120 Degree Phase Shift Osc.

fitler.

none.

R2R ladders.

Perhaps only ONE LUT is needed..use different taping sequence to the DACs.

none.

R2R ladders.

whatever,

Leave those cats alone, or you will be reported to the ASPCA!

For the toolchain I use the free gcc (compiler) + insight (debugger) + openocd (jtag).

For small single chip micros I don't usually bother with a dev board but there are inexpensive ones from the usual sources e.g. olimex. Openocd will work with any of the ARM JTAG programmers, priced from ~$20 to ~$200.

JF has a point about the learning curve of course!

As far as the chip is concerned it is slightly trailing edge and clunky now with respect to CPU performance and peripherals. I am looking at STM32 for some new designs. But it probably still has the edge over anything else on the analog side. I think they may be about to exhibit a Cortex M3 version according to some trade rag I just read, that could be interesting,

--

John Devereux

In this case the tools can cost less than the extra board space taken up by your parts. But you are absolutely right about the learning curve. It's how I would likely do it - It would take me far longer to do it your way - but that doesn't mean it's the best solution for the OP.

--

John Devereux

On a sunny day (Sat, 06 Nov 2010 21:36:02 -0800) it happened Robert Baer wrote in :

whatever,

Yes, this subject seems to be turning into a catfight :-)

oh. sorrreee.

John

Sloman

eem

s

co=3D

en

e fitler.

nnel.

umming

se none.

use R2R ladders.

In terms of minium parts nothing will beat a micro with internal oscillator. 3x cap-resistor for lowpass filtering pwm, connect each coil between two output pins to get bipolar drive.

-Lasse

: +---+ =A0+---+

|LPF|-->PHASE 1

---+ =A0+---+

=A0 =A0

+---+ +---+

---+ =A0+---+

|LPF|-->PHASE 3

+---+ =A0+---+ k 2

oard,

jtag isn't really required, must of ARMs have a buildin bootloader so all that is needed is a serialport at ttl levels.

something like an AVR might be easier since it is available in DIP and can be programmed with nothing but a few wires and a parallelport

-Lasse

True enough, the ADUC7k has this too. ADI provide some windows software you can use to transfer over a serial link. I never used it since I wanted the low-level debugging support provided by JTAG. The simpler adapters are very inexpensive now and adequate for small chips like this.

--

John Devereux

ote:

P. The

re only

ion between

t was

onships,

with

ns

y big

.

lt-

r

=A0sine

igh

ts

o)

l s

All of which would be fine, if the OP hadn't said "Power is via 4 x "AA" cells" right at the start of the thread.

This is inconsistent with the amount of switching that you need to generate three sinewave-outputs - even at 10Hz - with PWM, since you end up with a device that is being clocked at something like 20MHz.

The DDS approach, using DACs to generate the steps on the stair-case approximation to the sine-wave, can get by with a much slower-running processor. Analog Devices parts do not illustrate this point because they are designed to be able to generate very high output frequencies, and their fastest parts can be remarkably power hungry, including the only one they seem to sell that can generate three separate synchronised sine wave outputs in a single package

Which is why I suggested a roll-your-own DSP solution built around a Xlinix Coolrunner or similar low powered chip - which you could clock at 32,768Hz - and a couple of serial input DACs to generate the output waveform.

You are obviously fixated on situations where current drain isn't an issue.

-- Bill Sloman, Nijmegen

nsist

hout

Not if the OP opened the thread with "Power is via 4 x "AA" cells."

Only a hardware-ignorant software peddler could have missed that.

-- Bill Sloman, Nijmegen

On a sunny day (Sun, 7 Nov 2010 11:32:37 -0800 (PST)) it happened Bill Sloman wrote in :

Sorry, but even a PIC 18F14K22 running on 16 MHz internal clock with a 4 x PLL to 64 MHz only consumes a few milli amps, and works over the whole voltage range of 4 rechargables. That is why I can run PICs from the serial port DTR line, like this for example:

the last one runs on a 20MHz xtal, in case you had any doubt.

