32-bit Microcontroller for $1.00 -Guy Macon

It is true - I don't know much about the workings of google groups.

Not at all. At least I did not mean it that way, more in the sense of tricky and non-obvious.

messagenews:4884e753$ snipped-for-privacy@clear.net.nz...

I seem to recall being in this conversation and I think I made a prediction that the ARM7 core would see no new chip product introduction (other than ones already in the pipeline) by the end of the year (not sure which year though.. ;^). I got a big argument from you saying that there were a lot of programmers invested in the ARM7 and no one had a compelling reason to switch, so the chips would not be changing over from the lack of demand.

Time has passed and I think I was right in that the ARM7 core is no longer seeing new commercial chip introductions and that the major players have all licensed the M3 core. With all of the new chip products using the CM3 core, new embedded products will see the ARM7 fade away by necessity.

The question of how the ARM CM3 will fare in comparison to the alternative CPUs is a different matter. I don't think the various chip makers will make the ARM products vanish and I don't think the ARM chips will make the other products vanish. If a company is using multi-millions per year of a given MCU, then their choice will be based on things other than what most of us use to pick a chip. Automotive makers will use what ever chip gives them a $0.01 advantage. That is one of the reasons that the Asian companies can still push totally proprietary MCUs to the automotive makers.

I have to say, I don't really see the advantage to the auto makers to have multiple sources for their MCUs. If you are buying even just 1 million a year of a given chip, do you think the maker is going to phase out that chip??? Are they going to jack up the price on a chip and risk a relationship with a given multi-million or maybe even a multi-billion dollar customer??? Is the price competition for your socket any different with two companies supplying the same part or two different parts?

If you are that large of a customer, you can nearly buy a company like Atmel. The chip maker is not likely to do anything that will make the

Rick

messagenews:4884e753$ snipped-for-privacy@clear.net.nz...

That's a natural industry progression, but it is a double edge sword.

As volumes drop on the ARM7 production, devices will be pruned, and that will have a NEGATIVE impact on those users. (and the same thing will likely happen further along the time-line, on M3)

They will need to (mostly) respin PCBS/recode/recompile/requalify (remember, M3 is not binary compatible) - those are all $$$ steps.

- assuming they have a NEW device that actually meets their feature set. (Right about here, they will likely re-check their core choice too)

That's a lot of costs, over a lot of users, waiting in the pipeline...

A number of ARM7 users are migrating up to the ARM9, rather than sideways to the M3. (or moving to DSP, or other cores...)

That's correct from a strict logistics viewpoint, but Asians also operate on a less tangible area, and like to see things like commitment. Two of the largest players, offering a second source IS going to get attention from designers & managers.

-jg

messagenews:4884e753$ snipped-for-privacy@clear.net.nz...

That depends on how the software is written. I share a lot of code between different platforms and therefore I write everything as portable as possible. Ofcourse the low level stuff needs to be rewritten for a new target, but it is very much possible to write portable code in C without giving up speed or code size. It is just a matter of knowing what you're doing.

IMHO second source is only a requirement when doing small volume products that need to be maintained for a long time. Unfortunately this market is very uninteresting for most chip makers. If you are a car maker, you can buy so many devices it will be profitable to produce another batch.

I find much better the STM32 series.

I'm not sure I see the connection between a requirement for long time availability and second sources. Unless the connection is through the fact that this is mostly to support the auto industry, why would a second source make the life time of the product any longer? If anything it would reduce the volumes of each maker and push the part to be canceled earlier.

Rick

Some may want it but it is hardly as important as you're trying to spin it. The fact is there are virtually no second source devices out there today, so it is definitely not high on anyone's requirement list.

R&D, all they are doing is an extension

Well maybe you explain why it is a good idea to make X devices in one fab and Y in another when you could make X+Y in just one fab and get the advantage of higher volumes? It only becomes necessary to use another fab if one is maxed out (highly unlikely given the volumes for automotive are small compared to other market segments).

Given the high costs involved to align two processes and qualify identical designs on them and the resulting lower volumes I'd say it's a good way to ensure the devices will cost more than they really need to. Also we're not talking about true second source with price and performance competition (like when you could replace an Intel 286 or 386 with an AMD clone that was faster) but a joint venture by two companies.

