Discrimination vs. Notching

[Dave Rishar]

So, I've been led to believe that discrimination is to be avoided when possible due to a loss of depth, but that notching doesn't seem to affect depth.

Say I wanted a program that knocks out basically everything at 35 or so and below. I've seen a few programs like this, but they invariably retain the discrimination of 10 from the factory programs and notch out the rest. I understand that just a bit of discrimination is desirable (to keep the horseshoe telling the difference between ferrous and conductive if nothing else) but beyond this, what is the advantage of running the descrimination that high? Would it not be more effective to set discrimination at 3 or so and cover the rest with notching?

I feel like I'm missing something obvious here but for the life of me, I can't tell what.

 

Amazon Forum Fav 👍

Upgraded Night Vision Monocular Digital Infrared Monocular Telescope for 100% Darkness with 8X Zoom - Grab it through Amazon!

[cooper1841]

I cant wait to here this!!! I have been wondering the same thing. Are they not doing the same exact thing??

So, I've been led to believe that discrimination is to be avoided when possible due to a loss of depth, but that notching doesn't seem to affect depth.

Say I wanted a program that knocks out basically everything at 35 or so and below. I've seen a few programs like this, but they invariably retain the discrimination of 10 from the factory programs and notch out the rest. I understand that just a bit of discrimination is desirable (to keep the horseshoe telling the difference between ferrous and conductive if nothing else) but beyond this, what is the advantage of running the descrimination that high? Would it not be more effective to set discrimination at 3 or so and cover the rest with notching?

I feel like I'm missing something obvious here but for the life of me, I can't tell what.

 

[DDancer]

Someone will correct me if I'm wrong or a bit off.

Discrimination is an offset to the detector that alters the machiens sensitivity to low end, ie ferrous, responses to targets. It can be considered like an offset to your ground balance *though its not*. Discrimination does lower the detectors depth somewhat because you are simply masking out low end reactive targets... kinda like a chunk of junk masks a good target.
I mentioned ground balance as a reference because manually ground balancing a machine above or below true GB is also a form of discrimination as you are also adjusting the machiens sensitivity to the local soil type. This can be beneficial but takes time to get used too.

Notch discrimination affects the audio portion of the circuit. It does not put an offset in the machiens sensitivity, hence no affect on depth, but allows the machiens audio analyzer to pass or reject audio based on what you select not to listen to. And in the case of many newer detectors it will also leave things like VDI displays blank or intermitant even if there is a target.

My thoughts anyrate.

 

[U.K. Brian]

Exactly the same thing but a notch is selective and narrow. Regretably both reduce the performance of any detector so the old rule of using the minimum amount of discrim./notch still applies.

 

[parkgt]

Exactly the same thing but a notch is selective and narrow. Regretably both reduce the performance of any detector so the old rule of using the minimum amount of discrim./notch still applies.

Brian,

You don't list having a XP Deus, so are your comments about it just a guess or are they based on some experience with one?

Several creditable sources with many hundreds of hours on the Deus disagree with you and support the position but forth by Dancer that notch is only a audio function and has no effect on actual detection.

The CTX also has audio selectivity that does not effect detection.

 

[timpsu]

I copy here something that has been seen at our finnish metal detecting forum (by "Etsijä82") aka "Finder82" and also at an australian forum. I have this program saved but I haven't used it yet.... Ground balance index was 77 when this test was conducted so the soil was pretty mineralized.

Translated this test to English from an Austrian forum.
Pretty remarkable to see the effect discrimination has on depth.
Thought you might be interested in the results.
The test was done according to Ingo standards.
________________________________________

Today, we were at the "borehole" on our testsite with 2 Deus for another try-out according to the INGO standard, as a follow-up of the last test.
The objective was to record what effect setting the individual of parameters in the menu have on depthperformance.
Testpiece was (again) the 2 RM-piece (1939 Paul v. Hindenburg) and as my base setting I chose my personal program "FOREST" comprising of the following parameters:

- Disk 0
- 4 Tones, but effectively 3 Tones. T1&2 200Hz., T3 317Hz., T4 791Hz. Threshold T2/3 22, T3/4 80.
- Sens 95, Power 3
- 12 Khz.
- Reactivity 0
- Audio Response 2
- Silencer 1
- G.B Manual 80, following the measurement by the DEUS.

