Fisher Impulse AQ Discrimination Explanation

[Steve Herschbach]

The new Fisher Impulse AQ is doing something new with pulse induction discrimination. I think I have a pretty good idea of what the engineers are up to and so want to pass on my thoughts in case it will help people have a better idea of what to expect.

I have posted numerous articles in the past about coin detecting with ground balancing PI detectors. GBPI detectors split targets into two classes depending on the ground balance setting. The two target classes exhibit by different tones. All Minelab PI detectors do this, as do the Garrett Infinium and ATX, plus the White's TDI series based on the Eric Foster Goldscan.

White's on the TDI labeled these as low conductors and high conductors but that is not actually accurate. They are items with a low time constant and items with a high time constant as measured on a pulse induction metal detector. For more on signal decay and time constants see the excellent primer by Reg Sniff Understanding the PI Metal Detector. If you could measure the time constant of the various targets and line them up you would see size has a lot to do with it, small items versus large items regardless of composition.

White's TDI "Target Conductivity" switch (notice ground balance on required)

I caught on to this fast with the Garrett Infinium in 2004 and Minelab GP 3500 in 2005 and when the time came it was I who suggested the tone switch be incorporated into the White's TDI to allow the user to choose between one of the two classes of targets. Plus the TDI allowed the operator to directly set the ground balance, creating a crude discrimination system for savvy operators. It is actually possible to coin detect with a TDI and dig almost no trash, but you lose its PI depth advantage by setting the machine that way. I preferred using the tones with full depth settings myself with excellent results.

The two target classes vary depending on the exact ground balance setting, but generally one one hand you have low conductors and small ferrous targets, and on the other hand coins and large ferrous targets. This means if looking for small gold you dug all manner of small trash including small ferrous trash. Coins were better but you would still dig large nails and other large ferrous items. The TDI allowed a person to advance the ground balance to where most ferrous trash could be eliminated while still getting the coins, but the setting was so close to the coin range to do this that depth suffered as a result, and so was no better than a decent VLF. It's a long read but a lot can be learned by reading the following three links.

Coin Detecting with the Garrett Infinium

Beach Detecting with the Minelab GP 3500

Coin Detecting with the White's TDI

Some Commentary On White's TDI Tuning & Discrimination

It can be seen that as far back as 2005 I was predicting that high power PI detectors would find widespread use with beach hunters and relic hunters in particular. That prediction did later come true at Culpepper and other locations.

It was obvious to me that this crude discrimination offered more room for improvement. I even emailed Carl Moreland at one time and made mention of the concept of using two ground balance controls to delimit another class of targets beyond the two created with a single ground balance point. But really it is not ground balance, it is just a setting that measures a certain time constant. And that folks, is basically what the Fisher Impulse is doing. It is using at least two dividing points to separate items into three classes, lowest time constant, middle (gold ring) range, and highest time constant

The Impulse is eliminating the items with the shortest time constant and the longest time constant and focusing on the mid-range. This eliminates tiny trash and tiny gold, and large ferrous stuff and most coins. What is left is the sweet spot in the middle where gold rings occur plus a yet to be determined range of trash targets. There will be some, that is inevitable, but the vast majority of trash should be eliminated with this approach.

This picture shows the result of using the Garret Infinium and separating the short time constant targets from high time constant targets in tot lots. Low constants (nickels and zinc penny plus trash) on left and high constants (copper pennies, dimes, quarters) on right. Notice very little trash with the coins if you dug that class only. But that is deceptive due to the location, which lacked any nails.

Garrett Infinium targets separated by tone

So take a look at this result from a park, this time only the long/high time constant targets with the TDI... coins and large nails.

White's TDI low tone / high time constant targets

OK, so let's combine the two to get a clearer picture....

Left of red line low time constant versus right side high time constant targets

To reiterate, having only one point of separation presents issues. You get coins but you also get large nails. And the low time constant area is a mess, just way too much stuff including small ferrous. But in areas with few nails it actually works well for coins. The dividing line does vary with the ground balance setting but is roughly about zinc penny if ground balanced in moderate to high mineral ground.

The Impulse AQ is basically blowing off the high end to get rid of the nails, and the high conductor coins also get tossed out. Most jewelry hunters will be fine with that - digging coins in the surf is a waste of my time and effort.

