AVR Advantage [over ARM?]

Go To Last Post
44 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

On Thursday I attended my second annual 'Texas Instruments Technology Day' where TI puts on a bunch of free seminars and a lot of vendors set up booths to show off their stuff. Last year the keynote speaker was the CEO of Luminary Micro which had recently been bought by TI. They make low power ARM MCUs and their Stallaris products now make up nearly the entire ARM MCU line at TI.

This year I spent quite a bit more time looking at the Stellaris MCUs and it got me wondering what the advantage of AVRs really are compared to ARM. For nearly the same price as any of the higher end Atmel chips, I can get an ARM Cortex-M3 running at 50Mhz with USB, ethernet MAC+PHY in one package. It just seems like everywhere I look its ARM ARM ARM and when the price isn't very different then what's the appeal to go otherwise?

I understand that AVR32 is supposed to compete in this area, but I feel that coming out with a completely new architecture is a bit silly. I've also been using xmega quite a bit lately, but I don't understand why they wouldn't at least have versions with ethernet and/or usb.

Anyway, just some random thoughts after just waking up. :D Thoughts?

On a side note, we also got a code that lets us buy any sub $500 TI dev board for only $10. :P :P One time use only, otherwise I would share.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

For nearly the same price as any of the higher end Atmel chips, I can get an ARM Cortex-M3 ...

Not just Stellaris, STM32 and NXP LPC too.

The hope might have been that the newly available Atmel SAM3s might enter the fray - but they appear to be priced to attract a different market (or maybe so they don't impact AVR8s and AVR32s?)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That's a pretty sweet code. I attended the MSP430 / Stellaris day here in Canada about a week ago, and they didn't give us a deal like that. Lucky you.

I agree with your comments, which is why our new board has a STM32 Cortex-M3 on it.

I will say however that the AVR is a simpler part to use. Setup is easy - there's almost nothing to configure on an AVR to initially get it running, unlike an ARM. Not that ARM is particularly hard; just that for a beginner I think AVR is easier. The WinAVR toolset combined with AVRStudio is also excellent and free. Unlike most "unrestricted" ARM tools which are either expensive, or quite challenging to set up.

My personal opinion is that your comments reflect why we haven't seen anything happen with Xmega since the parts were released, more than 2 years ago. We haven't even seen the errata be fixed. Given the competition from Cortex-based parts, I can't imagine Atmel has had many significant new AVR design wins. The sales volume probably isn't sufficient to justify the expense of further development in this product line. I'd love to be proven wrong, but that's my guess, which is why I'm not pinning my hopes on any new AVR parts from Atmel.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I agree, ARM parts are getting more attractive by the day. Here's my take on what advantages AVR8s still hold

- Easier to get up and running. ARM parts have a lot of power, but that means a lot more registers to figure out.

- 5V operation. Sometimes that's what you need.

- High current outputs. Don't know of any ARMs that have all 20mA/pin outputs.

- DIP packages. Nice for one-offs, students, hobbyists.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

We sure re-plow this well-plowed field a lot.

Since I've gone farmer with the metaphor, lets use the example of my vegetable garden in my back yard. Say I could get a shovel for $2.00 and a Bulldozer for $4.00, I still wouldn't get the Bulldozer because the shovel is the most appropriate tool. Sure the Bulldozer has a shit load more power and features and only costs a couple of bucks more, but all I need is the shovel and it is a damn sight easier to use for may application.

I may eventually add the ARM to my arsenal. That will when either I have a real need for the extra power or the cost (including development tools) and ease or use are such that it would warrant me learning yet another development system. I don't think it is there yet.

When I worked for Philips Consumer Electronics I was told that they could afford to have an engineer full time employed to reduce the cost of a product by $0.10 per unit. When you start thinking in those terms then the relatively minor - to us - cost differential between the higher end AVRs and lower end ARMs helps understand why the AVR stock valuation is still booming.

Smiley

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

there's almost nothing to configure on an AVR to initially get it running, unlike an ARM

Quote:

Easier to get up and running.


