ST beat NXP again in the MCU battle

[westfw]

Has anyone ever doubted that Microchip's IO expanders or IR comm chips are actually pre-programmed PICs?

[dannyf]

It is not that hard to implement an I2C/SPI slave on a chip with hardware i2c. So it is entirely possible to code a PIC to behave like a i2c/spi slave.

Whether those devices are actually PICs, you may have to ask Microchip - I doubt it.

[westfw]

(well, their newer USB/Serial bridge chip turned out to be a PIC.  To the point where a programmer would identify and re-program it (saving a few cents off the cost of the same chip sold as a programmable device.) http://dangerousprototypes.com/2011/01/18/hack-open-source-usb-stack-on-mcp2200/ )

I oughta take one of those micros with a "parallel slave port" or whatever and write firmware so that it can emulate one (all) of those popular legacy IO controllers (MC6850, etc.)  It'd be nice to have that sort of functionality in a narrow 28pin package with assorted modern features...  It should be just about possible, speedwise...

[Stonent]

Hmm LQFP 32, 48 and 64. TSSOP 20. Nope, no DIP.

[poorchava]

Why would you need a DIP anyway? Production will use SMD if they hava a chance to, because of smaller footprint and cost, for prototyping you use a breakout board or something. Neither NXP nor ST gives a shit about hobbyists.

[timb]

The only problem with SMD to DIP breakout boards is it's a PITA having to keep 15,000 different variations on-hand just to prototype something. When I come up with an idea and I'm looking for a chip to test it with, I always have to go and look what boards I have on hand.

"Hmmm, twenty-seven and one-half pin TSS-FLIP-FLOSS small oatmeal package with 1028 micron lead spacing in a Wide Track Grand Prix body style... Let's see, let's see... Oh, here we... Wait, no... *Digs Through a Stack of Boards*"

[Kjelt]

No-one said this hobby is getting easier 
I sometimes bent the pins of smd's i can't handle directly (no breakout board or whatever) alternately up and down and glue it with a small raiser to the prototype board, then handsolder 0.1mm wires to the pins to the prototype board.
Only BGA's are a real PITA and I therefore don't use them.

[westfw]

Neither NXP nor ST gives a shit about hobbyists.

I claim that the existences of NXP ARMs in DIP packages suggests otherwise!  (because you're right - DIPs are pretty much only for hobbyists.)  Except that they probably aren't clear where the the "hobbyist" market stops, and the "educational" market (which is more important) begins.  (and understandably so.)

I can see some Marketing Guy at STM sitting there saying "our Discovery boards are MUCH more powerful than an Arduino, and cheaper too (after we sell them at a loss), and we've even given away a whole bunch of them.  How come OUR name isn't plastered all over the popular press and blogosphere, like Atmel's?  Why aren't we getting any damned MIND SHARE?"

(does anyone doubt that Arduino has been really good for Atmel, regardless of whether there are actual "devices sold" count for Arduino boards is "significant" ?)

(maybe it's because hobbyists PUBLISH.)

[mikeselectricstuff]

Neither NXP nor ST gives a shit about hobbyists.

I claim that the existences of NXP ARMs in DIP packages suggests otherwise! 

Nonsense. Hobbyist volumes are negligible. It's about volume markets like washing machines and other non space-constrained applications where single-layer PCBs are the norm, and there are other big, lumpy parts like relays which mean through-hole is more appropriate.
Any hobbyist-friendliness is just a spinoff from the requirements of real volume customers.

[Stonent]

That and Arduino is a good "get your foot in the door" for Atmel. That was one of the first things I thought of after getting an Arduino was, "Hey, I can just pop this chip out and it only costs 3 bucks vs 30!"

Atmel, PIC, NXP, Propeller, TI all have DIP microcontrollers. I think On Semi does as well.

[westfw]

Any hobbyist-friendliness is just a spinoff from the requirements of real volume customers.

So what IS the "real volume customer" requirement for an ARM in an 8pin or 28pin DIP ?

(NXP in particular seems to be issuing all sorts of experimental packages; at the opposite end from DIPs we have the LPC1102 in 16pin 0.5mm BGA (2.2*2.2mm total size)   I think they're just "fishing", and I worry about the longevity of these packages if they DON'T pick up some volume customers.  OTOH, TI seems to be pretty committed to offering (some of) their low-end MSP430 products in DIP packages; the MSP430F2001 has been out a long time now.)

[dannyf]

I think they're just "fishing",

BGAs are not uncommon in production. I do think offering DIP is fishing from a volume customer's perspective - it is hard to imagine a volume customer utilizing DIP today. SOIC, msop, qfn or lqfp are far more common.

[tszaboo]

Nonsense. Hobbyist volumes are negligible. It's about volume markets like washing machines and other non space-constrained applications where single-layer PCBs are the norm, and there are other big, lumpy parts like relays which mean through-hole is more appropriate.
Any hobbyist-friendliness is just a spinoff from the requirements of real volume customers.

