Driver to drive?

On Mon, 4 Aug 2014 05:58:02 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Saturday, August 2, 2014 12:32:27 AM UTC-4, Bill Sloman wrote:
On Saturday, 2 August 2014 12:12:27 UTC+10, dagmarg...@yahoo.com wrote:
On Thursday, July 31, 2014 2:53:08 PM UTC-4, Robert Baer wrote:

Yes, the Golden Rule: "He who has the Gold, Rules".

He who rules, gets the gold. (Just look at today's DC.)

It was articulated by some of the more plain-spoken of the founding tax evaders as "the people who own the country should run the country".

And here we have Bill Sloman, America-hating, throwing Molotov cocktails at innocents and honored dead, on invented charges and imaginary grounds. It's useless rebutting; none it ever happened, Bill has no basis for it, has no source--made it all up himself--yet fanatically believes it.

History suggests that they shouldn't - they don't spend enough on keeping the working classes fed, healthy and educated to maximise their own profit, let alone everybody else's. There have been times when they've been less greedy (and have made money out of it) but since Reagan got elected the US has been stuck with a particularly greedy and short-sighted exploiting class.

America doesn't have static "classes"--that's Marx's terminology (part of his strategy to incite peoples to violence and revolution). Most of the people in our lower income quintile aren't there ten years later, nor do those in the top stay either, generally. America spends among the most on education--and our poor suffer obesity, not famine. Yet Bill's jihad continues unaffected. If anything, intensified.

It's truly bizarre, disturbed. Unbalanced.

Mencken's hobgoblins, the dead, and decent, caring, civil people are the easiest to attack and safest to confront. And if you're attacking imagined injustices you're doing something noble, even if the people you disparage don't deserve it. Maybe that explains it.

No, you had it nailed in the previous paragraph. Slowman is truly
bizarre, disturbed, and unbalanced.

 

[John Larkin]

On Mon, 04 Aug 2014 13:48:43 -0400, krw@attt.bizz wrote:

On Mon, 4 Aug 2014 05:58:02 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Saturday, August 2, 2014 12:32:27 AM UTC-4, Bill Sloman wrote:
On Saturday, 2 August 2014 12:12:27 UTC+10, dagmarg...@yahoo.com wrote:
On Thursday, July 31, 2014 2:53:08 PM UTC-4, Robert Baer wrote:

Yes, the Golden Rule: "He who has the Gold, Rules".

He who rules, gets the gold. (Just look at today's DC.)

It was articulated by some of the more plain-spoken of the founding tax evaders as "the people who own the country should run the country".

And here we have Bill Sloman, America-hating, throwing Molotov cocktails at innocents and honored dead, on invented charges and imaginary grounds. It's useless rebutting; none it ever happened, Bill has no basis for it, has no source--made it all up himself--yet fanatically believes it.

History suggests that they shouldn't - they don't spend enough on keeping the working classes fed, healthy and educated to maximise their own profit, let alone everybody else's. There have been times when they've been less greedy (and have made money out of it) but since Reagan got elected the US has been stuck with a particularly greedy and short-sighted exploiting class.

America doesn't have static "classes"--that's Marx's terminology (part of his strategy to incite peoples to violence and revolution). Most of the people in our lower income quintile aren't there ten years later, nor do those in the top stay either, generally. America spends among the most on education--and our poor suffer obesity, not famine. Yet Bill's jihad continues unaffected. If anything, intensified.

It's truly bizarre, disturbed. Unbalanced.

Mencken's hobgoblins, the dead, and decent, caring, civil people are the easiest to attack and safest to confront. And if you're attacking imagined injustices you're doing something noble, even if the people you disparage don't deserve it. Maybe that explains it.

No, you had it nailed in the previous paragraph. Slowman is truly
bizarre, disturbed, and unbalanced.

But more important, he's tedious.


--

John Larkin Highland Technology, Inc

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Tom Miller]

"John Larkin" <jlarkin@highlandtechnology.com> wrote in message
news:qqivt95407ri9lgfqsp02o2d5m82akk0s8@4ax.com...

On Mon, 04 Aug 2014 13:48:43 -0400, krw@attt.bizz wrote:

On Mon, 4 Aug 2014 05:58:02 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Saturday, August 2, 2014 12:32:27 AM UTC-4, Bill Sloman wrote:
On Saturday, 2 August 2014 12:12:27 UTC+10, dagmarg...@yahoo.com
wrote:
On Thursday, July 31, 2014 2:53:08 PM UTC-4, Robert Baer wrote:

Yes, the Golden Rule: "He who has the Gold, Rules".

He who rules, gets the gold. (Just look at today's DC.)

It was articulated by some of the more plain-spoken of the founding tax
evaders as "the people who own the country should run the country".

And here we have Bill Sloman, America-hating, throwing Molotov cocktails
at innocents and honored dead, on invented charges and imaginary grounds.
It's useless rebutting; none it ever happened, Bill has no basis for it,
has no source--made it all up himself--yet fanatically believes it.

History suggests that they shouldn't - they don't spend enough on
keeping the working classes fed, healthy and educated to maximise their
own profit, let alone everybody else's. There have been times when
they've been less greedy (and have made money out of it) but since
Reagan got elected the US has been stuck with a particularly greedy and
short-sighted exploiting class.

America doesn't have static "classes"--that's Marx's terminology (part of
his strategy to incite peoples to violence and revolution). Most of the
people in our lower income quintile aren't there ten years later, nor do
those in the top stay either, generally. America spends among the most
on education--and our poor suffer obesity, not famine. Yet Bill's jihad
continues unaffected. If anything, intensified.

It's truly bizarre, disturbed. Unbalanced.

Mencken's hobgoblins, the dead, and decent, caring, civil people are the
easiest to attack and safest to confront. And if you're attacking
imagined injustices you're doing something noble, even if the people you
disparage don't deserve it. Maybe that explains it.

No, you had it nailed in the previous paragraph. Slowman is truly
bizarre, disturbed, and unbalanced.

But more important, he's tedious.


--

And boring. And always off topic for the group.

