What Is The Technical Difference Between Discrimination And Notch?

[iron_buzz]

I do understand that (at least in the Deus world) discriminated targets still gives an iron sound, and notch is silent.  Also that you can have multiple ranges of notching and that they don't have to start at (or below) 0.  I've got that.  But internally, programatically/technically why do we have both?  Does one affect depth or speed of processing more than the other?  Any other differences?

[Sinclair]

6 hours ago, iron_buzz said:

discriminated targets still gives an iron sound, and notch is silent

That's not necessarily the case, you can disable the Iron Audio completely. Notch silences audio for specific target IDs - no matter what's under the coil. Discrimination is intended to be used to separate between good/bad targets -> declare till which ID it's most likely ferrous/non-ferrous. It's more like a baseline.

Lets say you have a target composed of ferrous/non-ferrous material - with notch you would only hear the "good" non-ferrous part - with discrimination and Iron Audio on, you would hear both the ferrous and non-ferrous, no matter which target ID the non-ferrous part has.

Notch can help you to cherry-pick multiple specific target IDs independent of your discrimination settings (as long as your discrimination is set lower than the first desired good target ID of course..).

Afaik some filters use the discrimination range as a "bin" to distinuish between a good or bad target (B.Caps for example).

[iron_buzz]

5 hours ago, Sinclair said:

That's not necessarily the case, you can disable the Iron Audio completely. Notch silences audio for specific target IDs - no matter what's under the coil. Discrimination is intended to be used to separate between good/bad targets -> declare till which ID it's most likely ferrous/non-ferrous. It's more like a baseline.

Lets say you have a target composed of ferrous/non-ferrous material - with notch you would only hear the "good" non-ferrous part - with discrimination and Iron Audio on, you would hear both the ferrous and non-ferrous, no matter which target ID the non-ferrous part has.

Notch can help you to cherry-pick multiple specific target IDs independent of your discrimination settings (as long as your discrimination is set lower than the first desired good target ID of course..).

Afaik some filters use the discrimination range as a "bin" to distinuish between a good or bad target (B.Caps for example).

Yes, Iron audio can be reduced to zero.  But while I apprecite you taking the time to type out a thoughful reply, you are talking about how to use each.  I'm asking about the internal differences, not the functional/practical ones.  What, to a metal detector engineer or programmer is the difference?

[UKD2User]

1 hour ago, iron_buzz said:

Yes, Iron audio can be reduced to zero.  But while I apprecite you taking the time to type out a thoughful reply, you are talking about how to use each.  I'm asking about the internal differences, not the functional/practical ones.  What, to a metal detector engineer or programmer is the difference?

The truth is, nobody (outside those who need to know) knows - because the code is proprietary, and is protected by encryption.  There is clearly processing which creates a sound-based representation of a target, and separate (but obviously closely related) processing to generate as reliable a visual TID as possible. These processes may well go on in parallel, although the visual TID seems to be time-averaged, and therefore, later than the audible representation.

Notching and disc are both, on the face of it, processes which require a TID, of some kind - probably not the same as the one displayed visually - to have been calculated first - but I'm sure it's more complicated than that!

[Geotech]

Internally, there is no difference. Notch is just a narrow band of discrimination, and all discrimination is based on target phase. And tone ID works the same way, where tones are assigned based on the phase response. With disc & notch, you just assign it "no tone at all" or, maybe, assign it a grunt sound.

[CPT_GhostLight]

Great question, iron_buzz, I've been curious about this too. So if I understand Geotech correctly, Disc on the D2 examines the target phase and assigns anything lower than the max disc setting as ground or iron phase? And the 3 notches on the D2 can of course be set very narrow or very wide to bin certain targets phases.

And since everything in the digital world affects time cycles and bytes, it would be logical to surmise that the more of any kind of processing will cost more memory and processing usage. Of course with the super power of small processors and memory circuits these days, I would wonder at what point does using more processing affect performance?

I feel like I can use quite a bit of extra filters, notch and disc, etc without causing major performance degradation, but I seldom use much filtering out of an abundance of caution.

[Geotech]

In the old days, the response of any target that was discriminated (whether using a linear disc or notch) was just blanked out. That is, if you were listening to a threshold audio, it would go silent over a disc'd target. In digital detectors, you can do whatever you want; make it grunt like a pig, or give it a Price-Is-Right "loser jingle."

The processing price of discrimination is negligible. A detector can reasonably resolve target phases to around 1°, and you could easily notch targets down to that resolution. That is, you could have 180 notch settings and control each one, even giving each one a different tone if you like. The White's V3 did this.

[Skull diver]

Passionate speech ?

[UKD2User]

18 hours ago, Geotech said:

Internally, there is no difference. Notch is just a narrow band of discrimination, and all discrimination is based on target phase...

And magnitude of course (it's a vector, not scalar, quantity).