Prop Speed Amplifier - Reveal Your Thoughts

Humanoido

In many areas of science, we build amplifiers to increase something.
In Astronomy, we build larger telescope mirrors to gather more light and amplify the intensity.
In radio and sound, we use amplifying circuits to increase sound volume.
In race cars and mechanics, we amplify horsepower to gain power and speed.
In Physics, we stretch the rubber band more to amplify potential energy.
In heating and cooling, we amplify BTU for more effect.

On the LM386 amp chip, it can be wired for increasing power gains.
What about our propeller chip?

We can put two lenses together to increase the power of one.
Can we use one propeller chip to increase the speed of the other?

If one Propeller is 100 Mhz, can we build a 10x or 100x circuit?
How about an innovative prop chip amplifier?

Anyone dare to think about this and speculate about a working solution?
Use this thread to share your thoughts.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
humanoido
*Stamp SEED Supercomputer *Basic Stamp Supercomputer *TriCore Stamp Supercomputer
*Minuscule Stamp Supercomputer *Three Dimensional Computer *Penguin with 12 Brains
*Penguin Tech *StampOne News! *Penguin Robot Society

Post Edited (humanoido) : 7/14/2009 7:21:39 AM GMT

Clock Loop

humanoido said...

If one Propeller is 100 Mhz, can we build a 10x or 100x circuit?
How about an innovative prop chip amplifier?

Anyone dare to think about this and speculate about a working solution?
Use this thread to share your thoughts.

Your talking about a pll (phase lock loop) circuit. (which takes a clock, and makes steps inbetween the main clock "artificially"

When it comes to the prop...

I had this idea, and its in a thread I created which has to do with synchronizing multiple prop chips.

Basically you take a single prop, use it has a master clock which controls all other props.

The propeller is capable of 128mhz clock output, but CANNOT measure something that is this fast.

It can if you use that master prop, to OVERLAP a bunch of props.

Basically they all work in tandem, one after another to measure signals that are much higher than the 1/2 nyquist theorem.... limit of a single prop.
(80mhz or 100mhz divided by 2 = a max measurement speed of 40 to 50 mhz.)

With many props in tandem being controlled by a master prop, the maximum measurement speed that you can have IS the maximum OUTPUT clock speed of the prop, and for that matter any chip that can do this same technique.
(just generate two 80mhz signals, which are offset by 1 internal clock of 128mhz)
(so when one 80mhz signal is rising/falling, the other is mid clock)



So im saying that multiple props can measure a signal of 128mhz because each prop is being clocked to measure the signal in question at different times.

If a single prop is limited to 1/2 its clock (lets say 1/2 of 80mhz, = 40mhz)

So if you wanted to make a scope with a single prop, best case would be getting 40mhz max clock measurement.

But if you take 3 props, have one as a master clock(which also aggregates all data from other props), and put a clock out to each prop, but offset that clock by the smallest amount possible (1 clock cycle, at 128mhz) well...

One prop can be looking for a transition from low to high or high to low, while the other is processing the initial low to high or high to low trigger.

With enough props, I would imagine someone could get up to 128mhz of measurement resolution using chips that can only measure 40mhz a piece.

The reason this cannot be done with a single prop and a few cogs, is because this method requires each prop be clocked slightly different.

(one props clock is offset from the other props clock by only 1 clock at 128mhz)

If someone can stagger the clocks of each prop using a clock generator of much higher capability
(say 1ghz) (1,000 mhz / 40mhz (max measurement of a single prop)

it would take 25 props with clock inputs staggered 40mhz apart to measure a signal of 1ghz with complete accuracy.


Something like this?

Everyone should try this. Perhaps im missing something major when it comes to this idea, but i suspect I AM onto something.

Its a pretty big claim to say that one can measure a 1ghz signal using 25 props.

I say it can be done.


And along the same lines, with 25 props, you could get a signal output of 3.2ghz (but you still need a source clock chip capable of outputting 25 different clock lines that are staggered 1 clock in step of eachother.

I feel that i may be missing something very important on why this isn't possible, because one would think that this would have been done/suggested already. So I hope others chime in on this so I don't do wasting my time trying to build something that does this. (i don't have a scope, and wanted to build one, so i thought I could do it this way)

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
TERMS OF USE: MIT License

"Permission is hereby granted, free of charge, to any pers...........................
..............................OMITTED FOR FORUM...............................................
.................. OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. "

The dsp/fpga king is dead, long live the prop.

