field test unit no 001 "model T"

[hdphilip]

i'm searching for the adc your usuing, (of course,you'd expect that from me) at this level, shouldn't be to hard,

hats off to you, for a home brew test rig with a -220db "floor" pretty good, humm let see, -220db I'd be almost able to see packet bursts from the mars rover

Even with a balance mixer in front, -213 db spectrum analyzer ain't to shabby, i guess I've been out of the communication biz to long

Philip

[chudster]

OK, but practically speaking, you do not need MHz bandwidths unless you are not into discriminating sea water salinity Every practical target is much slower than that.

This is sounding like my previous postings about trying to understand the signal we are looking for. If we knew that we could break it into components and decide what sampling frequency is needed to represent it digitally.

I can taste salt water- don't need a detector. The problem is WHAT IS NEEDED? Aziz says that audio frequencies are good enough (< 24KHz) Moodz is up in the MHz region and picking up talkback from Alan Jones (OZ joke). On the balance of what I have seen, I chose 250KHz as being below AM radio and covering the frequencies of interest for targets.

I am beginning to learn that on the forums one or more of the following is true:
1) No one knows.
2) Some may know but don't want to say.
3) Some think they know but don't want to say because they may be wrong.
3) People are going in circles because of crap design and over or under engineering.

Sorry if it is a tough question but:

What is the bandwidth needed to cover up to the smallest targets of interest (presumably small gold targets)?

Its a simple question and if answered we probably could have saved Aziz 1000s of posts and a lot of lost time working in the audio spectrum rather than the low frequency radio spectrum- unless the audio spectrum really is good enough.

If the audio spectrum is good enough then we could save Moodz a lot of time and expense designing FPGAs to sample 24 bits at 1MHz of bandwidth.

Of course tinkering and posting for boasting is fun and maybe that is the real aim but if we aspire to really solve problems- that ain't science or engineering.

Chudster

[hdphilip]

moodz

ok i got it, Jeopardy style... (answers must be in form of a question)

what is... adc filter pro......... by TI ?
and.......... drum roll......

and what is a ADS1675 ?

a pretty sweet setup.....

philip

[moodz]

i'm searching for the adc your usuing, (of course,you'd expect that from me) at this level, shouldn't be to hard,

hats off to you, for a home brew test rig with a -220db "floor" pretty good, humm let see, -220db I'd be almost able to see packet bursts from the mars rover

Even with a balance mixer in front, -213 db spectrum analyzer ain't to shabby, i guess I've been out of the communication biz to long

Philip

eeek .... no the -220 db is just a "digital" scale its not achievable ( like 200 km/hr on your car speedo ) .. the FIR filter rolls down to -220 but reality is -120 to -130 db which equates to 1 microvolt of sensitivity. Though in practice with SAT the sensitivity is better than 100 nV in a narrow bandwith ( could be better ). Effective gain resolution => 2.5 volts / 100 nanovolts = 1 part in 25,000,000.

BTW high resolution ADCs make a nicer "log amp" than an actual log amp because of resolution vs dynamic range.

[hdphilip]

whew, ok i got it now,

when i look at your picturs, are those RF signals that are pointing down in the am radio band also?

[moodz]

This is sounding like my previous postings about trying to understand the signal we are looking for. If we knew that we could break it into components and decide what sampling frequency is needed to represent it digitally.

I can taste salt water- don't need a detector. The problem is WHAT IS NEEDED? Aziz says that audio frequencies are good enough (< 24KHz) Moodz is up in the MHz region and picking up talkback from Alan Jones (OZ joke). On the balance of what I have seen, I chose 250KHz as being below AM radio and covering the frequencies of interest for targets.

I am beginning to learn that on the forums one or more of the following is true:
1) No one knows.
2) Some may know but don't want to say.
3) Some think they know but don't want to say because they may be wrong.
3) People are going in circles because of crap design and over or under engineering.

Sorry if it is a tough question but:

What is the bandwidth needed to cover up to the smallest targets of interest (presumably small gold targets)?

Its a simple question and if answered we probably could have saved Aziz 1000s of posts and a lot of lost time working in the audio spectrum rather than the low frequency radio spectrum- unless the audio spectrum really is good enough.

If the audio spectrum is good enough then we could save Moodz a lot of time and expense designing FPGAs to sample 24 bits at 1MHz of bandwidth.

Of course tinkering and posting for boasting is fun and maybe that is the real aim but if we aspire to really solve problems- that ain't science or engineering.

