Sign In
Register
Help
GP3000 Technical
MultiQuote
#2
Robby_H 
- Member
-
- Group: Members
- Posts: 36
- Joined: 12-May 04
Posted 28 March 2005 - 05:49 PM
The GP series transmit a pulse train made up of one long pulse followed by 3 short pulses and this train is around 900 usec long. A complete analysis is made during this time and then further averaged. So depending on how you look at it, 1100 pps or 4400 pps.
#4 Guest_Tony_*
Posted 29 March 2005 - 06:40 PM
QUOTE (Robby_H @ Mar 28 2005, 05:49 PM)
The GP series transmit a pulse train made up of one long pulse followed by 3 short pulses and this train is around 900 usec long. A complete analysis is made during this time and then further averaged. So depending on how you look at it, 1100 pps or 4400 pps.
Hi Robby.
Could you be good enough to answer some questions I have about Ml pi detectors following your response to wirechief I would like to explain if you could how you calculate the cycle time for the 3 short and one long pulse. As I understand the short pulses are 40micro seconds and the long pulse 240 micro seconds, total duration of short and one long pulse=360 micro seconds. What I don’t know is for each pulse what is the sampling delay in micro seconds after pulse off and the width of the sampling window after each pulse. I guess that the sampling delay after pulse off would be about 50 microseconds giving 360+200=560 micro seconds+ additional sampling time after 50 microseconds. Thus for 2 pulse trains per cycle a total duration of at least 1120 microseconds. I have no idea if this calculation is correct or not.
The other questions that I would very much appreciate your help in answering concern the frequency spectrum of ML pi detectors. I guess by definition that the transmit frequency is broadband, what I don’t know is what is the frequency range of the transmit signal. Is it say Dc to at least 1MHz or more or less and what do you think that the amplitude vs frequency response curve for the TX signal would look like. I guess that I really want to know if there is a “average centre frequency” of maximum amplitude like for example in UWB radars.
Best regards.
Tony
#5
Robby_H
- Member
-
- Group: Members
- Posts: 36
- Joined: 12-May 04
Posted 29 March 2005 - 08:48 PM
Sheeze! Maybe this should be taken private or to the pi forum.
I have only bothered looking at the GP once and it was with an old HP cro on the field, as I couldn’t afford to lug the TEK digital around. I know the HP had a few problems and my figures may not be spot on. I got a 280usec pulse (you say 240) followed by three 40usec pulses and the off times were 240, 84, 84, 84. I now have access to all ML pi and should have another look see sometime. Nobody has publicly made it known what the sample times and lengths are and there are several different ways he could be attacking this. His US patent shows one attack and is worth a read but I believe it has errors. His old Australian patent makes his intentions much, much clearer.
He states that sampling should wait at least 3 times the coil’s clamped TC but believe he starts much earlier. I suspect that he starts his primary sampling at or very, very shortly after the knee of the curve to zero as seen before amplifying, not at the false knee seen after a pre-amp and uses very clever circuit techniques to almost overcome the obvious problems with the change in coil properties near mineralised ground.
Pi detectors are broadband by nature and this is why they pick up so much interference. I’m not sure what you are asking here and it might be worth an ask on the pi forum but the frequency component may well be calculated off the 11th harmonic determined by the rate of switch off but there isn’t any advantage in switching off any faster than you need for that particular design and Candy switches off his pulse much slower than he could have. A ML coil may have a resonant frequency of 500khz and a Q of 5 but the inductance and coil properties change dramatically in the presence of magnetic minerals so what you ask is a bit beyond me.
I have only bothered looking at the GP once and it was with an old HP cro on the field, as I couldn’t afford to lug the TEK digital around. I know the HP had a few problems and my figures may not be spot on. I got a 280usec pulse (you say 240) followed by three 40usec pulses and the off times were 240, 84, 84, 84. I now have access to all ML pi and should have another look see sometime. Nobody has publicly made it known what the sample times and lengths are and there are several different ways he could be attacking this. His US patent shows one attack and is worth a read but I believe it has errors. His old Australian patent makes his intentions much, much clearer.
He states that sampling should wait at least 3 times the coil’s clamped TC but believe he starts much earlier. I suspect that he starts his primary sampling at or very, very shortly after the knee of the curve to zero as seen before amplifying, not at the false knee seen after a pre-amp and uses very clever circuit techniques to almost overcome the obvious problems with the change in coil properties near mineralised ground.
