Have you ever used a beta version microcontroller?

[technix]

Today the two STC8A8K64S4A12-28I-PDIP40 chips I ordered have arrived. Those are brand new 8051-compatible microcontrollers designed by STC - so new that those chips are actually prerelease beta versions.

What should I be aware of when using such chips in experimenting?

[Kjelt]

Only one answer: Anything.

I worked a lot with first silicon microcontrollers and what you are actually doing is exploring and updating the errata sheet (those things that the dev team already found out in the few months of testing before releasing the first silicon).
If you ever worked with microcontrollers you know there is not a single microcontroller to be found that does not have an errata sheet.
So that is what you can expect.

[technix]

Only one answer: Anything.

I worked a lot with first silicon microcontrollers and what you are actually doing is exploring and updating the errata sheet (those things that the dev team already found out in the few months of testing before releasing the first silicon).
If you ever worked with microcontrollers you know there is not a single microcontroller to be found that does not have an errata sheet.
So that is what you can expect.

I had an "Exploring STC 8051 MCU" blog series. Should I try to move the same experiement performed on the IAP15W4K61S4 chip (a chip in full production) on this beta chip and document it in a continuation of the blog series?

[nctnico]

If you ever worked with microcontrollers you know there is not a single microcontroller to be found that does not have an errata sheet.

The NXP LPC11E6x series doesn't have an errata sheet.

Anyway working with beta silicon is a pain in the ass. I've seen several cases where chips where extremely faulty, the released version was different or didn't even get released. I'd wait for the official silicon to arrive and let other people do the testing.

[Kjelt]

I had an "Exploring STC 8051 MCU" blog series. Should I try to move the same experiement performed on the IAP15W4K61S4 chip (a chip in full production) on this beta chip and document it in a continuation of the blog series?

I don't have an answer for you on that question.
Ask your blog followers.

[Kjelt]

The NXP LPC11E6x series doesn't have an errata sheet.

That would be unique or you have to give it some time.
A micro without errata would be equal to software without a bug.

I agree on the PITA although working with proto uCs give you an advantage against competitors it can be very frustrating.

[8086]

If you ever worked with microcontrollers you know there is not a single microcontroller to be found that does not have an errata sheet.

The NXP LPC11E6x series doesn't have an errata sheet.

http://www.nxp.com/documents/errata_sheet/ES_LPC11E6X.pdf 

[JPortici]

If you ever worked with microcontrollers you know there is not a single microcontroller to be found that does not have an errata sheet.

The NXP LPC11E6x series doesn't have an errata sheet.

Anyway working with beta silicon is a pain in the ass. I've seen several cases where chips where extremely faulty, the released version was different or didn't even get released. I'd wait for the official silicon to arrive and let other people do the testing.

i would be just about as worried as if i had to use one with scary long list

[nctnico]

If you ever worked with microcontrollers you know there is not a single microcontroller to be found that does not have an errata sheet.

The NXP LPC11E6x series doesn't have an errata sheet.

http://www.nxp.com/documents/errata_sheet/ES_LPC11E6X.pdf 

I just looked and couldn't find it but appearantly it does have a bug!  I agree with JPortici that a chip with no errors worries me but the ARM microcontrollers from NXP usually have very few bugs in them.

[rsjsouza]

The NXP LPC11E6x series doesn't have an errata sheet.

That would be unique or you have to give it some time.
A micro without errata would be equal to software without a bug.

I agree on the PITA although working with proto uCs give you an advantage against competitors it can be very frustrating.

The absence of an errata list may indicate that either a device is not popular enough (and may risk getting the cut in the future) or is not scrutinized enough (and may have lingering unknown bugs).

[janoc]

I agree on the PITA although working with proto uCs give you an advantage against competitors it can be very frustrating.

Considering it is a 8051 MCU which are dime a dozen, I am not quite sure what sort of competitive advantage one could get there to justify the hassle.

[legacy]

Have you ever used a beta version microcontroller?

No, and never I will as I am already fine with all the can happen when you have a datasheet, and an errata-sheet to fix it.

[abraxa]

Today the two STC8A8K64S4A12-28I-PDIP40 chips I ordered have arrived. Those are brand new 8051-compatible microcontrollers designed by STC - so new that those chips are actually prerelease beta versions.

What should I be aware of when using such chips in experimenting?

Only one answer: Anything.