The AQ is adding another separating point somewhere in that left hand low time constant area, that eliminates the tiny stuff including I assume most wire type ferrous items and foil. Rings due to the round shape with a hole read better than most of the misc trash in the photo. However, some of those items will read in the ring range so do not expect the AQ to be trash free. Trash that reads close to the nickel range will probably come through, like pencil eraser ends and some tabs that read like nickels. If you think of the AQ as a nickel detector that will probably not be far off the mark. My biggest area of concern is zinc pennies, especially corroded ones. I hate those things but I suspect we will be digging them with the AQ.

All the above is just speculation based on everything I have learned using this class of detectors over the last 15 years. I have had no part in the AQ development. That said, I think most of what I am describing will prove to be pretty close to the mark as far as how the discrimination functions on the Impulse AQ.

And just a reminder. The Impulse AQ will let you turn off the discrimination to run in pure pulse mode, presumably with no ground balance engaged. It should be at similar to but more powerful than the TDI Pro with ground balance shut off. This will not be good for nugget detecting in most areas as it will light up hot rocks, but for many beach detecting situations should be giving us a detector like the famed Eric Foster Deepstar in a much more ergonomic and affordable package.

[Bill (S. CA)]

Steve,

Absolutely brilliant post.  I think most of us beach hunters would agree with you that what we are hoping for is that the AQ hits that "sweet spot" like no other machine.  How do you think it will handle platinum?

Bill

[Steve Herschbach]

Good question. Platinum rings can read real low, and smaller ones down in the foil range. I would not be shocked if some platinum gets lost with the AQ discrimination setting. The talk focuses on gold for a reason. Like I have mentioned in another post getting our hands on this machine and doing a lot of target testing will be critical to determine where the lines are drawn and if they can be moved at all.

Let's face it, marketing will talk about what the machine does well and will let other stuff go unmentioned. We will have to figure out the details ourselves. In a nutshell, where are the mid-range cutoffs both high and low? And what trash is still going to show up in the mid-range? I have an idea what to expect but only actual use will prove the reality of it all.

There are bound to be people who just freak out over the idea of missing anything. The fact is unless you are using a high power PI and dig every target we all miss stuff. The more discrimination you employ, the more inevitable it becomes. For me it is purely a practical matter involving how much time I have and in some cases how many targets I can physically dig. Moving the odds in my favor by eliminating worthless digs at the cost of some possible good targets missed is one of the most important strategies we employ in metal detecting, and getting it wrong too far in either direction can be costly.

I have often used VLF discrimination to eliminate zinc penny and higher targets while park detecting for rings. People will say "yeah, but some rings read like zinc pennies". Yup, that's true. But rather than fill my pockets with coins I prefer to concentrate on the mid to low range where the odds are better for rings and save a lot of time digging coins I am not looking for. On the very low end I may also pass on the real small foil and can slaw bits. This is what the AQ is also doing.

It sounds like lots of time has been spent optimizing where to put the cutoff points and I am going to trust the testers have chosen wisely. Nothing is perfect. I am all in for a PI that concentrates on the main sweet range while eliminating a lot of deep trash digging, the bane of any PI detector. And if the trash level is low anyway, just switch to all metal and dig it all. Best of all it's not an insane 7 lbs like my old ATX!

I expect the AQ discrimination system will have limits but none that will surprise me and I am willing to accept the limits to get some form of decent discrimination while retaining PI depth in black sand laden beaches. If the machine does as expected I don't see huge benefits for low mineral white sand beach hunters. The main benefit will be west coast and volcanic islands where a PI will shine. And in those areas heading for the trashy places that normally drive PI hunters away will be a good idea.

[ALEXANDRE TARTAR]

Hello,

I am physicist engineer to Fisher Research Laboratory, I designed the Impulse AQ.

Your subject is constructive, so I will read the post carefully and I will try to answer it tomorrow.

I am French and I need a little time to read and respond.... 😉 

Alexandre

[Steve Herschbach]

Welcome to the forum Alexandre! We are all excited about your new detector and speculating about how the discrimination system works. The goal on this forum is only to help people, so if anything does not seem helpful please let us know.