Chudster

Chudster there is an old but true story engineering story about a drill company that was so particulary chuffed a about a new range of very fine and hard wearing drills that they sent out samples to all thier competitors as a sort of a "boast" if you can call it that .... anyway one of these companies sent the drills back with a short note saying that they really liked the product but that they needed to do better ... the engineers looked at the drills that had been sent back and realise thier competitor had drilled a hole right through the centre axis of each drill ( ie like a pipe ) LOL.

Lets answer your question in percentages of people reading this forum ...

1) No one knows. 80% ( and Im being generous )
2) Some may know but don't want to say. 2%
3) Some think they know but don't want to say because they may be wrong. 12%
4) People are going in circles because of crap design and over or under engineering. 98%

...hey that adds up to more than %100 ....????

.... thats because even the guys who know what they are doing get caught up in endless discussions about hypothetical discussions that are often totally irrelevant to the problem. For example .... some guy publishes some coil results and then everyone wants to see target responses ...from the bench ....give me a break ... the real targets in the ground will probably not be anything like a bench target and they conveniently forget the biggest target ( ie the ground LOL ).
Another furphy is power consumption ....you publish some circuit and there is all this discussion about power consumption ....who cares ...a basic principle of good engineering is "make it work ... then make it better" not the other way round ... there are so many on this forum that try to make it better ... before it even works.

As for solving problems ... I have posted the experimental results ... the results I post do demonstrate what I believe are the measurment limits to ensure adequate experimental discovery and performance ... if later those limits are shown to be excessive thats OK however I wont have to kick myself because I placed limitations on the investigation that made me miss a vital piece of information. If you are a good engineer / tech you dont need my hardware details to replicate or better these results .... if you are a "catalogue engineer" then you are in trouble. LOL

The second part of this story is that there is a real danger that if you discover a "breakthrough" you cannot really rely on
publication in a so called "public" forum as this to either protect your idea as yours or even to place it in the public domain.
Not unless you can afford $1000/hour legal assistance ;-)

[Tinkerer]

This is sounding like my previous postings about trying to understand the signal we are looking for. If we knew that we could break it into components and decide what sampling frequency is needed to represent it digitally.

I can taste salt water- don't need a detector. The problem is WHAT IS NEEDED? Aziz says that audio frequencies are good enough (< 24KHz) Moodz is up in the MHz region and picking up talkback from Alan Jones (OZ joke). On the balance of what I have seen, I chose 250KHz as being below AM radio and covering the frequencies of interest for targets.

I am beginning to learn that on the forums one or more of the following is true:
1) No one knows.
2) Some may know but don't want to say.
3) Some think they know but don't want to say because they may be wrong.
3) People are going in circles because of crap design and over or under engineering.

Sorry if it is a tough question but:

What is the bandwidth needed to cover up to the smallest targets of interest (presumably small gold targets)?

Its a simple question and if answered we probably could have saved Aziz 1000s of posts and a lot of lost time working in the audio spectrum rather than the low frequency radio spectrum- unless the audio spectrum really is good enough.

If the audio spectrum is good enough then we could save Moodz a lot of time and expense designing FPGAs to sample 24 bits at 1MHz of bandwidth.

Of course tinkering and posting for boasting is fun and maybe that is the real aim but if we aspire to really solve problems- that ain't science or engineering.

Chudster

The smallest target we can detect?
The decay slope of a target with a time constant of 10us, results in a diminishing of the response signal by about 63% in 10 us. If the target is some distance from the coil, this can mean the peak response signal as 100uV, after 10us it is now 37uV.
Is the detector capable of sampling at 10us after TX switch OFF?
How wide is the sample?
If the sample is 10us wide, the target signal will be about 15uV at that point. The sample amplitude will then be near the average between 37+15/2=26uV.

Now you must extract this signal from a common mode of about 10mV+all sorts of noise.

A 1" square of thin aluminum foil has a TC of about 10us, for test purpose.

Now, do the calculation for a 1/2"x1" square of aluminum, which has a TC of about 5us.

Then cut this in half and you have a target of about 2us.

These are big targets, because what counts most, is the surface area presented to the coil, or the area that intercepts the magnetic field lines that we use to visualize a magnetic field.

Although these targets are big, meaning the response amplitude is high, the target TC is short, meaning the target can only be detected for a short time.

A target with similar surface area, but more thickness, has a longer TC.

But, a target of lower conductivity of similar size, has a shorter TC.