If wasting money appeals, maybe... I'm not holding my breath.

Wilco

The Logic and SRAM memory markets give some examples of how this works. A second-source seeds more design wins (in theory), and so ensures earlier critical mass. Logic suppliers are well used to this from their customers.

Then, a decade or two or three later, as the volumes tail off, all suppliers do NOT pull the plug concurrently.

Some jockeying occurs, and one supplier decides to 'take the tail business' - Tail-end SRAM has examples of this right now, not many suppliers, but the market size is still enough to keep at least one company interested. Same with 4000 series CMOS - we still have that in active designs and it is even on new design radar too.

You can be sure ST and Freescale are only doing this uC second sourcing, (which is extra effort) because it will result in higher total revenue. Design-ins will already be underway.

-jg

R&D, all they are doing is an extension

Talk to the big players that do this already.

Xilinx is one good example.

Some large corporates (and military too) see the bigger picture, and have quite strict risk management policies, that dictate an earthquake or other single point event, cannot be allowed to disrupt their total business.

When you are making cars, that processor suddenly gets extremely expensive if that ONE fab is taken out, halting your vehicle line!

Spanning two continents, and with multiple fabs.

I'd say that has some appeal to the risk-managers in the large corperations.

You see, there is a LOT more to chip selection decisions, than just "which core"?

-jg

Speaking of "less tangible": how exactly does one "see" commitment? One can believe in some company executive's statement of commitment --- or one might opt not to. But how do you see commitment other than after the fact, when it's no use to anybody?

messagenews: snipped-for-privacy@clear.net.nz...

R&D, all they are doing is an extension

Yeah, now there is a huge market, Military! Let's see, a few dozen processors in a missile that is made maybe 1000 per year, which gives you a volume of... well, let's be generous and say 100,000 a year. Yes, if it is specially qualified, the pricing is up there and I am sure the profits are too. But unless there is a specific promise to make that device for a long time, it gets the axe when the commercial version goes by the wayside. Didn't that just happen a few months ago that a Power PC processor was being dropped that the Military is using big time? When was the last time an earthquake took out fabs for more than a couple of weeks??? Obsolescence is the enemy of the long life part, not earthquakes.

Which large corporations? Do you really think that anyone other than automotive and Military care about multiple sourcing? In my career of about 30 years, the only time I have been asked about second sourcing is in DOD type jobs. Even those have mostly given up on the idea of second source and are trying to go the route of using more commercially oriented designs where possible. Some 15 or 20 years ago I worked on a design for a graphics terminal that was still going to be "old school", but they looked hard to going COTS. The only thing that stopped them was inertia. Now they buy all sorts of commercial products and only ask for a 5 year life... like what I am working on now... or not working on because I am typing this...

Yes, but not to the mass markets. I can't remember the last time that even my manager got involved in the decision of which CPU chip to use. I decide, I justify and I do the design. That's the way it should be unless there is some overriding factor... and that seldom happens.

Rick

I don't even know what you are trying to say at this point. The ARM7 is being driven out by the CM3. The ARM7 will still be made for some years and very few products will "respin" to use a new part. Most of those products will be out of production by the time the ARM7 CPUs are no longer made.

Are you trying to create a straw-man?

Yes, users migrate from the ARM7 to nearly every other processor. They also migrate *to* the CM3 *from* every processor. I think there is little doubt that the CM3 is here to stay and will be a huge success. Even if it is just marketing hype as some have said, it was successful marketing hype.

I guess we will see. Like I told Nico companies may care about a product life time, but few feel the need to have second sources. The reason for second sources has always been product availability and price. Availability can be an issue, but having multiple sources doesn't fix that. If the demand peaks, a dozen companies can be caught with their pants down. Just ask the memory vendors. Every four years or so DRAM/Flash prices jump and then after a bit they continue dropping.

Likewise, price gets no advantage from multiple sourcing. Everyone competes on price now with sole sourcing. You design in based on the price they will give you and in a year to work the next design and start again with the newer parts and new pricing.

Don't make this more complex than it is.

Rick

... snip ...