You guys should check out this new fangled programming language called "C". Sure there's a bit of Asm code called the "C runtime" that needs to be written just once for any processor and, yes, it's written in Asm. But the great thing about C is that the chances are someone else has already written the C runtime for you and in addition to the kind of things it does on AVR (cli, set stack, copy .data, clear .bss) it has the added complication of running PLLs up to speed, setting up SDRAM controllers so that the RAm is refreshed and .data/.bss can be accessed and so on. But the really great thing, as I say, is that there's a strong chance the silicon vendor or one of "the gurus" has already written this CRT for your CPU/board so you just write C like you do on any processor and it simply works.

As an example, in the case of NXP LPC there's a company called Codered who'll have done this for you so you don't have to - but they are really great guys and they supply it in source form so you can study it and see what they did (if that kind of thing interests you).

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

Sure the Bulldozer has a shit load more power and features and only costs a couple of bucks more, but all I need is the shovel and it is a damn sight easier to use for may application.

The point missing here is that the bulldozer not only costs the same as the shovel. It has the same size etc. It is a wee bit harder to learn to operate but the stuff you do need to learn, and the tools you need to master, can largely be re-used for your plow, your combined harvester etc..

For a hobbyist, I'd say that these are driving the usage of AVRs

- Several PTH parts
- Fairly straight-forward data sheets
- Good starter kits
- Substantial user base around the world
- AVRfreaks.net

For large pro users most of the arguments given here does not apply. For most (all?) it will be price. For some it will be things like power consumption (AVRs pretty strong, I think). For some it will be available features (AVRs quite weak here).

As of January 15, 2018, Site fix-up work has begun! Now do your part and report any bugs or deficiencies here

No guarantees, but if we don't report problems they won't get much of  a chance to be fixed! Details/discussions at link given just above.

 

"Some questions have no answers."[C Baird] "There comes a point where the spoon-feeding has to stop and the independent thinking has to start." [C Lawson] "There are always ways to disagree, without being disagreeable."[E Weddington] "Words represent concepts. Use the wrong words, communicate the wrong concept." [J Morin] "Persistence only goes so far if you set yourself up for failure." [Kartman]

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The shovel/bulldozer analogy doesn't really work anyway - in the realworld you'd probably have a problem digging a 6"x6" hole with a bulldozer that could be achieved using a shovel but in the ARM/AVR world either can dig a 6"x6" hole if that's what you want to do.

I'd make the analogy of Steve Austin's bionic arm+hand in the Six Million Dollar Man versus a "normal" human arm+hand. Except this time the ARM(sic!) doesn't cost $6m ;-)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

And there are ARM boards, like the NXP LPCXpresso, that cost less than AVR boards, are just as easy to use, and deliver a lot more performance:

http://ics.nxp.com/lpcxpresso/

Leon Heller G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Everyone is astonished at the price and feature set of 50MHz ARMs compared to Mega168s/Mega128s (the two most common AVRs). Everyone assumes that they too will be converting to ARMs in the not-too-distant future. But most people don't.
Inertia is a big reason. How many processor structures do you learn and use before you don't want to learn another one? Usually three. I've done five or six (6502/6800/68HC11, 8051, 8088/80286, PIC, AVR,...). I start reading the ARM data sheet and my brain turns to mud and says 'no!'.
Another reason is cost. The ARM chips cost $3-$5US, but they are tiny-pin surface-mount and require pre-made custom application-specific PCBs. Or prototyping boards with an ARM already mounted. So the cost isn't $2 per chip like the Mega168; it's $25-$40 per ARM prototyping daughterboard. And it's the cost of the tools. Not so much the software compilers as there are some GNU tools for the ARM, but the JTAGICE debuggers.

And you have to learn a new language: ARM-C. There is a fantasy that C is a learn_once:apply_everywhere language. That only applies to complex OS systems like Linux, Windows, and mainframes running abstract high-level applications. The closer that you get to the printed circuit board, the more different C gets from other Cs. Don't believe me? Try translating a 10000 line complex C program from 8051 variant to AVR.

And who needs 50MHz system clock speed? Most applications need either more RAM or ROM, or one of the peripherals that is in the Mega but not in the Tiny.
AVRs are designed for small applications developed by small teams or individuals (either in or out of a corporate framework). Like Arduino aps, sensor monitors/data loggers, or airbags. ARMs are designed for complex applications that will be produced by large teams making 1000000+ units of a design and selling in the $300-$100 range each. Like smart phones.
So, yes, AVR developers will continue to look at ARMs with envy and astonishment, but relatively few will cross over. They are different device types with completely different development frameworks.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

And it's the cost of the tools. Not so much the software compilers as there are some GNU tools for the ARM, but the JTAGICE debuggers.

