8 bit microcontroller market
in [Embedded]
|
Prev: Tiny Bootloader
Next: Link&Locate 86?
From: Joerg on 7 Sep 2006 13:23 Hello Dimiter, >>I guess it's targeted at the luxury sector. S-Class, 7-Series etc. > > I would be less sure of that. They do offer the 5200 - which is > flashless, > but full of stuff at <$20. (400 MHz 603e, plenty of serial/CAn ports, > MAC10/100, > USB, DDR,PCI etc.). > I don't expect this one to cost more. > $20 is a whole lot of money in automotive. I remember a few guys from a sports club who said that at work they had seen efforts in excess of a man-year to save one Dollar via a new water pump design. > >>Personally I don't like the idea of having to trust that the contents of >>those 3MB are bug free while cruising down an autobahn at 120mph. At >>least not for mission critical stuff in there. > > I might have similar feelings on that. All - _all_ of the code my DPS > has on > its disk is about 2 M. This includes kernel, filesystem, windows > support system, > device drivers, object definitions, system commands, utilities, > assemblers, > linker (it's 1 for all :), tcp/ip stack, my gamma spectrometry code > (which is probably about 1/3 of all that), you name it - several > hundred files located in several directories. > But then, they only have C to support the PPC so they provide all that > flash - which is a good thing itself, the code they intend to put into > it > may be not that good... > The last sentence brings it to the point. Who knows how reliable the code is? This is why I would never let Windows control anything critical. -- Regards, Joerg http://www.analogconsultants.com
From: Joerg on 7 Sep 2006 13:25 Hello Yuriy, > >> It was already sort of a microcontroller, >> even though it needs external ram/rom. Ten years ago - still >> a valid choice, but the arena is getting smaller and smaller >> for the 8051. I agree it can still be a very valid choice today, >> for typically very less demanding tasks. > > It would be hard to create architecture worse than x51 today. Limited > stack, single pointer, three different type of data memory, no thought > at all about high-level language, 12 clocks per cycle, etc. > By any rules it should be dead long ago... > That's what they also said about CD4000 logic, non-SMT packages, Volkswagen Beetles, and so on. Didn't happen. -- Regards, Joerg http://www.analogconsultants.com
From: Joerg on 7 Sep 2006 13:38 Hello Yuriy, >>>>> Programmer who can handle only 8051 core shall be fired for the >>>>> incompetence. >>>> >>>> That's not how managers in the industry think here in California and >>>> those are the folks who make most decisions. >>> >>> Poor managers. Micromanagement is a very, very counterproductive option. >> >> Has nothing to do with micro management. > > This is exactly what manager selecting CPU core is doing. > The managers do not select the CPU core. A good manager educates the staff in business matters. For example, how to be or become a good cost thinker, what the importance of second-source is and so on. Then good engineers will arrive at the proper CPU selection all by themselves :-) >>> BTW, how many managers were polled to get these results? >>> >> No need to for me. I live here :-) > > OK, this is a single person opinion. That's OK, too. > Ahm, I am not alone with this opinion. Not by a long shot. >> Our apps are a lot more demanding. Realtime stuff in assembler, other >> designs in C. > > Necessity to use assembler usually points to the inadequate processor > selection. > For hardcore realtime apps the is no alternative to assembler yet. No matter which processor. Also, there are times (many times) when the BOM budget or the battery budget does not allow a fancy chip. I just have one of these: A Blackfin would just hit the spot. But, it's too expensive and would deplete the batteries way too fast. IOW, with what you coined "adequate" processor you would not have a saleable product. Instead you would have an unhappy client or boss. >>> A lot of people have better food without ever heard such meaningless >>> words as 2n3904, etc. :-P >> >> So, you think a 2N3904 is meaningless? Interesting. Open some gadgets >> and look inside. My designs that contain a lot of such discretes >> generate comfortable profit margins for my clients. > > :) These people never heard word transistor either. > My clients know them very well. >> And here I don't mean toys only. Furnace controllers, HVAC >> controllers, etc. > > I.e. very simple stuff. > There is nothing simple about any mass product. Have you designed one? >>>> To the point where many fresh graduates are not even able to >>>> understand my designs anymore. >>> >>> Ironically, there are at least two possible reason for that... >>> >> Which would be? > > Either fresh graduates are incompetent, or the designs. > The design that is the least costly and fits the bill is the best design. Good old capitalism :-) So that leaves the grads. I wouldn't say incompetent but they do have a thorough lack of practical skills. Many can't even solder. It's quite pathetic that several (!) fresh EEs did not understand my module spec for a electronic circuit while an older chemical engineer understood it perfectly well. -- Regards, Joerg http://www.analogconsultants.com
From: Joerg on 7 Sep 2006 13:42 Hello Eric, > >>I'd never buy a car with dozens of controllers. > > It's getting hard to buy a car now that has less than a dozen > controllers, if you consider all of the electronic gizmos (power > mirrors, power steering, power windows, air bags, anti-lock brakes, > cruise control, fuel injection, electronic ignition, emissions control, > entertainment systems, ...). > True. However, some Internet research led me to just such a car in 1997. Except for fuel injection, ECU and the associated emission controls (lambda sensor etc.) it has none of the above. My wife's has electric windows and sure enough, one day, bzzzt .... errrrrrr ... KA-CRUNCH. $250 later it worked again :-( -- Regards, Joerg http://www.analogconsultants.com
From: Joerg on 7 Sep 2006 13:52
Hello Jim, >> >> It would be hard to create architecture worse than x51 today. Limited >> stack, single pointer, three different type of data memory, no thought >> at all about high-level language, 12 clocks per cycle, etc. >> By any rules it should be dead long ago... > > Of course, with a clean slate, and a huge amount of hindsight, and a > shift of the target goal posts, with new 2006 processes and design > tools, then it's no real surprise a different answer would result. > > You could say the same about almost any processor, the Pentium included. > > Just one tiny problem: the real world is not a clean slate, and there > are huge amounts of IP and training invested. > Fully agree. The other tiny problem: Mankind wants to be able to buy $9.95 thermostats, $19.95 sprinkler timers and whatnot at the stores. A big ARM processor or a fat DSP ain't in the cards there. Many things have to be produced in China for a grand total of a couple Dollars or less. > Include that, and the 80C51 has a long life ahead of it still. > An so do discrete parts. Look at "modern" TV sets: Many use non-SMT parts because they wanted to save a few cents on the circuit board and consequently chose, ahem, (Yuriy, close your eyes now) the good old phenolic board. There is a reason why smart companies like TI brought their new 430F2xx out in DIP. Hobbyisist are most certainly not the reason. -- Regards, Joerg http://www.analogconsultants.com |