 

[Bill Sloman]

On Tuesday, 5 August 2014 07:23:35 UTC+10, dca...@krl.org wrote:

On Monday, August 4, 2014 3:42:57 PM UTC, Bill Sloman wrote:

I'm an elite member of an elite profession. I was bright enough and self-disciplined enough when young to complete an elite education - admittedly in physical chemistry - and ten years later I was good enough at electronics to be accepted as a member of the IEEE despite not having had any explicit training in electronics.

Only in your own mind. You are unemployed. No job, no profession.

Well - I'm still active in the IEEE. which is exercising my profession

http://sites.ieee.org/nsw/committe/

I'm now on the local - New South Wales - IEEE branch committee as treasurer..

I do apply for electronics jobs from time to time, but I'm 71 and it seems very unlikely that I'll ever get an interview, let alone a job.

Profession - Merriam-Webster Online

www.merriam-webster.com/dictionary/profession

Merriam-Webster

a type of job that requires special education, training, or skill. : the people who work in a particular profession.

The first definition includes me. The second doesn't. The quote thus doesn't support your claim.

--
Bill Sloman, Sydney

 

[Bill Sloman]

On Tuesday, 5 August 2014 04:05:47 UTC+10, John Larkin wrote:

On Mon, 04 Aug 2014 13:48:43 -0400, krw@attt.bizz wrote:
On Mon, 4 Aug 2014 05:58:02 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:
On Saturday, August 2, 2014 12:32:27 AM UTC-4, Bill Sloman wrote:
On Saturday, 2 August 2014 12:12:27 UTC+10, dagmarg...@yahoo.com wrote:
On Thursday, July 31, 2014 2:53:08 PM UTC-4, Robert Baer wrote:

<snip>

> But more important, he's tedious.

John Larkin's definition of "tedious" is "not including comments experienced as flattering by John Larkin".

--
Bill Sloman, Sydney

 

[Bill Sloman]

On Tuesday, 5 August 2014 08:58:51 UTC+10, Tom Miller wrote:

krw@attt.bizz> wrote in message
newso20u953sg70b33d0s16j68dqbkv04cgqj@4ax.com...
On Mon, 4 Aug 2014 14:27:03 -0400, "Tom Miller"
tmiller11147@verizon.net> wrote
"John Larkin" <jlarkin@highlandtechnology.com> wrote in message
news:qqivt95407ri9lgfqsp02o2d5m82akk0s8@4ax.com...
On Mon, 04 Aug 2014 13:48:43 -0400, krw@attt.bizz wrote:
On Mon, 4 Aug 2014 05:58:02 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:
On Saturday, August 2, 2014 12:32:27 AM UTC-4, Bill Sloman wrote:
On Saturday, 2 August 2014 12:12:27 UTC+10, dagmarg...@yahoo.com
wrote:
On Thursday, July 31, 2014 2:53:08 PM UTC-4, Robert Baer wrote:

<snip>

> That and "here's how we did it back in the 60's". Or some reference to Wikipedia.

I do cite other sources from time to time. Anybody dim enough to be a krw groupie can be predicted to be incapable of detecting that.

> Like I say, boring followed by some personal slur towards those he disagrees with.

The "boring" is in the mind of the beholder. The personal slur does now seem to be the group etiquette.

--
Bill Sloman, Sydney

 

[dcaster@krl.org]

On Monday, August 4, 2014 11:56:34 PM UTC, Bill Sloman wrote:

a type of job that requires special education, training, or skill. : the people who work in a particular profession.



The first definition includes me. The second doesn't. The quote thus doesn't support your claim.



--

Bill Sloman, Sydney

Since the first part says " a type of job " Note that word job! It seems that the first definition does not include you.

Dan

 

[Maynard A. Philbrook Jr.]

In article <p56vt9hgqjdkfg8718mg5l5m3ablj8dj2e@4ax.com>,
jjlarkin@highNOTlandTHIStechnologyPART.com says...

My Schmitt_Blinker_2 works from 38 volts to 2000 with a 3 meg series resistor,
the upper limit set by a reasonable power dissipation limit in the resistor.

At 50 volts, the blink rate is just about 1 Hz. That will work for me.

So what have I forgotten?

That range is not shown in any of the files you have posted.

Show it by posting it with proper voltage ranges.

?-)


Drop dead.

Spoken like a true professional

Jamie

 

On Mon, 4 Aug 2014 14:27:03 -0400, "Tom Miller"
<tmiller11147@verizon.net> wrote:

"John Larkin" <jlarkin@highlandtechnology.com> wrote in message
news:qqivt95407ri9lgfqsp02o2d5m82akk0s8@4ax.com...
On Mon, 04 Aug 2014 13:48:43 -0400, krw@attt.bizz wrote:

On Mon, 4 Aug 2014 05:58:02 -0700 (PDT), dagmargoodboat@yahoo.com
wrote:

On Saturday, August 2, 2014 12:32:27 AM UTC-4, Bill Sloman wrote:
On Saturday, 2 August 2014 12:12:27 UTC+10, dagmarg...@yahoo.com
wrote:
On Thursday, July 31, 2014 2:53:08 PM UTC-4, Robert Baer wrote:

Yes, the Golden Rule: "He who has the Gold, Rules".

He who rules, gets the gold. (Just look at today's DC.)

It was articulated by some of the more plain-spoken of the founding tax
evaders as "the people who own the country should run the country".

And here we have Bill Sloman, America-hating, throwing Molotov cocktails
at innocents and honored dead, on invented charges and imaginary grounds.
It's useless rebutting; none it ever happened, Bill has no basis for it,
has no source--made it all up himself--yet fanatically believes it.

History suggests that they shouldn't - they don't spend enough on
keeping the working classes fed, healthy and educated to maximise their
own profit, let alone everybody else's. There have been times when
they've been less greedy (and have made money out of it) but since
Reagan got elected the US has been stuck with a particularly greedy and
short-sighted exploiting class.