?

From my POV it seems like the ARM JTAG things are cheaper than the AVR Dragon/JTAGICEs. Atmels absolutely low offer is the Dragon @ $40 - and it is a proprietary/closed/non-disclosed standard. E.g. the Stellaris family is OCD'ed by e.g. a FTDI2232 chip, and AFAIK the protocol is open. And I think thee.g. the LPC-Link is also cheaper than the Dragon.

As of January 15, 2018, Site fix-up work has begun! Now do your part and report any bugs or deficiencies here

No guarantees, but if we don't report problems they won't get much of  a chance to be fixed! Details/discussions at link given just above.

 

"Some questions have no answers."[C Baird] "There comes a point where the spoon-feeding has to stop and the independent thinking has to start." [C Lawson] "There are always ways to disagree, without being disagreeable."[E Weddington] "Words represent concepts. Use the wrong words, communicate the wrong concept." [J Morin] "Persistence only goes so far if you set yourself up for failure." [Kartman]

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

USB JTAGs can cost a lot less than the AVR JTAGICE. A parallel port JTAG can be built for $10.

High-end ARM9 and ARM11 devices are widely used in phones and the like, but the low-cost Cortex-M0 and -M3 parts are being used for the same sort of low-volume applications as the AVR.

A board like the $25 LPCXpresso can be plugged into a low-cost prototype board, or even a solderless breadboard.

Leon Heller G1HSM

Last Edited: Sun. May 29, 2011 - 07:44 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

it's $25-$40 per ARM prototyping daughterboard

You haven't looked at the STM32 and LPC Xpresso boards have you?

The Xpressos are effectively an "Arduino" and a "Dragon" for $22 http://search.digikey.com/script... .

The STM32 boards are <$12: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=497-10633-ND - try buying an AVR board (say an Arduino) for $12 !

And ARMs have "OpenOCD" (that is Open On Chip Debugger http://openocd.berlios.de/web/ ) which means you can build a parallel port wiggler for the same few dollars an AVR parallel port programmer costs. Because Atmel keep the AVR JTAG/dW protocols "proprietary" there's nothing like this and, in reality, to debug them you are looking at $50 or more.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If a Bulldozer was the size of a shovel I'd never be able to get my ass in the seat and I'd probably break off the controls at the first push. So much for bad metaphors.

Okay, if what you guys are saying is true, then why isn't there a mad rush to the LPC? If there is an ARM system that is arduino-easy then why aren't they selling like hot-cakes at MakerFaire? Where are all the ARM hobbyist projects? Why aren't ARMs being used for touch screens and automobile engine sensors rather than AVRs? Where is the equivalent to ARMFreaks?

Smiley

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

They are selling very well, just look at what people are doing with them on the mbed and LPCXpresso forums. mbed had a big stand at the last two UK Maker Faires, which was well-attended last year. I didn't go this year.

My LPC2000 Yahoo group is the equivalent of AVR Freaks, we have over 9,300 members.

Leon Heller G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

clawson wrote:
Quote:

there's almost nothing to configure on an AVR to initially get it running, unlike an ARM

Quote:

Easier to get up and running.


You guys should check out this new fangled programming language called "C". Sure there's a bit of Asm code called the "C runtime" that needs to be written just once for any processor and, yes, it's written in Asm. But the great thing about C is that the chances are someone else has already written the C runtime for you and in addition to the kind of things it does on AVR (cli, set stack, copy .data, clear .bss) it has the added complication of running PLLs up to speed, setting up SDRAM controllers so that the RAm is refreshed and .data/.bss can be accessed and so on. But the really great thing, as I say, is that there's a strong chance the silicon vendor or one of "the gurus" has already written this CRT for your CPU/board so you just write C like you do on any processor and it simply works.

Cheeky fellow! But you're really being too hard. You still need to understand the various flags and constants that can be set in the startup code, which is not much different from writing the code in the first place. And then you need to understand the other registers that control setup options not found in startup code. Just look at all the settings for configuring some ARM family GPIO pins for one example.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here's an ARM-based board with an Arduino footprint. There may be others, I didn't bother to search.

http://leaflabs.com/devices/maple/

Don Kinzer
ZBasic Microcontrollers
http://www.zbasic.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

AVR has been chosen on our school because it's simple to use and has clear datasheets.
I can't say the same about the 2 LPC (nxp) demo-kits I have here.

