Forums

WOW, 32-bit ARM flash controller for $1.47 from Philips

Started by An Schwob in the USA September 26, 2005
Hi everybody,

The LPC2000 forum on Yahoo http://groups.yahoo.com/group/lpc2000/ had a
posting today that I consider a huge step towards having 32-bit micros
in almost all applications. An ARM7 device running a stunning 70 MHz
starting at $1.47! It has strong communication features such as two
16550 compatible UARTS, 1 SPI, 1 SSP, 2 I2C, 8-channel 10-bit ADC and a
Real Time Clock that can be run from a 32 kHz additional clock.
Can't wait to get such a micro in my hands but the announcement talked
about samples in November.  There is however a prel. Data Sheet for
download available

Well, there is more to that device and if you are interested, check it
out here:
http://www.standardics.philips.com/news/lpc210x/

I know that some of you will probably flame me for this posting (to
those my apologies) but usually I do post technical answers to
questions here in the newsgroup, so I hope this will be interesting for
many users here.

An Schwob

An Schwob in the USA wrote:
> Hi everybody, > > The LPC2000 forum on Yahoo http://groups.yahoo.com/group/lpc2000/ had a > posting today that I consider a huge step towards having 32-bit micros > in almost all applications. An ARM7 device running a stunning 70 MHz > starting at $1.47!
It is only 8KF ( which on ARM is probably sub ~2K on 80C51 ), but the price is not the usual '2006 asymtope' price, but a 10K one, so that is impressive. Maybe this is why the LPC952 seems to have hiccuped :) Pin count is also good, for that price. 32 5V tolerant IO.... Dual Vcc is a minus... They could have copied the Infineon XC866, and put an On Chip regulator ? At this price/Icc level, it is VERY hard to find 1.8V regulators, that do not impact either Icc, or BOM $$$. This will cause some deep swallowing across the industry.... It has strong communication features such as two
> 16550 compatible UARTS, 1 SPI, 1 SSP, 2 I2C, 8-channel 10-bit ADC and a > Real Time Clock that can be run from a 32 kHz additional clock. > Can't wait to get such a micro in my hands but the announcement talked > about samples in November. There is however a prel. Data Sheet for > download available > > Well, there is more to that device and if you are interested, check it > out here: > http://www.standardics.philips.com/news/lpc210x/ > > I know that some of you will probably flame me for this posting (to > those my apologies) but usually I do post technical answers to > questions here in the newsgroup, so I hope this will be interesting for > many users here. > > An Schwob >
Jim,

agree in all but two of your statements.  8k Flash on ARM is like 2k of
8051 only if you do bit manipulations.  If you use Thumb mode and use
an ARM for serial protocol conversion or may be even better as a Math
Coprocessor, the code size on the ARM will actually be similar for the
protocol converter or MUCH smaller for the math Coprocessor. May be we
can agree to worst case 2k of 8051 code size, however best case for the
ARM could be like 64k of 8051 code ;-).

In regards to the ICC of a dual voltage supply, it is probably better
to have an external converter because the existing ones on
microcontrollers do increase the quiescent current of the micros more
than the external component. For cost reasons I would prefer the 1.8V
regulator to be on-chip as well.

The real value of these 3 devices announced by Philips is the huge
processing power for the money but after all, that depends on your
application again.

