Detector Unit Consistency (long Brainstorm)

[GB_Amateur]

This topic seems to come up about everytime a new detector is released.  E.g. "did I get a lemon?"  I don't recall much more than ridicule responses although maybe I missed something.  Reality leads me to a common question I ask (myself, but also you) and will continue to do so until I'm booted from this site 😁:  "How do you know?"

This question relates to Quality Control (QC) which is (hopefully) done by manufacturers at the component level and at the final (macro) product level.  I know almost nothing about digital electronics and just enough about analog to qualify for the old "a little knowledge is a dangerous thing."  But I think back when detectors were purely analog the individual unit consistency was a bigger deal, possibly getting more attention.  The components (capacitors, resistors, inductors) for sure had rather wide tolerance ranges back then and many were sensitive to temperature differences.  I think there was more use of adjustable components (e.g. variable resistors) to fine tune an individual unit before shipping.  (Of course I could be wrong on that....)

I recall stories here about the coil QC at White's and the extremes their coils (prior to selection at the facility) could exhibit.  And I think this continued there into the digital detector age.

So what about today?  There still are some analog components of the design but I think that's held to it's basic minimum, now letting the digital back end do the heavy lifting.  And we've all seen the results in terms of useful features.  (More raw depth? Well apparently not much if at all, but more usable/meaningful depth -- definitely.)  I also think modern analog components have much tighter tolerances than typical from the old days.  Voltage regulation is another area of improvement, I think, and there likely are many others.  But with all those improvements, what's the bottom line?

Time for a thought experiment.  I go to Gerry's place and get 100 brand new 'identical' detectors.  (I know he has that many Manticores sitting there, getting a laugh out of it watching us on his waiting list.  😄)  Stay in factory presets, ground balance to the same channel on all, set to the same gain, etc. (and keep them far enough separated that they don't interfere with each other).  Each one is subjected to the same test target with some consistent specification as to max distance ('depth') that is detectable.  What is the spread in those distances?  IDK.  Do you?

I (and probably you) would like to think that something similar, maybe even more rigorous, is done at the factory for every unit prior to release.  I have a feeling that is extremely naive.  Often in QC, spot checking is used but do they even do that?  When I see the sloppy glue job that phrunt showed a couple weeks back I don't get a warm fuzzy that the Manticore (for example) is getting sufficient attention post assembly.  (It's just cosmetic right?  What's the big deal?  No really....????)

Would it be feasible for detectorists like we here to actually create a reliable/repeatable (at a distance) test whose shared results would be meaningful?  One thing I'm pretty sure of is that it would be difficult to match that key 'yes/no' decision on where the signal disappears with target distance.  For sure you can't depend upon everyone's hearing!!  I thought about using the disappearing VDI as an indicator, and maybe that would work for some detectors (Deus 2?) but it wouldn't work for *my* Equinox, for example.  Then there's the dreaded and variable EMI degradation of signal....  Uniform test target?  People following instructions to a tee??  Starting to look tough and maybe that's why we're left wondering.  Anyone else (who's read this far) think of a solution?

 

[schoolofhardNox]

I would think (guess) that there are just test points on the boards, and that there is an acceptable range of tolerance that allows a unit to pass. They can't all measure out exactly the same. So maybe some machines are still in the acceptable tolerance range in certain areas and others that are spot on, just like the lab prototypes. But then there is the coil, and that is where I think the tolerance is most important. Put together a unit who's PC board just makes the lowest tolerance level and place a just acceptable coil on it and you have an under performer. Put together the perfect unit with a perfectly balanced coil and you have a machine that just works great. I don't know if that scenario is what is going on, but in my head, it makes perfectly good sense 😆

[F350Platinum]

On the "box" end of things, I'd say that digital processing takes a lot of the inconsistency away, but the coil is always going to be slightly different.

[phrunt]

Interesting point and I can see how analog detectors have wild variations, digital detectors should have far more repeatable results HOWEVER a part change which sometimes happens due to shortages, and the quality of components can all make a difference.  Not all parts are equal, the GPX 6000 is a prime example of this with its dodgy inductors, some detectors appeared to perform better than others, while some were just wild.  This was all because of a couple of dollar tiny components.

People who are/were into overclocking CPU's on computers know all about variations in electronics.  Some CPU's overclock really well, others of the same batch do not, because I had access to many CPU's from the production line I would grab 20 or so identical CPU's, setup the motherboard the way I wanted with the overclocking, put each CPU on and test how far I can overclock it and keep it stable and see which one of them runs the best, and keep the best one for myself, the others go back into the production line 🙂  They weren't faulty of course, but didn't overclock as well.    Some memory chips can handle the overclock better than others, upping the FSB of the CPU made the memory module run faster too, some brands were better than others but more importantly some individual modules ran better than others of the same brand.  This is all running the hardware outside of specification but shows there are inconsistencies in components.

Basically for those that don't know overclocking is running a CPU at higher clock speeds than it was designed to run, thereby making your computer faster sometimes giving you the performance or near enough to of a much more expensive CPU.

Back in the early days you could buy a Celeron 300A which run at 300MHz for $149 USD or so, it was on a Front Side Bus speed of 66MHz, the P2-450MHz was the high end CPU at the time and it was $669 USD but it's FSB was 100MHz.  You could overclock that little Celeron on certain motherboards where the manufacturer was a bit lazy not checking your CPU's designed bus speed just by changing the FSB setting in the bios from 66MHz to 100MHz, the Taiwan motherboard manufacturers quickly picked up on this and made boards for overclocking that made it simple for people to do this and marketed them for the purpose.  Once you changed your FSB to 100MHz your cheap little Celeron ran at 450MHz, the same clock speed as the much more expensive P2 450 which was Intel's flagship at the time.  A $500 or so saving back in 1998 was a decent chunk of cash for very near P2 450 performance.  There were earlier and later CPU's you could do similar with, the 300A stands out as it was legendary for just how well it overclocked to be so close to the much more expensive CPU.