Sure, when in a professional environment ARM will probably be chosen instead of AVR.

(And NXP is Dutch, which makes it better than any other :P )

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That ARM-based Maple "Arduino" is very expensive when compared to the Uno32.

Leon Heller G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I wouldn't use an ARM for an app. that can be done with an 8 / 16 bitter. It's a BIG waste of resources and the wrong tool for the job. Shouldn't engineers do their thing with as much efficiency as possible (No professional responsibility ) ? It seems that the ARM companies are selling below market to get rid of ANY non-ARM competition out of corp. greed. I think the only real issue is...in the end will they succeed in pushing it all the way through ( what can 8/16 bit companies do to protect themselves...) ? End-user companies don't care ( greed ) if their decisions to use an ARM in an AVR app. may put the little bitter MCU companies out of business, but if there was some trend that jeopardized their co. they'd care about that ( hypocrisy ). The right tool for the right job is the advantage of AVR over ARM. I take what little stand I can against what the ARM companies are doing in their pricing. "I might as well just use ARMs for all my work because the price...", is easy to do and say. But then none of us own one of these companies that are or may get in the ARM crosshairs.

Sony underpriced their TVs decades ago and put ( Magnavox ? ) out of business. Now THAT'S dirty dancin' :evil:

1) Studio 4.18 build 716 (SP3)
2) WinAvr 20100110
3) PN, all on Doze XP... For Now
A) Avr Dragon ver. 1
B) Avr MKII ISP, 2009 model
C) MKII JTAGICE ver. 1

Last Edited: Mon. May 30, 2011 - 12:59 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

indianajones11 wrote:
I wouldn't use an ARM for an app. that can be done with an 8 / 16 bitter.

Even if the ARM could do it better/faster/easier, or for lower cost?

Quote:
It's a BIG waste of resources and the wrong tool for the job. Shouldn't engineers do their thing with as much efficiency as possible (No professional responsibility ) ?

I like how much more efficiently ARMs do 16 and 32-bit manipulations. :)

What is "efficiency"? Quickest development time? Using already-owned & understood tools? Lowest cost? Fewest unused transistors on the die?

Quote:
It seems that the ARM companies are selling below market to get rid of ANY non-ARM competition out of corp. greed.

I think most "ARM companies" also have 8-bit offerings, so why would they compete against themselves?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

1) If the 8 bitter is handling the app., using the 32 bitter's a BIG waste.

2) When the app. needs alot of those, then an ARM is the tool to use.

3) We're talkin' about an 8 vs. 32 bit MCU, that's not a good enough definition of inefficient for you ? :wink:

4) People are talking about replacing with a 32 bitter, whether the ARM co. has 8 bitters or not.

1) Studio 4.18 build 716 (SP3)
2) WinAvr 20100110
3) PN, all on Doze XP... For Now
A) Avr Dragon ver. 1
B) Avr MKII ISP, 2009 model
C) MKII JTAGICE ver. 1

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

indianajones11 wrote:
1) If the 8 bitter is handling the app., using the 32 bitter's a BIG waste.

A waste of WHAT, exactly? Let's say, for example, I have an app that could fit into an ATmega328 ($2.33/100), or an LPC1111 ($1.29/100). Assume further that all other costs are identical between the two choices. What will I be wasting if I choose the LPC1111?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

kk6gm wrote:
Assume further that all other costs are identical between the two choices.

But that is just the thing, costs are never identical - that is what makes engineering hard ( and interesting ). If nothing else, the weight applied to different considerations will vary given application, timeline, etc. In many commercial situations price may be the main driving concern. In others, it may be peak performance ( for varying definitions of "performance" ). In still others it may be primarily expedience or ease of use ( which may not be the same thing ).

While I haven't moved to ARM I believe they are definitely becoming attractive across a wider field of applications. I still do much of my work with standard AVRs because I do not need the increased speed, I have the equipment and knowledge and - to some extent - I enjoy the challenge of 'fitting' the functionality into a, perhaps, tight package. Then again, I work at small volumes where development time ( which includes training time ) has a larger effect on final costs than bare chip price does.

