NXP announce ARMs down to 8 pins and $0.39

[Mechatrommer]

And yes, 5V is so 70s

you are drunk right?

[mrflibble]

And yes, 5V is so 70s

you are drunk right?

Drunken typo. He probably meant 60s.

Other than that it's quite a few years back now since I last used a 5 Volt part. Most of the breadboards are 3.3 or 3.0 Volt.

[madires]

In my experience, even if you go through all sorts of gyrations to prevent digital noise from getting into your ADC during conversions (like putting the chip to sleep while the ADC is converting), you still are not going to get the performance that you would enjoy from an external ADC.  It depends on what I'm doing of course-- if I'm just reading the position of a potentiometer or something, an on-chip ADC is fine-- but for anything that needs to be seriously accurate, I usually opt for an external ADC.

If feasable (by the project) use oversampling with the internal ADC and the noise even helps. With that method you'll get a resolution about 12 Bits from a 10 Bit ADC, or up to 16 Bits from a fast 12 Bit ADC. Statistics don't lie :-)

[nctnico]

In my experience, even if you go through all sorts of gyrations to prevent digital noise from getting into your ADC during conversions (like putting the chip to sleep while the ADC is converting), you still are not going to get the performance that you would enjoy from an external ADC.  It depends on what I'm doing of course-- if I'm just reading the position of a potentiometer or something, an on-chip ADC is fine-- but for anything that needs to be seriously accurate, I usually opt for an external ADC.

If feasable (by the project) use oversampling with the internal ADC and the noise even helps. With that method you'll get a resolution about 12 Bits from a 10 Bit ADC, or up to 16 Bits from a fast 12 Bit ADC. Statistics don't lie :-)

Indeed statistics don't lie but if you put it to the test then you'll see you won't get more resolution. Getting more resolution depends greatly on the spectrum and quality of the noise. Good noise is something you won't get for free.

[mrflibble]

If feasable (by the project) use oversampling with the internal ADC and the noise even helps. With that method you'll get a resolution about 12 Bits from a 10 Bit ADC, or up to 16 Bits from a fast 12 Bit ADC. Statistics don't lie :-)

Better make sure your noise is nice and gaussian then.

[RCMR]

On another note, but in the ARM cortex area, I finally switched from my FT2232 based JTAG with SWD adapter to this J-Link clone I got off of e-Bay.
http://www.powermcu.com/product-20.html

Woohoo... I just picked up one of those myself -- glad to hear they work okay.

[NiHaoMike]

What I don't understand is how to effectively use an ARM chip with only 6 I/O pins. Maybe stuff like encrypted data logging would need a lot of processing power but little bandwidth. Or connect the SPI to another processor (and maybe a connection to a peripheral) in order to use it as a coprocessor.

If it had a built in ADC and DAC (even 8 or 10 bit ones), it might make a great modem DSP for low speed wireless and PLC applications.

[poptones]

Wouldn't a DSpic, that comes in 8 pin packages and includes a DSP core and A/D and D/A make a better... low end dsp?

That said, I also don't get the point of a low cost chip that has no analog I/o. In fact, I can't understand making a modern 8 pin chip that doesn't have the ability to use ANY of the 5 or 6 usable pins as either analog or digital.

[NiHaoMike]

The dsPIC is only 16 bit (which is still plenty for a simple modem), whereas ARM is 32 bit. What Microchip has to compete with the low end ARMs is the PIC32, which is basically a 32 bit MIPS processor.

[Mechatrommer]

What I don't understand is how to effectively use an ARM chip with only 6 I/O pins.

one eg (4pins IO) i've done. radio remote control. you hook up with radio kit or module and few buttons... 2 or 3 serial lines to radio (depending on the module type) so you got 2 or 1 free pin for button press. there's module also will allow for 1 line only for radio, then you can make 3 buttons each with its own specific data to be sent to radio.

[T4P]

The PIC32 simply cannot win ARM on the processor grunt front, granted they do have a 28pin processor but NXP announced a 28-pin 0.6" DIP chip right? It's the LPC1114

[mikeselectricstuff]

What I don't understand is how to effectively use an ARM chip with only 6 I/O pins.

No different from any other 8-pin chip. In small MCU applications, processor speed is rarely a big issue - it's typically more about peripherals and devtools.
AFAIK two UARTS in an 8-pin package is unique - could be handy for protocol conversion type applications.
I don't understand why they are offering DIP8 but not SO8 though - OK the SSOPs are probably a similar size to a SO8 but for consumer type apps you may not want the pin density.

[Mechatrommer]

What I don't understand is how to effectively use an ARM chip with only 6 I/O pins.

In small MCU applications, processor speed is rarely a big issue

another project i've done pic10f206... 3 pins to read rotary encoder (led + wheel ball mouse type), 1 pin comm to main mcu (if you have more pins for comm is better), and this is certainly speed issue. the encoder is tucked to the 3000rpm. be it tucked to 10Krpm motor, i certainly need more MHz to work on. why i did the project? check out the price for "absolute position" rotary encoder. here one eg... http://www.ebay.com.my/itm/OMRON-Absolute-Rotary-Encoder-E6CP-AG5C-256P-R-E6CPAG5C-new-box-free-ship-/140854534118?pt=LH_DefaultDomain_0&hash=item20cb959fe6 thats the cheapest china "buy it now" currently in ebay. you can have better absolute encoder at quadruple the price. and i'm not sure how much rpm they can go to. and go figure how to interface this one... http://www.ebay.com.my/itm/ALLEN-BRADLEY-8-24VDC-GRAY-CODE-ABSOLUTE-ROTARY-ENCODER-845G-F3G8HC1024A-NEW-/271104847790?pt=LH_DefaultDomain_0&hash=item3f1f1bafae mine was using serial comm... and 1 pin only! because the space constraint i needed to use sot package (6 pins pic10f206).