I agree with Kjelt. To be a little more precise, though: you can't assume anything. Writing to any register may be non-working in any possible way. Data may not be written at all, maybe not all bits are written, the opcode may write the wrong register, and so on. Using the debug interface may be broken, too, leading you to read values that aren't actually true or a breakpoint may hit at the wrong address. Also, clock dividers may be off by one, a device reset may not actually reset all registers, hardware timers may not be stopped when the core is halted... all sorts of stuff is possible.

Sure, generally this stuff will work but in the case something *doesn't*, you could be in for a fun ride.

[westfw]

Are these actually still in "beta" chips, or are these chips left over from the completed beta test?  The latter is pretty common; once a company feels like their chip is working pretty well, and like they have a good handle on anything remaining that's wrong (do be sure to read the errata), they'll take the chips that were manufactured during the test period and start plopping them on development boards and/or selling/sampling them to "early adopter" customers.  IIRC, TI, Atmel, and Freescale are all "well known" for shipping development boards with "X" version chips ("engineering version")

This is not the same as actually being part of the beta test.  :-)

[Kjelt]

I agree on the PITA although working with proto uCs give you an advantage against competitors it can be very frustrating.

Considering it is a 8051 MCU which are dime a dozen, I am not quite sure what sort of competitive advantage one could get there to justify the hassle.

Yes ofcourse, in my case we had an exclusive deal with the manufacturer to help test and develop some special new peripheral and in the deal there was an exclusive period before releasing to other companies.

[technix]

Are these actually still in "beta" chips, or are these chips left over from the completed beta test?  The latter is pretty common; once a company feels like their chip is working pretty well, and like they have a good handle on anything remaining that's wrong (do be sure to read the errata), they'll take the chips that were manufactured during the test period and start plopping them on development boards and/or selling/sampling them to "early adopter" customers.  IIRC, TI, Atmel, and Freescale are all "well known" for shipping development boards with "X" version chips ("engineering version")

This is not the same as actually being part of the beta test.  :-)

The latest beta for PDIP40 package is Rev. C as I have here. The surface mount variants are at Rev. E but those have additional pins and associated pin muxes.

There will be a few more betas as STC planned to but have yet to another implement a CAN controller on this. Neither Rev. C PDIP40 or Rev. E LQFP64 have it yet.

[janoc]

Yes ofcourse, in my case we had an exclusive deal with the manufacturer to help test and develop some special new peripheral and in the deal there was an exclusive period before releasing to other companies.

I obviously understand that. However, I wonder what has been such a special sauce on a '51 MCU that you couldn't find elsewhere that it was worth becoming a silicon betatester and taking all those risks.

[Kjelt]

I never said it was a '51 in my case

[technix]

I agree on the PITA although working with proto uCs give you an advantage against competitors it can be very frustrating.

Considering it is a 8051 MCU which are dime a dozen, I am not quite sure what sort of competitive advantage one could get there to justify the hassle.

I am just beta testing the chips for the fun of it. And STC have rewards for people that discovered bugs for them.

[technix]

Reporting back: the chip successfully booted and the on-chip debug worked. Loading an empty program (infinite loop) and attach to the on chip debugger showed the code in a working state.

Since stcgal does not support this new chip yet I will not experiment too much on this.

[Feynman]

Basically, you should be aware of A LOT of undocumented errata in the silicon and in the documentation.

I myself never used beta or pre-release parts and never will. I try to avoid using parts that aren't out long enough to have good chances that at least the major flaws are documented.

Of course, every micro has its errata, but well documented errata are almost as good as no errata at all

[technix]

Basically, you should be aware of A LOT of undocumented errata in the silicon and in the documentation.

I myself never used beta or pre-release parts and never will. I try to avoid using parts that aren't out long enough to have good chances that at least the major flaws are documented.

There are chips to be used in a product and chips just to play around with. This beta microcontroller is solidly in the "chips to play around with" territory here and definitely will not end up in a shipped product. (Maybe I would spin a board with the chip designed in, but it will not be made publicly available until the final version of the chip is released, and the beta chip is almost certainly socketed so I can swap it out with a final one as soon as I get it.)

[Jeroen3]

I've worked on NXP engineering samples. You never know what might happen. You can be running code fine for a while, and then suddenly you've bricked it.
But it's ok, you mark it and you pop a new one from the tray in the BGA socket.