Martin Jay McKee

As with most things in engineering, the answer is an unabashed, "It depends."

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

mckeemj wrote:
kk6gm wrote:
Assume further that all other costs are identical between the two choices.

But that is just the thing, costs are never identical - that is what makes engineering hard ( and interesting ).


Quite true, but one has to make some bounding assumptions to isolate different areas of discussion. I'm trying to find out how a 32-bit device is a "waste" if tool and PCB costs are equivalent, and if the 32-bit part itself costs much less than the 8-bit part.

Now if somebody wants to argue that ARM tool costs are always considerably higher than AVR costs, or that ARM board costs are always considerably higher than AVR board costs, then I'm all ears.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I can only talk as a DIY hobbyist. DIP packages and 5 volt logic levels are what I want to continue to use. I certainly not against 32 bit processors if I can use them on a platform as easy to learn and use as the Arduino IDE (lighten up, again I'm a hobbyist, I do this for fun and challenge not to make a boss happy), but I think the lower voltage and current drive of most arms would increase circuit complexity for the projects I like to work on. I know DIPs will soon go the way of crank start cars but at my age I think I can survive on e-bay sellers of used parts to supply my needs if AVR and others abandon me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Also as a DIY hobbyist, I do feel an increasing loom of the "ARM" threat. Considering that I don't have much time with the AVR architecture, I do feel like coming late to the party. Everyone seems too eager to jump the ARM bandwagon.

However, if Atmel continues to refresh its current 8bit line up and introduce the much needed features like USB,SRAM, GLCD,fpu etc. as standard features, I do think it can survive the ARM onslaught.

Unfortunately, introducing the much needed features in its core lineup will likely cannibalize sales of other products.

I don't think Atmel even sees ARM on the radar. It's not until Atmel has recognized the ARM threat that Atmel will do something drastic.

When given the option over PIC, I think most AVR users will chose AVR because we likely want 1 chip with the most complete feature set instead of 5-7 chips.

I find it interesting that ARM has turned the tables. It seems like Atmel chips are the ones that are in the lacking...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MarioRivas wrote:

When given the option over PIC, I think most AVR users will chose AVR because we likely want 1 chip with the most complete feature set instead of 5-7 chips.

That's rubbish! PICs have always had just as many features as AVRs, and have more in many cases with their latest devices. What PIC application uses 5-7 chips where an AVR would only need one? Microchip wouldn't be number one in 8-bit MCU sales if that was the case.

Leon Heller G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Microchip wouldn't be number one in 8-bit MCU sales if that was the case

Not sure what years you're mentioning, but those "sales" numbers are not last year's.

ATMEL 2010: $1,644,100,000
MICROCHIP 2010: $947.73 M

>$700 is a big difference in sales between both companies.

Also, I mention "5-7" chips needed in reference to Microchips tendency to manufacture chips with a VERY limited feature set. Where as Atmel likes to diferentiate its uCs by amount of memory(ex. Atmega88 Vs Atmega168 Vs Atmega328), Microchips produces whole "families" lacking timers, USARTs, EEPROM, SPI, IC2, different CPU speeds, etc. Just look at "Family 12."

I will take A SINGLE Atmega328 over those THIRTEEN Microchip mcus any day!

I want it all!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You are still spouting rubbish!

Microchip has been number one in 8-bit MCU sales for the last few years and Atmel has been in fifth place.

Those PIC12 devices you mentioned are little 8-pin chips costing about 50c. It's rather silly of you to compare them to the far more expensive ATmega328.

One reason why PICs outsell AVRs is that customers can select devices with just the features that they need, reducing the cost. If they want a full-featured device in a particular family, it is available. They also make devices with more features than AVRs, like on-chip regulators, digital modulators, peripheral pin select, etc. A wide range of very low power devices is available, with power consumptions far lower than those offered by Atmel (down to 15 nA in sleep mode).

Dp you still maintain that PICs have fewer features than AVRs?

