfortunately in the latter case, mostly responsive in the sense of "we know our product's design kinda sucks"). Of course, people in general rarely post "I had a great experience" stories.
Atmel's parts are good, but their app notes are horrendous and there are risks with supply, die shrinks and discontinuation, as many here will testify. And sampling is worse than hit or miss. As a private party I would say it is impossible to get samples at all. As a large company it is possible to get 5pcs relatively easily, but special hoops need to be jumped through in order to get more.
Once again, I don't find the Atmel web site easy to quickly get a feel for the capabilities of the parts so let me ask you. I see on the AVR Solutions page that a number of the AVR32 devices have a "Hi-Fi stereo Audio DAC", but I don't see any mention of a corresponding ADC. Is that missing from the AVR32 parts? If I can find an affordable MCU in the right package with a audio quality stereo CODEC, that would solve a problem of mine. Is the AVR32 the solution to this problem?
Thanks for the pointer. Like I posted, Atmel includes a "Hi-Fi" quality stereo DAC in some of the AVR32 devices, so clearly there is some impetus to combine high quality audio with an MCU. I just need the ADC in addition. The fact that they make a part that combines two switching PSU with high quality audio converters says it is not too hard a problem that they couldn't combine it with the MCU. In fact, there are products that do just this from TI and ADI, but they are sold into a market where the processing is selected from a laundry list and I can't get anyone to tell me that I could use it for what I want to do. In fact, those devices would be perfect for my needs. I think the primary market is hearing aids.
This AT73C246 is a very odd beast indeed. But it has potential. Once it is actually on the market (as in showing up as stock somewhere) I'll take another look at it.