What Is The Technical Difference Between Discrimination And Notch?

[iron_buzz]

20 hours ago, CPT_GhostLight said:

 

And since everything in the digital world affects time cycles and bytes

 

That is not necessarily true, though, in all cases.  For example, I hear a lot of people claiming that the higher you set your discrimination, the more you limit depth, but to me, as a retired computer programmer, I would assume that the discrim level you set is simply a parameter passed to a function.

[iron_buzz]

1 hour ago, midalake said:

There are a few people that think running the notch rather than discrimination to your set point on a D2 has some secret sauce.
For instance, I start my Disc at 22 for beach hunting. Some seem to think that notching to 22 is better?  I don't know. Never experimented with it much but going to do some testing this year.   

Yeah, that is a major part of my reason for asking about this.  I don't know, but I do suspect, that a lot of the things we hear about discrimination may be holdovers from the days of analog processing.  I did some testing on my own to see if discrimination affected depth, and I was satisfied that it does not. But when I posted that finding to my local MD Facebook group, I got blasted with claims to the opposite, but not one who actually tested it themselves.

[midalake]

9 minutes ago, iron_buzz said:

did some testing on my own to see if discrimination affected depth, and I was satisfied that it does not.

You are 100% correct. The level of depth is not affected for the targets out of discrimination on a D2.  Although not tested I am confident depth not affected by level of notch either. 

[Geotech]

6 hours ago, UKD2User said:

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

Yes, both magnitude and phase are calculated from the response vector, but only phase is used in disc and notch. Magnitude is used for the depth meter and audio loudness.

 

1 hour ago, iron_buzz said:

For example, I hear a lot of people claiming that the higher you set your discrimination, the more you limit depth, but to me, as a retired computer programmer, I would assume that the discrim level you set is simply a parameter passed to a function.

That's one of those holdovers from the analog days. Those people are probably 60 or older. Back then, discrimination was set by rotating the demodulator phase until a given target started producing a negative response. It was then discriminated. Problem is, conductivities that were progressively closer to the disc point were progressively weakened. Newer designs do it digitally without altering any responses.

[Sinclair]

Depth is not directly affected.. some people might perceive it that way, because very deep targets, at the edge of the detectors max depth, often get IDed lower and thus fall into the low numbers ferrous range, which you - depending on your settings - won't hear anymore. Notch wouldn't do better in that situation, though.

Personally I almost newer use notch. I like to hear the iron if i want to.

[iron_buzz]

19 hours ago, Geotech said:

Yes, both magnitude and phase are calculated from the response vector, but only phase is used in disc and notch. Magnitude is used for the depth meter and audio loudness.

 

 

Which brings me to another, related question... Gold and Relic in the D II use Iron Amplitude Rejection instead of tradititional discrimination (although notching is still available, which I use often with those programs to make them useable).  The idea, if I understand it right, is to reject ferrous targets with the highest magnitudes.  I have wondered it it would also be possible to apply that same logic to non-ferrous targets for detecting in places where you want to ignore shallow coins and pulltabs and such so you can focus on the deep targets.  Of course, it would have to be used with the caveat that you could be missing important stuff on the surface but there are situations where that might be OK.   Is my thinking sound?  Could and should this be done?

[Chase Goldman]

8 hours ago, iron_buzz said:

Which brings me to another, related question... Gold and Relic in the D II use Iron Amplitude Rejection instead of tradititional discrimination (although notching is still available, which I use often with those programs to make them useable).  The idea, if I understand it right, is to reject ferrous targets with the highest magnitudes.  I have wondered it it would also be possible to apply that same logic to non-ferrous targets for detecting in places where you want to ignore shallow coins and pulltabs and such so you can focus on the deep targets.  Of course, it would have to be used with the caveat that you could be missing important stuff on the surface but there are situations where that might be OK.   Is my thinking sound?  Could and should this be done?

The short answer to your question is:  Probably not.

