C18 Compiler again

Okay. So you are arguing that uninformed design engineers are like windows computer buyers?

I can't go there. I am tempted, at times. But no.

Well, I think we agree on the "keep" part, of course. But the rest doesn't really dispute what I was addressing. The person I was writing to was summing up Microchip in a Microsoft comparison, implying that it's all just marketing. And I know Microchip's success is not _all_ marketing or even half marketing. I'll leave it there, though.

Jon

Earlier unpaid experiences (at university) were with IBM 1130 (asm and FORTRAN II), IBM 360 (BAL), PDP-8 (assembler), PDP-10 (FORTRAN and ALGOL-68), PDP-11 (assembler and later in

1977/78 in c), and HP 2116 and 2114 (all assembler.) There was a Bouroughs thing in there somewhere and an IBM System 3 and RPG II, as well, which was paid work. Starting in 1979, I was paid to work on the VAX-11/780 in MACRO-32, BLISS-32, C, BASIC, and COBOL. That would have been around the time when I also began using the 6502.

Jon

No, I am arguing that people with little or no experience of microcontrollers are likely to pick a device that appears popular. PICs turn up pretty quickly in any search on the web, or "learn to use microcontrollers" books.

Whenever someone starts looking at a new topic, they are "uninformed" by definition. Microchip targets this area very successfully.

I have no facts or figures to back this up, but I would expect that most PIC users used PICs as their first microcontroller. Relatively few will have switched from other devices to PICs.

I certainly agree that Microchip's popularity is not all marketing - I just think that their marketing strategy of targeting small users is a big contributor to their popularity. This is in no way a bad thing or a derogatory comment - good marketing is important to businesses.

It is not remotely like Microsoft's business practices. But even their success is not all due to marketing - some of it is due to coercion, bribery, and other illegal or unethical business practices.

The key to understanding Microchip PIC assembly is to view it as lower-level than "normal" assembly language - it's closer to a sort of microcode. If you view the "w" register as a sort of internal temporary storage within an ALU, rather than a "normal" register, then it all makes more sense. At least, it did to me when I worked with it about ten years ago.

This says it all. In practice Microchip has done a good job at every customer level. A lot of silicon companies could learn a lot about the way Microchip deals with their customers when they meet them at trade shows and promotional events and annual user conferences. All of these meetings have substance backed by user notes, reference designs and hard information. Microchip's after sales support is second to none.

They know better than most companies how their product can be used competitively and aggressively go after that market.

Regards,

w..

-- Walter Banks Byte Craft Limited

It was used as far as I know for essential one application area low speed switching and sequential applications. The only application I know for sure was traffic light control. At the time it was a low cost solution replacing mechanic timers and relay sequencers.

Regards,

w..

-- Walter Banks Byte Craft Limited

Exactly.

[To everyone: don't forget that unlike most people in this newsgroup, I am a hobbyist when it comes to embedded work so the issues which matter to me are very different from those which matter to you. You are looking for support for your commercial product; I am more concerned with been able to use the devices in a home development environment.]

When I started out, I didn't really have an opinion about which then current microcontroller to choose. I did however see PIC been pushed to the hobbyist in places like Maplin and I saw that Microchip had a lot of support in various ways for the hobbyist user.

The comparison to Microsoft is because I realised that Microchip knew, like Microsoft, that if you spent time and money targeting the hobbyist, you stood a higher chance of the hobbyist recommending your products when they got into a work environment.

Support for the hobbyist user, even when it costs money, can be regarded as a form of marketing as you are using that support as a way of trying to draw new people into using your products in the future.

However, I made sure I evaluated all the architectures which met my needs at the time so I ended up rejecting PIC in favour of the HC08 and then the AVR.

(So I did make an engineering type choice instead of a "what's popular" choice after all. :-))

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980's technology to a 21st century world

Since I've actually designed ONE actual, functional micro with logic gates and actually ran programs on it with a display and had it work (after much wrangling to fix my errors), I can with some modest authority generally agree with that. And I don't find it a problem. It works well enough.

Jon

I was going to ask, "How so?" to the first paragraph and then read this. So I guess I can't ask that question, now. It's seems now apocryphal and must simply be left there.

I certainly don't help you there. I have had long, long experience with other micros before using PIC micros.. going back to the MITS ALTAIR 8800 I built circa 1975.

Again, I guess I'm a data point of one against your point.

Jon

Me too. I was a BIG Motorola/Freescale, Zilog, and Philips guy for ~15 years before switching most of our products over to different PIC variants, and have done lots and lots of new products using PICs. A good cross compiler can hide a lot of the nasty stuff, and the newer parts have fixed some of the more annoying gotcha's. I like the fact that they still make a lot of DIP parts, and they seem to always have a pin for pin compatible upwards migration path. Something Motorola couldn't seem to get their heads wrapped around.

Jim

They support their tools and parts nearly forever and are a "how high" jumper when asked to jump. They don't question or grill me about how many parts I will buy from them before getting in gear and helping out and never ever hassle me. They simply apply their shoulders to my problem and move me forward, each and every time, without question or bother.