On the contrary, it is very important. However the inavailability for most products forces other choices. Without second-sourceing device production can easily be stopped by a single supplier. The result is last minute massive redesigns, followed by loss of that customer to that supplier.

Don't kid yourself. A company is not going to go the second source route knowing that the product lifetime will be shorter than normal for them and longer for their competition. Like you say, they will do the second source because there is a customer (or class of customer) who is asking for that. But that does not translate into a market wide demand for second sourcing. Considering that you are the only person here who is saying it makes a difference (and I haven't heard you say it makes a difference to *you*), I expect this is a NOOP to most applications.

It will result in additional revenue just as any product introduction will result in additional revenue over not introducing a product. Again, that does not translate into any sort of demand by the general market. Even you have only come up with the automotive sector as the potential customers for this.

In general, markets tend to focus on "favorite" CPUs. Comms market likes the Motorola (who are they these days??? Freescale? ON? I just can't keep up) CPUs, automotive likes the ARM (just ask TI). I don't really know the rest, but the bias tends to come from familiarity. An engineer from company A uses processor X, moves to company B and continues to use processor X instead of switching to processor Y. More engineers from company B use processor X and move to other companies in the same sector. It is not an absolute, but it does make a trend since frequently there is no *best* processor for a given app.

Second sourcing was big in the 70's. It went away in the 80's or so as product lifetimes got shorter. I don't see it coming back until product lifetimes get longer or the learning curves for devices get a lot steeper.

Rick

Agree with you there... I've seen co-workers who spend more time writing up documentations and holding meetings discussing the pros and cons of various microcontrollers or FPGAs they might use (on a board they were personally going to be pretty much 100% responsible for, everyone else only caring about what comes out of their, e.g., serial port or whatever) -- before writing one line of assembly/C/VHDL/Verilog/whatever -- than it would take them to just choose one themselves and finish that part of the project.

Rick, didn't you used to have a small company named Aquarius selling DSP boards? Or was that another Rick?

---Joel

No, but it only takes a couple of customers to cause DELIVERY of market wide of second sourcing. It is the result that matters, not how many customers drove it.

Second Sourcing matters to every designer. (or should)

eg: Right now, a customer has asked us for an extension in operating voltage, so we are trawling the net, looking for 'same package/same pinout' alternatives [ie a second source!] that we can drop onto the PCB, with no other changes.

BECAUSE suppliers choose to deliver common packages and pinouts, we are able to that (and Looks like a few suppliers will change, in the BOM). Had we chosen single-sourced parts across the design, we would be in full respin mode == Delays and $$$.

It is the microcontroller sector, that is somewhat lagging here, second sourcing [common packages and pinouts across vendors] is very much alive and well in many other sectors.

Some uC vendors ARE waking up to this wider across-family Pin-Compatible second/alternate source idea. Freescale do this now, NXP are going to do this, I believe. Large chunks of the 80C51 market operate this way.

Why ? It speeds their design ramp, and helps the customer. Can't see any losers in that ?

Another example, is Voltage-Sourcing - a feature subset of second sourcing. Half a decade ago, uC vendors were delivering an ever-multiplying range of Vcc's, and design of products was an exercise in 'moving goal posts'.

Now, we see more devices offered with 5V Vcc (sometimes 3.3V), and on chip regulators, that 'hide' the core Vcc.

Why ? Because customer demand MADE that happen - they want to MINIMISE the supplier dictates/gotchas in their systems.

Again, I see the Automotive Customers leading this trend. Others benefit from it.

I am quite happy to take advantage of it :)

Vendors that ignore these important trends are going to wake up one day, and wonder where their market share went....

-jg

That might have been true in the past when each manufacturer had a distinctive distributor. Now every distributor carries (almost) all manufacturers. So getting design wins is not a matter 'if the customer doesn't like distrib. A, then he might buy through distrib. B' anymore.

Well, a second source usually takes over when the original manufacturer quits or is about to quit. I deal with low volume long running products every now and then. After a couple of years a second source is required because the original manufacturer quits making certain parts.

That is not a second source. That is a new primary source. A second source gives you redundancy -- you can buy part B if part A is unavaulable, buggy, etc.