MCS-51 too old ?

[janengelbrecht]

Im doing a lot of ARM and AVR design. But still i get back to good old MCS-51. Mainly Atmel processors but also Cypress.
I do not dare to make any commercial designs for my company though based on the MCS-51....because....is it gone tomorrow so spare parts will disappear ?

[Bassman59]

Im doing a lot of ARM and AVR design. But still i get back to good old MCS-51. Mainly Atmel processors but also Cypress.
I do not dare to make any commercial designs for my company though based on the MCS-51....because....is it gone tomorrow so spare parts will disappear ?

Do you mean the original Intel 8051 (and its variants like 8751 and such), or do you mean any of the new devices based on the 8051 core, like the Silicon Labs parts?

[dannyf]

The particular chip is very old. However its many clones live on and many of them offer performance beyond most pic and avr, and even arm chips of recent vintage.

as long as you program in a high level language like c, the core is largely transparent to you. So instead,focus on the peripherals.

[janengelbrecht]

Okay so i´ll be home free....i can rely on Atmel to still be in the market and produce AT89C51, AT89S52 or other parts with same DIP40 layout for many years to come ?

[paulie]

as long as you program in a high level language like c

Haha... Now it's just too obvious so I won't even bother.

Strange that nobody has mentioned the real reason 8051 will live forever.... OPEN ARCHITECTURE. Copyrights/patents ran out years ago. 90% of the special purpose chips like RF transceivers, media players, etc have one hidden away inside. "The Chip That Wouldn't Die!" now playing at a theater near you. Second bill: "IC With a Million Apps".

[nctnico]

I wouldn't use any 8051 based CPU in a new design. The programming model is extremely limited compared to the flat memory model of many 16 and 32 bit microcontrollers. A lot of CPU cycles are wasted on shuffeling data around if you program these in C (*). And there is the next problem: there is no good free C compiler for the 8051 architecture. I've seen Keil's 8051 compiler generate pretty good code.

(*) These old 8 bit architectures are also allergic to pointers so you end up declaring all variables global instead of passing a pointer to a struct which is much nicer. Also function pointers in ram are a no-go so no polymorfism.

[dannyf]

i can rely on Atmel to still be in the market and produce AT89C51, AT89S52 or other parts with same DIP40 layout for many years to come ?

I wouldn't go that far: Atmel has trouble sourcing their own AVRs, let alone 8051 in dip.

However, there are plenty of high-performance 8051 vendors out there, Silabs for example.

[mikeselectricstuff]

Okay so i´ll be home free....i can rely on Atmel to still be in the market and produce AT89C51, AT89S52 or other parts with same DIP40 layout for many years to come ?

The only MCUs I'd rely on being available in DIP for the forseeable future would be PICs. And even then maybe not as big as 40 pin.

[dannyf]

being available in DIP for the forseeable future would be PICs.

You have to give Microchip credit for doing that.

[janengelbrecht]

Well i realize that DIP (well DIP40 then) is a gonner in a near future
Eventually the DIP28 will go the same way and ATMEGA328P will also disapear....well the day i can only get my hands on Surface Mount technology is the day i will leave the hardware design business

Thx for your input

[nctnico]

Maybe you should just give SMT components a try. They are surprisingly easy to solder. Much quicker than a DIP40 package.

[true]

Eventually the DIP28 will go the same way and ATMEGA328P will also disapear....well the day i can only get my hands on Surface Mount technology is the day i will leave the hardware design business

You are in "business" doing this yet you won't look at SMT as the way things are now?

It's like you are complaining the new interesting blocks only come in Lego form and you will give up building blocks because they don't make the cool things with Duplo. The industry has grown up; stop being middle aged playing only with kids toys. SMT even as a hobby is easy.

[janengelbrecht]

One could say im not in the microcontroller hardware design business... But still i sometimes makes small contributions in that area..my main focus personally is software design where i use prefabricated hardware mainly and small custommade pcbs for interfacing.....and of course one can come a long way using evalution boards in the development phase...this way it doesnt matter if the technology is smt or leaded components....but i realize now that the my taste for leaded components must either change or i will just have to focus onsoftware only The solutions i make are not for production in great numbers ( series are small and all is made for one customer only ) ...but still...to use technology that will be obsolete in a near future isnt the winning choice if the customer cant get spare parts one solution is of couse to buy a great number of components for the customer....that way...there will not be a shortage of spare parts and it is a possibility to continue using technology that is on its way out of the market And yes...i think i will ordre 3-400 AT89c55wd and other atmel stuff while it is still possible to get ones hands at theese

[nctnico]

I'd outsource the hardware design if you are having such constraints. That way you are much more flexible than having to use the same microcontroller until the end of time. Besides that I can't believe that a single microcontroller is a good fit for all your projects. I try hard to limit the number of different microcontrollers I use (and keep a few in stock) but I think I have about 6 different types (from the same series though).