Leon Heller G1HSM

Last Edited: Mon. May 30, 2011 - 05:11 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Assume further that all other costs are identical between the two choices
. I wouldn't ASSUME it, it either is or it ain't .

kk6gm wrote:
What will I be wasting if I choose the LPC1111?
For your cherry-picked example, not much. But the thread is about the price of ARMs in general, not the '1111, and that one can get a much more powerful MCU w/ more peripherals for around the same or less price of an 8 bitter. Also, the '1111 datasheet states it was designed for 8 bit apps. WHY water it down by surrounding the 32 bit core with an arch. that's appropiate for an 8 bitter ? WHY not just use the full-blown ARM chips ( I wonder if there's a 5 letter word to describe such situations... ) that are already out and NEVER come out with such a MIXED design ? Yeah, it brings the MCU in range of the cheaper 8 bitters, but took making it look like an 8 bitter to do it, except for the core.

kk6gm, the topic context and some technical know-how ( which I'm sure you have ) make it OBVIOUS what I mean when I use the word waste ( But if you, or anybody else doesn't think so,...we'll just leave it at that. :roll: ).

1) Studio 4.18 build 716 (SP3)
2) WinAvr 20100110
3) PN, all on Doze XP... For Now
A) Avr Dragon ver. 1
B) Avr MKII ISP, 2009 model
C) MKII JTAGICE ver. 1

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In the first place, I quoted the wrong ARM part. The LPC1114, not the LPC1111, has 32k of flash. It's ($1.75/100).

Beyond that, why do you call my example "cherry picked"? The ATmega328 is a very popular AVR, and a valid target for such comparisons. If you'd prefer, we can compare a Cortex M3 part like the LPC1313 ($2.35/100).

Like I said in an earlier post, I see many AVR advantages such as 5V capability, high output drive, DIP parts, ease of use, and (one I forgot), onboard EEPROM. But if none of those are a particular consideration, the fact remains that you can get ARM parts with equal or greater power, for equal or less cost.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Also as a DIY hobbyist, I do feel an increasing loom of the "ARM" threat. Considering that I don't have much time with the AVR architecture, I do feel like coming late to the party. Everyone seems too eager to jump the ARM bandwagon.

However, if Atmel continues to refresh its current 8bit line up and introduce the much needed features like USB,SRAM, GLCD,fpu etc. as standard features, I do think it can survive the ARM onslaught.

Unfortunately, introducing the much needed features in its core lineup will likely cannibalize sales of other products.

I don't think Atmel even sees ARM on the radar. It's not until Atmel has recognized the ARM threat that Atmel will do something drastic.

When given the option over PIC, I think most AVR users will chose AVR because we likely want 1 chip with the most complete feature set instead of 5-7 chips.

I find it interesting that ARM has turned the tables. It seems like Atmel chips are the ones that are in the lacking...

Mario - ARM don't make chips - they just license the processor design. Atmel has had an ARM license for some time and make quite a few products based on the ARM architecture. Your comments regarding PICs don't make sense - I think you would find it hard not to find a PIC with the same if not similar feature set for each AVR.

I haven't come across an ARM based offering with eeprom - that is a defining factor for many of my apps. Sure, you can use flash, buts that's a different beast.

As for using a 32bit cpu for an 8 bit app a waste? How about using a 8bit micro for a 4bit app? Or the converse, using a 8bit micro for a 16 bit application? Is that inefficient?
Ultimately, if we get the job done and get the cash, we've satisfied the business end of the deal.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The processes used for ARM devices don't support EEPROM.

Leon Heller G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

So it would seem, Leon. Freescale mention magic flexram for their Kinetis devices but I have yet to look closely at these.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

IMO:

As a hobbyist I am tempted by the possibilities to choose a ARM Cortex that has peripherals I need for a specific peoject built in. An ethernet interface is the most obvious example. This is a pro-ARM argument.

In my activities I'd like spend as much time as possible on firmware work, and as little as possible on hardware. Since there are no PTH ARM parts this argues against them.

For a hobbyist, typically doing one-off projects, a price difference in the order of a $ does not matter much, if at all.

As of January 15, 2018, Site fix-up work has begun! Now do your part and report any bugs or deficiencies here

No guarantees, but if we don't report problems they won't get much of  a chance to be fixed! Details/discussions at link given just above.

 

"Some questions have no answers."[C Baird] "There comes a point where the spoon-feeding has to stop and the independent thinking has to start." [C Lawson] "There are always ways to disagree, without being disagreeable."[E Weddington] "Words represent concepts. Use the wrong words, communicate the wrong concept." [J Morin] "Persistence only goes so far if you set yourself up for failure." [Kartman]

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ARM chips in SOIC are available from Stellaris/TI.

