Microchip coming closer

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

Hi.

This is not an Microchip advertisement but a whining about that Atmel hasn't been able to improve the AVR series in a rapid rate. Especially the ever demanding speed topic.

Microchip and the PIC series is the greatest competitor to Atmel/AVR. I have come to the conclusion that AVR has the greater instruction set, FLASH and a RISC core in ALL parts and Microchips don't. On the small devices (8 pins) Microchip has just OTP devices.

But now thay are releasing something that I think is ment to be a AVR killer. An 8 pin microcontroller based on the "RISC" 18F core. Damn it, it has a whole bunch of things that is equal to some AVR:s and some very are better. First I realy thought it was a true "killer" but reading the data sheet reveals that the so called RISC is still no more than a Reduced Instruction Set. Not the One-cycle-execution. The devices come in clock speed of 40 MHz BUT, the clock is as before, diveded by 4 to get the internal clock and when again with 4 but this is pipelined to make it possible to executo an instruction as on the AVR on 1 or 2 clock cycles.

Some important data:
- 2 or 4 k FLASH, double the size of AVR
- 40 MHz speed. Divided by 4 gives the same speed as the 2343 that is the fastest 8 pin AVR.
- Selectable internal clock speed. Similar to tiny 12 I guess
- 8x8 bit hardware multiplier. Whops, great indeed!
- 16 bit timer
- In circuit programmable. (As all AVR)
- 77 instructions. (Compared to 90 on tiny and 120 on 2323/43)
- µA current consumtion
- Comes in a NARROW SOIC 8. That is 150 mil with instead of AVR 200.
- NO ADC. An important disadvantage for many.
- 256 bytes RAM and 64 bytes EEPROM. That is more RAM but similar or smaller EEPROM than AVR.
- Cost ?????

So, if they could put all that in a such a small package I wonder if Atmel could keep Microchip to eat up all the lead that Atmel have so far. Atmel have clearly showed with their tiny series that they have pit a great effort to be the leading manufacturer of the smallest microcontrollers.

I really hopes that the AVR series will show even more and faster improvement in the future.

Regards, Bengt

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

I'm sorry Bengh, but who the hell will need 4K Flash / 40Mhz mC in 8pin package with just 6 IO pins?
It seems like bicycle with 200HP engine ;)
What will be this appilcation with need of so much power in so small package? Usually the 8 pin micros are used for remote keyfobs etc.

Best regards
Tsvetan

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

"Usually the 8 pin micros are used for remote keyfobs etc."

The continuation of that sentence could read "because they don't have the power to be used for anything else..."

6 IOs would be enough for, say, a serial protocol converter (we do something along those lines using an 8MHz 8535 - the product allow our field engineers to connect their laptop serial port to our other AVR-based products via their SPI connectors and talk directly to the unit rather than having to go via the RS485 network - very handy if you don't know what the unit network address is!). Whilst a lot of flash mightn't be required there, speed almost certainly is, unless you want to restrict the baud rates you're able to work at - at 115k2, you've only got 70us between bytes, and that isn't a lot of time to be translating from one protocol to another. Even at 38k4 we have to delay sending the converted message, not by much, but a delay is a delay...

Another project in the works is to use a Tiny15 for motor control - there we definitely would like a bit more flash. The code will fit into 1KB, but it'll be tight, even though we're just using a simple control scheme. Fortunately we don't *need* particularly precise control in this project, but if we did then it wouldn't be a lack of IO that held us back.

If Microchip think making these devices is a good idea, then they must have some applications in mind for them... For me, embedded systems designer heaven is where the micros have more memory, more speed and more onboard IO/peripherals. Meanwhile, back here in the real world I'll settle for 2 out of 3 :-)

Regards
Chris

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

hi

im not complety agree whit you

recently atmel whit the ralase of the new mega
part is on the good way :-) more powerfull part
tha came whit more flash en more MHz ;-)
whit the new 16 mhz 232 it verry killer for serious project

but the big question is if ther part whil be in the
market en aviable ????

the 161 is stil not aviable at www.pios.com end all other
after near of one year of life .....

end more stable dev tool for cover the completly
range it need a brunch of stok

end by the way the microchip competition is just good
to force atmel to respect ther smal costomer enf keep
it uptodate whit new chip en tecknologie

marc laonde
alphatronic

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

Good points all...some random thoughts from a current PICer, AVR wannabee..

Microchip has a great history of expanding their product line with all manner of on-chip devices.

Atmel is leaps ahead by having the GCC compiler available (kudos to those avr-gcc folks :). Lots more of the on-chip ram and FLASH too.

If Atmel's JTAG developer is available, it should compete with the ICD-II from Microchip which isn't yet available (was August, now October), however, I think Microchip is keeping the ICD at $99)

Ordered both but still haven't received the Network board from Atmel or Microchip.

Why wasn't Atmel at Embedded Systems Conference East Coast ? Micro chip was there, showing their networked evaluation board, and 14 new chips...If atmel was there, I didn't see their booth.

The competition is good news for everyone :)

Ralph

admin's test signature
 

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

dear friends GOD bless you all

this is real pic micros have lots of flash counterparts that mach the atmel avr micros in all forms, except speed.

i have few project with atmel avr, with the at90s2343 and at90s2313, the 2313 work very good but the 2343 have few problems,specialy with the internal clock which don't work properly,this don't happen with the pic, atmel told me to use the attiny22, but the attiny22 is no longer in production, here in los angeles you can get any micro but the tiny22 you can not get it any where. also they told me about a new at90s2343-1pc which should work right with the internal clock, however no one in USA have this chip for sales i did check all distributors, this don't happen with the pic micros even when they are the most popular micros in the world.

i like the avr very much, however atmel have to speed up and cach microchip in they own game, also other have better micros than atmel and microchip:
like signal news 8051 with flash and jtag, also yesterday i did check a chip from fairchild semiconductors, they have a chip 8 pin with the same or better than at90s2343 and attiny15 togethers.
just look what is out there
thank you very much
marcial

admin's test signature
 

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

>im not complety agree whit you

>recently atmel whit the ralase of the new mega
>part is on the good way :-) more powerfull part
>tha came whit more flash en more MHz ;-)
>whit the new 16 mhz 232 it verry killer for serious project

Can you say Microchip DSPic. 16bits with some nifty DSP clipons.

