The TPU uncovered? (A PROBABLE technique.)

[pauldude000]

MAN OH MAN!

First, I thunk I have had a mental breakthrough, and so I shall share the concept. IT IS BASED UPON MR MARKS OWN WORDS... IE the clues were there all along.

1. Sine or Square waves?

Either pure sine waves, OR square wave DC will create a rotating magnetic field. Sine waves created using DC input voltage are rather difficult to construct, and require complex circuitry to attain. Square wave DC though..... nothing could be easier OR more probable. 

A. Simple circuitry (small boards low parts count).
B. Easily battery powered (1 9v works great.) 
C. Simple frequency selection/waveform modification.

2. How to create a rotating magnetic field.

Tesla invented this, which he demonstrated with his "Spinning Egg", and is now used in every three phase system motor/generator on the planet.

2 or more coils, and for our purposes three, spaced properly and fed polyphase signals of the proper degree of offset create a rotating magnetic field. Three coils, require a spacing of 120 degrees offset (points of an equilateral triangle -two 30-60-90 triangles back to back, superimposed on the coil.)

I was mislead by my own thinking. the frequencies ARE a function of the center frequency, However, I was thinking linearly, and not properly.   It is NOT 0deg 90deg and 180deg; but instead is 90Deg -  0 Deg  90 Deg +

The three coils are fed --->

first coil, 90 degrees - (90 degrees preceding original waveform)
second coil 0 degrees (original unmodified signal)
third coil 90 degrees + (90 degrees leading original waveform.)

I also held the preconceived notion at one time that the waveforms represented a SINGLE frequency, modified to get three phases of the same frequency. However, this is not necessarily true. I have built enough stereo speakers so I should have realized what was going on. Phase shifts can ALSO be gained by using three separate SPECIFIC frequencies, all denoted from the center frequency. For instance, Steve gave all the information in the world with "5khz with some hash". Using square wave DC, and the three frequency patterns mentioned, you would have that exact definition, also the output would be "5khz but with a DC component". Both would be EXACTLY true, but in themselves misleading. I never once remember him saying AC.....

Using 5khz, which applies to a (the) 17" TPU (smaller coil higher frequency), we find that the first coil is fed with half the center frequency, or 2500HZ, the second with 5000HZ, and surprisingly the third with 7500HZ.

Why not 10Khz? 10Khz is 180Deg out, NOT 90deg.

ALL THREE FREQUENCIES ARE FED INTO A SINGLE WIRE SIMULTANEOUSLY. You will learn why below.

3. How to feed the necessary frequencies to the appropriate coils, and do so using feedback from the output.

This hit me like a thunderbolt. Since the output is pulsed DC, at three frequencies, all that is required to "tap" the output is a blocking diode, with a resistor or thermistor to limit current to the control coils. This IS FED TO A THREE WAY CROSSOVER, which splits one signal into three, and routes said signals to the appropriate control coil inputs!!!!!!!!!!

I THOUGHT THAT STUPID "CONTROL" IN THE CENTER OF THE 17" TPU LOOKED FAMILIAR!

The oscillator circuits feed three mosfets, each connected by matched resistors to positive, fed through diodes into a single wire into the crossover. ALL TPU CIRCUIT GROUNDS ARE SHARED.
 
4. How to calculate said frequencies.

The frequencies are HARMONICALLY PHASE SHIFTED and therefore are algorithmic, and NOT linear, to achieve 90 degree separation. The formulas for each coil are based upon a center frequency A.

coil 1 = 1/2 A --> (as expected 1/2 harmonic of A. This is A-90 Degrees.)
coil 2 = A --> (as expected, since is center frequency.)
coil 3 = A + (1/2 A) --> (A + 90 Degrees, instead of 2A as expected.)

All control coils to be wound bifiliar for maximum efficiency and coil energy. (read Tesla's explanation)


I hope that this post floors you, as much as the realizations floored me.

I may try to do some accompanying graphics later.

Paul

[eldarion]

Phase shifts can ALSO be gained by using three separate SPECIFIC frequencies, all denoted from the center frequency.

Hi Paul,

This is one of the best posts I have seen around here in a very long time!

Can you explain the quoted section a bit further?  I am an EE, not an audio person, and have never heard of that type of phase shifting before.  Are you talking about a phase shift occurring in the crossover itself?

Great find on the 3-way crossover!  Look, two inductors: http://hyperphysics.phy-astr.gsu.edu/hbase/audio/cross.html  They even look similar to parts of the "control" circuit.  "No mass circuitry"...

Maybe Steven just added a simple 9-volt powered blocking oscillator to the crossover to get the whole thing started...

Eldarion

[pauldude000]

Maybe my useage of "phase shift" in reference to the square wave DC frequencies was not quite an accurate phrase to use here, as no actual shifting is done to the square wave center frequency. (that would yield 1 frequency with a leading and trailing overlap, as viewed on a scope.)

I am referring to two different frequencies coupled together with a center main frequency. To the magnetic field they would APPEAR to be phase shifted, due to the degree of offset, and the fact that we are talking square wave DC.   

This type of phasing is necessary when designing crossovers, to make sure that A. the frequencies overlap, and B. that the signals all reach the drivers in phase with each other, to prevent distortion due to signal time lapse.

I never thought it might have application elsewhere, so I never considered it before. However, Steven was big into audio...

Paul

[sparks]

      The 3pass filter has to be fed from 3 oscillators to initiate the phasing then after that the output which will have all the original freqs in it is fed to the filter?

[Dansway]

 

@pauldude000

Excellent.

This is indeed very good meat for the table.

~D