Test:

Using the above mentioned values, the 2RM piece was placed in the ground so that it could just be detected. This was at 23 cm ( the ground was bonedry and the coil not fully charged)
Now every individual parameter was changed and the resulting depthperformance recorded. You should read the results as follows: on the left the setting, on the right the performance in cm's
With depthperformance I mean: a usable signal you would dig.

We started of with the Sensivity.

Sensitivity:

80 / 18cm
85 / 20cm
90 / 21cm
95 / 23cm
98 / 23cm

Frequency:

4 Khz / 23cm
8 Khz / 22,5cm
12 Khz / 23cm
18 Khz / 22cm


Now instead of the big 2RM piece, we tried a silver Hohlpfennig weighing 0,12gr, because frequency should have a noticable effect on the small finds capability of a detector.

4 Khz / 7cm
12 Khz / 12cm
18 Khz / 12,5cm

After which we proceeded again with the 2RM piece and with Recoveryspeed.

Recovery:

0 / 23cm
1 / 23cm
2 / 23,5cm
3 / 22cm
4 / 20cm
5 / 19cm

Audio Response:

0 /23cm
1 / 23cm
2 / 23cm
3 / 23cm
4 / 23cm
5 / 24,5cm

Discrimination:

0 / 23cm
1 / 20cm
2 / 19,5cm
3 / 17,5cm
4 / 17,5cm
10 / 16,5cm
20 / 16,5cm
40 / 16,5cm
60 / 16,5cm

Ground (Deus itself measured a value of 7Cool:

77 / 25cm (but it was totally impossible to detect at 77, due to an incredible amount of false signals)
78 / 24cm
80 / 23cm
85 / 22,5cm
90 / 22cm

Unfortunately I totally forgot about the Silencer settings, but I will test this soon.
I couldn't change the Powersetting, despite the 12 kHz I couldn't change it in the Expertmode.
Obviously the Powersetting is coupled not only with the 4 kHz, but als with some other parameter.
I haven't found out which other setting is blocking the Power parameter in the Expertmode.
May be another Deus user can help me out.

Edited: 30.05.2011:

PS: - just tested the effects of Silencer- and Powersettings

With a freshly charged coil I can now detect the 2RM piece at 26 cm - so it pays off to have a fully charged coil - yesterday with Audioresponse at 5
in program FOREST I could only detect it at 24,5cm.

Power:

I tried all POWER settings more than once, but I can't, for the life of me, find any difference in depthperformance!
This surprises me greatly, I have taken my time with this test - the difference is truly insignificant.
I will use setting "1" in the future to save on batterylife and ask myself: why is it made adjustable at all?

Silencer:

The Silencer setting also doesn't have any noticeable effect on depthperformance - whether set in -1 or 4, performance is the same.
Setting 4 resulted in a less crisp signal and is therefore better not used.
Interesting would be to know if the Silencer has any effect on Recoveryspeed..


I translated it for Rogers relics forum and here's some more:
________________________________________
This Dierk's a smart un because he continues to do testing and shows that with discrimination set a 0 you get 4 inches
more depth on some coins than in standard GMP setting.
So what he does is make a setting with discrimination set at 0 and 4 Tone.

The first tone is set at 200hz (lowest tone possible)
The second also at 200 hz (lowest tone possible)
The third at 480hz (somewhere in the middle range)
The fourth at 791hz (highest tone possible)

Why does he set the first two tones at 200hz?

Because now he can set the first audiothreshold at 0 iso 6.8 or 4.6 or 10 which would automatically set the discrimination at 6.8 or 4.6 or 10 and would give you loss in depth.

The 2T/3T threshold is set at 23, which gives an iron tone to any target below 23
The 3T/4T threshold is set at 80, to seperate silver targets from copper ones.

So now you end up with an all metal mode with up to 4 inches more depth than GMP.

How does this work now?

You have an all metal mode with visual ID and Audio ID, so it sounds out on every metal target, but seperates the audio in
low tones: i.e. Iron, middle tones: Copper etc,and hight tones: Silver etc.


Not a good setting for Iron infested sites, but if the ground is clean and you want maximum depth with some sort of audio ID it's a really good idea.