America doesn't have static "classes"--that's Marx's terminology (part of
his strategy to incite peoples to violence and revolution). Most of the
people in our lower income quintile aren't there ten years later, nor do
those in the top stay either, generally. America spends among the most
on education--and our poor suffer obesity, not famine. Yet Bill's jihad
continues unaffected. If anything, intensified.

It's truly bizarre, disturbed. Unbalanced.

Mencken's hobgoblins, the dead, and decent, caring, civil people are the
easiest to attack and safest to confront. And if you're attacking
imagined injustices you're doing something noble, even if the people you
disparage don't deserve it. Maybe that explains it.

No, you had it nailed in the previous paragraph. Slowman is truly
bizarre, disturbed, and unbalanced.

But more important, he's tedious.


--

And boring. And always off topic for the group.

Once in a while he drones on about the Baxandall oscillator he's been
building for a decade.

 

[Glenn]

(Answer to sci.physics )

On 04/08/14 23.51, benj wrote:

On 08/04/2014 03:20 PM, jimp@specsol.spam.sux.com wrote:
....
A Fourier transforms data in the time domain to the frequency domain
which
includes the frequency of zero, i.e. the DC component.


Babbling insane nonsense!

A time "domain" is real and true.

"Frequency domain" is an imaginary fantasy that does not exist.

(Even though the so-called "zero frequency" is related to the DC
component in spite of the fact that zero frequencies don't exist...
and for that matter DC components don't exist either)

Hi Ben and Jim

You are actually both right.

All mathematics are actually a fantasy - mind spin.

Even a measured time domain signal is a fantasy - why? Because it is not
the original signal.

So, everytime we measure and use mathematics we have to be aware that we
MODEL. The measured signal is just a (crude) model of the original signal:
https://en.wikipedia.org/wiki/Mathematical_model#Model_evaluation

And when we use (discrete) FFT, we have to very careful to interpret the
"frequency axis".

The sampling proces has to be done very carefully - why?:

Firstly:

You have to sample your signal with sinc-functions! You "simply"
convolute your original signal with sinc-functions at some chosen
timings. A equally why to sample is to hard low-pass frequency limiting
so the sampled (by delta-functions) signal do not have frequencies, that
will be "aliased" down.

https://en.wikipedia.org/wiki/Sinc_function
https://en.wikipedia.org/wiki/Sinc_filter

In reality the above sampling is not possible. That is because a
sinc-function is not mathematically compact supported. And you can not
make a hard low-pass frequency - then you have to wait for eternity for
your filtered signal.

A trade-off is to:
* ensure that the original signal do not have to high frequencies - or
sample often enough per second.
* Use a soft low-pass filter, that also dampens in-band frequencies.

(Will be elaborated at Thirdly)

-

But note that the process FFT-aliasing do exist!

It is YOU (or YOUR process procedure) that have NOT sampled correctly,
if "aliasing" happens!

-

Secondly:

When you use FFT you actually "wraps" the first of the signal together
with to last of the signal.

In other words: The signal is used by FFT as a ring signal.

That is also a problem, because most of the time this inadvertently
"inserts" a jump at the signal ends: First-to-last.

And a "jump" in the "time domain" means "additions" of many frequencies
in the "frequency domain".

One of the "solutions" is to weighting the "time domain"-signal with a
"window"-function:
https://en.wikipedia.org/wiki/Window_function
My first choice would be the Hann(ing) window. But other trade-off exists.

-

Thirdly:

The signal you sampled at "First" was only the "real"-part of signal.
The "imaginary"-part is normally set to zero, and that is an error (if
not dealt with otherwise)!

The "imaginary" part of the signal is actually as real as the
"real"-part of the signal!

To sample the "imaginary"-part of the signal you have to take the
original signal and filter it by a (hard/soft) low pass filter, that
besides the low pass function also twists all in-band frequencies 90 degree.

When using the full "frequency domain" you can not "reverse engineer"
the "imaginary"-part from the discretized "real"-part!

In other words: The conception of "positive" and "negative" "frequency
domain"-coefficients is WRONG!

-

If you design your FFT to only be fed with a "real"-part, you actually
has to chose your low-pass-filter accordingly. You additionally also
have to filter out all frequencies from (and with) half "frequency
domain"-frequencies and above - why?

Because there is not enough information in the sampled signal to
distinguish between the two halves of the "frequency domain".

After you FFT a "real"-part-only signal you have to zero frequencies
from (and with) half "frequency domain"-frequencies and above.

Then you have to multiply all the rest of the "frequency
domain"-frequencies coefficents except the first ("DC"-coefficient) by two.

-

Forthly:

After all the former quirks:

What you "see" in the "frequency domain" is not the actual frequencies
in your signal.

The base functions used in your FFT are only a finite frequencies.

If you used 1024 coefficients in the "time domain", you only get 1024
(512 if "time domain"-part only) distinct "frequency domain"-frequencies.

And actually the FFT only can "recognize" those special frequencies.

FFT can not handle an original frequency with an "odd" frequency - e.g.
other than the "frequency domain"-frequencies base functions (sines and
consines).

When "odd"-frequencies (frequencies between two base function
frequencies) are in the original signal, the REAL ALIASING happens!

You can test it for yourself.

Try sampling one base function frequency (no windowing!) - after FFT you
precisely get only one coeffient in the "frequency domain"!

If you sample a single frequency that is between two base function
frequencies, your FFT (no windowing!) will smear your original frequency
all over your "frequency domain".

-

One compromise is to use windowing in "time domain" so both
base-function-frequencies - and frequencies that is between two
base-function-frequencies - are smeared approximately equally - in the
"frequency domain".

-

Conclusion

Using discrete FFT is a riddled with a lot of compromises.

And a lot of people using FFT do not recognize that they implicitly made
all these compromises.