Leon Heller G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

But the thread is about the price of ARMs in general, not the '1111,

Ah that explains your ill considered arguments above. The thread is exactly about the ARMs that are competitive with AVR8's and AVR32's. That is the Cortex M0 and M3 devices. In my experience you get more features for less money with the M0/M3's. All this talk of "too powerful" is complete tosh. Say I have a car with a 300bhp engine capable of a top speed of 170mph I can still use it to pop down to the local shops to buy a loaf of bread (in fact I did exactly that about 10 minutes ago). The great thing about it is that when I DO want to accelerate to leave virtually every other car on the road standing I can (and do ;-)).

Just to give another comparison. An NXP LPC 1343 is a 48pin CortexM3 with USB device facilites for ~$6. I guess the closest AVR8 model is ATmega32U4. It too costs ~$6. One can run at 16Mhz the other can run at 72MHz. One has 2.5K of RAM the other has 8K of RAM. The only thing I can see in the AVRs favour is the 1K of EEPROM so I guess it depends whether your application needs to use NVM or not.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

One can even use the same software tools for both AVR and ARM. IAR and Rowley use the same IDE for both families.

NXP has the tiny little 2mm square wafer-scale LPC1102 with 16 bumps; that is probably smaller than any AVR. I'm putting one on a 0.3" DIP14 PCB.

Leon Heller G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ultra-low power consumption is rather easy to achieve with an AVR. Not possible with an ARM, be it a cortex-m3 or whatever.

Bulk pricing is lower for AVR-s, I am afraid.

Don't bury the AVR for no reason, it has its place.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There are ARM Cortex-M3 and -M0 devices with very low power requirements:

http://www.energymicro.com/

They will use less power than AVRs in many applications:

http://www.energymicro.com/techn...

Leon Heller G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Wow, I did not check this thread for a day after posting. I see I have set off a few sensitivities here! :D

I do agree that AVR is much more appealing from a hobbyist and student point of view. I think that cutting your teeth on a 8-bit ISA is quite a bit easier than a 32-bit.

But I was really trying to consider many of the large scale applications being put out today. Through my school years we really only focused on 8-bit stuff unfortunately. Up until very recently I was of the opinion that most of the simple embedded devices use an 8-bit MCU and that 32-bit was the exception. However I feel that I am finding the opposite is true.

I was able to meet with a few reps from Medtronic (HUGE medical device manufacturer if you're not familiar) and they were saying basically any device they make with an MCU in it is ARM. They sell a pacemaker with an ARM core embedded in it for goodness sake.

I also just started doing some firmware design for a company that makes commercial scales, and another that makes pretty simple climate control panels for the home. In both cases they are using 32-bit Coldfire MCUs running a full RTOS. The nicer ones are even running a linux kernel. I did not expect that level of computing at all.

Maybe I'm just not connected to the right people, but I feel that just as I'm becoming really comfortable with my 8-bit AVRs I am discovering that they will not cut it it most cases. Of course much of what I've learned can be applied to 32-bit MCUs. I love what I've learned with the AVRs and they are great, but I think I'm going to be focusing on ARM over the next year as I complete my graduate program.

PS: Happy happy memorial day to my fellow freaks! May you have a wonderful day of grilling! :D :D :D

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

One way to look at the matter is to consider that an MCU core often takes up only a small portion of the total silicon die. Flash and SRAM and peripherals account for much more of the die area. That is what is driving the move to 32 bits (along with expectations of ever-increasing functionality in even the simplest of devices). Where an app can fit into just a few k of flash and a few hundred bytes of RAM and a few simple peripherals, then that 8 bit die can be considerably smaller than a 32 bit die.

Your decision to add ARM knowledge to your AVR knowledge is exactly right, IMO. Those two architectures will give you a good, balanced foundation for a large range of embedded work.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

Of course much of what I've learned can be applied to 32-bit MCUs.

Surely the most important point of all? The fact is that an SPI is an SPI and a UART is a UART and whether you learn it on AVR or ARM you should be able to reuse (most of) that knowledge on the next CPU. If you program in C then the actual Asm language is fairly immaterial so it's not really a question of "AVr or ARM" but BOTH.
Quote:

Wow, I did not check this thread for a day after posting.

You want to see the previous discussion on this same subject - it petered out after 27 pages.