Frequency Domain and Time Domain relationships

[mikebg]

NOISE AT NARROW BAND amplifier

Noise at narrow band amplifier.

In post #53 is showed the output of a broadband preamp when the AGC maintains its gain
near to the threshold of saturation with noise.
http://www.geotech1.com/forums/showt...898#post165898

The given below image shows the output of a narrowband amplifier whose gain is controlled by
the same AGC. You can see one sinusoid whose frequency is in the center of received bandwidth. One
inexperienced researcher would think that this is oscillation due to some parasitic positive feedback. To see whether this is a generation, you should gradually reduce the gain. If the amplitude also
decreases gradually, this is not generation. Decrease of gain interrupts generation suddenly.

The noise we see on the output of narrow band amp only looks like a sine wave.
If we look with caution, should be apparent the chaotic amplitude modulation and the angle modulation. The maximum modulating frequency is equal to the bandwidth. Angle modulation of sine wave by narrowband noise is difficult to discern in preamp for metal detector if the bandwidth is 16Hz. The reason for this is that modulation
index appears too small compared to the center frequency 4kHz or more. Amplitude modulation is
more noticeable especially if the bandwidth is below 16Hz. Then the vision has not merged images
so the chaotic variation of the sinusoide is visible.

[Aziz]

Yes, another your: "It must fail or I will eat the hot rocks." But now fractalised hot rocks, instead of "neuron-networked".

I didn't eat them. So there must be something "majik".

I don't blame you for not knowing the data mining concepts. Many scientists using the tools to find relationships buried in the raw data.
Aziz

[WM6]

How dare you?

If you don't understand something, which is obviously beyond your scope, then simply be quiet please.
You all must learn a lot. Much a lot.

The art of processing is to seperate the wanted information out of noise and other unwanted information in a data set. It's the data mining engineering.

Aziz

Yes, another your: "It must fail or I will eat the hot rocks." But now fractalised hot rocks, instead of "neuron-networked".

[ODM]

Well, if the Beatles went to India for some inspiration and diving into their psyche with experts and chemistry, it's not far off?

[Aziz]

Aziz guys at LongRangeLocators are desperately awaiting you.

How dare you?

If you don't understand something, which is obviously beyond your scope, then simply be quiet please.
You all must learn a lot. Much a lot.

The art of processing is to seperate the wanted information out of noise and other unwanted information in a data set. It's the data mining engineering.

Aziz

[WM6]

I hope, some of you got inspired and got some "majik" into the brain by looking at the video. BTW, the brain is also a fractal design.

Aziz guys at LongRangeLocators are desperately awaiting you.

[Aziz]

I hope, some of you got inspired and got some "majik" into the brain by looking at the video. BTW, the brain is also a fractal design.

You probably become fractal expert now and you will focus your eyes to find repetative patterns (pattern recognition by eyes).

Take these "new eyes" and look at the frequency domain response. You must see it.


Ok, if you have problems in identifying the useful information, you can use other numerical methods:
- neural networks
- statistics

Good luck.
Aziz

[Aziz]

Only when you have taken real hardware to Beggary Hill or Wedderburn, Australia, or one of the recognised test patches that have been created out there. Have witnessed tests done, on real nuggets, or non-ferrous test pieces under ironstone 1 - 3ft thick. Do a quick visual scan round first to see if there is a man on a hill not far off with a pair of binoculars. Take a couple of beefy minders too, (the sort that have a slight bulge under the left armpit) in case it really is WBGB.

Eric.

Eric I think my WBGB would turn into WWGB (World's Worst GB). *LOL*

Anyway. Some magic words to everyone just to encourage for finding a better GB.

We have to hunt in the hidden dimension. Frequency domain will help you to find the hidden dimension (I'm on-topic now ). We are living in a fractal world (universe). We all face with (repetitive) patterns, which are based on simple principles.

I shouldn't had been working too much on pattern recognition and image processing systems. I have trained my eyes obviously.

Here is a nice video to stimulate some brains:
Fractals - Hunting The Hidden Dimension
http://www.youtube.com/watch?v=s65DSz78jW4
Must see the interesting video.
(Note: 30m:44s fractal antennas!!! )

I'm going to look for FGB (fractal GB) ..

Cheers,
Aziz

[Ferric Toes]

And you can believe me to have the WBGB technology too.