[Chase Goldman]

The following is a discussion of existing PI detector features that I find to be useful for relic hunting that goes beyond the target discrimination break-point discussion that Steve articulated so well above.  Beside the high mass-high conductor/low mass-low conductor discrimination breakpoint setting - another useful feature implemented in the ML GPX and to a lesser extent (or less successfully, based on my experience) on the Garrett ATX is iron identification, rejection, or blanking.  The GPX implements a blanking function on probable "high" (low tone) corroded iron signals with the peak of the signal tone blanking out.  Not 100% effective as it can also provide "false positives" on non-ferrous targets but it is nevertheless a pretty useful feature.  Similarly, the ATX has an iron grunt check, which is both ergonomically inferior (requires you to engage a separate button during your swing) and I have found it to be less reliable (almost to the point that it is a 50/50 proposition if you do get an iron grunt).  Blanking has it's drawbacks of course.  Besides reliability, iron reject/blanking on the ML can only be utilized with DD coils and can become overwhelmed in thick iron situations. Anyway, Alexandre or Rick (once you get your machine), if either of you can shed light on whether such a feature will exist on any or all iterations of the Fisher Impulse, that would be enlightening. 

Finally, compared to the Garret ATX, I find that the as you gain experience digging innumerable nails, the tone "shape" of the GPX provides a wealth of information.  For example, the classic "W" double peak high tone (low conductor/small mass target) is often a clue that you are swinging along the long axis of a nail.  Turn 90 degrees and that nail sounds like a sweet high tone indicative of round brass or small lead.  Bent nails, however, are very difficult to ID, they sound great.  Higher mass minie balls give a nice smooth low tone that will not break unlike a large rusted nail or other piece of iron.  The ATX uses more of a processed doorbell high-low/low-high tone, so it cannot convey much additional target information other than the high conductor/high mass or opposite characterization.  So I am interested whether the tone implementation on the Fisher Impulse target audio will convey analog-signal-like target characterization (e.g., irregular or regular shape) information.

Thanks for reading.  Excited to see the Fisher Impulse in action.

[Rick K - First Member]

In all this discussion, it ought to be remembered that the AQ is very tightly focused on recovering gold jewelry at salt beaches. That doesn’t mean that it will not be useful for other applications, but to the extent it may prove not ideal for those, the answer will be that tight focus and optimization for a purpose means that other purposes may not be as well served.

[Tmox]

Thanks Rick, your opinions on gold coins on the beach? Will it depend on the size of the coin?. It's the excitement before Christmas!

[Steve Herschbach]

2 hours ago, Chase Goldman said:

another useful feature implemented in the ML GPX and to a lesser extent (or less successfully, based on my experience) on the Garrett ATX is iron identification, rejection, or blanking.  The GPX implements a blanking function on probable "high" (low tone) corroded iron signals with the peak of the signal tone blanking out.  Not 100% effective as it can also provide "false positives" on non-ferrous targets but it is nevertheless a pretty useful feature.

Long story short this has never been mentioned at any as a potential feature on the Impulse AQ. Perhaps it may be seen on the land prospecting version since the AQ methodology is not very application to ground balanced nugget hunting.

The main problem with the ferrous blanking as implemented in both the Minelab PI detectors and the ATX is it only works on shallow targets, and not on the targets where you need discrimination the most - the really deep ones. I don't mind digging 6" nails, it is the two foot nails that I find annoying. This inability to function at depth made the GPX iron blanking pretty useless for most of my applications where deep ferrous is a common occurrence, whether on a beach or in tailing piles. That, plus numerous instances of my seeing quite large and valuable gold specimens rejected as ferrous by the GPX meant I never used it. It might be acceptable to miss bullets due to improper identification but not $500 gold specimens.

Now don't get me wrong, the surface ferrous blanking function can be very useful for some people and some applications. I can see where it would be a great benefit for relic hunting in particular. I'm just pointing out it has some serious limitations such that most gold prospectors at least never use it except in rare circumstances.

This is typical. This location produced some real nice gold nuggets. I got a bare whisper signal with the GPX 5000. I had a Gold Bug Pro on hand to act as a pinpointer and discrimination tool, but no signal from it until well over a foot. It started reading ferrous but by that time I was close anyway, and wanted the potentially target masking nail to be gone. So after quite a bit of effort in packed rocky soil out comes a large nail at about two feet. That is what I hope the Impulse AQ will help with. On the beach in particular the effort expended goes up exponentially with depth as the hole is likely caving and filling almost as fast as it can be excavated. I have had to dig some huge pits employing a PI in the surf with mask and snorkel only to retrieve a large ferrous item.

