A
Anonymous
Guest
Here is one of the log plots that I mentioned. I can
Find's Treasure Forums
Welcome to Find's Treasure Forums, Guests!
You are viewing this forums as a guest which limits you to read only status.
Only registered members may post stories, questions, classifieds, reply to other posts, contact other members using built in messaging and use many other features found on these forums.
Why not register and join us today? It's free! (We don't share your email addresses with anyone.) We keep email addresses of our users to protect them and others from bad people posting things they shouldn't.
Click here to register!
Need Support Help?
Cannot log in?, click here to have new password emailed to you
Changed email? Forgot to update your account with new email address? Need assistance with something else?, click here to go to Find's Support Form and fill out the form.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
A
Anonymous
Guest
Eric, what are we looking at here? Is this the decay curve after the first stage? How close were these targets to the coil?
A
Anonymous
Guest
Hi Charlie,
Yes, this is the decay curve after the first stage, which is a log amp. I can't remember how close the objects were but no more than two or three inches from the coil. PI's are famous for their great range but if you placed an object at the point where it was barely detectable with a standard system, the signal would be totally buried in noise. It is only by subsequently integrating over a series of samples that the object signal can be extracted. Recovering sufficient signal at late times on the decay curve may be a problem. What do you DSP experts think?
Eric.
Yes, this is the decay curve after the first stage, which is a log amp. I can't remember how close the objects were but no more than two or three inches from the coil. PI's are famous for their great range but if you placed an object at the point where it was barely detectable with a standard system, the signal would be totally buried in noise. It is only by subsequently integrating over a series of samples that the object signal can be extracted. Recovering sufficient signal at late times on the decay curve may be a problem. What do you DSP experts think?
Eric.
A
Anonymous
Guest
Eric, thanks. Thats what I thought. When I look at signals from targets at the limit of detection range I can not average enough to get a satisfactory signal to noise ratio. Another approach is to do horizontal averaging or boxcar integration. In this method you take many more samples of the signal per second than you need and then average them in horizontal "bins". This is the approach taken by the TEK TDS7xx scopes. The net result can be a significant enhancement in signal to noise ratio because not only are more bits of resolution obtained, but the process produces a low pass filtering action with the 3dB point roughly equal to sample rate /2. Even with this is difficult to see differences in the decay curve for distant objects. I have not tried this in about a year. Since then I have modified the detector considerably and it does have a better s/n. I'll go try a test right now and see what the results are.
A
Anonymous
Guest
Excellent plots Eric. That's exactly what I had anticipated it looking like. What kind of spectral noise density did the log amp have, and what was the approx. ending gain?
- Carl
- Carl
A
Anonymous
Guest
Group,I have just measured the difference in the return pulse (width) at the limits of the detector range using headphones. This pulse gets only 7nS wider at the onset of detection! This occurs long before there are any measurable amplitude effects. This may be due to the XL500 coil inductance which gives a pulse 26uS wide without any targets present. Perhaps if there were less coil inductance and less amplifier bandwidth the recovery would occur sooner and the risetime would be slower and with less noise. I don't know. Anyone else measured this? What were your results?
A
Anonymous
Guest
Hi Charles and All,
By
By
A
Anonymous
Guest
Eric, Thanks for the feedback. I did mean the back emf/saturation time sum. What I noticed was a very slight widening of this total pulse long before you could perceive any change in the shape of the decay.What effect AM I observing here? Is this due to very slight changes in the decay or some kind of intereaction like mutual inductance between the target and the coil?
A
Anonymous
Guest
Hi Charlie,
Interesting observation, the widening of the saturation pulse. I would expect this to change by a small amount due to the mutual inductance between object and coil. A few things to try:- Does it increase both for ferrous and non-ferrous metal? If it is mutual inductance, it should reduce for a chunk of aluminium or copper and increase for an iron or steel object. The other way of looking at it is to reduce the coil drive pulse until the back emf is lower than the avalanche voltage of the TX transistor (or use a higher voltage device), then look for changes in the peak voltage. It should increase for ferrous and reduce for non-ferrous.
Eric.
Interesting observation, the widening of the saturation pulse. I would expect this to change by a small amount due to the mutual inductance between object and coil. A few things to try:- Does it increase both for ferrous and non-ferrous metal? If it is mutual inductance, it should reduce for a chunk of aluminium or copper and increase for an iron or steel object. The other way of looking at it is to reduce the coil drive pulse until the back emf is lower than the avalanche voltage of the TX transistor (or use a higher voltage device), then look for changes in the peak voltage. It should increase for ferrous and reduce for non-ferrous.
Eric.
A
Anonymous
Guest
are not you fellows getting lost in bull#####?
I do not accept that a response has to come from the generating coil. There is sorts of things happening in there! Design a TX coil that shuts- up quickly. Dampen it so it does so. Then go and look and see what happens - with another coil.