Nuggets have a shorter TC than pure gold, because of their shape and structure.

Therefore we found that we can use a titanium target, which has much less conductivity than pure gold for testing. It needs a fast detector to be able to detect a 1 gram titanium target at 10".

So, what is the bandwidth you need? How fast is your detector? if the earliest sample you can take is at 10us, see above.

Not a good explanation? Yeah, I know, I will never make it as a teacher.

Tinkerer

[moodz]

whew, ok i got it now,

when i look at your picturs, are those RF signals that are pointing down in the am radio band also?

Thats a spectrum display .. the "pointing down" signals are just artifacts due to noise and errors resulting from mathematical limitations of the algorithm ... there can be effects due to sampling / coherancy and algorithmic effects that can cause such dips. True Gaussian noise should not exhibit such artifacts.

[chudster]

Chudster there is an old but true story engineering story about a drill company that was so particulary chuffed a about a new range of very fine and hard wearing drills that they sent out samples to all thier competitors as a sort of a "boast" if you can call it that .... anyway one of these companies sent the drills back with a short note saying that they really liked the product but that they needed to do better ... the engineers looked at the drills that had been sent back and realise thier competitor had drilled a hole right through the centre axis of each drill ( ie like a pipe ) LOL.

Lets answer your question in percentages of people reading this forum ...

1) No one knows. 80% ( and Im being generous )
2) Some may know but don't want to say. 2%
3) Some think they know but don't want to say because they may be wrong. 12%
4) People are going in circles because of crap design and over or under engineering. 98%

This accords with my experience. By the way I didn't want to cause offence, I meant crap design parameters. There is some good design evident, I just question the design parameters.

[QUOTE]
...hey that adds up to more than %100 ....????

.... thats because even the guys who know what they are doing get caught up in endless discussions about hypothetical discussions that are often totally irrelevant to the problem. For example .... some guy publishes some coil results and then everyone wants to see target responses ...from the bench ....give me a break ... the real targets in the ground will probably not be anything like a bench target and they conveniently forget the biggest target ( ie the ground LOL ).
Another furphy is power consumption ....you publish some circuit and there is all this discussion about power consumption ....who cares ...a basic principle of good engineering is "make it work ... then make it better" not the other way round ... there are so many on this forum that try to make it better ... before it even works.

As for solving problems ... I have posted the experimental results ... the results I post do demonstrate what I believe are the measurment limits to ensure adequate experimental discovery and performance ... if later those limits are shown to be excessive thats OK however I wont have to kick myself because I placed limitations on the investigation that made me miss a vital piece of information. If you are a good engineer / tech you dont need my hardware details to replicate or better these results .... if you are a "catalogue engineer" then you are in trouble. LOL
[\QUOTE]

I don't have any issue with what people choose to share of their work or not. It is their decision. I don't have any reason to doubt the authenticity of your work. You need though to get some value from the time and effort to post be it for peer review, ideas for new directions, second party result confirmation etc. and folks need to get some value from reading them.

The second part of this story is that there is a real danger that if you discover a "breakthrough" you cannot really rely on
publication in a so called "public" forum as this to either protect your idea as yours or even to place it in the public domain.
Not unless you can afford $1000/hour legal assistance ;-)

Well minelab has certainly spent its money well if it gets us to change approach so that we don't collaborate to push MD technology to the limit.

I point out:

1) If you fully describe an invention on a forum it is public disclosure. Except for a company being able to improve it and patent that improvement, its validity can be challenged because of prior publication.
2) BW defence could have been: all the details of the QED are available publicly. ML is being both lazy and vexatious with this action and they should deal with the infringement by whoever as it chooses seeking their particular flavour of the QED to see if it infringes. I don't consent to the court order to provide this stuff for that reason. The fact that BW did not plead this points to a different situation.
3) People should be able to patent if they want. Of course the patent process causes delays in disclosure but as has been tacitly reasoned its either play the game by patenting or to publicly publish or as some may prefer more crudely "Give it to the Chinese" so they can mass produce them for $10 a piece and we can all be happy. The problem is when we do neither well.

IANAL but I am looking at putting together a project/ thread charter to mitigate the risk of a MD company doing to others as they have to BW and was discussing that with Philip for his thread. Despite the QED folks behaviour in rattling the bear's cage, some response to the action of ML needs to happen or we all run to our holes and they win by QED collateral damage.

