120 Degree Phase Shift Osc.

On a sunny day (Sun, 7 Nov 2010 15:56:35 -0800 (PST)) it happened Bill Sloman wrote in

OK, first about 'approximation' even at 8 bit, ever looked with the scope at an 8 bit sine wave? Looks pretty good actually.

Second 'a few mA' I mentioned, is if you do things in a software loop, and even then it is probably less. Maybe using an internal timer and wakeup 256 x 10 = 2560 x per second, could further reduce current consumption. in the wakeup you then put 3 8 bit output words on ports, and update 3 pointers in a sine table, not a lot of work.

3 R2R ladders, now we are probably talking about micro Amperes.

In the PWM case, there are PICs with 3 hardware PWM generators too. My suggestion to you is, before you really show more ignorance, to get some microcontroller, some programmer, buy a soldering iron too, and build some projects, to get a feel. Or you could at least bother to read some PIC datasheets, Microchip is very proud on their low power consumption features, and lists the current of each end every hardware part of their chips. You can get down to a few uA using sleep mode, as my lightning detector shows:

runs of 3V, years on some penlights, probably battery self discharge happens before the PIC empties it :-)

You mention 'orders of magnitude' current consumption difference for your DDS, so how much does you DDS use?

On a sunny day (Sun, 7 Nov 2010 16:15:10 -0800 (PST)) it happened Bill Sloman wrote in :

I think modern 8052 versions come in CMOS with low current consumption.

l Sloman

e

or example:

it.

A 32768Hz watch oscillator isn't going to draw much current, nor a Xilinx Coolrunner clocked at 32,768Hz. There are a whole bunch of low- power memory chips around

looks promising - with a 90uA standby current. You'd need to add a bit of circuitry to the cpld to get a short read pulse - a 30.5usec wide read pulse would be extravagantly long.

The Maxim MAX5535 dual 12-bit DAC

might look at bit slow for this application, with a 660usec settling time to 0.5 lsb for a full scale step, when one would like to up-date every 30.5usec, but the DAC outputs would be changing by 0.1% at most, so you might get away with it.

is almost as low-powered, and does settle respectably fast.

Gven a free choice, I wouldn't use a Maxim DAC (or any other Maxim part) but Farnell doesn't seem to stock anybody else's.

-- Bill Sloman, Nijmegen

lSloman

oesn't i=3D

-

can

you wit=3D

u say

elighted

can't

you

Perhaps, except that James Arthur was planning on having the chip microcontroller do everything, including the fine timing, which isn't compatible with "idling some of the time".

10MHz might be good enough. It depends on the size of the step you can tolerate in your staircase approximation to the sine wave.

ge.

how

But if you were buying a microprocessor as intellectual property in a programmable device, why on earth would you pick the 8051 architecture? And a DDS system doesn't need a whole microprcessor to generate the digital approximations to a sine wave, so why pay license fees for something you don't actually need? =A0

ven

And - as usual - krw doesn't back up his claim with any kind of example. He knows that he will be shown up as a posturing windbag if he tries.

krw's own incredible ignorance has led him to imagine putting a 8051 structure into a programmable chip, when anybody who had his head on straight would have noticed that you don't need anything as complicated as a full-fledged micro-controller to do the job that Mark Weaver wanted done.

The 8051 is of particular interest to people who have worked on electron beam testers. When Intel licensed the original 8051 to Siemens, back in the 1970's, Siemens used their lashed-up electron beam tester to see what was actually going on inside the chip and found the one under-sized transistor that was bottle-necking the Intel layout. When Siemens adjusted the layout to get rid of the bottle- neck, they could specify their 8051 at three times the clock frequency that the Intel part could support. This got lots of people interested in elecron beam testing ....

