Five Frequency Times Eight

[Chase Goldman]

1 hour ago, GrievousAngel said:

I was relating to the -3db frequency point of a given target (assuming all other being equal, e.g., shape, orientation, size, conduction value, etc. ). As I understand it, the target's reflected signal is maximum at the ~ -3db frequency level. This is assuming the transmitted frequency is also set to the target's frequency -3db level point. This happens to be the r-component of the signal, which is defined at it's  half-power point.

The terminology you are using implies to me that you seem to be looking at this like radar.  RF being reflected off metal.  That's not what is going on here.  It is induction balance.  You need to be thinking transformer.  In very simplistic terms, the transmit coil magnetic field  induces eddy currents in the target  (and ground) that are then magnetically coupled to the receive coil.  The resulting phase shift is indicative of the electromagnetic properties of the target.  Not sure why you are repeatedly referencing the half power (-3db) "cutoff" frequency.  I must be missing something.  HTH

[GrievousAngel]

On 1/16/2022 at 3:24 PM, PimentoUK said:

Getting complicated, now.

Detectors don't just measure the reactive component of a target. If they did, a 13kHz machine ( decent all-rounder Fisher F75, for example ) would be hopeless at finding 1 kHz targets ( US silver dollars / half-dollars ), as the phase lag is about 86 degrees, with the reactive component way less ( 7% ) than the resistive component.

"I would tend to think this has a lot to do with Multi-IQ design and resulting algorithms"

Now you're bringing multiple freqs into it ... and it sounds like you don't really understand the why's of multi-freq.

Using multiple frequencies has very little to do with "hitting the target with a range of freqs hoping that one will hit the spot". It's primarily about working out the ground signal, so it can be largely eliminated, thus making the target more visible.

"Now you're bringing multiple freqs into it ... and it sounds like you don't really understand the why's of multi-freq." by PimentoUK.

Maybe so, but please refer to this article: Selectable Frequency and Multiple Frequency from this site. In which part is taken from an article by George Payne. See this link: http://jb-ms.com/Baron/payne.htm (2002).

My comments was taken directly from these two articles. The -3db point yields max returned signal of a given center frequency. I then became interested and wondering if the -3db half-power data was somehow used in the different weighting algorisms used in Minelabs Multi-IQ. I would guess, very likely.

I am hoping to get comments on this specific concept. Maybe I put this thread in the wrong forum.

Thanks, Billy

[Chase Goldman]

18 minutes ago, GrievousAngel said:

"Now you're bringing multiple freqs into it ... and it sounds like you don't really understand the why's of multi-freq." by PimentoUK.

Maybe so, but please refer to this article: Selectable Frequency and Multiple Frequency from this site. In which part is taken from an article by George Payne. See this link: http://jb-ms.com/Baron/payne.htm (2002).

My comments was taken directly from these two articles. The -3db point yields max returned signal of a given center frequency. I then became interested and wondering if the -3db half-power data was somehow used in the different weighting algorisms used in Minelabs Multi-IQ. I would guess, very likely.

I am hoping get comments on specific concept. Maybe I put this thread in the wrong forum.

Thanks, Billy

Refer back to the other thread where you posted a similar question.

[GrievousAngel]

On 1/16/2022 at 3:24 PM, PimentoUK said:

Getting complicated, now.

Detectors don't just measure the reactive component of a target. If they did, a 13kHz machine ( decent all-rounder Fisher F75, for example ) would be hopeless at finding 1 kHz targets ( US silver dollars / half-dollars ), as the phase lag is about 86 degrees, with the reactive component way less ( 7% ) than the resistive component.

"I would tend to think this has a lot to do with Multi-IQ design and resulting algorithms"

Now you're bringing multiple freqs into it ... and it sounds like you don't really understand the why's of multi-freq.

Using multiple frequencies has very little to do with "hitting the target with a range of freqs hoping that one will hit the spot". It's primarily about working out the ground signal, so it can be largely eliminated, thus making the target more visible.

