uC lifetime

We can't completely out of AVR yet. But we are working on it. First step is to phase out the high end AVR and replace it with a low end AVR and a custom demux chip for I/O. The customer wants AVR arch and we just can't change it totally.

AVR is the case of a fine technical arch screwed over by Atmel managers.

Understood. I've a product using Atmel AT91S2313. And I've a boat load of parts laying about that I use for hobby stuff (assembly coding only, for now.) But I don't specify or design it into any commercial instrumentation, now. I would consider doing so for a commercial product designed for the hobbyist market, though. So it's not a complete zero. Just know your application and market and match it with the company.

Jon

One useful precaution for this may be to grab any open source tools you can find, with all the documentation, and save them away for later. Then if you have a problem and your commercial compiler won't work on Windows 13, you still have the open source tools as a backup. It might not be pretty and you might have to spend a day recompiling them, but it might get you out of a pickle when the existing toolchain has bitrotted away.

Alternatively install and test tools on a virtual OS, then file away the OS disc image for later so you can boot it in an emulator straight off, instead of having to worry about how you arrange the future equivalent of finding Windows 2.1 on 5.25" floppies.

Theo

You should see my office area. I've got 80386 machines with 8-bit and

16-bit ISA buses, that still run. I stopped using Windows after XP and can't see how I will ever proceed beyond that. I still install and use Win98SE, quite often. Boxes of full-product DOS 5.0 systems here, as well. And I use them. I have MPLAB installation files that Microchip no longer makes available -- and they still make a lot of them available. I have every revision of compiler tool I ever use, kept on several types of media. I still have a functioning Lattice C compiler toolset and I use it for a product that is STILL selling to this very day.

Doesn't change any of my experience with Analog. In this case, it was an important hardware device I needed to use that was no longer supported. Not software. It ceased to function. Luckily, I had ONE more of them. ONE! But I got lucky. It might have turned the other way, though. They retire old DSPs (for example, I started using the ADSP-2111 when it was still just sampling and I watched as it became "unavailable.") And their new tools will not support some older devices.

Anyway, it's a different world with Microchip. I've never even been scared, let alone truly jeopardized. They really know what it takes out in the field.

Of course, their PIC10/PIC12/PIC14/PIC16/PIC17 families are bare-metal ALUs and some of their peripherals could use some sprucing. But that doesn't rank even 10th on my top list of concerns. It's so low down that I can't recall it noticing it as a concern to weigh amongst others. It's off the radar scope.

Which brings up another thought. Silicon bugs. Unlike, for example, Texas Instruments which for all I can tell _never_ fixes any, Microchip keeps a long laundry list of them and actually _works_ at correcting them over time. Take a look at the A3 silicon errata for their PIC18F2525 part and compare it with the B5 stepping's errata, for example. They seriously work to repair errata in future steppings.

Basically, they work hard for us. Really, really hard. And it shows like gang-busters.

Jon

AVR is about 3 times faster then PIC18, and AVR core is nearly perfect for C programming. There is a class of problems (like small control systems, modems, encoders/decoders, etc.) for which the AVR is suited well, but, as you said, the stupid practices made Atmel an unreliable vendor.

VLV

that's interesting, which part is affected?

Oliver

--
Oliver Betz, Munich
despammed.com might be broken, use Reply-To:

[...]

Ack. See

- they assure a life cycle of at least 10 years for standard parts and 15 years for "automotive" controllers.

This - and the availability of many 5V parts - is a huge advantage for my long-lived industrial applications with analog signal processing operating in noisy environments.

I remember only one case where a Motorola EOL caused some headache - the MC68HC811E2 got no drop-in successor.

I don't think so. The process is known and you have to migrate to newer parts to get smaller, cheaper chips (and new bugs, of course).

"Old" HC12 devices also ran much faster than specified. The data sheet is very conservative.

Oliver

--
Oliver Betz, Munich
despammed.com might be broken, use Reply-To:

Vladimir Vassilevsky wrote

Funny you should say that...

I have been using the 90s1200 for about 10 years in production volumes. We heep about 5 years' of strategic stock.