Post Edited (Clock Loop) : 7/13/2009 1:21:45 PM GMT

Beau Schwabe

Clock Loop,

Instead of trying to sync other Propellers off of a staggered 128MHz clock, you might gain more resolution by creating an external "ring oscillator" and driving the other Propellers at various ring taps each signifying a specific amount of clock delay with relation to the next tap. This way you can resolve values in-between the clocks from one Propeller to the next.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Peter Jakacki

@Clock

There is this term called "pipe dreams" but actually many people come up with ideas while they are on the loo which is more like an s-bend pipe dream. Have you been taking your laptop into the loo again?

Ok, friendly humor aside you haven't addressed the many many hardware and software issues that are part of the "real world" [noparse]:)[/noparse]

Fast clocks skew badly enough across a tiny piece of silicon so it takes very careful design to make a common clock to serve an array of chips spread out on a board. But then fully synch'd Props aren't much use to anyone for high speed signal processing as each Prop will only see what the other sees. Now you have to take the clock and make it skewed to the right degree as I think you mentioned and as Beau alluded to. But none of these schemes will somehow magically enable a 500ps (0.0005us) pulse to be detected let alone processed. I haven't even begun to consider the issues let alone how you might develop software that somehow magically knows what to do with each chip and then co-ordinate the operation.

Time to flush this one but that doesn't mean you shouldn't keep on "thinking" though. It always helps though to test out and examine various aspects of our ideas and understand some of the issues before we say it can be done and then also see how we can put it into practice.

By the same token thanks for taking the time to express your ideas in the spirit of this forum (but can you include some code next time for us leechers and lurkers and general gadabouts).

*Peter*

P.S. BTW, you can see from my logo that these are my initials "p b j" but rotate the image 90 degrees clockwise and you can see I'm wearing glasses and yes, I have a wry grin!

Post Edited (Peter Jakacki) : 7/13/2009 11:04:23 PM GMT

localroger

@Clock Loop -- You could do that much easier and cheaper with a shift register or counter capable of operating at the higher frequency.

To do the clock trick, the simplest (and most workable) method is to put an inverter on the oscillator and drive one prop from one side of the inverter and the other from the other. Beware -- the inverter must be VERY fast, or its latency will seriously mess up your scheme.

You could probably in hobby land extend this to 4x by using a 160 MHz clock and flip-flop to divide it, with inverters and gates to give you four phases. A 74xx series J-K flipflop capable of working at the prop's voltage and working at that speed would also do the trick. I doubt you will find such a thing in through-hole mounting though.

As for higher than that -- well, I have better hopes for my Gold Amplifier. See, I have a few Krugerrands and I've figured out that if I can make each of them act like several, I can turn them in at the exchange for a tidy profit and use the profit to buy more Krugerrands, amplify them, and after a few cycles I can just buy my own massively multiparallel supercomputing complex.

Clock Loop

Beau Schwabe (Parallax) said...
Clock Loop,

Instead of trying to sync other Propellers off of a staggered 128MHz clock, you might gain more resolution by creating an external "ring oscillator" and driving the other Propellers at various ring taps each signifying a specific amount of clock delay with relation to the next tap. This way you can resolve values in-between the clocks from one Propeller to the next.

I guess I didn't specify it very well, and code might have helped..., because thats what I was trying to do with the "MASTER" prop... use it AS a ring oscillator... but you mention "more resolution" so perhaps your saying the prop is not good for this purpose.

Perhaps this image attachment will help a bit,..

And thanks for your replys! I have wanted to try this and will someday, with all its problems, but thats why we are all here, to experiment and succeed!

Beau Schwabe

The "Ring oscillator" method is often used internally to a Chip where pipe-lining is done to do exactly what you are describing. An example was a 10/100/1000 Ethernet Chip set that I worked on that used this method where the ring oscillator was capable of operating in the 3 to 4 GHz range. There is no reason though that the same or similar technique can not be done externally although a bit slower... descrete's would be faster here than IC's in this case. Of course there are a few signal lines that are shared between each stage to help keep things in sync.

