pricing strategy?

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

I'm quite new to micro controllers so this may be a stupid question but I just don't understand the pricing strategy around these products.

I have started out (and continuing ) my first project using the arduino. Outstanding starting point with easy on-ramp experience.

I found the atmega a bit cramped for space and timers, so I thought I would give the atmega2560 a try. Nice improvement, but when I looked at prices on mouser and digikey I have a hard time understanding the price differential. It would be easy to understand if they were the only solution, but the xmega seems far more capable and a lesser price, and many ARM chips are substantially cheaper with more capabilities and faster and with the same claims of lowest power.
I thought I would try the xmega ATxmega256A3U, but that would require me to use some other development environment. I've tried in vain to understand the avr studio 6.0, this seems like some monster with poor documentation, and very bad performance. (BTW "“ this product is in desperate need of a splash screen that comes up when starting to let the user know it is starting"“ not 5 minutes after - this would prevent starting multiple session. I'm not complaining "“ if I had paid $$$ for it I would complain)
Back to the point "“ is the atmega2560 more advanced? Or is no longer in production and the price is for a scarce supply? If I were to build a product based on this chip, would I find it at the end of life before I start?

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

Quote:

Back to the point – is the atmega2560 more advanced? Or is no longer in production and the price is for a scarce supply? If I were to build a product based on this chip, would I find it at the end of life before I start?

2560 has been around for 5+ years and is "current". It costs a lot because it has a lot of flash. Unless you plan on loading it with "dense" data like audio samples, bmp images, fonts or similar then a 2560 is way beyond a single engineer's reach to fill with code. It could take literally man years to write that amount of tested/debugged code.

Most hobbyists go with the ATmega1284P as the "ultimate" device. It comes in easily bread/vero-boardable 40pin DIP packaging, has 128K of flash and a whopping 16K of SRAM (double any other large AVR).

But, yes you are right, with Cortex M0 and M3 it's now not unusual to have 128/256/512KB of flash for just a few dollars and in realms that need that kind of memory the ARM's start to look far more price competitive.

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

Quote:

when I looked at prices on mouser and digikey I have a hard time understanding the price differential.

Compared to what? A mega328-based Arduino?

First of all, what are you comparing? Qty. 1 prices? Those aren't really meaningful for any "value" comparisons. E.g., DigiKey has a $25 minimum order, so even a single $1 chip is $30+ delivered to your door.

Now, examine some of the ARM-mentioning threads here. In particular, a recent "are 8-bit micros dead?".

Wouldn't you expect a bigger AVR to be more expensive than a smaller AVR? Yes, from time to time there are anomalies in the pricing structure but in general bigger is more expensive. '328 to '2560 you have 8x the flash, (IIRC) 8x the SRAM, ~3x the I/O, and double most of the peripheral subsystems.

Also, the new "A" variants are in general a bit cheaper. Are you comparing '328A prices to non-A '2560?

Indeed the new Xmegas with USB give good value.

It depends on the needs of your app and the quantity used in production. I use a lot of different AVR8 models for the same reasons you mentioned--I'm familiar with them and already have development chain in place. If I were to start with a fresh sheet of paper for a qty. 10000 design, then I might switch families if the features and/or cost warrant the learning curve and obtaining new tools.

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

It is not the flash size, it is that no one uses an atmega2560 so there are no scales of economy behind that chip.

On ARM 256k is common and well priced, once you run a RTOS, an ethernet stack, or graphics, then 256k starts looking small. You can not really do those things on an AVR, well you can but it is not very common as the computational horsepower is a bit low to do them well.

Keep in Mind that ARM op codes are 16 or 32 bits long, so you would need somewhere in the neighborhood of 384-512k of flash for a 256k AVR program assuming you do very little to port the code.

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

Quote:

Keep in Mind that ARM op codes are 16 or 32 bits long, so you would need somewhere in the neighborhood of 384-512k of flash for a 256k AVR program assuming you do very little to port the code.

On Cortex M0 the minimum opcode size is 16 bit. On AVR tiny/mega/Xmega the minimum opcode size is 16 bit. So what's the point you are making?