TL/DR version:

I am speculating here, because I don't have insider knowledge of XP's implementation of IAR but I can make some semi-educated guesses of what's going on.  My understanding is that what IAR is doing is differentiating the rate of the target's amplitude change (i.e., how fast it reaches its maximum amplitude) vs. the reactive (X) based phase change (which is what is used to determine TID) to separate ferrous from non-ferrous targets.  Put another way, but greatly oversimplified, ferrous target signal amplitudes peak at a different rate than non-ferrous targets and you can use that characteristic to ID ferrous beyond the phase shift TID method that is typically used in discrimination or more sophisticated notch implementations. So, my limited understanding (which could be mistaken) is that it is not based on the max amplitude but how quickly it reaches max amplitude.  Geotech can confirm or refute whether this is a real thing but I believe this characteristic is also used by pulse induction detectors to crudely differentiate ferrous from non-ferrous targets.  On PI detectors it can be fooled by strong shallow targets that result in a strong amplitude response and it also can be fooled by highly mineralized soil conditions, because of the ferrous composition of the mineralization (magnetite or maghemite).  On the Deus/Deus 2, I have not seen IAR be fooled by strong shallow targets but I have certainly seen it struggle in highly mineralized ground conditions, especially in the Deus 1 implementation.  It seems that XP has refined IAR significantly in Deus 2, but it is still not as reliable as traditional discrimination and it is essential that Notch 00 is used in conjunction with it in Relic and Gold Field modes to eliminate ground feedback at 0 TID and below (which is also explicitly noted by XP in the Deus 2 manual snippet below that describes notch).  Non-ferromagnetic metals do not exhibit significant temporal (time varying) peak magnitude effects in a manner that would enable differentiating between various non-ferrous metals, so it cannot be used other than as a means to crudely differentiate ferrous from non-ferrous, from what I can tell. 

So that just leaves disc/notch for undesirable non-ferrous targets and the caveats associated with using those filters as desirable targets will invariably be occasionally filtered out as well. 

I have recently detected some sites heavy in aluminum trash, and the only other tool I have found to have some use is the AUDIO FILTER setting.  Setting AF to high settings tends to give aluminum and other light junk metals a kind of "brassy", distorted audio, especially when full tones audio is used and the effect varies when using PCM vs. Square vs. Hi-Square audio, as well.  I tend to interrogate a suspected aluminum target by switching to a custom full tones program that I have set up adjacent to my preferred main search mode (which is either Relic or General, depending on site conditions and target objectives).  But it usually just sets my expectations, accordingly (i.e., low), but I seldom use it solely to make a dig, no-dig decision because, it is also far from foolproof.

Most of the above is empirical field experience rather that theoretical.  I tend to try to not overthink it too much and just get out there and dig all the non-ferrous and much of the iffy ferrous to get a feel for the type of junk at a particular site and until I get tired or worn down and then I will start to cheat with the filters and audio tools.

I haven't really seen anyone refer to the XP Deus 2 manual in this thread, so I thought it would be helpful to use XP's own words against them.  

Below are excerpts from the Deus 2 manual describing the various filters like Discrimination, TID, B Caps, Notch, Multi-Notch, Silencer, and IAR, its limitations, how it is designed to either provide awareness of small shallow ferrous or deep big ferrous (though I have not really seen it do this in practice, I typically just set IAR at 5 as it gives the highest "grunt" response on ferrous regardless of depth or footprint - and run Relic as basically a "two-tone" ferrous/non-ferrous setup with Target ID).

A couple things to note in the snippets below:

XP describes Notch as a supplement to Discrimination.  The reason being is that Discrimination utilizes Iron Volume and if you use notch instead in the lower TID regions, you basically "lose" a tone region if you don't utilize any disc when using multi-tones.  You can use this to your advantage, but if you are unaware of this distinction, it can also be a downside.

Notice how XP describes the 23 - 24 notch on the Sensitive, Sensi FT, and Fast programs as a means to reject soil "wet zones".  I always thought is was for foil.

XP says IAR can result in rejection of lower conductivity desirable targets such as gold nuggets, so use it with caution and also notes that notch can be used in conjunction with IAR.

 

[Steve Herschbach]

On 11/9/2023 at 8:53 AM, Geotech said:

Yes, both magnitude and phase are calculated from the response vector, but only phase is used in disc and notch. Magnitude is used for the depth meter and audio loudness.