Because of the sensitive nature though this project should be design or application specific, perhaps an outlined goal of what exactly it is that you need to accomplish should be put together first.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Clock Loop

Beau Schwabe (Parallax) said...
Because of the sensitive nature though this project should be design or application specific, perhaps an outlined goal of what exactly it is that you need to accomplish should be put together first.

Sensitive? Like as in, clock limits due to high speed? Or more like FCC issues due to noise interference?

My main goal behind this was two fold, one, i have always wanted to make a scope. And 2, I want to process(input) digital audio(spdif stereo) using props.

Yea, could be done using other chips, but i like the challenge doing it with hardware that can't do it alone.

Beau Schwabe

Sensitive in terms of relatively high speed delay lines and syncing multiple processors together.

I'm not familiar with spdif, but it doesn't seem terribly fast... 48kHz/192 DataBits/Stereo = about 18.4 Meg ... just a guess . A single Propeller and 4 or so synced cogs could probably do it, maybe two Propellers.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Clock Loop

Beau Schwabe (Parallax) said...
Sensitive in terms of relatively high speed delay lines and syncing multiple processors together.

I'm not familiar with spdif, but it doesn't seem terribly fast... 48kHz/192 DataBits/Stereo = about 18.4 Meg ... just a guess . A single Propeller and 4 or so synced cogs could probably do it, maybe two Propellers.

Your getting my hopes up....and presenting me with yet another challenge... after I post my schematics and code for my FX/sequencer musical prop device, i will be lookin into your challenge...

Thanks for the info, and sorry if I hijacked the thread...

Humanoido

I wonder if this concept can be utilized in any way to amplify one propeller chip using external circuitry? My thinking was to develop a prop system that is not additive, (previous Basic Stamp supercomputer projects develop an increase in power by adding processors), but rather use a multiplicative approach. I have never seen any supercomputers using this idea, yet we have ways to multiply the number of ports by using extender chips and we can increase the standard clock to a higher level by upping the crystal and modding the software. The multiplicative approach could be used on a next generation supercomputer. The idea is to build a very simple circuit to demo the concept. But how to get started on this? Can we agree the focus of this is strictly about clock pulse timing or are there other methods to consider?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
humanoido
*Stamp SEED Supercomputer *Basic Stamp Supercomputer *TriCore Stamp Supercomputer
*Minuscule Stamp Supercomputer *Three Dimensional Computer *Penguin with 12 Brains
*Penguin Tech *StampOne News! *Penguin Robot Society

Phil Pilgrim (PhiPi)

humanoido,

What you're after cannot be done without reliance on the supernatural. I'm afraid the constraints imposed by the laws of physics consign us to a humdrum reality wherein we cannot just wave our hands, wish something to be true, and — voila! — it happens. Magic Propeller amplifiers? 'Fraid not!

Okay, now that I've got the morning snarkies out of my system, I guess you could consider the Prop II to be an "amplifier" for the Prop I, insofar as it expands on the Prop I's design and increases its speed. But this is done within the constraints imposed by solid state engineering, by using smaller silicon features, along with a necessary decrease in operating voltage — not from something external. The only other option is to overclock a cooled Prop I. Overclocking has been tried with very limited success, but I'm not sure that cooling has been part of the formula in any of those cases.

-Phil

waltc

I've always believed in using the right hardware for a job since I don't like to be tortured by my hobby but I've also have a soft spot for ideals born in the outhouse like this one.

Go for it!

Humanoido

Phil Pilgrim said...
I guess you could consider the Prop II to be an "amplifier" for the Prop I...

Prop II? Where is it available?

waltc said...
...I've also have a soft spot for ideals ...

The idea is some thinking out of the box.
The objective is to find as many ways possible to reach the solution.
To give another example of thinking out of the box to increase power,
when working with super cooled CCD camera imagers, to multiply
performance, the noise was reduced by cooling to -107 deg. F.
However, thinking out of the box, a way was invented to achieve
the effect of super cooling using only electronics with a
circuit external to the camera. It was the first virtual cooled CCD
cameras. Before virtual cooling was invented, as with many such
inventions, people either never thought of the idea or believed it was
impossible.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
humanoido
*Stamp SEED Supercomputer *Basic Stamp Supercomputer *TriCore Stamp Supercomputer
*Minuscule Stamp Supercomputer *Three Dimensional Computer *Penguin with 12 Brains
*Penguin Tech *StampOne News! *Penguin Robot Society

Leon

What has that got to do with your concept of "amplifying" chip performance; which isn't possible, BTW.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Post Edited (Leon) : 7/16/2009 10:41:45 PM GMT

Phil Pilgrim (PhiPi)

humanoido said...
Prop II? Where is it available?