Heading for China with a shovel

And my reward - did not help it was wedged under that large rock in the hole

[ALEXANDRE TARTAR]

Foreword: For reasons of technological confidentiality, it is obvious that I could not discuss certain technical details. I can only share my knowledge in a relative way in order to give a comprehensive understanding satisfactory to the reader. So I'm going to do a little popular science oriented to "PIs".

All metal detectors based on an impulse principle depend mathematically on the ‘exponential decay’. This exponential decay is multiple, it is a series of exponentials nested in each other.(Ground effect + target + etc). Each exponential decay corresponds to a time constant.

Each target has its unique time constant. I mean by target, rings, gold nuggets, ground effect ...

Target size is indeed an issue when looking at this for the first time.

But once the size of the targets is removed from the equation (yes, this is easily doable) the time constants are homogeneous and identifiable without drifts.

Analogue analysis is not enough to deal with this kind of problem, digital processing is essential.

However, analogue analysis can be greatly improved to meet certain research expectations without losing too much sensitivity. This is what is done in the IMPULSE AQ and has never been done before.

The IMPULSE AQ is one of the first models in a range of PIs, these future models will be more and more upscale. I'm not talking about gadgets here, but technology. This detector will however already be classified among the high end of the market for beachcomber. The IMPULSE AQ has a purely analogue analysis.

Although this analysis is analogue, it differs greatly from the other models.

This analysis has been improved and it compensates for the majority of the time constant identification problems of gold jewellery.

The IMPULSE AQ is not a coin shooter, it is only intended to detect gold jewellery and this on all types of sand saturated with sea water.

Although some coins or silver rings have long time constants, not all of them will be rejected in "Mute" mode. But as mentioned above, this is not a coin shooter, so that is not our goal.

People who want to detect coins and objects that are not gold can find on the market a multitude of metal detectors, examples Equinox 800, BBS etc ...

These prospectors will then be confronted with prospectors who will have the IMPULSE AQ and will lose 20 to 30 cm on a 2 gr ring in 18K gold in the best case.

In the worst case, they will not even be able to detect a large gold ring on volcanic sand.

This is another of the characteristics of the IMPULSE AQ, this model is provided with a specific mode which makes it possible to detect in volcanic sand saturated with sea water with a reduced loss of sensitivity.

The IMPULSE AQ cannot however be used to search for gold nuggets, other models are provided for this, they are more sensitive and do not need to compensate for the conductivity of seawater.

These systems are designed differently…

I cannot say more about the various technologies under development at this time.

The IMPULSE AQ has the following modes:

- ALL METAL

- TONE

- MUTE

- VOLCANIC SAND

 The discrimination by analogue analysis of the time constants of the IMPULSE AQ allows it to place almost all of the gold jewellery in "high pitch"

 Here is an example of a time constant from smallest to largest:

10k --> 14K --> 18K --> 22K --> 24K

0.3 gr --> 30 gr

All jewellery below 22K whatever the weight will be detected with a minimum loss of sensitivity (10%)

22K jewellery will have a greater loss of sensitivity depending on their weight (20%).

For 24K jewellery, some will be detected in low pitch depending on their weight.

But in all circumstances, the sensitivity in MUTE / TONE / or VOLCANIC SAND mode is extremely superior to all VLF, and BBS on the market. (from simple to double)

In ALL METAL mode, there is of course no analysis of time constants so, no loss of any kind.

Iron is detected in all-metal mode by its characteristic double beep (provided the sound is at the limit of detection). This double beep is also present in Tone mode.

The IMPULSE AQ has many other advantages that competitors do not have.

It was once said in the forums that it was impossible to descend below 10µs of pulse delay on the sand saturated with sea water. This figure is of course to be taken lightly, because It is not difficult to design a PI with a short pulse delay but it is difficult to design it with a strong coil current!

The IMPULSE AQ has a powerful coil current and it is now placed at 7µs of pulse delay, which gives it a increased sensitivity on all types of jewellery sizes.

There has been a lot of work at this level to decrease the influence of the conductivity of the salt, maintain a very high stability and a small pulse delay.

For all these years of research, the IMPULSE AQ has been a tremendous development platform for my future technologies, already completed for the gold nuggets market.

To be continue.... (I am sick today)