120 Degree Phase Shift Osc.

I haven't seen those before -- or if I have, nowhere near often enough to have remembered them. Quite interesting...

Not a lot. :-)

Was that IC design or discrete circuit design? I'm starting to think that MOSFETs as "just like a BJT but with horizontal lines" might be an IC designer thing. Heck, they use all sorts of weird symbols... two overlapping circles for a current source? I wonder who came up with *that!*

---Joel

That's the way we draw mosfets. Very intuitive. The "standard" form is confusing and too busy.

John

A Democrat ?:-) ...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!

Makes me paranoid. Destroys lungs. Silly drug.

John

Sloman

fitler.

channel.

summing

none.

use R2R ladders.

Any idea how long each IRQ runs?

John

happenedBillSloman

:

The 100 MHz, 32-bit ARM, in sleep but keeping up the 100 MHz PLL, is a couple of mA. Deeper sleeps are much less, but take longer to come out of. The ~~1000 page datasheets have the exact numbers. Our referenced project keeps the clock running at 100 MHz, interrupts at 100 KHz, and runs an average IRQ around 3 us. Power consumption is, as expected, about 1/3 of what it would be at 100% up, namely about 38 mA instead of 90. Idd is pretty much linear on clock rate and on IRQ execution duty cycle.

No, sensible and on-topic is ambitious enough.

But Inland? Who dat?

John

Go for it. You have the time.

John

I think she used HTML, java of some sort, CSS, and something else. Mostly from the "XXX For Dummies" books.

Weird that she should do this. She majored in softball and beer pong, with a minor in psychology.

John

You probably taught her something about doing any technical job asked of her for the person handing her a paycheck... :-)

I think it looks pretty nice and it loads nice and fast. I find it a bit odd that at the menu along the top everything's in lower case (e.g., products -> delay/pulse generators) whereas on the big pictures they're titled in camel case (e.g., "Digital Delay & Pulse Generators"), but I'm going to assume that was a purposeful choice of artistic expression on her part.

How long did it take her to do all that? And did she have to do some Photoshop work as well, or did you already have those tight-cropped pictures of your products around?

---Joel

IBM standardized symbols. They liked boxes for most things, because a box could be drawn on a chain printer. ;-)

Both, though anything going into an automated system was just a square box, like:

+------+ C | B | | E | |

The first ones I saw weren't in ICs.

My father used two overlapping circles for a current source so I sorta doubt it's an IC thing. ;-)

Sloman

fitler.

summing

none.

R2R ladders.

When I did mine, I formed the SIN wave in the table and each entry was the number of times the out put repeated it self.. I used the Integer type to indicate if I was counting the off cycles or the on cycles. when the accumulator reached zero, I would then move to the next index in the table..

For every entry of the IRQ handler, I first make sure to finished the accumulator value, before fetching a new entry from the table.. This allowed me to keep the counter for the IRQ handler set to a constant value. I could have +128 and -127 frames*IRQ_Clock. I found it to be sufficient for generating a constant 60hz phase A,B and C with a very clean sine...

If that resolution isn't good enough, you can form larger size tables.. to use 16 bit for example..

Very easy to do with uC's

the Delta Sigma method works ok also. With that, you can use 8 bits per entry as a bit stream instead of doing a count down like I did.

P.S. WIth AVR's, the clock cycle for each instruction is constant and thus, you can actually perform a software timing loop that is dependable. This is ok as long as the processor isn't doing other random duties. No so easy with PIC's how ever.

I suppose. I didn't fight, or at least win, many artistic battles.

Probably two brat-months all together. I shot most of the accessory pics, and my production manager, a better photographer than me, shot most of the serious product pics. They used some photoshop and some Linux thing, gimp?, to tweak the pics.

John

s

I surveyed a few, gave you the result, and you didn't believe me. I did the same thing for the 83c751, wasted half an hour altogether, told you the result (as related to my older design), and you didn't believe that either. All to convince you of things you should've known in the first place, and could've easily looked up yourself.

This is all out of proportion. Mine was a modest suggestion, not at all worthy of such carping or microscopic quibbling as it's received, but AFAICT, it was easily more than enough that everyone else got it. That was my goal--a suggestion, not a design, not a seminar for the uninitiated.

I also mentioned a few variations, like using a uC with PWM hardware, or adding a few resistors to make a DAC, and so forth, for those who prefer those, or those for whom engineering time is more critical than parts cost.

Mine was no giant brainstorm nor original idea deserving a monument larger than, say, a Nobel Peace Prize, nor requiring an inquest rivaling the Inquisition, or the Neurnberg trials. It was a perfectly ordinary, obvious approach, quickly appreciated by all. I really just wanted to be the first in this thread to say...