Now where you put the thresholds is up to personal preference of course because local targets can be different, but I think its really smart how he got a setting with maximum depth possible and still retain audio ID so you don't have to swing and look at the display the whole time.

He got (compared to standard GMP setting):

10 cm extra depth on a 50 eurocent coin = 4 inches
7 cm extra deph on a small 10 eurocent coin = 3 inches
and als 7cm extra on a 2 RM coin as discribed in the austian test = 3 inches.

And this is according to INGO standards, so no airtesting!

I think that's a lot and it shows that coin size is not important as the 10 eurocent is a very small coin and the 2 RM is a BIG coin.
But a low discriminationsetting is all important if you want depth out of the DEUS.

With the first threshold at 0 it's no longer possible to set the Iron volume, so it's at level 5 all the time, but thats a small price to pay for 4 inches extra depth.

This text is very important to read, if you want to maximize the depth capabilities of the Xp Deus.

 

[Dave Rishar]

Interesting. I think that I get what you're saying, Dancer.

Timpsu? I've seen that post on a few forums and I'm thankful that people are doing tests like this. I believe that on one of the forums, it was established that this testing was done with V1 or V2 and that some of it may not apply to V3. I'm certainly keeping it in mind though.

So besides possibly missing very small rings and earring studs (which I'm willing to do if it also means that I'll miss ten pounds of can slaw at the same time), are there any drawbacks to taking this approach? This isn't meant for all sites, just our local trashy parks. I know that it's possible to play around with the tones instead to make the junk sound different, but I'd rather not hear it at all.

 

[DDancer]

Hrm, interesting.

Thanks for the support.

My explanation was pretty much just the simplified version of how I understand Notch'VS'Discrimination. *Forgive the pointy head* In older analog machiens discrimination was *and may still be in digital versions* an adjustment to one or more filters after the first stage of a detector... first stage is called the All Metal mode. After this stage all signals are processed thru Filters... and their digital equivalents~ this is where discrimination settings have their affect.

One can prove this by simply jumping between All Metal and Discriminate *After Ground Balancing* to see what signal response is. Granted All Metal does not give the tones or responses one expects in Discriminate... why.. well that's because All Metal is literally Raw Data... no processing so no tones. There is also no adjustment for discrimination aside from manually adjusting the Ground Balance in All Metal mode. Once you start adjusting for Discrimination you are adjusting the filtered response of the detector... It still receives the raw data, ie All Metal plus Ground Balance, but in your decision to adjust the threshold of your detectors response thru the filters you are essentially dumbing down your detectors sensitivity, and depth, to reject low conductivity elements *like ground balance does to reject low conductivity influence due to soil conditions*.
In my explanation Notch discrimination simply negates audio response to signals in the selected band on the output end of the detector circuits... these exist after the filters and discrimination offset. Notch does not affect discrimination or GB... its just an audio filter so hence no affect on detector performance.

Does this hold true for newer microprocessor controlled detectors... or multi-frequency detectors?? cant say. But the basics usually prove true no mater how the machine works.... its just a matter of making the electrons dance for ya.

How can you prove this?? Well here's an experiment. Bury a nickel deep as your all metal mode can detect it in clean soil you can ground balance on easily. Now turn on you discriminate and leave the notches all open and discrimination all the way down *or as far as you can go* and do not re-balance it.... Can you still detect the nickel? Probably can **I did years ago playing with this idea** now notch it and watch you detectors response. Bet you get a broken sound or pips and pops of audio and VDI.... now remove the notch and increase the discrimination... watch the nickel disappear... switch to all metal and watch it re-appear. It really works

 

[Texas Looter]

Brian,

You don't list having a XP Deus, so are your comments about it just a guess or are they based on some experience with one?

Several creditable sources with many hundreds of hours on the Deus disagree with you and support the position but forth by Dancer that notch is only a audio function and has no effect on actual detection.

The CTX also has audio selectivity that does not effect detection.

He did not say that it affected the depth of the machine but it affects the perfromance which it does. Notch is a filtering process and any additional filtering affects the performance of any machine. Now the amount of performance it affects is debateable but there all the same.
Notch also affects the ability to detect two targets close together if one of them is notched out and the other not.
Another thing that affects performance in some ways is the use of multiple toneswhich is lso a filtering process.But that is another can of worms.