Detectors don't just measure the reactive component of a target. If they did, a 13kHz machine ( decent all-rounder Fisher F75, for example ) would be hopeless at finding 1 kHz targets ( US silver dollars / half-dollars ), as the phase lag is about 86 degrees, with the reactive component way less ( 7% ) than the resistive component.

Of course they don't. Returned data is in the form of X (inductance-like current momentum) and R (inversely related to conductivity). Of course that is a simplified analogy.

I am not disagreeing with all you have said. You are correct. Billy

[GrievousAngel]

On 1/16/2022 at 3:51 PM, Chase Goldman said:

The terminology you are using implies to me that you seem to be looking at this like radar.  RF being reflected off metal.  That's not what is going on here.  It is induction balance.  You need to be thinking transformer.  In very simplistic terms, the transmit coil magnetic field  induces eddy currents in the target  (and ground) that are then magnetically coupled to the receive coil.  The resulting phase shift is indicative of the electromagnetic properties of the target.  Not sure why you are repeatedly referencing the half power (-3db) "cutoff" frequency.  I must be missing something.  HTH

Actually it is similar to radar, but the transmitted magnetic waves cutting throw the target's eddy currents is what produces the returned 'radar-like' signals. I think I said that correctly. Billy

[Chase Goldman]

4 minutes ago, GrievousAngel said:

Actually it is similar to radar, but the transmitted magnetic waves cutting throw the target's eddy currents is what produces the returned 'radar-like' signals. I think I said that correctly. Billy

Actually, it's two completely different principles with perhaps a few similarities.  In the case of the metal detector, it is near field magnetic coupling using the ground/air and target as the core of a transformer comprised of the transmit (primary) and receive (secondary) windings of the transformer.  In the case of radar, you have far field Radio Frequency electric field transmission and reflections off the target.  But it's really not worth arguing over the down in the weeds details other than to note that there are some relevant differences that factor into how you design a metal detector vs. a radar transmitter-receiver.

[PimentoUK]

Quote:

"The -3db point yields max returned signal of a given center frequency"

No it doesn't. It's just the reactive element that is greatest. Using my prior example of the F75 ( 13 kHz machine ) detecting the silver Dollar: The coin in fact gives a very strong signal to the detector, as you would expect a large lump of metal to do. It's just that the phase lag is very much towards the '90 degree' region, as the detector freq is massively higher than the target freq. So it's not hard to detect, if you look at the correct signal(s).
What affects the targets 'detectability' is the processing the machine uses to determine what's there. The F75 has two different modes: 'DE' = default, and 'JE' = jewelry. Fisher don't spill any beans about what the difference is, of course. My hunch is the JE mode does full vector addition of the I and Q signals to determine target strength,

ie. (signal = I2 + Q2) , whereas DE mode does a linear addition, (a I + b Q ) , which is probably similar to what classic analogue machines do.
So you can now see that if you have multiple frequencies ( ie. 2 or more )  to interrogate the target & ground, you have plenty of ways of analysing each freq and combining them in a way that produces a target/no target outcome.

As you have a specific interest in Minelab's 'Multi- IQ' techniques, it would be useful for you to understand some basics of what the machine actually does.
Cut 'n' pasted from other posts of mine:

Both the 600/800 models use 7.8kHz, 18.2kHz and 39 kHz in Multi-IQ .
Beach modes have an additional 13kHz in the mix.

The detector transmits a complex square-edged waveform, that repeats every 385 microseconds. In that waveform are 15 cycles of 39kHz, 7 cycles of 18.2kHz and 3 cycles of 7.8kHz. That is how the operating freqs are related to each other : 7.8k : 18.2k : 39k are ratios of 3 : 7 : 15.
Beach modes ( and apparently all Vanquish's ) use 4 frequencies, 7.8kHz, 13kHz, 18.2kHz, 39kHz in a 3 : 5 : 7 : 15 ratio.

The single-frequency options do exactly what is claimed, they are 5 / 10 / 15 / 20 / 40kHz.