-

An alternative sampling and transform tool:

Use compactly supported Wavelet and scaling base functions to sample the
original signal - and Complex Wavelet Transform.

https://en.wikipedia.org/wiki/Discrete_wavelet_transform#The_Dual-Tree_Complex_Wavelet_Transform_.28.E2.84.82WT.29
http://eeweb.poly.edu/iselesni/pubs/CWT_Tutorial.pdf
https://web.archive.org/web/20131116101914/http://eeweb.poly.edu/iselesni/pubs/CWT_Tutorial.pdf
http://www.math.sc.edu/~blanco/IMI/DTCWT0.pdf
https://web.archive.org/web/20120904164219/http://www.math.sc.edu/~blanco/IMI/DTCWT0.pdf
"...
TROUBLE IN PARADISE: FOUR PROBLEMS WITH REAL WAVELETS
....
ONE SOLUTION: COMPLEX WAVELETS
....
The second school seeks a redundant representation, with both wr(t) and
wi(t) individually forming orthonormal or biorthogonal bases. The
resulting CWT is a 2x redundant tight frame [26] in 1-D, with the power
to overcome the four shortcomings.
....
In this article, we will focus on a particularly natural approach to the
second, redundant type of CWT, the dual-tree approach, which is based on
two FB trees and thus two bases [55], [57].
....
From Figure 3, we see that we can reach quite close to the ideal even
with quite short filters.
....
As a result, the dual-tree CWT comes very close to mirroring the
attractive properties of the Fourier transform, including a smooth,
nonoscillating magnitude (see Figure 1); a nearly shift-invariant
magnitude with a simple near-linear phase encoding of signal shifts;
substantially reduced aliasing; and directional wavelets in higher
dimensions. The only cost for all of this is a moderate redundancy: 2x
redundancy in 1-D (2^d for d-dimensional signals, in general).
....
In the following, we describe three methods for FIR dual-tree filter
design. Fast implementations of some of these filters have been recently
described in[1].
....
MATLAB software for the dual-tree complex wavelet transform (and related
algorithms) is available at the following locations on the web:
http://taco.poly.edu/WaveletSoftware/,
http://www-sigproc.eng.cam.ac.uk/~ngk/, and http://dsp.rice.edu/.
...."

http://taco.poly.edu/WaveletSoftware/ ->
http://eeweb.poly.edu/iselesni/WaveletSoftware/
https://web.archive.org/web/20131203145028/http://eeweb.poly.edu/iselesni/WaveletSoftware/

http://www-sigproc.eng.cam.ac.uk/~ngk/ ->
http://www-sigproc.eng.cam.ac.uk/Main/NGK
https://web.archive.org/web/20140109175106/http://www-sigproc.eng.cam.ac.uk/Main/NGK
Downloadable teaching material (PDF and zip files):
http://www-sigproc.eng.cam.ac.uk/Main/NGK#JumpMaterials
DESIGN OF Q-SHIFT COMPLEX WAVELETS FOR IMAGE PROCESSING USING FREQUENCY
DOMAIN ENERGY MINIMIZATION:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.65.9830&rep=rep1&type=pdf
https://web.archive.org/web/20140805094418/http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.65.9830&rep=rep1&type=pdf
Complex Wavelets for Shift Invariant Analysis and Filtering of Signals:
http://faculty.washington.edu/dbp/s530/PDFs/ACHA-Kingsbury.pdf
https://web.archive.org/web/20140805094354/http://faculty.washington.edu/dbp/s530/PDFs/ACHA-Kingsbury.pdf

Glenn

 

[Glenn]

(Answer to sci.physics )

On 04/08/14 23.51, benj wrote:

On 08/04/2014 03:20 PM, jimp@specsol.spam.sux.com wrote:
....
A Fourier transforms data in the time domain to the frequency domain
which
includes the frequency of zero, i.e. the DC component.


Babbling insane nonsense!

A time "domain" is real and true.

"Frequency domain" is an imaginary fantasy that does not exist.

(Even though the so-called "zero frequency" is related to the DC
component in spite of the fact that zero frequencies don't exist...
and for that matter DC components don't exist either)

Hi Ben and Jim

You are actually both right.

All mathematics are actually a fantasy - mind spin.

Even a measured time domain signal is a fantasy - why? Because it is not
the original signal.

So, every time we measure and use mathematics, we have to be aware, that
we MODEL. The measured signal is just a (crude) model of the original
signal:
https://en.wikipedia.org/wiki/Mathematical_model#Model_evaluation

And when we use (discrete) FFT, we have to very careful when
interpreting the "frequency axis".

The sampling proces has to be done very carefully - why?:

Firstly:

You have to sample your signal with sinc-functions! You "simply"
convolute your original signal with sinc-functions at some chosen
timings. A equally way to sample, is to hard low-pass frequency
limiting, so the sampled (by delta-functions) signal do not have
frequencies, that will be "aliased" down.

https://en.wikipedia.org/wiki/Sinc_function
https://en.wikipedia.org/wiki/Sinc_filter

In reality the above sampling is not possible. That is because a
sinc-function is not mathematically compact supported. And you can not
make a hard low-pass frequency - then you have to wait for eternity for
your filtered signal.

A trade-off is to:
* ensure that the original signal do not have to high frequencies - or
sample often enough per second.
* Use a soft low-pass filter, that also dampens in-band frequencies.

(Will be elaborated at Thirdly)

-

But note, that this type of "aliasing", when FFTing a signal that
orinally had too high frequencies, do NOT exist!

It is YOU (or YOUR process) that have NOT sampled correctly, if this
type of "aliasing" happens!

In other words - the crude error happened because of the wrong sampling
proces!

-

Secondly:

When you use FFT you actually "wraps" the first of the signal together
with to last of the signal.

In other words: The signal is used by FFT as a ring signal.

That is also a problem, because most of the time this inadvertently
"inserts" a jump at the signal ends: First-to-last.

And a "jump" in the "time domain" means "additions" of many frequencies
in the "frequency domain".

One of the "solutions" is to weighting the "time domain"-signal with a
"window"-function:
https://en.wikipedia.org/wiki/Window_function
My first choice would be the Hann(ing) window. But other trade-off exists.

-

Thirdly:

The signal you sampled at "First" was only the "real"-part of signal.
The "imaginary"-part is normally set to zero, and that is an error (if
not dealt with otherwise)!

The "imaginary" part of the signal is actually as real as the
"real"-part of the signal!