Just as I was getting around to replace the 1200 with the recommended replacement (2313 or something like that?) I found out that that one had been discontinued also. So next year I will look into it again.

Fortunately it is just a bit of code that reads a dipswitch and runs a timer based on the setting...

Atmel used to be such a great company...

D Yuniskis wrote

Good to see you back Don :)

In article , snipped-for-privacy@seen.com says... ...

NXP is just as bad, best I had was a USB hub chip that had loose specs for TTL and Schmitt inputs, but NOTHING told you which inputs were what type.

--
Paul Carpenter          | paul@pcserviceselectronics.co.uk
    PC Services
 Timing Diagram Font
  GNU H8 - compiler & Renesas H8/H8S/H8 Tiny
 For those web sites you hate

Vladimir Vassilevsky skrev:

The AT91 product line is fairly stable, and has been so since it started.

The AVRs 0.5u was droppped but there are other examples like the ATmega8 which has seen production for quite some time now. When Atmel announced the new AVR "A" range, it was announced that the older parts would remain in production.

Seem to remember a $10 bet on that an XMEGA would be obsolete within one year. Time to collect soon ?

Ulf Samuelsson Personal Opinion - as always.

Vladimir Vassilevsky skrev: >

The AT91 product line is fairly stable, and has been so since it started.

The AVRs 0.5u was droppped but there are other examples like the ATmega8 which has seen production for quite some time now. When Atmel announced the new AVR "A" range, it was announced that the older parts would remain in production.

Seem to remember a $10 bet on that an XMEGA would be obsolete within one year. Time to collect soon ?

Ulf Samuelsson Personal Opinion - as always.

The forthcoming AVR32 based AT32UC3C runs at 5V, so there are exceptions to that rule.

--
Best Regards
Ulf Samuelsson
These are my own personal opinions, which may
or may not be shared by my employer Atmel Nordic AB

Peter skrev:

The recommended replacement is the ATtiny2313. There is an ATtiny2313A around the corner. While the ATtiny2313 will be moved to "not recommended for new designs", (because for new designs, the ATtiny2313A will be better) I believe that the intention is to keep the ATtiny2313 in production for a substantial time. This is unlike the move from the AT90S2313 to ATtiny2313, where the older part was obsoleted soon after the introduction.

--
Best Regards
Ulf Samuelsson
These are my own personal opinions, which may
or may not be shared by my employer Atmel Nordic AB

Oh, I don't know. Intel has come out at least twice that I can remember (I think it's more like three or four) loudly touting "We're an embedded systems company now! It's usually during or right after an economic downturn when their processor sales slip, and it usually only takes a year or two of strong PC sales before they start touting how they're "going to refocus on our core business" and drop their embedded parts like hot potatoes.

Which is a pity, because when they care to Intel makes nice embedded parts.

Yea verily.

--
www.wescottdesign.com

I have no problems getting samples from TI or freescale, but not from Atmel. The Atmega32u6 specs are out for several months. Digikey does not carry it. Mouser needs 1K order and two to three months for delivery. Still waiting for my sample request months ago, both from Atmel and Mouser. The customer will buy 10K to 100K to ..., but I am not buying 1K just to see if it works.

While waiting for Atmel, the customer wants an alternative solution. If they really want AVR, they would have to buy out Atmel. They certainly have enough cash to do it. By the way, I am sure they have PIC,MSP, HC,... reference designs, in addition to our AVR design. So, we can't use another arch, but they might.

Remarkable to see such a similar story to mine playing out with you, regarding Atmel. I've posted it here before, so I won't duplicate it again.

Jon

I work for a multibillion dollar company and *I* still cannot get samples from Atmel. I get samples from the distributor, but I can only have 5pcs at most of any one device. I can get >5pcs only by finding someone to sign a PO (impossible; takes six months of begging and procedure and paperwork) or by getting the distributor to submit multiple orders staggered 8 days apart since anything closer than 8 days would result in the orders being denied.

I am willing to buy small amount of 25, 50 or 100, but not 1000. If the part is not available for months or years, why did Atmel put out vaporware release?