-- Bill Sloman, Nijmegen

an't

you

They've clarified tha a uC can draw far less power than the AD9959 which I identified as the "obvious" DDS chip for someone who wanted three outputs phase-shifted by 120 degrees. I also pointed out that it drew way too much power for the application and went on to recommend combining a couple of lower-power Analog Devices DDS chip, or rolling your own DDS system with a Xilinix Coolrunner for the digital part and three serial DACs to generate the three sinusoids, which still looks like the lowest power solution.

But not for something that was producing a continuous 10Hz sine wave.

Since you've just demonstrated that you didn't actually read my original post in this thread, whence your - false - claim about the power hungry DDS chip - and you've now chosen to pretend to expertise that you clearly haven't got by adducing your experience with a burst- mode single-side-band transmitter as a example of low-power design, your example of "understanding what you are talking about" doesn't inspire confidence.

Since exactly this effect has discouraged me from going for PWM to synthesise sine waves in the past, your ignorance is informative.

I didn't think that using a uC to generate a PWM approximation to a sine wave qualified as a low power solution. Modern microcontrollers draw less current when running at 20MHz than the parts that I was looking at five or more years ago, and - granting that Mark Weaver is going to be using 11mA to drive his coils - PWM does now seem to be a viable, if slightly current-extravagant solution. Using a uC to generate a 10Hz sine wave with a DAC is obviously perfectly feasible, though you don't need a uC to do the job - something like the Xilinx Coolrunner can handle the digital part of the job perfectly adequately, and probably rather more economically, since you are only switching the transistors that you actually need, rather than feeding a whole microcontroller.

Clearly, you don't have colleagues who are sensitive to detail, and your own grasp of detail has its weaknesses - I didn't recommend the AD9959, but rather pointed out that despite it's ostensible suitablity for the job it drew too much current to be a contender, and went on to recommend less current-extravagant solutions.

John Devereux ARM processor, with the built-in DACs, could do the DDS job without any difficulty. It's not going to be the cheapest possible solution, nor the lowest power solution, but the part - programmed to do DDS - would not draw anything like the current being fed into Mark Weaver's three coils, so it is sufficiently low-power for all practical purposes, and programming the ARM chip would be a lot easier

- and thus cheaper - than programming a pld to do the same job, so Mark Weaver would have to be contemplating high volume production before the extra parts cost would disqualify the ARM processor.

-- Bill Sloman, Nijmegen

sn't insist

n

ou without

say

ighted

n't

ou

Don't be silly.

And how to they get that? By synchronising to a faster clock, which is to say you are still feeding a fast counter even when the bulk of the microcontroller is asleep.

Which is to say that you get to define "low power" as less than your isolated power suplies could deliver. Which does make life a little easier than it might be - if you'd run into a problem you'd have bought a bigger power supply.

Improving the executuion speed of compiled code by examining the lower level code generated. When I did it, we looked at the assembler code.

I did Theory of Computation 1 as an extra undergraduate course while I was a graduate student, and sat in on the lectures on assembler programming from the Theory of Computation 2 course, which was taught from the pont of view that you programmed in a higher level language (mostly Fortran back then) but read the critical bits of the compiled code with a view to improving the execution speed. This was when I was writing assembler for a PDP-8, which looks - in retrospect - very like a microprocessor.

I can't say that I enjoyed sitting through lectures - you got to learn about a lot of stuff that you would never need, but an undergraduate's idea of what one will need to know is rather restricted. What you learn is also pretty restricted, but universities are also good at teaching where to find the information that you didn't get in your lectures. Being a graduate student exposed me to a lot of real problems. I've not used my glass-blowing and optical polishing skills since then, but what I learned about electronics has proved more useful.

-- Bill Sloman, Nijmegen

Bad guess. I've worn out several copies of Mastering the Art of French Cooking.

Really? I've got to admire the - limited - skills you have before you can notice mine?

And some of the peer-reviewed off-topic stuff that I trust is actually within my personal expertise - as a physical chemist with a Ph.D. You distrust it without having any relevant expertise, because you are a sucker for denialist propaganda delivered to you in the right-wing media that you seem to devour uncritically.