Well, it is not like theese machines are using anything else than 68000 or 8051. And to be honest I think it is never justified to use that paper one sided PCB. I know how much a 2-4 layer PCB costs in high volume, which can completely solve the design in SMD. And TH components are more expensive, a 1uF cap costs 6 cents instead of 0,4 in TH package. The population cost is 1 cent for SMD in contrast to a Chinese worker. 99% of the components are SMD, and they just dont make the good stuff in TH for good reasons.
I really dont know, how they justify their obsoleteness, and no-one seems to reveal the true story. I'm guessing that they calculate the cost of throwing out equipment, or they decided to use something for 20 years, and they are stuck with it. Accounting cannot handle it maybe.
Also, you should not bring up Arduino as an example for a good project. It was a student project, never intended to be used this broad.
http://www.adafruit.com/blog/2010/03/18/the-first-arduino-ever-made/
They have a market aim. 0,5mm BGA is obviously hand held market, while QFP packages are for general. DIP is only justified for millitary/space/harsh.
And it is really not like you need to do so much old school prototyping with ARM. Pinouts and firmware is compatible (usually for a single company) so just buy one of the demoboards for a price of a pizza, and use that.

[poorchava]

another thing: once they have the silicon designed and tested they can put it into any package they want. Cost of introducing another package is not very high in comparison to silicon development.

In the low end market price is everything, so unless they can sell those microcontrollers in massive volumes for all the possible stuff including ADC, port expander, i2c-uart bridges and whatnot at lower prices than actual dedicated chips that do those things, they will not score a big success.

[westfw]

you should not bring up Arduino as an example for a good project.

"Good" has nothing to do with it.  Arduino's *results*, for Atmel, are enviable...

[mikeselectricstuff]

And to be honest I think it is never justified to use that paper one sided PCB. I know how much a 2-4 layer PCB costs in high volume, which can completely solve the design in SMD.

Large stuff like washing machines often have large PCBs as controls,LEDs etc. are spread out over a front panel, so a paper based, single layer PCB can save significant cost.
 

[dannyf]

Arduino's *results*, for Atmel, are enviable...

I don't know what "enviable" is. But I am willing to say that the sales through all Arduino channels likely means non-substantial to Atmel (a $1bn+ revenue company).

I don't know how many arduino boards are sold (250k/500k/1 million? per year) but let's say that 1 million ardiuno boards / avrs are sold. Let's also say that each arv amounts to $3 sale to Atmel (likely far less than that). So that means total revenue from the arduino boards is less than $3 million at the top end, or 0.3% of atmel's total revenue.

Let's say further that we are off by a factor of 10x -> arduino sales amount to 3% of atmel revenue.

Big deal? no.

Enviable? depending on how enviable you are.

[nctnico]

You shouldn't measure absolute sales but derived sales. How many products started as an Arduino prototype? Same goes for Microchip. Lots of hobbyists use PICs so PICs are likely to end up in designs.

[dannyf]

Lots of hobbyists use PICs so PICs are likely to end up in designs.

A "design" of one product is a design; A "design" of 1 million product is also a design.

All of them are designs, except that some designs are more equal than other designs.

[chickenHeadKnob]

There is another question(s) we should be asking. Is all this seemingly intense competition in the arm mcu market going to leave bankruptcies? I worry about ST micro because I have chosen to use their chips. I heard ST isn't that healthy. The second thing is does the semi-company carry too many offerings? There must be an inherent cost in maintaining such large catalogs.

[dannyf]

Other than the big 2 (renesas and freescale, and potentially infineon), the industry is highly fragmented and mostly consists of generalists. So consolidation is likely.

Surprisingly, of them all, renesas is the sickest and least likely to die. I wouldn't worry too much about ST. But the future is quite uncertain for Microchip, and Atmel to a lesser extent, in my view.

[Kjelt]

No idea how I have to read these Q3 figures but looks like:
 
Atmel has 356 million net revenue, 5.4 million net income
Microchip has 492 million net sales, 99 million net income
ST has 2013 million net revenue, -142 million loss

Hmmm this was last quarter, no idea what kind of things they did to get a loss but sure looks like ST has 5 times more revenue then Atmel or Microchip but that is then not on microcontrollers but other components. Microchip looks good, Atmel bad, ST disaster.
Hey but maybe a financial wizzard can explain this differently 

[tszaboo]

ST has 2013 million net revenue, -142 million loss

I'm guessing they still have to fire all remaining employee from the remaining of ST-Ericson. That company was a disaster.

[westfw]

the future is quite uncertain for Microchip,

Hmm.  Despite what we might call "technical inelegance" Microchip has out-performed most other microcontroller vendors on "the bottom line."

the big 2 (renesas and freescale, ...

Have had negative net incomes on their annual balance sheets for several years in a row, now.  While Microchip has been consistently profitable.
(OTOH, MCHP has gone $419M, 337M, 127M for 2011, 2012, 2013, while Freescale went -$410M, -102M for 2011, 2012, so perhaps things are turning around.)

[dannyf]

Microchip has out-performed most other microcontroller vendors on "the bottom line."

That's probably true. But how's it relevant in a discussion about "future"?

Have had negative net incomes on their annual balance sheets for several years in a row, now. 

Not sure what "negative net incomes on annual balance sheets" means.

While Microchip has been consistently profitable.

Many of us, on net income basis, outperform most of the technology stocks.

Unfortunately, that hardly makes us more valuable or prospecting than those technology stocks.