To sample the "imaginary"-part of the signal you have to take the
original signal and filter it by a (hard/soft) low pass filter, that
besides the low pass function also twists all in-band frequencies 90 degree.

When using the full "frequency domain" you can not "reverse engineer"
the "imaginary"-part from the discretized "real"-part!

In other words: The conception of "positive" and "negative" "frequency
domain"-coefficients is WRONG!

-

If you design your FFT to only be fed with a "real"-part, you actually
has to chose your low-pass-filter accordingly. You additionally also
have to filter out all frequencies from (and with) half "frequency
domain"-frequencies and above - why?

Because there is not enough information in the sampled signal to
distinguish between the two halves of the "frequency domain".

After you FFT a "real"-part-only signal you have to zero frequencies
from (and with) half "frequency domain"-frequencies and above.

Then you have to multiply all the rest of the "frequency
domain"-frequencies coefficents except the first ("DC"-coefficient) by two.

-

Forthly:

After all the former quirks:

What you "see" in the "frequency domain" is not the actual frequencies
in your signal.

The base functions used in your FFT are only a finite frequencies.

If you used 1024 coefficients in the "time domain", you only get 1024
(512 if "time domain"-part only) distinct "frequency domain"-frequencies.

And actually the FFT only can "recognize" those special frequencies.

FFT can not handle an original frequency with an "odd" frequency - e.g.
other than the "frequency domain"-frequencies base functions (sines and
consines).

When "odd"-frequencies (frequencies between two base function
frequencies) are in the original signal, the REAL ALIASING happens!

You can test it for yourself.

Try sampling one base function frequency (no windowing!) - after FFT you
precisely get only one coeffient in the "frequency domain"!

If you sample a single frequency that is between two base function
frequencies, your FFT (no windowing!) will smear your original frequency
all over your "frequency domain".

-

One compromise is to use windowing in "time domain" so both
base-function-frequencies - and frequencies that is between two
base-function-frequencies - are smeared approximately equally - in the
"frequency domain".

-

Conclusion

Using discrete FFT is a riddled with a lot of compromises.

And a lot of people using FFT do not recognize that they implicitly made
all these compromises.

-

An alternative sampling and transform tool:

Use compactly supported Wavelet and scaling base functions to sample the
original signal - and Complex Wavelet Transform.

https://en.wikipedia.org/wiki/Discrete_wavelet_transform#The_Dual-Tree_Complex_Wavelet_Transform_.28.E2.84.82WT.29
http://eeweb.poly.edu/iselesni/pubs/CWT_Tutorial.pdf
https://web.archive.org/web/20131116101914/http://eeweb.poly.edu/iselesni/pubs/CWT_Tutorial.pdf
http://www.math.sc.edu/~blanco/IMI/DTCWT0.pdf
https://web.archive.org/web/20120904164219/http://www.math.sc.edu/~blanco/IMI/DTCWT0.pdf
"...
TROUBLE IN PARADISE: FOUR PROBLEMS WITH REAL WAVELETS
....
ONE SOLUTION: COMPLEX WAVELETS
....
The second school seeks a redundant representation, with both wr(t) and
wi(t) individually forming orthonormal or biorthogonal bases. The
resulting CWT is a 2x redundant tight frame [26] in 1-D, with the power
to overcome the four shortcomings.
....
In this article, we will focus on a particularly natural approach to the
second, redundant type of CWT, the dual-tree approach, which is based on
two FB trees and thus two bases [55], [57].
....
From Figure 3, we see that we can reach quite close to the ideal even
with quite short filters.
....
As a result, the dual-tree CWT comes very close to mirroring the
attractive properties of the Fourier transform, including a smooth,
nonoscillating magnitude (see Figure 1); a nearly shift-invariant
magnitude with a simple near-linear phase encoding of signal shifts;
substantially reduced aliasing; and directional wavelets in higher
dimensions. The only cost for all of this is a moderate redundancy: 2x
redundancy in 1-D (2^d for d-dimensional signals, in general).
....
In the following, we describe three methods for FIR dual-tree filter
design. Fast implementations of some of these filters have been recently
described in[1].
....
MATLAB software for the dual-tree complex wavelet transform (and related
algorithms) is available at the following locations on the web:
http://taco.poly.edu/WaveletSoftware/,
http://www-sigproc.eng.cam.ac.uk/~ngk/, and http://dsp.rice.edu/.
...."

http://taco.poly.edu/WaveletSoftware/ ->
http://eeweb.poly.edu/iselesni/WaveletSoftware/
https://web.archive.org/web/20131203145028/http://eeweb.poly.edu/iselesni/WaveletSoftware/

http://www-sigproc.eng.cam.ac.uk/~ngk/ ->
http://www-sigproc.eng.cam.ac.uk/Main/NGK
https://web.archive.org/web/20140109175106/http://www-sigproc.eng.cam.ac.uk/Main/NGK
Downloadable teaching material (PDF and zip files):
http://www-sigproc.eng.cam.ac.uk/Main/NGK#JumpMaterials
DESIGN OF Q-SHIFT COMPLEX WAVELETS FOR IMAGE PROCESSING USING FREQUENCY
DOMAIN ENERGY MINIMIZATION:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.65.9830&rep=rep1&type=pdf
https://web.archive.org/web/20140805094418/http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.65.9830&rep=rep1&type=pdf
Complex Wavelets for Shift Invariant Analysis and Filtering of Signals:
http://faculty.washington.edu/dbp/s530/PDFs/ACHA-Kingsbury.pdf
https://web.archive.org/web/20140805094354/http://faculty.washington.edu/dbp/s530/PDFs/ACHA-Kingsbury.pdf

Glenn

 

[Glenn]

(Answer to sci.physics )

On 04/08/14 23.51, benj wrote:

On 08/04/2014 03:20 PM, jimp@specsol.spam.sux.com wrote:
....
A Fourier transforms data in the time domain to the frequency domain
which
includes the frequency of zero, i.e. the DC component.


Babbling insane nonsense!

A time "domain" is real and true.

"Frequency domain" is an imaginary fantasy that does not exist.

