Preferred MCU manufacturer selection

There are some AVR's close to that size, that have far fewer pins so they can be assembled without special equipment beyond normal SMT tools. I haven't looked into exotic packages for that reason. Certainly you can get them as bare dice.

There is no such thing as too small, too simple, or too cheap. They are comparatives. If processor A is smaller, simpler to program, and cheaper than processor B, then that is often enough to say "processor A is preferable" regardless of the actual parameters.

I wouldn't say the correlation is loose.

It has 20 solder balls on the bottom of the package and you need special assembly equipment and PCB fabrication to use it. It doesn't seem practical for low cost, small quantity applications. I'd like to know why they don't make 8-pin ones like the 8 bitters.

It also has "10 power modes to help developers conserve power and to provide a flexible power on environment." I've never heard of an

8-bitter where the programmer has to deal with 10 power modes.

--

John Devereux

Nice but that's a PDIP, a rather big package. I remember noticing it before and wondering why there is no small SMT version. More code space than 4k would also be nice.

I've heard ARM code density also suffers because, all immediate constants including offsets are stored as 32 bits. You end up needing 8 bytes for a call or jump (the instruction itself, and the target address). But I'm not sure of this.

That's only partially true for ARM, and even less so for Cortex. Cortex is a mixed 16/32 bit ISA, and in a typical program, most instructions will be 16 bit, including instructions that can work with 8 bit immediate constants, and branches with short offsets, as well as branch link instructions that are 32 bits, with bigger offsets as part of the instruction. Overall, Cortex programs are quite compact.

But even on 32 bit ARM there's a branch instruction, which is 32 bit, but includes a 24 bit offset in the instruction itself.

It was an LPC23xx series, an ARM 7 variety.

Yes, until some manufacturer comes with a complete line up of Cortex-Mx MCUs from 8-pin up to 100-pin in "friendly" SMT versions, I will probably stick with my approach.

--------------------------------------- Posted through

I don't get your point? There definitely *is* too small if you can't handle the soldering or too simple if it won't do the tasks you need or too cheap if it doesn't have the capabilities you need.

Regardless, there *is* too big, etc. and if you don't specify how small, etc. how can the "requirement" be met?

No, but I did.

Have you looked? I'm pretty sure they do. I don't recall which maker, but someone had come out with a small part like you are talking about some years ago, I think it was NXP. Something like an 8 pin QFN I think.

Lol. You just haven't looked very hard. You don't have to use all 10 modes. You can just use two or even one if that meets your requirements.

What is your point?

--

Rick

I bet you just haven't looked very hard. The ARM is the new 8051. I doubt there is much of anything you can't find in terms of package and capabilities. For one, NXP has a huge range of devices from large to small. Atmel seems to sell more devices, but I have the impression their product line is not as broad, but I haven't looked in a while, so that may no longer be true.

Digikey would be a good place to sort through the package and device combinations.

--

Rick

Fine. :-)

I'll have a ARM in PDIP, ~28 pins, 128K flash, 32K SRAM with USB host and device please.

Simon.

PS: This is a good natured response. :-)

PPS: The description matches parts available in the PIC32MX range.

PPPS: I would _really_ love to see a ARM MCU available in that configuration at around the PIC32MX price point.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP 
Microsoft: Bringing you 1980s technology to a 21st century world

I've looked (maybe not everywhere), and not found anything suitable. Certainly there's fewer to choose from, compared with 100's of 8-bit parts in all kinds of configurations, in packages like that.

You can run the 8 bitter at full speed and still be at very low power. It looks to me like the ARM parts need intricate programming to switch between power modes so they use power only during bursts of calculation or whatever. The Stellaris board (Cortex M4F admittedly) seems to have a bunch of proprietary rom code to help deal with this, but then I don't know what the heck it is doing.

Have you looked? Try Nuvoton, I think they do PDIPs. There was a German company making ARMs for the automotive market but I'm not sure even they were making PDIPs. I mean, come'on, this is the *real* world. ;)

--

Rick

Ok if you can't find the "perfect" fit in an ARM use an 8 bitter. My point is that unless you really need that "perfect" fit which is usually price rather than package, ARMs can suit nearly any MCU need.

That's not true... but then I guess you need to define "low power". If you only need 12 MHz operation from an 8 bitter, I expect you could run the ARM at 1 MHz and get the same performance. So which is lower power then? I dunno, it would be a foot race I bet.

So why use power modes if you don't like them? Run with a slow clock rate and still outperform the 8 bitter at "full speed".

--

Rick

device please.

Nuvoton do have Wide Vcc parts, but PDIP they do only in their 8051 family. Cortex parts are 33-128 pins, all SMD. Some 0.4mm pitch, 4mm x 4mm.

Nuvoton also plan to expand their 8051 offerings (as too do SiLabs) so there is plenty of life in 8 bit devices yet, especially at the low-pin count, wide supply end of the market.

Infineon have Industrial TSSOP16 M0's, which is quite easy to handle - you are right that DIP is more marketing gimmick that mass-volume design driver. (interestingly Infineon have many Automotive parts, but the M0 they call Industrial )

device please.

is plenty of life in 8 bit devices yet, especially at the low-pin count, wide supply end of the market.

I never said the 8 bitter is dead. My point is that for most projects that aren't highly cost sensitive, the 32 bit world is a one stop design tool. But when cost is a big deal, yes the 8 bit devices can save a few pennies that add up. And of course those apps are the ones that run the numbers up on the sales figures. So no, the 8 bit parts are still used in quantity, but is there any reason to use them on a project that has a lifetime run of

Oh, yes, I've looked. :-)

I do understand the comment about the "real world" :-), but it also raises a question: why are Microchip bothering to produce a 32-bit MIPS core in PDIP and with full blown USB host thrown in as well ?

It obvious why they continue to produce 8-bit devices in PDIP, but why bother to design and produce a PDIP variant of a whole new 32-bit core unless there was a viable market for it ?

BTW, the PDIP ARM you mention in the other posting is the LPC1114. 32K flash,

4K SRAM, but no USB. There's also a 8-pin LPC810 in PDIP; 4K flash, 1K SRAM.

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP 
Microsoft: Bringing you 1980s technology to a 21st century world

Presumably they want to sell an upgrade path to their existing customers. The PIC32s are 3V parts and have only a limited number of 5V tolerant pins, but do you know if they're otherwise pin-compatible with some 8- or 16-bit parts?

-a

Unfortunately the small pinout chips seem entirely unobtanium. The LPC81x family isn't stocked anywhere (at least according to FindChips). The LPC1114 is (at its smallest) 28 pins; and the Infineon series (noted by j.m.granville) is "coming soon" according to the Infineon website.

So far I haven't found a truly small form-factor ARM chip that is available from distribution, despite the rhetoric. I would love to learn otherwise - what did I miss?

Have you tried talking to anyone in distribution or at the manufacturer? I figured being a small guy I wouldn't get much attention, but I was surprised to find that both Digikey and the manufacturer are interested in my problems. Digikey in particular pulled out all the stops once to get parts in for me to meet my deadline. I ended up actually exchanging some emails with the buyer in charge of inventorying the part.

Certainly the manufacturer would be able to tell you their plans for supplying a given device and I expect you may have some input to their plans.

Sometimes people surprise you.

--

Rick