"Use a PIC."

Mission accomplished.

-- Cheers, James Arthur

ill Sloman

:

I seem

This

it co=3D

)..

e

ly

etween

ips,

ig

imple fitler.

channel.

.. summing

, else none.

able use R2R ladders.

I love those AVRs. I've never used one, but I bought a sackful and a development system so that I can.

-- Cheers, James Arthur

ppenedBillSloman

:

an

ty, and doesn't i=3D

oint out -

o be. JA can

out-think you wit=3D

t when you say

who was delighted

he was.

. If you can't

s.design, you

M a 10Hz

" cells."

d that.

use "a

ng about

tion

w mA at

n't

one at

P's

n

You are the liar. My claim was that using a uC to generate a pulse- width-modulated approximation to a sine wave would draw too much power. Using a uC to drive three DACs to generate a stair-case approximation to a sine wave is obviously perfectly practical, if a little extravagant, since this is essentially the DDS solution, and - as I pointed out in the third post in this thread - you can realise the digital part of direct digital synthesis in a programmable logic part, such as a Xilinx Coolrunner chip and save yourself the extra transistors tied up in more complicated sturcture of a uC.

You have constructed a straw man and call me a liar when I point out that you have grossly distorted what I actually claim. You probably are too stupid to understand what you are doing - which would also explain why you can't grasp the distinction between constructing an approximation to a sine wave by pulse-width modulation (by switching an output between the positive and the negative rail) as opposed to constructing it by programming a DAC to output out a staircase approximation to the desired sine wave.

The former solution needs a clock running at something like 20MHz to control the on and off periods sufficiently accurately to get 10-bit precisison on a 50usec wide on-to-off-to-on cycle, while the latter can get by with a 20kHz clock (if realised in PLD with three parallel outputs feeding three parallel-input DACs), though a practical DDS- style solution would more likely ran at closer to 300kHz and drive serial DACs.

-- Bill Sloman, Nijmegen

happenedBillSloman

:

an

ote:

loman

ip

isn't

So scaled down to 20kHz interupts and a 20MHz clock that would be

7.6mA, except that Mark Weaver's application would call for four interrupts every 50usec - one to set all three pulsewidth modulated outputs high, and three separate interrupts to set each one low again at the right point in the 50usec to generate the right average output voltage.

This is comparable with the roughly 11mA that his three pairs of relay coils would draw (on average). The 32-bit ARM is almost certainly a massive over-kill for the application, but if it can get that close, a more appropriate choice of procressor should do quite well enough to be practical.

tick to

An ambition that you too would do well to adopt. Your off-topic recipes seem to be okay, but what you post about evolution and global warming falls a long way short of sensible.

Sorry, Highland Electronics ...

-- Bill Sloman, Nijmegen

If Mark Weaver asked me to, I probably would.

-- Bill Sloman, Nijmegen

ons

gh

Since you didn't address the power consumption question at all, I didn't see any reason to take you seriously.

Actually you quoted a current drain for the "80c51 family" without specifying a manufacturer. You number didn't tie up with the NXP data sheet for that family, so tell us the manufacturer, or explain why you were quoting from an out-of-date data sheet.

But I had looked it up for myself in

and you number didn't match what I could find in figure 21 of that document, which strikes me as adequate grounds for scepticism.

Not quite as unspecific as "use a micro" but not specific enough to be of much practical use, given the spectrum of capabilities now covered by Microchip Technology's offerings.

-- Bill Sloman, Nijmegen

t a

es

I'm a Popperian - all scientific facts are falsifiable hypotheses, like Newton's hypothesis about gravity (now falsified and replaced by Einstein's hypotheis about general relatvity).

The hypothesis that human beings are burning enough fossil carbon to signficantly raise the CO2 level in the atmosphere is about as well- founded as a scientific hypothesis can get. The hypothesis that this is raising the surface temperature of the earth is equally well- founded, but less quantitatively precise. The global warming through the 20th century was 0.74 =B1 0.18=B0C (1.33 =B1 0.32=B0F),but the estimate= s of how much it will rise over this century range from 1.1 to 6.4=B0C (2.0 to 11.5=B0F) though most of them come out around 4=B0C.

The difficulty is that we don't know how much fossil carbon we are going to burn in the remaining ninety years of this century, and we don't know enough about the positive feedbacks involved to be entirely happy that we've got them right.

Arctic ice cover is a particularly worrying point. The ice pack that now covers most of the Arctic Ocean in summer (when it matters) has a particularly high albedo, and absorbs relatively little light (and heat) from the sun. The summer ice pack is now shrinking rather more rapidly than expected, and the ocean surface that takes its place absorbs most of the light and heat that hits it.

-- Bill Sloman, Nijmegen