(Even though the so-called "zero frequency" is related to the DC
component in spite of the fact that zero frequencies don't exist...
and for that matter DC components don't exist either)

Hi Ben and Jim

You are actually both right.

All mathematics are actually a fantasy - mind spin.

Even a measured time domain signal is a fantasy - why? Because it is not
the original signal.

So, every time we measure and use mathematics, we have to be aware, that
we MODEL. The measured signal is just a (crude) model of the original
signal:
https://en.wikipedia.org/wiki/Mathematical_model#Model_evaluation

And when we use (discrete) FFT, we have to very careful when
interpreting the "frequency axis".

The sampling proces has to be done very carefully - why?:

Firstly:

You have to sample your signal with sinc-functions! You "simply"
convolute your original signal with sinc-functions at some chosen
timings. A equally way to sample, is to hard low-pass frequency
limiting, so the sampled (by delta-functions) signal do not have
frequencies, that will be "aliased" down.

https://en.wikipedia.org/wiki/Sinc_function
https://en.wikipedia.org/wiki/Sinc_filter

In reality the above sampling is not possible. That is because a
sinc-function is not mathematically compact supported. And you can not
make a hard low-pass frequency - then you have to wait for eternity for
your filtered signal.

A trade-off is to:
* ensure that the original signal do not have to high frequencies - or
sample often enough per second.
* Use a soft low-pass filter, that also dampens in-band frequencies.

(Will be elaborated at Thirdly)

-

But note, that this type of "aliasing", when FFTing a signal that
orinally had too high frequencies, do NOT exist!

It is YOU (or YOUR process) that have NOT sampled correctly, if this
type of "aliasing" happens!

In other words - the crude error happened because of the wrong sampling
proces!

-

Secondly:

When you use FFT you actually "wraps" the first of the signal together
with to last of the signal.

In other words: The signal is used by FFT as a ring signal.

That is also a problem, because most of the time this inadvertently
"inserts" a jump at the signal ends: First-to-last.

And a "jump" in the "time domain" means "additions" of many frequencies
in the "frequency domain".

One of the "solutions" is to weighting the "time domain"-signal with a
"window"-function:
https://en.wikipedia.org/wiki/Window_function
My first choice would be the Hann(ing) window. But other trade-off exists.

-

Thirdly:

The signal you sampled at "First" was only the "real"-part of signal.
The "imaginary"-part is normally set to zero, and that is an error (if
not dealt with otherwise)!

The "imaginary" part of the signal is actually as real as the
"real"-part of the signal!

To sample the "imaginary"-part of the signal you have to take the
original signal and filter it by a (hard/soft) low pass filter, that
besides the low pass function also twists all in-band frequencies 90 degree.

When using the full "frequency domain" you can not "reverse engineer"
the "imaginary"-part from the discretized "real"-part!

In other words: The conception of "positive" and "negative" "frequency
domain"-coefficients is WRONG!

-

If you design your FFT to only be fed with a "real"-part, you actually
has to chose your low-pass-filter accordingly. You additionally also
have to filter out all frequencies from (and with) half "frequency
domain"-frequencies and above - why?

Because there is not enough information in the sampled signal to
distinguish between the two halves of the "frequency domain".

After you FFT a "real"-part-only signal you have to zero frequencies
from (and with) half "frequency domain"-frequencies and above.

Then you have to multiply all the rest of the "frequency
domain"-frequencies coefficents except the first ("DC"-coefficient) by two.

-

Forthly:

After all the former quirks:

What you "see" in the "frequency domain" is not the actual frequencies
in your signal.

The base functions used in your FFT are only a finite frequencies.

If you used 1024 coefficients in the "time domain", you only get 1024
(512 if "time domain"-part only) distinct "frequency domain"-frequencies.

And actually the FFT only can "recognize" those special frequencies.

FFT can not handle an original frequency with an "odd" frequency - e.g.
other than the "frequency domain"-frequencies base functions (sines and
consines).

When "odd"-frequencies (frequencies between two base function
frequencies) are in the original signal, the REAL ALIASING happens!

You can test it for yourself.

Try sampling one base function frequency (no windowing!) - after FFT you
precisely get only one coeffient in the "frequency domain"!

If you sample a single frequency that is between two base function
frequencies, your FFT (no windowing!) will smear your original frequency
all over your "frequency domain".

-

One compromise is to use windowing in "time domain" so both
base-function-frequencies - and frequencies that is between two
base-function-frequencies - are smeared approximately equally - in the
"frequency domain".

-

Conclusion

Using discrete FFT is a riddled with a lot of compromises.

And a lot of people using FFT do not recognize that they implicitly made
all these compromises.

-

An alternative sampling and transform tool:

Use compactly supported Wavelet and scaling base functions to sample the
original signal - and Complex Wavelet Transform.

https://en.wikipedia.org/wiki/Discrete_wavelet_transform#The_Dual-Tree_Complex_Wavelet_Transform_.28.E2.84.82WT.29
http://eeweb.poly.edu/iselesni/pubs/CWT_Tutorial.pdf
https://web.archive.org/web/20131116101914/http://eeweb.poly.edu/iselesni/pubs/CWT_Tutorial.pdf
http://www.math.sc.edu/~blanco/IMI/DTCWT0.pdf
https://web.archive.org/web/20120904164219/http://www.math.sc.edu/~blanco/IMI/DTCWT0.pdf
"...
TROUBLE IN PARADISE: FOUR PROBLEMS WITH REAL WAVELETS
....
ONE SOLUTION: COMPLEX WAVELETS
....
The second school seeks a redundant representation, with both wr(t) and
wi(t) individually forming orthonormal or biorthogonal bases. The
resulting CWT is a 2x redundant tight frame [26] in 1-D, with the power
to overcome the four shortcomings.
....
In this article, we will focus on a particularly natural approach to the
second, redundant type of CWT, the dual-tree approach, which is based on
two FB trees and thus two bases [55], [57].
....
From Figure 3, we see that we can reach quite close to the ideal even
with quite short filters.
....
As a result, the dual-tree CWT comes very close to mirroring the
attractive properties of the Fourier transform, including a smooth,
nonoscillating magnitude (see Figure 1); a nearly shift-invariant
magnitude with a simple near-linear phase encoding of signal shifts;
substantially reduced aliasing; and directional wavelets in higher
dimensions. The only cost for all of this is a moderate redundancy: 2x
redundancy in 1-D (2^d for d-dimensional signals, in general).
....
In the following, we describe three methods for FIR dual-tree filter
design. Fast implementations of some of these filters have been recently
described in[1].
....
MATLAB software for the dual-tree complex wavelet transform (and related
algorithms) is available at the following locations on the web:
http://taco.poly.edu/WaveletSoftware/,
http://www-sigproc.eng.cam.ac.uk/~ngk/, and http://dsp.rice.edu/.
...."