Pi detectors are broadband by nature and this is why they pick up so much interference. I’m not sure what you are asking here and it might be worth an ask on the pi forum but the frequency component may well be calculated off the 11th harmonic determined by the rate of switch off but there isn’t any advantage in switching off any faster than you need for that particular design and Candy switches off his pulse much slower than he could have. A ML coil may have a resonant frequency of 500khz and a Q of 5 but the inductance and coil properties change dramatically in the presence of magnetic minerals so what you ask is a bit beyond me.
#6 Guest_Tony_*
Posted 30 March 2005 - 12:07 AM
QUOTE (Robby_H @ Mar 29 2005, 08:48 PM)
Sheeze! Maybe this should be taken private or to the pi forum.
I have only bothered looking at the GP once and it was with an old HP cro on the field, as I couldn’t afford to lug the TEK digital around. I know the HP had a few problems and my figures may not be spot on. I got a 280usec pulse (you say 240) followed by three 40usec pulses and the off times were 240, 84, 84, 84. I now have access to all ML pi and should have another look see sometime. Nobody has publicly made it known what the sample times and lengths are and there are several different ways he could be attacking this. His US patent shows one attack and is worth a read but I believe it has errors. His old Australian patent makes his intentions much, much clearer.
He states that sampling should wait at least 3 times the coil’s clamped TC but believe he starts much earlier. I suspect that he starts his primary sampling at or very, very shortly after the knee of the curve to zero as seen before amplifying, not at the false knee seen after a pre-amp and uses very clever circuit techniques to almost overcome the obvious problems with the change in coil properties near mineralised ground.
Pi detectors are broadband by nature and this is why they pick up so much interference. I’m not sure what you are asking here and it might be worth an ask on the pi forum but the frequency component may well be calculated off the 11th harmonic determined by the rate of switch off but there isn’t any advantage in switching off any faster than you need for that particular design and Candy switches off his pulse much slower than he could have. A ML coil may have a resonant frequency of 500khz and a Q of 5 but the inductance and coil properties change dramatically in the presence of magnetic minerals so what you ask is a bit beyond me.
I have only bothered looking at the GP once and it was with an old HP cro on the field, as I couldn’t afford to lug the TEK digital around. I know the HP had a few problems and my figures may not be spot on. I got a 280usec pulse (you say 240) followed by three 40usec pulses and the off times were 240, 84, 84, 84. I now have access to all ML pi and should have another look see sometime. Nobody has publicly made it known what the sample times and lengths are and there are several different ways he could be attacking this. His US patent shows one attack and is worth a read but I believe it has errors. His old Australian patent makes his intentions much, much clearer.
He states that sampling should wait at least 3 times the coil’s clamped TC but believe he starts much earlier. I suspect that he starts his primary sampling at or very, very shortly after the knee of the curve to zero as seen before amplifying, not at the false knee seen after a pre-amp and uses very clever circuit techniques to almost overcome the obvious problems with the change in coil properties near mineralised ground.
Pi detectors are broadband by nature and this is why they pick up so much interference. I’m not sure what you are asking here and it might be worth an ask on the pi forum but the frequency component may well be calculated off the 11th harmonic determined by the rate of switch off but there isn’t any advantage in switching off any faster than you need for that particular design and Candy switches off his pulse much slower than he could have. A ML coil may have a resonant frequency of 500khz and a Q of 5 but the inductance and coil properties change dramatically in the presence of magnetic minerals so what you ask is a bit beyond me.
Thank you very much for your reply,sorry to be such a pain in the arse!
Best regards to you,
Tony
#9
Robby_H
- Member
-
- Group: Members
- Posts: 36
- Joined: 12-May 04
Posted 30 March 2005 - 08:59 PM
Yes Timedomain, this is covered in Candy's patent if Tony wishes to plough thru it and best of luck to him!
If using his method of transmission, one of the effects from the rise and fall times of current during the short pulse compared with that of the long pulse is probably overlooked by most but becomes rather obvious in practice and he doesn't appear to disclose this at all or at best skips over it.
The Oz patent is simpler to grasp and proposes (1) a method similar to Eric Foster's pi in that it subtracts a late sample from the early one in order to obtain GB. He then goes on to propose (2) another method where a sample taken after one pulse length is compared with a sample taken after a noticeably different pulse length. He then proposes that the two ideas can be combined in one pi.
Combining the two means that (1) can be attacked in different ways and (2) opens up new possible variations that could be used.
Typo in my reply... should have been "damped TC" not "clamped TC".
.
If using his method of transmission, one of the effects from the rise and fall times of current during the short pulse compared with that of the long pulse is probably overlooked by most but becomes rather obvious in practice and he doesn't appear to disclose this at all or at best skips over it.