"Where?", is easy: Parallax. "When?", OTOH, is a question to which the most authoritative answer to date is, "When it's done."

BTW, can you provide a link to a tutorial on "virtual cooling"?

-Phil

heater

I think, possibly maybe, that Intel has been using their own products for a decade or more to help design the next generation of their product. Sounds like an amplifier to me.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.

Peter Jakacki

H, we ARE outside the box (and it still looks like a box), and we are waving back at you while you are at the window SAYING "think out of the box" [noparse]:)[/noparse]

This is the Propeller forum and it is an amazingly active forum that is constantly coming up with new ways to do things that would not seem possible, so in a way it is doing the impossible. Bear in mind that none of these who have come up with innovative ways of doing things were prompted to do so by others wishful thinking or simply wished it were possible. Doing, not dreaming makes it a reality. If it's your dream then you show us how it's done and we will be very very very amazed.

*Peter*

Humanoido

Phil Pilgrim said...
can you provide a link to a tutorial on "virtual cooling"?

The full article about VR Cooling is not available online. It appeared some years ago
in the ink printed issue of OTM.

As I remember, I took a non-cooled CCD imager and modified the circuit board.
Then I attached a potentiometer that set the level of the VR cooling externally and
connected it to the observatory telescope. It was possible to dial in the rates of cooling
which was calibrated on a dial scale.

The article included instructions and details to accomplish this and had the
necessary schematics and photos. The results were spectacular. I remember, after
performing the modification to the camera, at the telescope things just popped into view
such as the fainter moons around the planet Saturn, lots of very faint stars in real time,
and many other phenomena.

I used this modified camera with VR cooling for many years and it was part of the
observatory research programs which achieved many astronomical discoveries.

While we are on this subject, way back when they used only glass filters, I published
an article on creating a computer driven filter to achieve similar results using only
software and no glass. It was another "thinking out of the box" type invention that
no one had thought about at that time.

So again, I ask everyone to ponder and speculate on the possibility of amplifying
the power of Propeller chips using both conventional techniques, and some new
techniques not previously thought about.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
humanoido
*Stamp SEED Supercomputer *Basic Stamp Supercomputer *TriCore Stamp Supercomputer
*Minuscule Stamp Supercomputer *Three Dimensional Computer *Penguin with 12 Brains
*Penguin Tech *StampOne News! *Penguin Robot Society

Peter Jakacki

H, we don't need the article because you have actually done this "VR" cooling yourself so then you must be thoroughly familiar with the details, so tell us.

I know that Peltier effect thermoelectric coolers can be used with CCDs but that would not be virtual cooling. Perhaps it was in "VR" land that you cooled this CCD.

*Peter*

Humanoido

Peter:

Oops! I should say "Virtual" cooling and not "Virtual Reality" cooling! The cooling was not in VR land, (a very good point to make) since it duplicated the real effects of cooling by adding to the camera board. I started using Peltier devices after that camera. Well, I do not remember all those details - it was a long time ago in a space far far away. Better yet, just access the original article at the public library. Ok, whoever gets there first can provide the details. This may be one way to amplify the power of the Propeller, since cooling, virtual or otherwise, has the effect of making processor speed increases possible.

So far, we have the following:

* Cooling (virtual, peltier, di)
* Overclocking: Crystal increase, software change
* Chip evolution, upgrade
* Change operating voltage
* Some kind of external clock interlacing

I think this list should be ten times longer. This is almost the year 2010. We have great technology at our disposal. We've landed a man on the Moon. We are doing our own Near Space Programs and can build our own satellites. This Forum includes some of the greatest minds from across the world....

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
humanoido
*Stamp SEED Supercomputer *Basic Stamp Supercomputer *TriCore Stamp Supercomputer
*Minuscule Stamp Supercomputer *Three Dimensional Computer *Penguin with 12 Brains
*Penguin Tech *StampOne News! *Penguin Robot Society