Chudster

[chudster]

Hey Tinkerer,

Thanks for your response. Its helpful for the advancement of stuff with some real data to chew on.

In summary with a 2uS signal we need 1MSPS or higher to have a decent chance of detection or smaller targets. Moodz working at 2MSPS is 'overkill' but not unreasonable as its good to oversample (where is that antialiasing filter though Moodz). At 48KSPS, Aziz's soundcard seems OK for cannonballs with very long time constants (160uS) but not a tiny nugget.

Note we are talking about direct sampled solutions here. We can do all sorts of analog processing to have a half analog half digital solution (including integration) to mimic an analog detector with lower sample rates.

Chudster

[moodz]

[QUOTE=chudster;157970]This accords with my experience. By the way I didn't want to cause offence, I meant crap design parameters. There is some good design evident, I just question the design parameters.

...hey that adds up to more than %100 ....????

.... thats because even the guys who know what they are doing get caught up in endless discussions about hypothetical discussions that are often totally irrelevant to the problem. For example .... some guy publishes some coil results and then everyone wants to see target responses ...from the bench ....give me a break ... the real targets in the ground will probably not be anything like a bench target and they conveniently forget the biggest target ( ie the ground LOL ).
Another furphy is power consumption ....you publish some circuit and there is all this discussion about power consumption ....who cares ...a basic principle of good engineering is "make it work ... then make it better" not the other way round ... there are so many on this forum that try to make it better ... before it even works.

As for solving problems ... I have posted the experimental results ... the results I post do demonstrate what I believe are the measurment limits to ensure adequate experimental discovery and performance ... if later those limits are shown to be excessive thats OK however I wont have to kick myself because I placed limitations on the investigation that made me miss a vital piece of information. If you are a good engineer / tech you dont need my hardware details to replicate or better these results .... if you are a "catalogue engineer" then you are in trouble. LOL
[\QUOTE]

I don't have any issue with what people choose to share of their work or not. It is their decision. I don't have any reason to doubt the authenticity of your work. You need though to get some value from the time and effort to post be it for peer review, ideas for new directions, second party result confirmation etc. and folks need to get some value from reading them.




Well minelab has certainly spent its money well if it gets us to change approach so that we don't collaborate to push MD technology to the limit.

I point out:

1) If you fully describe an invention on a forum it is public disclosure. Except for a company being able to improve it and patent that improvement, its validity can be challenged because of prior publication.
2) BW defence could have been: all the details of the QED are available publicly. ML is being both lazy and vexatious with this action and they should deal with the infringement by whoever as it chooses seeking their particular flavour of the QED to see if it infringes. I don't consent to the court order to provide this stuff for that reason. The fact that BW did not plead this points to a different situation.
3) People should be able to patent if they want. Of course the patent process causes delays in disclosure but as has been tacitly reasoned its either play the game by patenting or to publicly publish or as some may prefer more crudely "Give it to the Chinese" so they can mass produce them for $10 a piece and we can all be happy. The problem is when we do neither well.

IANAL but I am looking at putting together a project/ thread charter to mitigate the risk of a MD company doing to others as they have to BW and was discussing that with Philip for his thread. Despite the QED folks behaviour in rattling the bear's cage, some response to the action of ML needs to happen or we all run to our holes and they win by QED collateral damage.

Chudster

No offence taken ... I have already played with the lower speed sampling and for experimental purposes in PI & my advice is that they are too slow ....after all if you have a digital CRO do you run the sample rate at 100K ??? nope ... there is lots of interesting detail that is provided at the higher sample rates. If you think you know the "secrets of PI" then you can use the appropriate sampling rate as in this case it is being applied in context of what is known ... but .... as it has become obvious to me ... not everything is known about PI technology as some would have us believe ... so a little more bandwidth does not hurt for experimental purposes ... in this case it adds about $40 in cost.

"Patent examiners only read other patents for prior art ( and sometimes not all that thoroughly ) ..." .. quote from an actual patent examiner .... so dont hold your breath if you think a blog entry is public disclosure ... ( ie a collection of rambling text and diagrams VS a carefully worded and diagrammed technical specification comprising a patent application ). At the bare minimum it would have to be tested / examined and that costs money .... big money ... and for what ? no net gain if you are a hobbyist. If you are litigated against by one of these establishment player ... the real question is ... not if you can win .... but can you afford to lose ?

The same thing happened in software .. they tried / still are trying to nail Linux for years because it threatened the big monopolies .... so if you come up with an "open source" detector that threatens "the establishment" then dont expect you wont get a knock on the door.