So while digital detectors are far more likely to be very similar to each other, it's not guaranteed they will be and because they expect them to be so close in performance they see no need to test them for inconsistency.  Those old videos of Whites testing coils on detectors to ensure they were up to spec and sorting the good from the bad, that is VERY unlikely to happen at places like Plexus.

[Steve Herschbach]

With old analog circuits and coil assembly there absolutely were hot machines and cold machines. And lots of substandard coils. The old stuff was basically hand made, and new production practices have at least lead to more consistency in production. However, I do think quality control used to be better, partly because detector sales used to be a low volume business and so at some companies every detector was checked before it shipped. Now they crank them out in high volume and just ship them, and figure all is well as long as the warranty failure rate falls within an acceptable amount. Part of that equation is that some people will never file warranty that really should have. Many detectors get minimal use, and can sit in closets until they get sold used, and then somebody else owns the problem.

I'm convinced that many "this is better than that" fights on the internet are caused when somebody is unknowingly using a substandard detector comparing to somebody else with a machine working correctly.

[Norvic]

Too many variables for there to be a conclusive test other than consumer popularity of a product. eg. the current thread on the Trash Can Lid coil, some consumers got good results some didn`t or the current thread on the QZ80 detector, some will buy and swear by them. Tolerance of electronic components and even coils is far finer than tolerance of the individual consumer. 

[abenson]

Have been thinking about this subject in relation to the Manticore and specifically coils. A lot of talk on youtube about both sensitivity levels and iron falsing in regards to the Manticore. Some people are saying they can run the Manticore at very high levels of sensitivity with great results. Others are saying they can't get past 23 without problems. And both of those groups are also reporting both excessive iron falsing and no falsing issues. My Manticore gets erratic at sensitivity levels above 23 no matter what or where I'm hunting. But I have I would say about the same levels of iron falsing I did on the 800. So is my machine/coil good or bad compared to the same person that gets excessive falsing at the same sensitivity levels. Or is the person that can run 32 sensitivity levels with non-issue and no falsing got a good machine. I do not know. And maybe it's the users inexperience?

[GB_Amateur]

1 hour ago, abenson said:

Have been thinking about this subject in relation to the Manticore and specifically coils. A lot of talk on youtube about both sensitivity levels and iron falsing in regards to the Manticore. Some people are saying they can run the Manticore at very high levels of sensitivity with great results. Others are saying they can't get past 23 without problems. And both of those groups are also reporting both excessive iron falsing and no falsing issues. My Manticore gets erratic at sensitivity levels above 23 no matter what or where I'm hunting. But I have I would say about the same levels of iron falsing I did on the 800. So is my machine/coil good or bad compared to the same person that gets excessive falsing at the same sensitivity levels. Or is the person that can run 32 sensitivity levels with non-issue and no falsing got a good machine. I do not know. And maybe it's the users inexperience?

My answer is: E) All of the Above.  Settings the same?  EMI environment the same?

Have one person, knowledgeable about settings and careful in technique with a properly constructed Faraday cage test multiple detectors+coils of the same model (e.g. Manticore).  It's doable, but it aint easy.

[phrunt]

I just think it's EMI, some places are bad for EMI, I'd imagine a bulk of the USA has quite bad EMI.  Huge population all spread out, lots of reasons for high EMI.

In city or town areas around here I can't run past 25 on the Manticore, drive outside of them and I can run 35 with only the occasional burp and pop, no ID falsing.

Here is a bit of a map around here, the little red circle down the bottom is a town of about 300 people, mostly holiday homes and probably half don't live there full time.   The red circle up higher is a town of about 15,000 people, again many holiday homes sitting empty a lot of the year, the yellow bits are part of the 15,000 population but where it starts to get more spread out with dense housing estates mixed in.  In the red and yellow areas I'm under 25 sensitivity on the Manticore, down to 16 to get it running nice.  Outside of those areas I'm sitting on 35 just fine.  There is almost no population outside of my circles as there is very little flat land and I think that's the key, very low population, no government induced EMI from military or anything, and just a lot of area with nothing but mountains.  Zoom the map out further and to the left are just a lot more mountains, to the right and below it starts to go to hills and then farm land, to the north all mountains, all low population.

Minelab seem pretty good and consistent at making VLF coils, I don't even recall reading about someone having an under performing or faulty coil other than ears snapping.  It's PI coils they seem to struggle.   I think differences in sensitivity levels people are running their Manticores is more likely related to EMI than to variable electronic components.

Have you tried beach mode yet in areas where you can't get past 23 Abenson? I'd be interested to hear if it helps.

[GB_Amateur]

13 minutes ago, phrunt said:

Here is a bit of a map around here,

What's the scale of your map?  And what's your cell coverage?  Power lines? (I know you've talked about and shown us these in some of your nugget hunting grounds.)

Do you have any USA coins (or if not, how about an Aussie large cent -- I happen to have one of those)?  I think you're a good candidate for some standardized air testing.  We can compare notes (as soon as I get mine -- I hear Gerry has applied for a Cabelas outlet...).