Dialog box shut: Now Microchip is set to gobble up Atmel Microcontroller gurus Atmel look set to be acquired by Microchip – and not by the UK's Dialog Semiconductor. Atmel, a $3.3bn business based in San Jose, California, designs and makes simple and low-power AVR, ARM and 8051-compatible chips for all sorts of electronic gadgets and gizmos. It is probably best known to people as …
They're just the gateway drug. I started with Arduino-alikes (Teensy's), quickly outgrew the ecosystem and now design with the bare AVR chips and code in avr-gcc. But the fact they you don't have to start off that way is what got me in. Yeah, I know PICs have a development environment, but the momentum behind Arduino, and the fact I could install the entire development suite on the Mac with one download was what convinced me to give it a try in the first place.
I have barely dabbled with Arduino, and only used a PIC when someone else selected it, but:
---
Yeah, I know PICs have a development environment, but the momentum behind Arduino, and the fact I could install the entire development suite on the Mac with one download was what convinced me to give it a try in the first place.
---
struck me as pretty funny. There was indeed a PIC development environment, that I could download to a Mac, but if it were not for the open-source command-line tools, I would have gone utterly mad trying to develop some conceptually simple, but complex in detail, PIC software. I hear that the "bigger" PICs are easier, and that staying within the Church of Windows makes the IDE a bit better, but yeah, Arduino is a lot easier to "get something done". PIC wins for "get something done for a really low price, not counting the psychiatric benefit for the programmer".
Except that PICs really aren't that cheap.
The silicon costs about the same as a bottom-end ARM, and once you add the compiler (you're not writing anything of real note in pure asm), PIC ends up quite pricey.
Not to mention that quite a few of the PIC application notes are outright wrong.
-- you're not writing anything of real note in pure asm --
_You_ might not be writing anything of note in asm, but both my PIC projects (latest on a 16F84) were pure asm. One was a "rescue" of a project whose hardware was "frozen" and whose software was semi-interpreted, and, Surprise! didn't come within an order of magnitude of the performance spec.
The other was a diving-catch replacement for a trivial bit of hardware because <redacted other vendor> notified us a few weeks before delivery that our chips were going to be a little late. Six months or so.
I cannot speak for the later/bigger PICs, but using a compiler on the small ones is madness. One would indeed be better off with an ARM, Atmega, or heck, even an 8051 (yeah, I have those scars too).
Not every nail need be driven with a steam hammer.
The irony is, when I first tried to get into microcontroller development, I got an Atmel starter kit (late 90's I think it was), with the "Keil" software, which was windows only.
Having got the starter kit, I found there to be virtually no support in the open source world for development and programming of Atmel uC's. All the hobby electronics people were using PIC uC's, and the open source assemblers/simulators were all for the PIC uC's.
So after lots of frustration, I gave up, stuck the atmel in the attic, and bought a Maplin PIC starter kit, which I then started using. The OSS software was clunky, and didn't support most PIC's (the "SDCC" software still only really supports PIC16 last I checked) but it worked, and allowed me to write C for the uC on my Linux workstation.
Fast forward a couple of years, and a lot more investment in PIC based hardware, and just when I got comfy with PICs, and Atmel "Arduino" is now the best dev environment for Linux development of uC's, and have spawned a lot of OSS tools for development. Conversly the PIC OSS scene has stagnated somewhat.
So now have been asking myself, whether to move across to Atmel again, as there is a lot more vibrancy and progress in Atmel OSS. Although who knows what will happen, now that Microchip are buying Atmel. Maybe both families will get some love?
Considering the first AVRs only escaped into the wild in '97, to have any amount of third-party support by the end of the 90's was no small achievement. I was first introduced to them at the end of '98 when starting my first job, and spent a shade over 6 years really getting to grips with a few of the family and the idiosyncracies of the available compilers. The IAR one wasn't too bad, apart from their forgetting to let us know about an upgrade we should have received as part of our service agreement until *after* I'd submitted a bug report, but the one we needed to use for Tiny15-based projects was so full of weirdness that it became second nature to read the asm output after each compile just to make sure the output code was going to do what you intended it to. ISTR ditching the C code for that project after it got allocated to me as part of a department reshuffle, and just doing the whole thing in asm...
Haven't spent nearly as much time working with PICs, though I can see why they've garnered so much support from both amateur and professional users alike. Going to be interesting to see what comes out of this takeover, and not just from the PIC vs AVR perspective.
Just use assembler, sounds about right. I was going to help a friend with PIC programming last year but I didn't know where to start with the toolchain. Decided PIC* is the Wordpress of embedded (blind leading the blind and all that). Just use AVR assembler... if I get around to it... if Microchip doesn't ruin everything.