Some still think that the CTX is not affected by disc. but i assure you it is as are all detectors.

Any detector run wide open will hit and find more targets, just up to the operator to be able to tell what they are.

I dont have a Deus listed either or any other detector but not sure it means that i dont own a detector.

Good hunting.

 

[Silver Searcher]

I find the Deus a very frustrating machine, one that likes coke very much it also has the same trates as the Xp gmp...coil to headphones lapse, and target loss.

SS

 

[parkgt]

He did not say that it affected the depth of the machine but it affects the perfromance which it does. Notch is a filtering process and any additional filtering affects the performance of any machine. Now the amount of performance it affects is debateable but there all the same.
Notch also affects the ability to detect two targets close together if one of them is notched out and the other not.
Another thing that affects performance in some ways is the use of multiple toneswhich is lso a filtering process.But that is another can of worms.

Some still think that the CTX is not affected by disc. but i assure you it is as are all detectors.

Any detector run wide open will hit and find more targets, just up to the operator to be able to tell what they are.

I dont have a Deus listed either or any other detector but not sure it means that i dont own a detector.

Good hunting.

I don't think I said that he said it had an affect on depth.

I have seen no objective proof offered anywhere that notching impacts detection on the Deus. The circuit you say is filtered is an audio function not a filter on what is the detection circuit. You may in fact be right but I would like to see it proved. If it has been that hard to observe and report it must be an insignificant impact if any. Same hold true for the CTX as far as I have been able to see.

Always eager to learn more substantiated information about the detectors I use, just not a fan of guess work or unfounded conjecture.

 

[Texas Looter]

I don't think I said that he said it had an affect on depth.

I have seen no objective proof offered anywhere that notching impacts detection on the Deus. The circuit you say is filtered is an audio function not a filter on what is the detection circuit. You may in fact be right but I would like to see it proved. If it has been that hard to observe and report it must be an insignificant impact if any. Same hold true for the CTX as far as I have been able to see.

Always eager to learn more substantiated information about the detectors I use, just not a fan of guess work or unfounded conjecture.

Guess work and unfounded conjecture?

No guess work here.
Just a lot of testing.

If you think that using notch will not cost you targets then proceed onward. Or maybe you can set up some test and see for yourself when good targets to close to bad targets drag your id's down into your notched area and you either hear nothing or a signal that you would not dig.

Same with the CTX.

 

[parkgt]

Guess work and unfounded conjecture?

No guess work here.
Just a lot of testing.

If you think that using notch will not cost you targets then proceed onward. Or maybe you can set up some test and see for yourself when good targets to close to bad targets drag your id's down into your notched area and you either hear nothing or a signal that you would not dig.

Same with the CTX.

What you describe is not a filtering effect on the machines ability to detect the target. But rather a choice of what detections to report.

Setting notches too close to desired targets can result in what you describe. That is a much different from the blanket unexplained statement that "Notch is a filtering process and any additional filtering affects the performance of any machine". There are many other settings on the Deus that are more likely to affect performance in much larger ways than any brought on by careful notching.

I don't notch that close to desired targets with the Deus or set CTX discrimination that tight either. So no I don't think they are affecting the performance. And yes, I have tested both machines in my test garden with coins buried with adjacent nails.

With the CTX you can even still see the discriminated target on Target Trace even when you don't hear them being reported.

Your results with other machines may be different.

 

[Texas Looter]

Few more words on the subject then I'm out of here.

First of all there are a lot more trash in the ground other than nails. I have had pull tabs pull Ids on the Deus down 14 points and as much as 7 or 8 on the CTX.Also first chance you get please provide me with a photo of a 10" target showing target trace on your CTX. Mine won't show target trace past 7"s or so.

As info.on my original post I stated that the effect of the filtering from notch was debatable but there all the same. I never said the it impaired the machine to the point of not using it.

 

[Dave Rishar]

So here's another weird idea. Why not dispense with discrimination and notching entirely and just use tones for the job?

You can set the tones on the Deus as low as 100 hz. As it turns out, between previous careers and my current job, I've lost enough of my hearing that I can only barely detect this; some of you probably can't hear this tone at all. So, minimal disc, no notches, and assign 100 hz to unwanted target ranges. The machine does no filtering and performance should not be affected, but you won't hear the result. (And if you do, probably only barely; human hearing only goes down to about 31 hz or so on average.) The upper range only goes up to 800 hz and nearly everyone should be able to hear that, so save that for targets that you want your attention drawn to.