Cheers,
Aziz

Only when you have taken real hardware to Beggary Hill or Wedderburn, Australia, or one of the recognised test patches that have been created out there. Have witnessed tests done, on real nuggets, or non-ferrous test pieces under ironstone 1 - 3ft thick. Do a quick visual scan round first to see if there is a man on a hill not far off with a pair of binoculars. Take a couple of beefy minders too, (the sort that have a slight bulge under the left armpit) in case it really is WBGB.

Eric.

[Aziz]

No BS Talking

Hi all,

Warning! It gets blatant off-topic with blatant fugging advertising is coming up..
You have been warned!

--------------

I have finally found a good example on my hard disk to show you, that I'm not talking B$ to you.
Below is a Wifi omni antenna (2.4 GHz) with 5 elements. Impedance 50 Ohm, almost 9 dbi antenna gain. SWR almost 1. It has a radial pattern coverage. Ideal for Wifi stations.

If you wanna play with it, I'm putting the source code, the symbolic antenna definition code, and the binary output code of the antenna model (see zip-file). But I'm not putting the magnificiant super duper compiler.
Have a look at the symbolic code. You can optimize the parameters with it.

Oh man!, I've forget a lot of things I have made at that time. So many different antennas... But I like to fiddle with metal detector designs today.



omni2.zip

And you can believe me to have the WBGB technology too.

Cheers,
Aziz

[Davor]

Considering op amps of the era, it must be a bit of all these reasons. Even today op amps with 15nV/sqrt(Hz) are advertised as low noise, say TL071, and it's input referenced noise is equivalent to ~15k resistor. With that in mind a 13k in inverting branch would not spoil anything. Low gain may be related to poor THD performance. Point is that you are simultaneously amplifying target and unbalanced Tx residual (so called air signal) with several orders of magnitude between them. THD becomes increasingly worse with output amplitude, so there you have it.

There are ways around these problems, and there are constraints. IMHO the biggest constraint is a stock coil that dictates unbalanced mode of operation. You simply can't apply a balanced Tx on a plain vanilla stock coil to cancel the "air signal" because one coil side is grounded. Just by cancelling the "air signal" you'd have a rig that performs immensely better.

[ODM]

Again: Was the particular IC type listed somewhere? I'd take that into account as a design point before saying the passives around it are the culprit.

Was there bandwidth to be spared with a "low" gain, phase response requirements, or noise intermodulation being kept at bay?

[Carl-NC]

Maybe, maybe not... redesign the C-scope preamp with a high gain, you still probably won't see any overall performance increase. I haven't done this with the C-scope, but I have done it with other detectors, and generally found no real improvement. Sometimes even deprovement, as a high gain can screw things up.

There are many ways to skin a kitten, and unless you are already close to the UD (as Moodz says) sometimes it's just over-engineering.

[mikebg]

I suspect if you take a C-scope 1220B and replace R28=100, R29=769, C6=13u, and C7=8.45n you will not get any performance increase at all. You'll just have a hard time finding the caps.

Carl, why you think that I will redesign the preamp of C-Scope 1220 for low noise in this way and will remain its gain 7.7?

When you calculated instead me the new values of components for preamp, you're not paying attention to point 4 in my post:

"4. Every competent designer knows that sensitivity of a metal detector depends on gain of preamp."

Resistance 13k for R28 is bad design not only for low noise, but for sensitivity. For example, Coinmaster has a competent designed preamp with high gain.

I posted in the forum examples how should be designed low noise preamp for RX:

http://www.geotech1.com/forums/showt...357#post116357.

Resistance of R28 should be chosen equal to or less than the equivalent noise resistance of the integrated circuit, which is about 1kohm. The IC generates much more noise than the RX coil.

Although I posted in the forum examples for low noise design of preamp, I have in several posts emphasized that due to imperfect block diagram (principle) of the conventional metal detector, its sensitivity is not limited by EMI and internal noise of components, but by existing AIR & GND signal in input.

[Davor]

We have a lot of EMI noise, which dominates over the resistor noise in some cases. The anti-interference configuration however can reduce this EMI noise pickup by 20 dB - 30 dB (or even more).

There is a noticeable hump of 20dB or so in the VLF region against the LF. This may be attributed to ionospheric activity, and mostly of horizontal polarisation, hence perfectly picked by our coils.
The other point is that our coils' resistivity dictates resistor noise well below 1nV/sqrt(Hz), but the front ends are not that good. I'd expect noticeable improvement using anti EMI coil configurations even at moderate suppressions of ~20dB or so if you push input referenced noise to about 1nV/sqrt(Hz).
Guess why I'm pushing the Davor-Sergey coil configuration so hard.

BTW, I have solid experience using NEC2. Too bad it is not so good for constructions far smaller than the wavelength.