It will be interesting to see how the new Megas are priced as the last 163's are not a long way from the 32bit H8S2148's that we use. The only reason we use Megas is for current consumption.

Issues aside I eagerly await being able to get to work with the later devices. The MLF package looks great, the JTAG will be excellent. Hell we might even finally get to buy an AVR with 2 Async Comms ports.

As stated earlier the 2343 limits some uses by being in such a massive package, the world has moved well beyond 50mil pitch packages eg uMax or 8pos CSP anyone??.

Crying shame that Atmel are sabotaging the AT94 AVR FPSLIC by not providing an entry level devel kit as many other SOC proponents are doing. It is allowing other manufacturers to gain market momentum by allowing other products the chance to become ubiquitous. I remember Microchip doing this a loooong time ago when they took a horrid outdated piece of silicon detritus & proceeded to Market the heck out of it thus taking off out of nowhere to become the successful company of today.

Cheers

Don

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

Interesting.. I have an app that's really squeezing a 2343.
Is it's internal ram all chopped up in pages, banks etc as normal for them, or its it flat, like an AVR?

If it's the usual Mchip mess, I don't think I could port my project.

However, I'm crushed to death in the 2343, and I don't see any AVR relief in sight.
(In an 8 pin package)

admin's test signature
 

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

Hi Dave,

for direct addressing, it is banked and you need to set the banksel register.
If you use one of the 3 pointers inside (FSR0-2), you see the RAM as linear space of 4kB. Program space is no more banked. If you need byte acess to program space, do it with the TBL register.
Other advantage, fast interrupt mode, saves all necesarry registers in their shadows and instruction set very very similar to AVR.

Andy

admin's test signature
 

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

Dave,

I forgot something important, if you acess directly to RAM, you have allways something like zero page on HC11 of first 128 bytes, just set the parameter in instruction. It is very usefull for real multitasking. In some way, similar to AVR 32 registers. And stack is no more problem too. You have acess to stack pointer and it's contents and automatic stack under/overflow flow control.

Andy

admin's test signature
 

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

A funny thing about the internal clock problem for 2343 is that I have a design (not a new chip) that uses the internal clock without any problems. It runs from a Litium 3 V battery giving a speed at about 340 kHz. I have had no problems what so ever with the clock. Maybe it is layout sensitive. It is a very small 2-layer board with a ground planes. The ground plane could be important.

My guessing about the Tiny22 (it is obsolete) that is almost exactly the same as 2343 but in the tiny core (90 instruction) is that they desided to improve the 2343 instead of replacing it with the tiny22.

Maybe it is no suprise that Atmel didn't showed up at Embedded Systems Conference East Coast, they have the unusual policy of giving no information in advance what so ever. Datasheets of parts never heard of suddenly just shows up. But I think they have been better on that so maybe i am exaggerating. For example, the Product overview pictures now include an "available" column.

Dave: Are you using C or assembler? If you are writing in C don't forgot to write your own start-up code, there is a lot to save there if you only have 1 k program memory. (2kbyte=1k program, 16 bit)

/Bengt

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

I have less problems with parts 'Appearing' these days. When the part becomes public it is ready to sample. This way when we commence a design based on a new part, we know that we will be able to buy some which are similar to the part in the spec sheet. I concurr that it is nice to know what is over the horizon as it makes jumping ship a little less likely when Microchip or similar release a new whiz bang device. The DSPic would be less of an interest for us if the FPSLIC, for example, had some entry tools available.

Not too long ago we got bitten by another manufacturer who did a very major change in a bit which we had started designing around based on an early spec sheet. Lots of work lost. Serves us right...

Cheers

Don

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

My 2c worth !

Every now and then a part comes along which, by some fluke or other, becomes so popular and widely used that the economies of scale are staggering and after 5 or 10 years the thing becomes available for only a few cents. Good examples would be the BC548 transistor, the LM324 op amp, 7805 regulator, etc. Theres absolutely no way that a new transistor offering can compete with a BC548 on its own territory, because a new transistor would have to be priced at 30+ cents to justify its development and marketing whereas the BC548 is orders of magnitude cheaper.

Continuing the example you can't go past the PIC12C508 ... its so much cheaper than the Atmel Tiny devices, and people have been using them for a decade or so !! I'm sure Microchip are shipping as many now as they were back then. Anyway the Atmel parts have lots and lots of nice features but obviously you have to pay for them. I have found that most jobs can be done with a cheaper part. I looove the AVR but I like to keep the blinkers off.

cheers,
Nick

admin's test signature
 

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

Nick,

I have found that most jobs can be done with a cheaper part.

True, but another thing to take into account is that for a small run, the development costs are significant. If the cleaner architecture of the AVR saves you 25% of your development time (=budget), this might swing the whole thing in favor of the AVR.

(In case my boss is reading this: see, I did learn something on that project management course... :-)

Sean.

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

Yes, well Sean, you're right, and it would also stop the folks out in the factory from yelling at me for taking the EPROM eraser home... grin
cheers,
Nick

admin's test signature