*Yes... one could compare Arduino to Wordpress... but not AVR.
Just use assembler, sounds about right.
+1 for "just use assembler". My first foray into MCUs was with Motorola 68HC11, and it was a joy to program for in assembler. No, no newfangled EEPROMs. Old fashioned EPROMs with a stint under UV light if they needed erasing to be reprogrammed.
Admittedly it was lot pricier (back in the 80s) than what PICs etc these days.
Friend of mine hates Atmel. They got him fired.
He had a project with a serious bug in it, which he traced back to the actual Atmel silicon, duly notifying Atmel of his discovery. His boss reviews the project, hears the explanation that its the chips fault, calls Atmel to confim the story, for them to deny all possible knowledge that there could be a bug in their chip. Friend (consulting) is fired from the project.
A month later Atmel publish a datasheet errata with a potential workround, for the bug my friend reported.
There are some quite different PICs. Some are ARM.
The 16 and 18 Series are the alternative to 8051 and AVR/Ardunio They are best programmed using JAL V2.
jal.sourceforge.net
justanotherlanguage.org
www.casadeyork.com/jalv2/
NXP do M0 ARM chips about same price as PIC and AVR. But some projects are so simple that a PIC or AVR is cheaper to develop and maintain.
https://en.wikipedia.org/wiki/JAL_%28compiler%29
The huge advantage is that switching PICs and setting the device options are easy. The team also have figured out most of the mistakes in datasheets. There is even a PCB that takes Ardunio shields (which are physically a daft idea!).
JAL libraries are good too, USB, FAT, SD Cards, Graphic and Text panels, Maths etc. You can swap between bit banged GPIO or dedicated I2C, Serial etc HW with JAL libraries.
An Ardunio is for people that can't solder veroboard or order cheap custom PCB from China. AVR and PIC are equally easy to use without that ecosystem.
Firstly, AVR processors are all over the place, the Arduino is only a very minor use for it.
I know several hardware designers who will walk away if they are subsumed by PIC makers.
Microchip are notorious for only having partially implemented features on their processors and a habit of changing behaviour without any notice whatsoever.
ATMEL may not be perfect but the other lot are awful, at least it isn't Jennic buying them I suppose
You know a bunch of idiots then.
Microchip silicon isn't substantially more or less buggy than anyone else's. There customer support is much better than most and Atmel,.
Atmel have a history of shitting on customers setting their fabs to make what is most profitable at the time leaving longer term customers with Atmel design ins to go fish.
They will also make parts obsolete leaving customers with expensive re-design and re-qualifaction. Microchip still make and sell parts from the 80's.
Companies change hands all the time, change minds even more often. I understand having an emotional response to ones culture being threatened, but decent severance goes a long way toward salving that. If you quit in a huff you miss out on that possibility. (<-- reply to originator of this subject)
That said, since inmypjs mentioned it, I guess it's OK to reveal that the vendor name I redacted in a previous comment was Atmel. To be _very_ clear, though, if you have been in this industry long enough you have a fair chance of being similarly screwed (broken delivery contracts, defective chips "delivered' but unusable, pricing jumps mid contract, re-labeled fallouts...) by any semi vendor. So, Atmel just joined the crowd with Intel, TI, Motorola, Signetics, and NatSemi. So far.
Microchip sells more than 3 million microprocessor chips a day, More than have been used to build all the Arduinos, clones, and derivatives ever made.
Arduino sales are trivial for Atmel and utterly trivial in the global semiconductor market.
When it comes to being gobby on the web and 'new media' the Arduino proponents are rather less trivial.
I got hooked by the $10 dev board with free environment offer. They're neat little devices, most are sub £1, and go like stink.
I've looked at other devices, but to get to the same level of decent IDE, plus in-circuit debugging, usually costs north of £100, so not much good for home messing about.
They're nominally 3V parts (a pain for using all those Arduino modules out there) but they actually work from 1.8V to 3.6V, neatly allowing them to be powered directly from a couple of AAs.
...for which I've had to learn Atmel's Studio 7, and worked with the free compilers, which are adequate, I looked at Microchip's offering.
Seems their licence depends on what part family you're using (8,16 or 32-bit), but turns out about €500 for ONE user, ONE year, ONE architecture. Atmel gives me 8 and 32 bit AVR, and all their ARM stuff - free.
Microchip's strategy would put me off a future upgrade for sure.
So, no thanks, MPLAB.
I'm sticking with Atmel, at least as long as their new Overchipped Microlords will let me...
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