The "big notch" program did me a solid today - two solids, actually (as per my post in the Today's Finds forum) but we may be able to do better. I'm going to give this tonal trickery a try and see how it works tomorrow.

And XP, if you're reading this thread, I'd like V3.3 to have a 50 hz tone available, as I can't hear that under any circumstances and most other adult males would not be able to either. It would make for a useful "discrim" tone.

 

[parkgt]

Dave,

That is an interesting idea, I will try my damaged hearing to see if it would work for me.


TL,

Going back to Dave's original question, it seems that notching to 30 is unlikely to cause any issues. Your pull tab observations are interesting and I will watch out for that.

Thanks,
Greg

 

[U.K. Brian]

The trouble with many of these people with hundreds of hours on any detector is they just don't find very much or they would give up their jobs and be out detecting every day to pay the bills. I sold my last XP last year and have many other detetectors that the site software won't allow me to post up due to limited capacity. Everyone interested in getting the best result from detectors should create a test bed consisting of 10 or 12 coins of the SAME type. Then its a five minute job to run down the row and see what is in fact happening after each adjustment has been made.
I should admit that I've only used the Mk 1.0 and 1.2 which had the generation 4 software so things may have changed.

 

[Dave Rishar]

My "deaf tones" program was an abysmal failure. I can hear 100 hz tones as produced by the Deus just fine. Not only that, but they're difficult to ignore for some reason - not at all like the low tones on an E-Trac set up for two tone ferrous. That lasted all of about ten seconds.

Sticking with "Big Notch" for crapped-up parks for the time being. I ran it side-by-side against Basic 2 and while it's just a hint chirpier, I'm only running Big Notch at silencer = 2. I had no real opportunity to test the differences in depth in a meaningful manner, as it's all shallow stuff in this park. I know that years of expertise have gone into those factory programs and I'm sure that they're excellent for 95% of the detectorists out there, but I'm equally sure that none of those guys have been working the parks around here. (Or have they? Basic 2 has a pretty aggressive silencer setting.) It's been good to me. It hasn't been "retire and do this for a living" good, but I would think that you'd need beaches that see large amounts of swimmers for that, and while we have the beaches, we don't have the swimmers; the water is only around 55F in the summer and quite a bit colder in the winter here, and the folks with money typically stay the hell out of it. I don't believe that it can be done in the dirt regardless of machine or skill level, and dirt's almost entirely what I do.

However, I can respect the opinion of a man with his convictions based in experience. If I were to explain my particular circumstances, and you felt that part of your detecting experience covered said circumstances, would you be willing to make some recommendations for machines? I don't see myself getting rid of my Deus any time soon, but I do have another high end machine that may wind up gathering dust and could possibly be sold off to fund a new acquisition. I'm still in the learning stage here, so I'm all ears.

 

[U.K. Brian]

Some love the Deus others don't. My two "anti" things were that I didn't want a wireless coil with its inbuilt battery because at first you had to return it to the manufacturer if flat or failed and I go abroad alot. They tackled this with a kit so you could DIY and they uprated the batteries. Also I never considered the wet beach performance that good.
They do seem to listen, so both software and hardware improvements (like the grip and camlock upgrades), have been made.
I would have thought you had picked a pretty good machine for the use your putting it to. In extreme rubbish conditions, after a reduction in coil size, my next move is to use a reduced discrimination level but this could be a good way to get you banned from your local parks.
In the old days (when batteries were made of wood) you had the option of off-resonance machines which had a circuit in the coil that was forced to resonate at a frequency higher than its natural resonance frequency. As the coil approached say a silver coin or gold ring the amplitude of the output signal increased but unwanted metals like a ringpull's output signal would remain the same or even fall slightly. Result, the coin or ring is accepted and heard whilst the ringpull would be in effect discriminated out. The price paid for this improved discrimination was lack of depth and it was depth that sold detectors.

 

[Dave Rishar]

Cherry picking coins is easy, at least in the US. Now cherry picking gold - that would be something I'd give up some depth for. As much as I like the Deus, I still have to dig trash if I want gold...lots of trash.