[miguelvp]

Is there any fear that Cypress is going to discontinue the PSoC 3?

Other than that all of those cheap MP4 players have an 8051 core and I don't see those disappear, granted they are all SMD SoCs.

[rob77]

8051 will be here for quite a long time yet. it's too popular to fade out even atmel introduced the S versions with ISP interface as a replacement of the C variants. so those will be around for a long time yet.
the DIP40 is another story.... that package will disappear very soon - but you still can use adapter boards to make the TQFP 44 fit into a DIP40.

[dannyf]

i will just have to focus onsoftware only The solutions i make are not for production in great numbers ( series are small and all is made for one customer only )

The place I used to work for (and now work for mostly in names only) is like that - entirely software focused but with hardware capabilities as an accommodation to customers.

It is not a bad place to be, particularly if you are in a high cost country.

[Zeta]

I remember there were a chineese company (STC or something like that) doing cheap 8051 clones in dip binary compatible with old chips are they still arround? can't find their aliexpress store.

[westfw]

You're probably thinking of SST?
Microchip acquired Silicon Storage Technology in April 2010, and has been selling a dozen or so varieties of 8051s compatible with the ones NXP discontinued ever since.  http://www.microchip.com/pagehandler/en-us/family/8051legacy/products/home.html

There are 8051s in a huge number of "legacy" devices.  If I were an 8051 developer, I'd be worried that Si Labs bought Energy Micro and is working on ARM chips.  Si Labs looked to be one of the few manufacturers truly *developing* 8051 chips, and if they've lost interest, "legacy" might be all that's left.

[Wilksey]

SMT is all that is in production design these days isn't it? Very rarely do you find TH components unless it's a crystal or "special" component.

Anyway, every IC has a chance of obsolescence, my last nightmare was the EXAR 2206 monolithic function generator IC,someone may has well have been prodding me in the ball sack with a sharp pointy needle for all the pain that caused!

But with modern technology taking over, I was able to perform the same function with a PIC within the same package size and almost same layout.

So you can't say that they will never remove a chip from their line, they have no real justification either (not that it seems they need to justify it), you can't say it's solely because of lack of demand because there was uproar in the community when it was removed, but it has been around 20+ years etc etc.

The 8051 is still used today, some modules for things like ISM Radio, WiFi, BlueTooth, etc use a Turbo 8051 core, and the Keil and IAR compilers etc are still available for the chips.

If you are going to use it in a project or product and you are worried, I would say do what I did with the 2206 and buy a shed load from a component sourcer and store them!  Don't forget spares!

[dannyf]

Yes, STC is still around and doing quite well selling 8051 clones on the cent,

[CC58]

The 8051 availability can only get worse.  I'd stick with something new like ARM M0/M3 processors.  The high volume production makes them very cheap and availability won't be a problem.  The technology advances are huge. 

[amyk]

I remember there were a chineese company (STC or something like that) doing cheap 8051 clones in dip binary compatible with old chips are they still arround? can't find their aliexpress store.

http://www.stcmcu.com/

Yes 8051 is still around and probably far more common than PIC AVR or ARM - in fact a lot of devices that have ARMs in them also have an 8051 or several; e.g. touchscreen controllers in phones/tablets.

[MarkL]

(*) These old 8 bit architectures are also allergic to pointers so you end up declaring all variables global instead of passing a pointer to a struct which is much nicer. Also function pointers in ram are a no-go so no polymorfism.

Both of these things can be done in SDCC, or do you mean it's just inefficient on the 8051 architecture?

[SeanB]

8051 code is around in literally every small device, like USB peripherals, USB sticks and pretty much all small personal devices. Used because the core is royalty free, and the tools are free and easy to use. Used in PLC units as well as a slow robust system. even hard drives have one in there, along with a 6502 core.