So are you saying that a successful "open" detector project can be done ?

[moodz]

Hey Tinkerer,

Thanks for your response. Its helpful for the advancement of stuff with some real data to chew on.

In summary with a 2uS signal we need 1MSPS or higher to have a decent chance of detection or smaller targets. Moodz working at 2MSPS is 'overkill' but not unreasonable as its good to oversample (where is that antialiasing filter though Moodz). At 48KSPS, Aziz's soundcard seems OK for cannonballs with very long time constants (160uS) but not a tiny nugget.

Note we are talking about direct sampled solutions here. We can do all sorts of analog processing to have a half analog half digital solution (including integration) to mimic an analog detector with lower sample rates.

Chudster

FYI .. the antialiasing filter response can be seen in the spectral rolloff.

[chudster]

This accords with my experience. By the way I didn't want to cause offence, I meant crap design parameters. There is some good design evident, I just question the design parameters.




No offence taken ... I have already played with the lower speed sampling and for experimental purposes in PI & my advice is that they are too slow ....after all if you have a digital CRO do you run the sample rate at 100K ??? nope ... there is lots of interesting detail that is provided at the higher sample rates. If you think you know the "secrets of PI" then you can use the appropriate sampling rate as in this case it is being applied in context of what is known ... but .... as it has become obvious to me ... not everything is known about PI technology as some would have us believe ... so a little more bandwidth does not hurt for experimental purposes ... in this case it adds about $40 in cost.

I think you are 100% on target. If you look at past posts I put a stake in the ground that 1MSPS was likely to be a practical limit to the start of sampling. It makes sense when experimenting, as you are, to go higher than that.

"Patent examiners only read other patents for prior art ( and sometimes not all that thoroughly ) ..." .. quote from an actual patent examiner .... so dont hold your breath if you think a blog entry is public disclosure ... ( ie a collection of rambling text and diagrams VS a carefully worded and diagrammed technical specification comprising a patent application ). At the bare minimum it would have to be tested / examined and that costs money .... big money ... and for what ? no net gain if you are a hobbyist. If you are litigated against by one of these establishment player ... the real question is ... not if you can win .... but can you afford to lose ?

It will not stop the patent being granted if the examiner is not aware of it- and you are right that they mainly seek prior art from patents. Companies though are happy to sit on their patent as a threat but drop like a hot potato or don't start action if there is a valid threat to that patent's validity from prior public disclosure. There is an old adage that a patent's value is never known until a court tests it. I suggested in a past post that if BW wanted ML to go away, finding prior art would be a great strategy. There is also an obligation on the patent holder to disclose any prior art it knows about and if a company employee read a forum post they are on shaky ground if they don't disclose that to a patent examiner if it was relevant.

The same thing happened in software .. they tried / still are trying to nail Linux for years because it threatened the big monopolies .... so if you come up with an "open source" detector that threatens "the establishment" then dont expect you wont get a knock on the door.

But yet linux lives....

So are you saying that a successful "open" detector project can be done ?

I don't doubt it. You would need to get the legals right and perhaps avoid some of the alleged grandstanding of the QED folks. An issue with any hardware though is getting funding to make it as business people want a return and could get scared by past patent action. BTW, have you seen www.kickstarter.com ?

[Carl-NC]

Sorry if it is a tough question but:

What is the bandwidth needed to cover up to the smallest targets of interest (presumably small gold targets)?

Depends on how you approach the design. A simple traditional PI running at a pulse rate of 1kHz only needs a 1kHz ADC sample rate to grab a single flyback sample. This can be used to detect a 500us-tau target, a 50us-tau target, or a 5us-tau target, depending on where the sample is taken. Keep in mind that the sample-to-sample signal variation is very slow, only 10's of Hz in BW.

If you also want to grab a ground sample, then the required sample rate depends on the spacing and ground sample width. This could suddenly jack the rate up to 100-200kHz. Then if you want to grab multiple samples for channelizing the responses you might want to grab data at 1-2us intervals, so now up to 1MHz. But even sampling at 1us, the signal BW is pretty low.

[moodz]

Sorry if it is a tough question but:

What is the bandwidth needed to cover up to the smallest targets of interest (presumably small gold targets)?

Chudster

The bandwidth required is 1000 hurts at 10000 db ( dollar bills ) with at least 2 high power lawyers and a PA ( patent attorney ) section. And the small targets of interest are hobbyists like yourself.