This one runs 64 MHz intern, also consumes only a few mA, but the LCD and it's backlight need some more:

You really need to play with some mof=dern stuff to get the feel of it.

You can do a LOT with a 64 MHz clock, as scope_pic shows.

t

True. If you see him socially, you aren't looking at his comments as insights that will be preserved and generally available until Google goes bankrupt.

Right. Like I should have said, you don't often notice when he says something stupid. Infrequently enough that you don't damage his somewhat fragile ego.

-- Bill Sloman, Nijmegen

rote:

m

Mark

aration.

s. I am

way to

apart from

oltage,

e much

e the

through

all

log

rate four

hips.

.pdf

waves

ld Hill

find it

DDS

sis-dds...

own a

logic

60Hz

wired to

20 degree

or 3,4

nd 0).

nd three

each,

(64

aller,

not

l part of

bits

ction of

e

even PWM

driving

=A0 =A0 =A0 =A0Vcc

| =A0 =A0 =A0 =A0 =A0 =A0 |

-+-[DIODE>]-+ =A0

=A0 =A0 | =A0 =A0 =A0 =A0 =A0| =A0

+--[COIL]--+ =A0 =A0

=A0 =A0 =A0 =A0 =A0 =A0| =A0

DIODE>]-+ =A0

=A0| =A0 =A0 | =A0 =A0 =A0 =A0 =A0| =A0

=A0 =A0 +--[COIL]--+ =A0

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0

-----+-[DIODE>]-+ =A0

=A0 =A0| =A0 =A0 | =A0 =A0 =A0 =A0 =A0| =A0 =A0 =A0 =A0 =A0 =A0

=A0 =A0 +--[COIL]--+

=A0 |

|

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0

ist

ut

This from the guy who posts recipes .... Since you only hark back to electronics when you've been caught with your pants down, you'd be the hopelessly transparent twit.

-- Bill Sloman, Nijmegen

On a sunny day (Sun, 7 Nov 2010 11:43:07 -0800 (PST)) it happened Bill Sloman wrote in :

Says clueless pussy.

It would be easy to use a low-power CPU to make 3-phase PWM sinewaves [1] for well under a milliwatt power consumption. It could sleep about

99.8% of the time.

Hey, Bill, tell us about your experiences implementing low power embedded CPU designs.

This is my latest:

13 NXP ARM processors on one board, 12 of them in isolated i/o channels. The isolated ones don't have enough power supply to run full time, so we let them sleep most of the time and do most of the work in an ADC IRQ service routine, using heavily optimized C to not do too many truly stupid things. The drill is: Write C code. Examine assembly intermediate code. Figure out why it's so stunningly inefficient. Repeat until there a 4:1 or so improvement.

How's the new web site look? The Brat did it in raw HTML. We'd appreciate feedback, especially any comments on bugs with various browsers. IE versions were a nightmare.

John

[1] which is not to say that that's what the OP wants.

On a sunny day (Sun, 07 Nov 2010 11:56:52 -0800) it happened John Larkin wrote in :

What? It loads in a flash? What? It does not start my system swapping to disk for minutes? What? No java and java script? What? No silly movies running all over the screen? Congratulations, very efficient, very fast, and very clear.

Opera OK so far.

John Larkin Inscribed thus:

Looks nice in "Chrome, Firefox, Opera, Konqueror" a bit weird in "Lynx" but nobody uses that... :-) Quick & snappy. I like it better than the old one ! It seemed dated in comparison.

--
Best Regards:
                     Baron.

OP. The

are only

ition between

but was

tionships,

d with

eans

bly big

ts.

s r

ht

uilt-

tor

=A0sine

high

ents

ago)

eal

z,

te

n y

t's

Never used PICs but something like an atmega uses ~10mA at 20MHz

~5mA for each coil + 10mA mcu, times 40 hours =3D 1000mAh standard alkaline AA is ~1500mAh

-Lasse