Here's a test program showing M3 code (I don't have an M00 compiler):

root@ws-czc1138h0b:~/arm-2007q1/bin# cat test.c
#include 

int main(void) {
  printf("hellO");
}
root@ws-czc1138h0b:~/arm-2007q1/bin# ./arm-none-linux-gnueabi-gcc -nostartfiles -mthumb -mcpu=cortex-m3 -g test.c -o test.arm
/home/clawson/arm-2007q1/bin/../lib/gcc/arm-none-linux-gnueabi/4.2.0/../../../../arm-none-linux-gnueabi/bin/ld: warning: cannot find entry symbol _start; defaulting to 000081a4
root@ws-czc1138h0b:~/arm-2007q1/bin# ./arm-none-linux-gnueabi-objdump -S test.arm 

test.arm:     file format elf32-littlearm

Disassembly of section .text:

000081a4 
: #include int main(void) { 81a4: b580 push {r7, lr} 81a6: af00 add r7, sp, #0 printf("hellO"); 81a8: 4802 ldr r0, [pc, #8] (81b4 <.text+0x10>) 81aa: f7ff eff6 blx 8198 <.text-0xc> } 81ae: 46bd mov sp, r7 81b0: bd80 pop {r7, pc} 81b2: 0000 lsls r0, r0, #0 81b4: 000081b8 .word 0x000081b8

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

At a recent seminar touting ARMs they told me that one of the plusses for ARM (Cortex M3/4) vs. AVR was the code density.

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

My question is why would someone choose the atmega2560 rather than the xmega series.

My thought is that the atmega series is less capable and should have been priced less than the comparable sized xmega. This is what leads me to question the life of the atmega (at least the 2560) and being mystified at the pricing strategy.

(I was only looking at prices for 100 - to 2500 units this maybe changes as you get into the > 100,000, but didn't take the time spec it out with all of the emails for custom quotes.)

As my application is a simple monitoring device (which does read/write to SD cards and does connect to the internet and is approaching the code limit)

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

erg144 wrote:
My question is why would someone choose the atmega2560 rather than the xmega series.

In the embedded world, the answer is often: it depends. :)

The choice of micro is very dependent on the application itself.

The product life-span also is considered.

Cost too.

All of these have to be balanced with other factors too.

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

I guess once you fill up the flash in the biggest mega avr, the decision to go to a bigger size cpu has been decided.

Imagecraft compiler user

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

Apart from the economic drivers of supply and demand, at the end of the day chip prices (in volume at least) are set by silicon area. An old, smaller (K capacity) chip on a large scale geometry like 0.18um may actually occupy more silicon area than that of a higher capacity chip manufactured using a tighter geometry such as 0.12um. In this sense I guess 2560 is probably "longer in the tooth" than things such as the A suffixed AVRs and probably most M0 and M3 ARMs.

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

Just to be clear - I'm filling my atmega328. Was looking at one time to move to 2560 till I saw the price. While that price would be ok if the 2560 owned the market (Only chip on the block), it just seemed very strange that the price was more than what I thought was a more capable chip.

Who was the target market for the atmega2560? - seems like it would only be existing systems that could not afford to change. This is what led to the comment about the life span of this device. I would not want to base a product on a limited life chip.

For this current iteration of my project, I have elected to use multiple atmega 328 (one required, one optional) that communicate over I2c. This reduces the chip price, but increase the board expense.

Thanks for all the thoughtful responses.

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

I think the main 2560 customers are AVR users that ran out of flash in their 128K models...

Imagecraft compiler user

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

Atmel seem to have about a 10 year life cycle for most models or if something is superseded earlier than this they tend to offer some kind of backward compatibility mode in new devices. Xmega are a bit different in that they are all effectively the same chip - just with some bits fitted or not.

The next logical step after 328 is surely a 64K chip and if you ever manage to fill that then a 128K chip then if you ever manage to fill that a 256K chip. It seems an odd choice to jump straight from 32K to 256K. If you want to do that (in volume) expect to pay about 8 times the price (though perhaps a bit less because of economies of scale).

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

clawson wrote:
Quote:

Keep in Mind that ARM op codes are 16 or 32 bits long, so you would need somewhere in the neighborhood of 384-512k of flash for a 256k AVR program assuming you do very little to port the code.

On Cortex M0 the minimum opcode size is 16 bit. On AVR tiny/mega/Xmega the minimum opcode size is 16 bit. So what's the point you are making?

Here's a test program showing M3 code (I don't have an M00 compiler):

root@ws-czc1138h0b:~/arm-2007q1/bin# cat test.c
#include 

int main(void) {
  printf("hellO");
}
root@ws-czc1138h0b:~/arm-2007q1/bin# ./arm-none-linux-gnueabi-gcc -nostartfiles -mthumb -mcpu=cortex-m3 -g test.c -o test.arm
/home/clawson/arm-2007q1/bin/../lib/gcc/arm-none-linux-gnueabi/4.2.0/../../../../arm-none-linux-gnueabi/bin/ld: warning: cannot find entry symbol _start; defaulting to 000081a4
root@ws-czc1138h0b:~/arm-2007q1/bin# ./arm-none-linux-gnueabi-objdump -S test.arm 

test.arm:     file format elf32-littlearm

Disassembly of section .text:

000081a4 
: #include int main(void) { 81a4: b580 push {r7, lr} 81a6: af00 add r7, sp, #0 printf("hellO"); 81a8: 4802 ldr r0, [pc, #8] (81b4 <.text+0x10>) 81aa: f7ff eff6 blx 8198 <.text-0xc> } 81ae: 46bd mov sp, r7 81b0: bd80 pop {r7, pc} 81b2: 0000 lsls r0, r0, #0 81b4: 000081b8 .word 0x000081b8

I always thought that most of the AVR op codes were 8 bit, but 16 makes more sense.