 

That's one of those holdovers from the analog days. Those people are probably 60 or older. Back then, discrimination was set by rotating the demodulator phase until a given target started producing a negative response. It was then discriminated. Problem is, conductivities that were progressively closer to the disc point were progressively weakened. Newer designs do it digitally without altering any responses.

As one of those old timers let me clarify that. It is not so much depth that is limited but the ability to detect the target, which can be perceived as the same thing.

Is there anyone here that does not get jumpy or dancing target id numbers on weak signals, especially those in highly mineralized ground? The problem is that targets often do not have nice tight distinct single number responses. The number varies with how you swing over the target. In effect a target produces a range of possible responses.

By limiting those responses you limit the ability to detect the target. Let's say you read that a nickel is 13 on an Equinox. So you block out everything but 13. Well, you go past those oddball nickels that read other that 13 on the first swing. Whether you were limiting the depth on those targets or limiting the ability to find them at all is a fine distinction. Any notch on any number can inadvertently block a desirable signal.

One of the finer points of a detector with a wide target spread like the White's V3i was that you could put a notch of a single number on a certain pull tab response. It would not eliminate the tab but cause it to break up, letting you know it was probably a tab with a trashy response while minimizing the overflow/slop dangers of applying, for instance, and range of three or four blocked numbers to cleanly reject the target.

Again with most detectors this is most commonly seen with the zinc penny. As it corrodes it produces a wide range of responses. To get rid of all those pesky pennies you have to notch out a range. But that same range can include Indian Head pennies or jewelry items, so how aggressive do you want to get? 

Every employment of discrimination is nothing more than a gamble, a way to change the odds when dealing with limited time. But rest assured that in mineralized ground almost all targets can occur in such a way as to read ferrous or to shift target id into some other trash reading. This is just gold nugget detecting 101 but it applies as much to every type of detecting there is. Overthinking discrimination is fraught with danger because it is at best a dull tool. So this old timer says not that the more discrimination you employ the more you limit depth, but that the more discrimination you employ to more you increase your odds for leaving good targets in the ground.

I rarely ever notch or outright reject anything, preferring to use full tones and simply dig targets that sound good. I'm a fairly proficient cherry picker with a PI detector just by reading the tonal responses. IAR is a way to try and replicate in electronics what a person like me is doing with their ears. Most round targets and most non-ferrous targets produce different signals than most ferrous targets. Not all by a long shot, but enough to sway the odds in your favor.

[Chase Goldman]

51 minutes ago, Steve Herschbach said:

I'm a fairly proficient cherry picker with a PI detector just by reading the tonal responses. IAR is a way to try and replicate in electronics what a person like me is doing with their ears. Most round targets and most non-ferrous targets produce different signals than most ferrous targets. Not all by a long shot, but enough to sway the odds in your favor.

Exactly.  To piggy back on this; non-ferrous, symmetric targets at depth usually (but there are always exceptions) change tonal pitch in a smooth manner whereas small and large non-ferrous change pitch in a more abrupt or "unstable" manner even if the pitch change amplitude is the same.  A smooth valley or hill vs. a more abrupt pit or pinnacle.  That is our brains interpreting the different time varying pitch and amplitude changes and rates.  Using the Axiom which uses "on demand" ferrous checks vs. continuous ferrous blanking.  I can pretty much tell by the abruptness of the target signal audio changes in pitch and volume amplitude whether I am likely over ferrous and engaging the ferrous check just sort of provides a confirmation.  Then I still need to make a dig decision because none of this is 100% foolproof and factors such as the local junk you have previously dug, the junk density factor, your fatigue factor, and hunches that will come into play.  This is all integrated in your head in real time, and if you have done it long it enough it becomes second nature without you realizing you are doing it.  You just do it.

[Geotech]

5 hours ago, iron_buzz said:

I have wondered it it would also be possible to apply that same logic to non-ferrous targets for detecting in places where you want to ignore shallow coins and pulltabs and such so you can focus on the deep targets.

This is called "surface blanking." It has been featured on a few detectors but I don't recall the specific models. Garrett did it years ago, and I think either Compass or Discovery as well.