I have duplicated Minelabs transmission pattern. I like edge-wound coils in kynar wire. They are very quick. I'm playing with Litz Rx coils and balanced-input amplifiers - before switching the RX on and off. Why all this talk of 709 amplifiers?? Have a look at Bill Whitlock's paper upon balanced-input amplifiers. Why this thrust towards speed?
There's nothing there - you turn your Rx on - then take a sample approx. 50 micro-seonds later. I don't see any droop on my responses. Noise yes, but it seems to all dissappear when you have you have gone through your first low-pass filter. (I am currently playing with LM833s]
g.
I do not accept that a response has to come from the generating coil. There is sorts of things happening in there! Design a TX coil that shuts- up quickly. Dampen it so it does so. Then go and look and see what happens - with another coil.
I have duplicated Minelabs transmission pattern. I like edge-wound coils in kynar wire. They are very quick. I'm playing with Litz Rx coils and balanced-input amplifiers - before switching the RX on and off. Why all this talk of 709 amplifiers?? Have a look at Bill Whitlock's paper upon balanced-input amplifiers. Why this thrust towards speed?
There's nothing there - you turn your Rx on - then take a sample approx. 50 micro-seonds later. I don't see any droop on my responses. Noise yes, but it seems to all dissappear when you have you have gone through your first low-pass filter. (I am currently playing with LM833s]
g.
A
Anonymous
Guest
[No message]
A
Anonymous
Guest
Hi Graeme,
Interesting comments, but likewise, I cannot accept your statements. The generating coil must have a response. Just like a metal object, a coil has inductance and resistance and therefore a time constant. At switch-off, the coil is looking into the damping resistor which is in parallel with the first amplifier input impedance and it is basically the coil L and this value of R which determines the switch off speed. The R that is required is determined by the self resonant frequency of the coil L and the associated parasitic capacitances. Hence low capacitance coils with short cables switch faster. After switch-off is initiated, the first thing you see is the coil decay which starts at a high voltage level. Then as the coil signal is decaying toward zero the object signal appears tacked on to the bottom end. The general idea is to design the circuit so that the coil decay is at least five time faster than the decay of the fastest object you want to detect. The point at which you start sampling is after the coil response has gone to zero and only the object signal is left.
Looking at the signal with another coil does not help as the decay from the transmitter coil is still coupled into it by the transformer principle.
If you are sampling at 50uS then I am not surprised that you don
Interesting comments, but likewise, I cannot accept your statements. The generating coil must have a response. Just like a metal object, a coil has inductance and resistance and therefore a time constant. At switch-off, the coil is looking into the damping resistor which is in parallel with the first amplifier input impedance and it is basically the coil L and this value of R which determines the switch off speed. The R that is required is determined by the self resonant frequency of the coil L and the associated parasitic capacitances. Hence low capacitance coils with short cables switch faster. After switch-off is initiated, the first thing you see is the coil decay which starts at a high voltage level. Then as the coil signal is decaying toward zero the object signal appears tacked on to the bottom end. The general idea is to design the circuit so that the coil decay is at least five time faster than the decay of the fastest object you want to detect. The point at which you start sampling is after the coil response has gone to zero and only the object signal is left.
Looking at the signal with another coil does not help as the decay from the transmitter coil is still coupled into it by the transformer principle.
If you are sampling at 50uS then I am not surprised that you don
A
Anonymous
Guest
Eric,
No criticisms were intended - just Kiwi vernacular on my part. I appreciate your reassoned response. I am a dabbler who simply strives for comprehension, and I was delighted to recently discover this site.
I like two coils - they can be individually critically dampened - and I like many paralled conductors. [Candy patented the use of Litz wire for coils]Coupling nulling is a critical physical adjustment, negating the transformer action you mention? I just don't like the idea of whacking the input and relying on diode protection. Balanced amplifiers with little gain are an attempt not to introduce crud. I open the Rx path after the balanced-input 15uS after cessation of each Tx pulse, and close it at the commencement of the next Tx pulse. No clipping, no saturation. My dabblings are heavily based on Candy's patent - 5,576,624 - and I would welcome contact from others pursduing his ideas.
He uses a 250uS Tx pulse, followed by four 60uS pulses. Various sampling periods at various delay timings follow. i.e. the 250uS Tx pulse is sampled after 50uS and the 60uS Tx pulses after 15uS. Tell all the Aussies using Minelab detectors they are not seeing everying! The multiple short Tx pulses no doubt provide the averaging described by Charlie Conger. I note your log plots started at 50uS.
What has interested me is the respective sensitivities of the differing Tx pulses. The long 250uS Tx pulse heavily favours silver over gold, while the shorter 60uS pulses heavily fav our gold over silver. Essentially what the Minelab provides is two detectors in one - with auto-selection between them. Cute.