An Schwob


Jim Granville wrote:
> An Schwob in the USA wrote: > > Hi everybody, > > > > The LPC2000 forum on Yahoo http://groups.yahoo.com/group/lpc2000/ had a > > posting today that I consider a huge step towards having 32-bit micros > > in almost all applications. An ARM7 device running a stunning 70 MHz > > starting at $1.47! > > It is only 8KF ( which on ARM is probably sub ~2K on 80C51 ), but the > price is not the usual '2006 asymtope' price, but a 10K one, so that > is impressive. > > Maybe this is why the LPC952 seems to have hiccuped :) > > Pin count is also good, for that price. 32 5V tolerant IO.... > > Dual Vcc is a minus... They could have copied the Infineon XC866, > and put an On Chip regulator ? > At this price/Icc level, it is VERY hard to find 1.8V regulators, > that do not impact either Icc, or BOM $$$. > > This will cause some deep swallowing across the industry.... > > > It has strong communication features such as two > > 16550 compatible UARTS, 1 SPI, 1 SSP, 2 I2C, 8-channel 10-bit ADC and a > > Real Time Clock that can be run from a 32 kHz additional clock. > > Can't wait to get such a micro in my hands but the announcement talked > > about samples in November. There is however a prel. Data Sheet for > > download available > > > > Well, there is more to that device and if you are interested, check it > > out here: > > http://www.standardics.philips.com/news/lpc210x/ > > > > I know that some of you will probably flame me for this posting (to > > those my apologies) but usually I do post technical answers to > > questions here in the newsgroup, so I hope this will be interesting for > > many users here. > > > > An Schwob > >
An Schwob in the USA wrote:
> Jim, > > agree in all but two of your statements. 8k Flash on ARM is like 2k of > 8051 only if you do bit manipulations. If you use Thumb mode and use > an ARM for serial protocol conversion or may be even better as a Math > Coprocessor, the code size on the ARM will actually be similar for the > protocol converter or MUCH smaller for the math Coprocessor. May be we > can agree to worst case 2k of 8051 code size, however best case for the > ARM could be like 64k of 8051 code ;-). > > In regards to the ICC of a dual voltage supply, it is probably better > to have an external converter because the existing ones on > microcontrollers do increase the quiescent current of the micros more > than the external component. For cost reasons I would prefer the 1.8V > regulator to be on-chip as well.
Look at how the XC866 handle this - they use dual regs, a tiny one for RTC and Ram-keep and another for Core operate. Result is a single device with wide supply operation, plus avoids the kludge of 5V pullups when driving MOSFETS.
> > The real value of these 3 devices announced by Philips is the huge > processing power for the money but after all, that depends on your > application again.
Or the peripherals.. This must be closest to the cheapest uC with Dual UART - and a lot of RAM for the price too.... Other oversights : ## No sign of an OnChip CalOsc ? [ Maybe the CCO can free run, for med. precision ] ## Peripherals do not seem able to run at full CCLK, for best granularity ? It is common now to see uC with peripherals clocked FASTER than the core, so one that cannot go faster than CK/2 seems a poor resource use. Surely the Silicon can go at the CCO freq, so running the timers from that, would give fine PWM control. -jg
Jim,
now I did look at the 866 which has no power consumption listed yet. On
chip voltage conversion usually results in rather miserable power down
values.
Comparing a 26.7 MHz 2-cycle 8-bit core to a 70 MHz (almost) single
cycle 32-bit core does not seem too appropriate.  If a peripheral runs
half CPU clock on the ARM ist is still much faster than full speed on
the XC866. The data sheet is already 7 months old and does not include
values for any power consumption. The timer and analog features seem
rather similar while the communication features are much better on the
LPC2103.
Don't know the pricing of the 8k Flash or 16k Flash but would be very
surprised if the 866 with Flash is lower cost than the ARM.

An Schwob in the USA wrote:
> Jim, > now I did look at the 866 which has no power consumption listed yet. On > chip voltage conversion usually results in rather miserable power down > values. > Comparing a 26.7 MHz 2-cycle 8-bit core to a 70 MHz (almost) single > cycle 32-bit core does not seem too appropriate. If a peripheral runs > half CPU clock on the ARM ist is still much faster than full speed on > the XC866. The data sheet is already 7 months old and does not include > values for any power consumption. The timer and analog features seem > rather similar while the communication features are much better on the > LPC2103. > Don't know the pricing of the 8k Flash or 16k Flash but would be very > surprised if the 866 with Flash is lower cost than the ARM.
IIRC the indicators for xc866 are around 1e, but typically vague on exact code size, and volumes. I was not trying to equate the two cores, just point out that deeply shrunk uC using On Chip regulators ARE being done, and on chips that target low costs. On the XC866, it allows them to have a 2.5V core, but still offer full 5V port operation - so you can direct drive Power MOSFETS for example. Atmel also realised the problems with dual supply uC, and their newest SAM7 variants include a 1.8V regulator. So I was surprised Philips went back to split core, especially as they DO have LPC213x members that are single supply ? As the price drops, the external regulator becomes a larger % of the BOM. -jg