My personal opinion is that:
Park1 and Field1 place the emphasis on the 7.8 kHz, and use the other two freqs to help reduce ground signal.
Park2 and Field2 place the emphasis on the 18.2 kHz, and use the other two freqs to help reduce ground signal.
Gold1/2 place the emphasis on the 18.2 kHz and 39 kHz , using the 7.8 kHz to reduce ground signal. (I have a Eqx600, so have no experience of what the Gold modes do differently )

In addition, I speculate that the 'intelligent' technique ML are using is this:
Example: In Park1 , they are also still analysing like Park2 and Gold2, as a 'background' process. If a target gives sufficient response at these medium / high weighted freqs, it will trigger an audio response. So you would still get 'hits' from some small low-ID targets, that would otherwise have been ignored if the machine had been running as a 'pure' 7.8 kHz ground-compensated machine.
Ditto, gold modes would likely 'hit' some larger/high-conductor targets that 39 kHz on its own wouldn't be so hot on.

As for the 'Beach modes':
It's known the transmitted freq mix is 7.8 kHz / 13 kHz / 18.2 kHz / 39 kHz for both modes.
The processing of these signals is going to be firmly aimed at reducing the salt-water dependant ground signal.

[Chase Goldman]

21 minutes ago, PimentoUK said:

No it doesn't. It's just the reactive element that is greatest.

Actually, according to Billy's reference (using X-R components), it's the frequency at which the resistive R component of the target signal that is greatest which also corresponds to the reactive X component being at it's half power (-3dB) point.  Just quoting the reference here, so don't kill the messenger.  I have yet to read a reference where this is clearly explained without contradictory or ambiguous statements and terminology.  In fact, in the quoted passage from the Payne reference which appears in this post, Payne appears to confuse the relationship between optimal target frequency (max R) and target conductivity stating the relationship exactly opposite of what we know is true.   Specifically, that higher conductors have a lower optimal frequency and lower conductors have a higher optimal frequency (Payne states the opposite, but provides the correct relationship in the his examples of specific target behavior).  Anyway...there you go.  I tend to agree with your assessment of how SMF detectors utilize the different transmitted frequencies to extract relevant target information, bias to specific target types, and to provide ground and salt cancellation and handling.  The secret sauce being the signal processing, not which and how many frequencies are pumped into the ground.

[Geotech]

6 hours ago, Chase Goldman said:

Actually, according to Billy's reference (using X-R components), it's the frequency at which the resistive R component of the target signal that is greatest which also corresponds to the reactive X component being at it's half power (-3dB) point.

This is correct. As you go up in frequency the R response increases, then decreases. The X response always increases. And the composite vector response always increases.

The reason the R response matters is because in VLF detectors the R response is what we look at to determine when a target is under the coil. When the detector is ground balanced the R channel has no ground signal at all but the X channel does. A target shows up in both the R channel and the X channel but it is the R channel that is target-only.

High conductors tend to have a fairly weak R signal at higher frequencies but a very strong X signal. If you have really good differentiators then you can probably get enough ground suppression in the X channel to separate out big silver. Back when VLF was first introduced there was only one demodulator. In TR-Disc mode you would tune the demod phase to discriminate. In VLF mode you would tune it to ground balance. There was no separate X channel which is why the early VLFs ran at such low frequencies: 1-3kHz. It was the only way to get the R response of silver out of the mud.

 

[Skull diver]

I'm glad to read this post, cause I'm maybe not far from an answer that I'm looking since a lot of time.

Apparently, and maybe I'm wrong, multiple times on my Ctx when diving, I feel the machine unstable despite of a lower sensitivity or any setting included saltwater filter.

I'm always in doubt, as a jewelry hunter, if a mono 18 to 24 Khz frequency can be better in salt underwater use, instead of a forced multifrequency range.

I think, compared to a 70Khz machine for nuggets, a medium, maybe slightly over 17Khz frequency can do a good compromise in terms of depth and sensitivity to gold.

In few words...Can a range of useless for the purpose frequencies, cause the erratic conduct of the machine?