[SeanB]

Most likely the die does not fit the package yet in one dimension. A few revs and it will shrink to fit inside

[westfw]

What I don't understand is how to effectively use an ARM chip with only 6 I/O pins.

You're not supposed to use it "effectively."  You're supposed to do the same sort of thing you do with the digital-only 8bit chips like a PIC12F509 or an ATtiny13.  Only you (are supposed to) rejoice because you don't have those tiny 1k-instruction limits any more.  The fact that the CPU is vastly overpowered for most of what you can do in 6 pins is supposed to be irrelevant, because the cost is the same as a less powerful CPU anyway.

Three uarts and 6 pins, eh?  That's sorta nice.  "other vendors" have been beaten up for not having any uarts in their small parts.

[mikeselectricstuff]

Three uarts and 6 pins, eh?  That's sorta nice.  "other vendors" have been beaten up for not having any uarts in their small parts.

The 8 pin one has only 2 UARTS. Something I noticed when skimming the data but I've not looked into in detail is that the UARTS share a fractional baudrate divider - I hope this doesn't mean they all have to run at the same rate, but just that only one can have funky division ratios. Also I don't think they have fifos.

[alm]

This would still be an improvement compared to the very limited peripherals available on 8-pin PICs/AVRs. I believe that the PIC10 series also has a quite ancient architecture not well suited for programming in C, this ARM core would allow you to use the same development tools as on the bigger parts.

[NiHaoMike]

one eg (4pins IO) i've done. radio remote control. you hook up with radio kit or module and few buttons... 2 or 3 serial lines to radio (depending on the module type) so you got 2 or 1 free pin for button press. there's module also will allow for 1 line only for radio, then you can make 3 buttons each with its own specific data to be sent to radio.

Why would you want an overkill ARM core when a lower power 8 bit core does the job just fine? (I suppose an ARM core might make sense if you have a highly encrypted rolling code system...)

Maybe having "the world's fastest 8DIP processor" is a good enough reason...

I wonder what the license fees are for the ARM cores. They obviously must be really low for those low end cores if the chip is going to sell for 39c.

[andersm]

Why would you want an overkill ARM core when a lower power 8 bit core does the job just fine?

Why would you insist on an 8-bit core if the ARM MCU does the job equally fine? (As Mike suggested, better tools is a good reason to use ARM.)

[mikeselectricstuff]

Why would you want an overkill ARM core when a lower power 8 bit core does the job just fine?

Why would you insist on an 8-bit core if the ARM MCU does the job equally fine? (As Mike suggested, better tools is a good reason to use ARM.)

Why would you assume 8 bit is lower power ? ARM is probably on a smaller geometry, can typically run at lower voltage and does more per cycle - this may or may not be useful - for bit-twiddling an ARM is little benefit, but any maths greater than 8 bits will be more efficient. Many power-sensitive apps will be sleeping and only drawing power when processing something - a faster processor can go back to sleep more quickly, so any increase in power draw may be compensated for in less time drawing it. 
8-bits will never go away, but more choice is always welcome - there will always be new applications that can make use of it.

[T4P]

I don't know about NXP but Energy Micro's parts have fierce clock gating.

[free_electron]

here's a few pointer as to why they put arm beast in 8 pins package

- lower power design than conventional cores
- smaller geometries. thus more yield per wafer. rerunning the mask for an old design in smaller geometry costs the same price as new technology in smaller geometry. so why keep holding on to the old ?
- arm is very good to run 'c' language. believe it or not butmost 8 bit machines are deplorable when it comes to running 'c'. The design of the languae requires a pile of ram and a register based architecture. not something most 8 bitters have... so the implemetation is crappy and you need all kinds of trickery in the compiler to 'emulate' a register based machine. swap core : problem solved.
- one core scales up from 8 pins to 144 pins and beyond. 1 compiler toolchain. ideal for the codemonkeys

[BravoV]

Maybe someday someone will build a cluster computing from this DIP sized Arm chips by stacking them with so called "Beowulf Shield" just for fun. 

[cloudscapes]

On one hand, it's always nice to see more fast micros, especially in breadboardable DIP!

But on the other hand, I fail to see many uses for an 8-pin micro of this speed. What do you do with 6 free pins? Control relays, preset saving, read some potentiometers, a bit of arithmetic, etc. I don't think 6 free pins isn't going to be a central "brain" of anything that an 8bit avr or pic can't do. Fast micros like PIC32 and ARM shine when you've got all this stuff interfaced to it and have it orchestrating a circuit.

I dunno, I may be wrong.

[T4P]

It's not about the speed though, not that you could use much with massive breadboard inductance 
It's tough to squeeze code into 1KB of space for the cheapest 8pin DIP ATTINY or even worse (512Bytes) for PIC 
And the fact that the chips only have 64Bytes of ram is a bit disconcerting ...
You could simply do much more with 6 pins with 32bit because of the addressable size as well as smaller manufacturing processes NXP can use for ARM