The Oz patent is simpler to grasp and proposes (1) a method similar to Eric Foster's pi in that it subtracts a late sample from the early one in order to obtain GB. He then goes on to propose (2) another method where a sample taken after one pulse length is compared with a sample taken after a noticeably different pulse length. He then proposes that the two ideas can be combined in one pi.
Combining the two means that (1) can be attacked in different ways and (2) opens up new possible variations that could be used.
Typo in my reply... should have been "damped TC" not "clamped TC".
.
#12
Wirechief
- Advanced Member
-
- Group: Members
- Posts: 281
- Joined: 27-March 05
Posted 08 May 2005 - 07:10 PM
Hello 1Dredger, I'm just fascinated with the theory and electronics of metal detectors. Nobody has to worry about me trying to get their trade secrets because I'm just a technician not an engineer. I have no intention of trying to build anything but would'nt mind being able to tinker a bit. Heck you can't even get schematics of these instruments, I'm a Tesoro dealer and they don't share any tech stuff at all with me. I understand why they don't and won't blame them for it. I am really interested in the P.I. stuff. Wish I could afford a GP3000 but will have to wait a long time for that. Anyway thanks for your interest 1Dredger, God Bless. Wirechief.
#14
grubstake
- Advanced Member
-
- Group: Members
- Posts: 1,108
- Joined: 20-January 04
Posted 08 May 2005 - 07:59 PM
Well, if the inside is like my 2200sd, everything is spray painted on the circut boards, so you can't see any #s or anything else on the chips, or any ID or any kind on the electronics. But I know when I hit a nugget, under the coil, it go's hello! hello1 and I dig!. Grubstake
#15 Guest_Guest_*
Posted 08 May 2005 - 08:10 PM
Goto
http://www.thunting....ile=patents.dat
And look at US5576624 and others by Candy and Minelab.
If this doesn't take you to the correct page then go metal detectors/patents etc.
Like I said before, this isn't the easiest to understand. His original Oz patent made his intentions much clearer but I have never seen this on-line anywhere.
http://www.thunting....ile=patents.dat
And look at US5576624 and others by Candy and Minelab.
If this doesn't take you to the correct page then go metal detectors/patents etc.
Like I said before, this isn't the easiest to understand. His original Oz patent made his intentions much clearer but I have never seen this on-line anywhere.
#16 Guest_Guest_*
Posted 08 May 2005 - 08:35 PM
QUOTE (Guest @ May 8 2005, 08:10 PM)
Goto
http://www.thunting....ile=patents.dat
And look at US5576624 and others by Candy and Minelab.
If this doesn't take you to the correct page then go metal detectors/patents etc.
Like I said before, this isn't the easiest to understand. His original Oz patent made his intentions much clearer but I have never seen this on-line anywhere.
http://www.thunting....ile=patents.dat
And look at US5576624 and others by Candy and Minelab.
If this doesn't take you to the correct page then go metal detectors/patents etc.
Like I said before, this isn't the easiest to understand. His original Oz patent made his intentions much clearer but I have never seen this on-line anywhere.
Go to the Australian patents office and look up the following patents:
200179440 A1
200179439 A1
200179426A1
200179376A1
They are available for download as PDF files. They will however be old hat by next year as the next generation of Pi detectors hits the market both from ML and some totally new players.
#18 Guest_Guest_*
Posted 08 May 2005 - 09:11 PM
QUOTE (Guest @ May 8 2005, 08:48 PM)
What do you think the next generation will offer??
Better depth,better sensitivity, correct elimination of the magnetic ground signal components without compromising depth(other than due to losses caused by the magnetic minerals themselves,not much we can do about this) or failing to detect some conductive targets that have very short decays(like most of our present Pi detectors do) because the algorithms used to remove the ferrite signal components are at best a compromise and at worst wrong and the sampling delays are too long.correct rejection of ferromagnetic signals,better immunity to external interferences and the capability to detect large conductive targets at very much greater depths by using much longer transmit pulses and sampling at much shorter times after pulse turn off and by using much better coils. The next generation coils will soon be available in the US
#19
Wirechief
- Advanced Member
-
- Group: Members
- Posts: 281
- Joined: 27-March 05
Posted 09 May 2005 - 03:03 PM
Hello Guests and Grubstake, I sure thank you all for the info and I will checkout those patent websites. That's too bad that they paint the circuit board, looks like it would make it hard on their repair techs. I'm not interested in trying to copy anyone's work but I am curious and want to learn as much as I can about these detectors. I wonder how long it takes a 1 ounce gold nugget to decay out completely? Is there actually enough eddy currents in the nugget once the detector turns back on that it induces a field into the coil? That has to be one sensitive instrument. If I have this wrong someone please correct me. Thanks guys for your help. Wirechief. 