http://taco.poly.edu/WaveletSoftware/ ->
http://eeweb.poly.edu/iselesni/WaveletSoftware/
https://web.archive.org/web/20131203145028/http://eeweb.poly.edu/iselesni/WaveletSoftware/

http://www-sigproc.eng.cam.ac.uk/~ngk/ ->
http://www-sigproc.eng.cam.ac.uk/Main/NGK
https://web.archive.org/web/20140109175106/http://www-sigproc.eng.cam.ac.uk/Main/NGK
Downloadable teaching material (PDF and zip files):
http://www-sigproc.eng.cam.ac.uk/Main/NGK#JumpMaterials
DESIGN OF Q-SHIFT COMPLEX WAVELETS FOR IMAGE PROCESSING USING FREQUENCY
DOMAIN ENERGY MINIMIZATION:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.65.9830&rep=rep1&type=pdf
https://web.archive.org/web/20140805094418/http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.65.9830&rep=rep1&type=pdf
Complex Wavelets for Shift Invariant Analysis and Filtering of Signals:
http://faculty.washington.edu/dbp/s530/PDFs/ACHA-Kingsbury.pdf
https://web.archive.org/web/20140805094354/http://faculty.washington.edu/dbp/s530/PDFs/ACHA-Kingsbury.pdf

Glenn

 

In sci.physics Glenn <glenn2233@gmail.com> wrote:

(Answer to sci.physics )

On 04/08/14 23.51, benj wrote:
On 08/04/2014 03:20 PM, jimp@specsol.spam.sux.com wrote:
...
A Fourier transforms data in the time domain to the frequency domain
which
includes the frequency of zero, i.e. the DC component.


Babbling insane nonsense!

A time "domain" is real and true.

"Frequency domain" is an imaginary fantasy that does not exist.

(Even though the so-called "zero frequency" is related to the DC
component in spite of the fact that zero frequencies don't exist...
and for that matter DC components don't exist either)

Hi Ben and Jim

You are actually both right.

All mathematics are actually a fantasy - mind spin.

Even a measured time domain signal is a fantasy - why? Because it is not
the original signal.

The rational world refers to that as "error bars".

So, everytime we measure and use mathematics we have to be aware that we
MODEL. The measured signal is just a (crude) model of the original signal:
https://en.wikipedia.org/wiki/Mathematical_model#Model_evaluation

And when we use (discrete) FFT, we have to very careful to interpret the
"frequency axis".

The sampling proces has to be done very carefully - why?:

Firstly:

You have to sample your signal with sinc-functions! You "simply"
convolute your original signal with sinc-functions at some chosen
timings. A equally why to sample is to hard low-pass frequency limiting
so the sampled (by delta-functions) signal do not have frequencies, that
will be "aliased" down.

https://en.wikipedia.org/wiki/Sinc_function
https://en.wikipedia.org/wiki/Sinc_filter

http://en.wikipedia.org/wiki/Spectrum_analyzer



--
Jim Pennino

 

In sci.physics Glenn <glenn2233@gmail.com> wrote:

(Answer to sci.physics )

On 04/08/14 23.51, benj wrote:
On 08/04/2014 03:20 PM, jimp@specsol.spam.sux.com wrote:
...
A Fourier transforms data in the time domain to the frequency domain
which
includes the frequency of zero, i.e. the DC component.


Babbling insane nonsense!

A time "domain" is real and true.

"Frequency domain" is an imaginary fantasy that does not exist.

(Even though the so-called "zero frequency" is related to the DC
component in spite of the fact that zero frequencies don't exist...
and for that matter DC components don't exist either)

Hi Ben and Jim

You are actually both right.

All mathematics are actually a fantasy - mind spin.

Even a measured time domain signal is a fantasy - why? Because it is not
the original signal.

Ever heard of "error bars"?


--
Jim Pennino

 

[benj]

On 08/05/2014 05:51 AM, Glenn wrote:

(Answer to sci.physics )

On 04/08/14 23.51, benj wrote:
On 08/04/2014 03:20 PM, jimp@specsol.spam.sux.com wrote:
...
A Fourier transforms data in the time domain to the frequency domain
which
includes the frequency of zero, i.e. the DC component.


Babbling insane nonsense!

A time "domain" is real and true.

"Frequency domain" is an imaginary fantasy that does not exist.

(Even though the so-called "zero frequency" is related to the DC
component in spite of the fact that zero frequencies don't exist...
and for that matter DC components don't exist either)

Hi Ben and Jim

You are actually both right.

All mathematics are actually a fantasy - mind spin.

Even a measured time domain signal is a fantasy - why? Because it is not
the original signal.

I'm Sorry Glenn but you are totally WRONG! Jimp is NEVER right!

As for time domain being "real" I was talking about the original signal
which is traced back to actual phenomena, such as voltages which is
traceable back to charge positions in time and space etc. In other words
"reality" as much as we understand it.

<snip remainder of Glenn's post explaining in great detail just exactly
what I was talking about>

Glenn now you've gone and spoiled the party!

 

In sci.physics benj <none@gmail.com> wrote:

On 08/05/2014 05:51 AM, Glenn wrote:
(Answer to sci.physics )

On 04/08/14 23.51, benj wrote:
On 08/04/2014 03:20 PM, jimp@specsol.spam.sux.com wrote:
...
A Fourier transforms data in the time domain to the frequency domain
which
includes the frequency of zero, i.e. the DC component.