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

Did you check the prices of the >=256kB devices ( at 1000 quantity)?

For 256k devices:

Renesas RL78(2,92$):
http://search.digikey.com/us/en/...
STM CM3(3$)
http://search.digikey.com/us/en/...
Atmel CM3(3,24$)
http://search.digikey.com/us/en/...

This code size is where most 8-bit families (including CM0) do not reach/compete. The limit is 128kB:

STM 8S(1,55$)
http://search.digikey.com/us/en/...
Atmel XMega(2,16$)
http://search.digikey.com/us/en/...
STM CM3(2,21$)
http://search.digikey.com/us/en/...

Edit: Oops, I didn't notice that Digikey does not offer reduced 1k pricing with Atmel's SAM3 and XMegas - those two quotes are incomparable with the remaining 1k quotes then. That is why I have filtered Digikey once again, but with a quantity of 100 pieces, and picked one of each family.

>=256kB of flash category:
Atmel Xmega(3,23$)
http://search.digikey.com/us/en/...
Atmel CM3(3,24$)
http://search.digikey.com/us/en/...
Renesas RL78(4,28$)
http://search.digikey.com/us/en/...
STM CM3(4,4$)
http://search.digikey.com/us/en/...

And in >=128kB category:
Atmel Xmega(2,16$)
http://search.digikey.com/us/en/...
Microchip PIC18(2,22$)
http://search.digikey.com/us/en/...
STM 8S(2,26$)
http://search.digikey.com/us/en/...
Atmel CM3(2,3$)
http://search.digikey.com/us/en/...

So when someone needs small, like ~100 pcs quantity then this second listing applies.

No RSTDISBL, no fun!

Last Edited: Tue. May 1, 2012 - 05:41 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You are quoting 100-off prices. I think OP got "sticker shock" from the 1-off price since he wants to replace one 328 with one 2560 (or whatever).

1-off pricing is a whole new science and often has no discernible bearing on volume pricing. Some parts are only -10% at 100-off while others have stupid ratios like -70% or even -90% or something.

On the face of it $17 for a single 256K chip sounds like a lot! It makes something like this look like pretty good value:

http://www.ebay.co.uk/itm/New-Ar...

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

Actually I thought:

Quote:
If I were to build a product based on this chip, would I find it at the end of life before I start?

it is not about one chip, but more about a pricing strategy of the distributors/manufacturers.
Quote:
OP got "sticker shock" from the 1-off price

:)

No RSTDISBL, no fun!

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

Quote:

I think OP got "sticker shock" from the 1-off price

OP clarified:
Quote:

(I was only looking at prices for 100 - to 2500 units

And indeed the '2560 is multiples of e.g. '328 in qty. 100. And Brutte pointed out that other lines than AVR8 might be a good value in that type of configuration (8K+ SRAM; 128K+ flash).

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

Quote:

And indeed the '2560 is multiples of e.g. '328 in qty. 100

X5.8 in fact

yet it is X8 the flash.

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

Quote:

yet it is X8 the flash.

Certainly. There is more "stuff". "Is the 8-bit micro dead?" Atmel needs to keep the pricing structure for the AVR8 so that when you get more you have to pay more.

And as pointed out there are new microcontroller families with a 'flatter' pricing structure.

Quote:
(Brutte)
This code size is where most 8-bit families (including CM0) do not reach/compete. The limit is 128kB:
...
Atmel XMega ...

Hmmm -- Four bucks ($4) for ATXMEGA256A3U with 16k SRAM, 256k flash, 50 I/O, and USB isn't too bad.

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

Quote:
Hmmm -- Four bucks ($4) for ATXMEGA256A3U with 16k SRAM, 256k flash, 50 I/O, and USB isn't too bad.

Atmel Xmega(3,87$/100pcs)
http://search.digikey.com/us/en/...

What I meant is that this is a (shrinking) island on the 8-bit market. Simply most small 8-bit families do not reach that far so the XMega and perhaps a bunch of other outsiders do not compete with themselves but with 16 and 32-bitters in this region. And there is no way ATXMega256 could compete with those chips - check prices of other USB chips:
Atmel CM3, 256kB(4,1$/100pcs)
http://search.digikey.com/us/en/...
or:
Toshiba CM3, 512kB(5$/100 pcs):
http://search.digikey.com/us/en/...