Really? Bad as it may be, it still does seem to get under your skin.

My wife would only comment if I'd made less-than-really-good pasta. I've spoiled her.

-- Bill Sloman, Nijmegen

Looks fine with Firefox 3.6.12 ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
 Ding Dong! That Wicked Old Witch Pelosi is Dead! Hurrah! Hurrah!

Another bug in Outhouse? Who'd have thunk it?

[...]

If low power is the overriding concern, we should probably go back to a purely analog solution, or analog+switched capacitor filter perhaps.

A state variable filter will give quadrature outputs which can then be mixed to give the three phases (using the formulae presented earlier in the thread).

This could be implemented with micropower parts, the current draw would be neglible compared to the relay coils.

--

John Devereux

And cute. You forgot to mention cute.

He enjoys it. Weird, huh? The idea that education, scientific advancement, and social progress will improve human relations and bring peace to the world is obviously wrong. Sloman is over-educated and lives on the dole in a socialist utopia, and he's mean and vain and sour anyhow. Human nature is pretty much wired in.

John

But tree-skiing is such fun.

John

Just took another look. James' sig is malformed, it does not have a space after the --

Although Agent is smart enough to snip it.

- t ,

I wonder who that "we" refers to?

But you need close tolerance capacitors to get equal phase shifts around each leg of the phase shift oscillator, and 10Hz tends to call for big capacitors, where anything better than +/-10% is hard to find.

You may want to think it out again.

-- Bill Sloman, Nijmegen

Same version here - 3.6.12 and it works fine today. Yesterday, it didn't. Wierd. Views 1, 2 & 3 were fine, view 4 showed only about

10% of the image height, at the top. Must be my problem, as it works fine for others, and works correctly for me today.

Thanks, Ed

Or maybe the "brat" fixed it upon first complaint ?:-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
 Ding Dong! That Wicked Old Witch Pelosi is Dead! Hurrah! Hurrah!

an

But not good enough for Analog Devices, whose DDS chips don't go below

10-bit DACs and often go to 14-bits. It's going to depend on what Mark Weaver can live with. More bits in the DAC make the sine wave easier to filter - not only are the steps smaller, but they are also closer together, shrinking the time constant of your filter as well as the attenuation it has to provide.

Ditto with Maxim low power DACs. Building your own DAC is a bit silly.

Did that years ago.

ry proud on

very hardware part of their chips.

I do, from time to time, whenever I get interested in an application where I might use their chips. Since the technology keeps moving on, there's not a lot of point in doing it when you don't have a specific application in mind.

hows:

ens before the PIC

But it isn't keeping track of a 10MHz clock. The current consumption of CMOS parts is usually dominated by the current drawn charging up the stray capacitances on it logic connections, then discharging them again when the logic state changes. Putting most of a microprocessor to sleep while only a timing counter runs does save power, but the timing counter is going to draw more than microamps.

DDS, so how much does you DDS use?

The difference between a DDS generating its sine waves by PWM and one that rewrites a DAC is between a counter running at 10MHz and a processor running at 32,768Hz - maybe two orders of magnitude, grantng that there is more going on at 32,768Hz than there is at 10MHz.

-- Bill Sloman, Nijmegen

s

could

That'll work. He'll want the attenuate the cos(wt) term a bit to match the other two amplitudes.

George H.

nsist

hout

I don't attack him, I attack the silly idea that he posts from time to time. Amongst other things, he goes to the trouble of retailing every last lame denialist slogan that he runs into.

His claim that our current atmosphere is short of CO2 because levels were higher in the geological past - over 1000ppm through the Mesozoic

- is pretty typical, as is his claim that this is cruel to plants. He fails to understand that we - and our favourite plants - have spent a few million years adapting to the current CO2 levels and that we wouldn't be well-adapted to the conditions that prevailed in the Mesozoic, but then he doesn't know much about evolution and adaption either.

-- Bill Sloman, Nijmegen