Hard to find that elsewhere.

Jon

That's either a point in favor or against Microchip, depending on your point of view. I like it, so I consider it "a good thing." Texas Instruments, for all I can tell when looking, _never_ fixes any silicon bugs. Microchip keeps a long laundry list of them and actually _works_ at fixing the important ones over time. To see that, look at A3 silicon errata for the PIC18F2525 part and then compare it with the B5 stepping's errata, for example. Then take a look at the MSP430F149, namely SLAZ017D, and see if they ever fixed anything even like the CPU4 bug, regardless of stepping.

Jon

Not necessarily. The shop I work with started with H64180, aka Z180. When they needed peripherals to be hung off the H64180, plus a little raw speed, they/we picked on PIC16 at the top clock rate. One wouldn't say they ever switched, though. The Z180s are still being used for the big part of the application, and PIC for the accessories.

Latest development is with PIC32 (which aren't really PIC,) and Mini-ITX motherboards (which aren't really Z180.) And USB is replacing SPI.

Mel.

It /is/ apocryphal - it's just my impression based on people I've talked to and things I have read over the years. The posts here from experienced developers who actively moved to PICs are eroding that impression a little, but they don't actually contradict it. As you say, you are a data point against this theory, and so are others here, but most people in this group are experienced and informed developers - it's a biased sample.

But it's all just a theory. If you don't agree, then that's fine. It's not as though we are arguing about facts, such as the best way to format C code :-)

I've heard this "they make DIP package parts" as a reason for using PICs many times. To my mind, this re-enforces the impression that these devices are aimed at small and hobby developers (with an aim to getting a long term professional customer in the future).

As a professional developer, I haven't had use for a DIP package microcontroller for over a decade, except for OTP devices. They are very rarely of use for serious prototyping or development - after all, none of the other components on a typical card are DIP any more, so you have no choice but to make up a proper card anyway. If you just want to try out some ideas, you use a ready-made evaluation card or development board. And if you really want a microcontroller in a DIP format for testing, there are endless varieties of ARMs and other microcontrollers mounted on a DIP-40 board package.

Hehe. Okay.

Anyway, I guess it's just that I never see all that Microchip marketing (any more than I see all the other companies doing it) to "people with little or no experience of microcontrollers," as you wrote earlier. Not much different from the mix I see from other companies competing against Microchip's market, anyway.

You'd mentioned web pages. Sure there are web pages for PIC and web pages for AVR and web pages for the BASIC Stamp, and Motorola, etc. People out there use stuff and where they are able they write about what they do and learn about. The companies themselves sometimes set up and fund the web server for user groups, too, from time to time (perhaps Atmel comes to mind, too?)

You'd mentioned books, too, as though that is another place that Microchip also competes hard and maybe outcompetes. There are books written by authors who choose what they want. For example, just to provide a random encounter I just had buying a book last week, David Cook's Intermediate Robot Building book discusses at length the Atmel ATtiny84. He also does, at the very end in "Choosing a Microcontroller" subsection, talk about AVR 8 bit micros generally, the Parallax BASIC Stamp, with a one sentence nod to the PIC. That's not atypical, either. I certainly don't "feel" or "sense" any author-bias in book publishing related to Microchip. If anything, somewhat towards the opposite is probably the case if my book shelf of such books is an indication.

If Microchip is "targeting this area very successfully" (re: ignorant beginners) as you say, it's not manifest to me. Atmel does at least as well, so far as I can tell, and probably better.

I think their real strength, the one that actually is the telling reason for the profit dollars they make where others don't do nearly so well, is found elsewhere.

As you say, that's just the view of one person who is already obviously an outlier data point just by the fact that I still read and sometimes post in a newsgroup. That all by itself makes me as rare as hens' teeth. Along with the rest of us anachronisms. ;)

Jon

Yep, that is a big draw for beginners, but we still use a lot of DIP packaged MCUs. Not a lot in the big scheme of things, but 100K+ per year, and of those 80% are PICs.

I guess this is one of those *depends* things. Why buy a surface mount part and then pay for the DIP adapter, when you could just buy the DIP part, but really, that is neither here nor there, I use the part that fits the job at hand and sometimes a DIP part is what I pick. We don't do anything that would qualify as anything NEAR bleeding edge, but it does have to work day in, and day out, for decades at a time and the simple fact is that the "old iron" guys have kind of abandoned me. In fact I have just finished up a job for a customer using a DIP PIC MCU, one of their Ethernet ICs in a DIP28 package and using a Halo FastJack with through hole mounting, with a shit load of other through hole parts. It is a very limited run, and I can just dip solder the 1000 or so pieces that the guy will need. It is what the customer wanted, and that is where the cheese comes from.

Before PICs I worked on 68HC05, AVR, 8051.

I first used PICs because of a good combination of price and size at the low end.

I am now working with PIC18s that have a good price for the feature set and memory.

There are some aspects of the architecture that force awkward code constructs with the chosen compiler, but have filled several niches well.

--
Thad