Babbling insane nonsense!

A time "domain" is real and true.

"Frequency domain" is an imaginary fantasy that does not exist.

(Even though the so-called "zero frequency" is related to the DC
component in spite of the fact that zero frequencies don't exist...
and for that matter DC components don't exist either)

Hi Ben and Jim

You are actually both right.

All mathematics are actually a fantasy - mind spin.

Even a measured time domain signal is a fantasy - why? Because it is not
the original signal.

I'm Sorry Glenn but you are totally WRONG! Jimp is NEVER right!

OK, now I understand why you jumped in with a pile of childisn nonsense...



--
Jim Pennino

 

[Bill Sloman]

On Wednesday, 6 August 2014 08:21:13 UTC+10, k...@attt.bizz wrote:

On Mon, 4 Aug 2014 21:22:41 -0700 (PDT), Bill Sloman
bill.sloman@gmail.com> wrote:
On Tuesday, 5 August 2014 10:45:34 UTC+10, k...@attt.bizz wrote:
On Mon, 4 Aug 2014 17:38:12 -0700 (PDT), "dcaster@krl.org"
dcaster@krl.org> wrote:
On Monday, August 4, 2014 11:56:34 PM UTC, Bill Sloman wrote:

a type of job that requires special education, training, or skill. : the people who work in a particular profession.

The first definition includes me. The second doesn't. The quote thus doesn't support your claim.

Since the first part says " a type of job " Note that word job! It seems that the first definition does not include you.

The IEEE accepts his dues. He's *somebody*.

They might accept dues from me if I were silly enough to offer them. Since I joined the IEEE in 1980 and was born in 1942, I qualified for Life Member status in 2011.

"Basic dues and assessments are waived for those achieving Life Member status."

So you're a deadbeat. That surprises no one here.

I don't pay them anything, but I am active in the IEEE

http://sites.ieee.org/nsw/committe/

I'm the treasurer for the local - New South Wales - branch of the IEEE.

With electronic banking it's not an onerous job, but the IEEE does make it more time-consuming by insisting that I report every transaction on their Netsuite program, throwing in .pdf images of every last claim form and receipt.

At the moment Netsuite and our bank agree about the amount of money we've got, but it took a while to get them reconciled.

--
Bill Sloman, Sydney

 

[rickman]

On 8/5/2014 5:05 PM, jimp@specsol.spam.sux.com wrote:

In sci.physics benj <none@gmail.com> wrote:
On 08/05/2014 05:51 AM, Glenn wrote:
(Answer to sci.physics )

On 04/08/14 23.51, benj wrote:
On 08/04/2014 03:20 PM, jimp@specsol.spam.sux.com wrote:
...
A Fourier transforms data in the time domain to the frequency domain
which
includes the frequency of zero, i.e. the DC component.


Babbling insane nonsense!

A time "domain" is real and true.

"Frequency domain" is an imaginary fantasy that does not exist.

(Even though the so-called "zero frequency" is related to the DC
component in spite of the fact that zero frequencies don't exist...
and for that matter DC components don't exist either)

Hi Ben and Jim

You are actually both right.

All mathematics are actually a fantasy - mind spin.

Even a measured time domain signal is a fantasy - why? Because it is not
the original signal.

I'm Sorry Glenn but you are totally WRONG! Jimp is NEVER right!

OK, now I understand why you jumped in with a pile of childisn nonsense...

I was waiting to see just why this conversation was cross posted to
s.e.d and I've figured it out. It was about to leave the domain of
rational discourse and devolve into petty bickering. Yup, s.e.d is the
perfect place for it.

--

Rick

 

On Mon, 4 Aug 2014 21:22:41 -0700 (PDT), Bill Sloman
<bill.sloman@gmail.com> wrote:

On Tuesday, 5 August 2014 10:45:34 UTC+10, k...@attt.bizz wrote:
On Mon, 4 Aug 2014 17:38:12 -0700 (PDT), "dcaster@krl.org"
dcaster@krl.org> wrote:
On Monday, August 4, 2014 11:56:34 PM UTC, Bill Sloman wrote:

a type of job that requires special education, training, or skill. : the people who work in a particular profession.

The first definition includes me. The second doesn't. The quote thus doesn't support your claim.

Since the first part says " a type of job " Note that word job! It seems that the first definition does not include you.

The IEEE accepts his dues. He's *somebody*.

They might accept dues from me if I were silly enough to offer them. Since I joined the IEEE in 1980 and was born in 1942, I qualified for Life Member status in 2011.

"Basic dues and assessments are waived for those achieving Life Member status."

So you're a deadbeat. That surprises no one here.

 

[Joe Gwinn]

In article <66m2u9d4qhjn26gihbva4i9l2ohpgoqp4f@4ax.com>,
<krw@attt.bizz> wrote:

On Mon, 4 Aug 2014 21:22:41 -0700 (PDT), Bill Sloman
bill.sloman@gmail.com> wrote:

On Tuesday, 5 August 2014 10:45:34 UTC+10, k...@attt.bizz wrote:
On Mon, 4 Aug 2014 17:38:12 -0700 (PDT), "dcaster@krl.org"
dcaster@krl.org> wrote:
On Monday, August 4, 2014 11:56:34 PM UTC, Bill Sloman wrote:

a type of job that requires special education, training, or skill. :
the people who work in a particular profession.

The first definition includes me. The second doesn't. The quote thus
doesn't support your claim.

Since the first part says " a type of job " Note that word job! It
seems that the first definition does not include you.

The IEEE accepts his dues. He's *somebody*.

They might accept dues from me if I were silly enough to offer them. Since I
joined the IEEE in 1980 and was born in 1942, I qualified for Life Member
status in 2011.

"Basic dues and assessments are waived for those achieving Life Member
status."

So you're a deadbeat. That surprises no one here.

I'm an IEEE Life Member as well. They do waive the fees. After 50
years of paying.

Joe Gwinn