I would say this situation is very similar on the opposite end of the market, with LPC1110-ish 32-bitters. These do not compete among themselves, but with much cheaper 8-bitters there.

No RSTDISBL, no fun!

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

So how do the arm dudes get better yield? Smaller features? Can't Atmel just order up some AVRs from from the same fab and process that the ARM guys use and get a 66MHz AVR?

Imagecraft compiler user

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

This is a BIG thank to whoever 10K+ is. For some reason, I thought I would need to either user the 328 or the 2560. I never really looked at the atmega1284. This chip is only a bit > $1 more, has enough memory and the 2nd serial, give me the opportunity to breadboard the design.

Great to have folks on this forum that are so knowledgeable and willing to help.

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

erg144 wrote:
This is a BIG thank to whoever 10K+ is.

The "10K+" is just an "award" for writing more than 10K posts. You can see the user name above that - there is certainly more than one of them.

Sid

Life... is a state of mind

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

After using a number of different AVR devices in projects, I now use the Mega1284 for everything. I recently bought a 100 of them from Digikey for a little more than $4 each. It makes software development easier. I don't have to remember as much...which is good because my memory is not so good anymore. It has all of the capability I need.

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

What about XMega128D3? It is only half of the price of Mega1284.

No RSTDISBL, no fun!

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

Quote:

The "10K+" is just an "award" for writing more than 10K posts.

As I was the one who recommended 1284P I suspect it was me. :oops:

@OP, make sure you get 1284P and not 1284. (more "goodies").

Quote:

What about XMega128D3? It is only half of the price of Mega1284.

When it comes in DIP packaging(*) for bread-boarding it may be of interest to hobbyists ;-)

(*) having said that one of the goodies I got when I past 50K posts(+) was some DIP packaged Xmega in fact. Wonder if they'll ever make it past engineering samples?

(+) and yes Dean I will take those photos one day ;-)

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

I really like the xmega chip set, but lack some of the tooling (really small sdm parts), and know-how. I'm still rather new to this stuff and stumbled onto the arduino and kinda stuck there for for this version of my project. The software is easy and the hardware was at least learn-able at my level.

If I ever do a version 2 of my project, I will certainly look at different chips. I don't need the extra clock speed and associated design challenges, but I do like the built in USB and I would really like a DAC and extra timers. (Only do a little with A/D). I've done a bit of DAC with PMW, but sure seems like it took a lot of code to do some simple things. (Maybe my attitude will change as I get more experience).

Again - thanks for all of the replies.

NOTE TO 10K+ I thought you got that name due to resistance. :)

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

Quote:

NOTE TO 10K+ I thought you got that name due to resistance.

My name is really Cliff Lawson abbreviated to "clawson" at the top left of every one of the 52,750 posts I've made here.

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

erg144 wrote:
NOTE TO 10K+ I thought you got that name due to resistance. :)

Cliff, did that one go over the dome? :wink:

R1 = 10K+

"I may make you feel but I can't make you think" - Jethro Tull - Thick As A Brick

"void transmigratus(void) {transmigratus();} // recursio infinitus" - larryvc

"It's much more practical to rely on the processing powers of the real debugger, i.e. the one between the keyboard and chair." - JW wek3

"When you arise in the morning think of what a privilege it is to be alive: to breathe, to think, to enjoy, to love." -  Marcus Aurelius

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

Pah. In Cliff's case - resistance is futile.

He *will* be assimilated.

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

I'm a software engineer, what do I know about things like "resistance"?

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

I'm an electronics engineer - what do I know about 'software'?

(which, er, takes us right back to the 'how do we teach' thread in AOB...)

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

Give me ready solution for moon rocket: hardware, circuit and software. I tried IAR but it gave errors. ATmega456 does not work. Y?

Plz hlp, ts rlly urgent!

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

Anyone?

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

Is this a school assignment?
Or just a one off hobby project?
Show us what you've done so far.
Hardware and software.
Humans have built rockets for decades, no need to reinvent the wheel, make Google your friend.
Good Luck, interesting project.
Btw you probably could do this with one of the Tiny models if you do it in assembler.

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

JohanEkdahl wrote:
Give me ready solution for moon rocket: hardware, circuit and software. I tried IAR but it gave errors. ATmega456 does not work. Y?

Plz hlp, ts rlly urgent!


Did the North Korean government ask that one. :wink:

"I may make you feel but I can't make you think" - Jethro Tull - Thick As A Brick

"void transmigratus(void) {transmigratus();} // recursio infinitus" - larryvc

"It's much more practical to rely on the processing powers of the real debugger, i.e. the one between the keyboard and chair." - JW wek3

"When you arise in the morning think of what a privilege it is to be alive: to breathe, to think, to enjoy, to love." -  Marcus Aurelius