More contributions to this site please!!!
g
No criticisms were intended - just Kiwi vernacular on my part. I appreciate your reassoned response. I am a dabbler who simply strives for comprehension, and I was delighted to recently discover this site.
I like two coils - they can be individually critically dampened - and I like many paralled conductors. [Candy patented the use of Litz wire for coils]Coupling nulling is a critical physical adjustment, negating the transformer action you mention? I just don't like the idea of whacking the input and relying on diode protection. Balanced amplifiers with little gain are an attempt not to introduce crud. I open the Rx path after the balanced-input 15uS after cessation of each Tx pulse, and close it at the commencement of the next Tx pulse. No clipping, no saturation. My dabblings are heavily based on Candy's patent - 5,576,624 - and I would welcome contact from others pursduing his ideas.
He uses a 250uS Tx pulse, followed by four 60uS pulses. Various sampling periods at various delay timings follow. i.e. the 250uS Tx pulse is sampled after 50uS and the 60uS Tx pulses after 15uS. Tell all the Aussies using Minelab detectors they are not seeing everying! The multiple short Tx pulses no doubt provide the averaging described by Charlie Conger. I note your log plots started at 50uS.
What has interested me is the respective sensitivities of the differing Tx pulses. The long 250uS Tx pulse heavily favours silver over gold, while the shorter 60uS pulses heavily fav our gold over silver. Essentially what the Minelab provides is two detectors in one - with auto-selection between them. Cute.
More contributions to this site please!!!
g
A
Anonymous
Guest
Hi, Where do I get Bill Whitlock's paper from? Thanks
A
Anonymous
Guest
Hi Graeme,
Welcome to the classroom. This is a place that we can all learn from our various discussions
Welcome to the classroom. This is a place that we can all learn from our various discussions
A
Anonymous
Guest
Hi Fred,
Go to www.jensen-transformers.com and look at "applications - white papers - AES papers.
G.
Go to www.jensen-transformers.com and look at "applications - white papers - AES papers.
G.
A
Anonymous
Guest
Eric,
Thankyou. Candy's patent re litz wire is 04890064. I've just finisged re-reading your letter in July/Aug '96 W.W. re stranded wire!! Probably both stranded and litz wire are an improvement over a single conductor. I use 10x paralled strands of litz in an effort to get R as low as possible. Similarly, my play with paralled strands of silver-plate kynar wire was an effort to get both low R and low C. The mosfet, IR830, certainly works and the coil noticeably warms.
Yes! I have external diodes across my input. The spikes you mention are the reason why I amplify a little, then turn the Rx on - the spikes are heavily attenuated and I can then chase the gain for, a to my mind, a cleaner signal to sample.
I cannot help re probing a Minelab as I have never seen one. Candy, thankfully, discloses in 5,576,624 a complete Tx and Rx timing diagram. He doesn't disclose how he does it though! It takes me 16 ics to duplicate his results.
Your comments are prompting me back to the bench to try nudging the sampling periods closer to the Tx closure. Fascinating stuff.
g
Thankyou. Candy's patent re litz wire is 04890064. I've just finisged re-reading your letter in July/Aug '96 W.W. re stranded wire!! Probably both stranded and litz wire are an improvement over a single conductor. I use 10x paralled strands of litz in an effort to get R as low as possible. Similarly, my play with paralled strands of silver-plate kynar wire was an effort to get both low R and low C. The mosfet, IR830, certainly works and the coil noticeably warms.
Yes! I have external diodes across my input. The spikes you mention are the reason why I amplify a little, then turn the Rx on - the spikes are heavily attenuated and I can then chase the gain for, a to my mind, a cleaner signal to sample.
I cannot help re probing a Minelab as I have never seen one. Candy, thankfully, discloses in 5,576,624 a complete Tx and Rx timing diagram. He doesn't disclose how he does it though! It takes me 16 ics to duplicate his results.
Your comments are prompting me back to the bench to try nudging the sampling periods closer to the Tx closure. Fascinating stuff.
g
A
Anonymous
Guest
I've studied all the patents on I think just about every metal detector. Whew! What I got out of the Candy ones is that the short pulses are an attempt at measuring the ground mineralization. The shorter sampling is to catch the signal from black sand which decays quicker. The four pulses are to help build up this signal which is weak compared to the single long pulse which is used to look for the goodies. (And it saves the battery) But the goodie signal needs the black sand signal removed from it, so these amplitudes (after being integrated/filtered) are each digitized and sent to a look up table {built up by trail and error}(some DSP huh?) to figure out if there is a goodie in the black sand and how big it is. If there is it makes a noise, and then you get to figure out which noise means what with your DSP. Add some equations to confuse the issue and patent it.
I use 20 strands of 30 A.W.G. (American Wire Gauge) Served litz wire, but then my coil is a 52 inch octagon.
JC
I use 20 strands of 30 A.W.G. (American Wire Gauge) Served litz wire, but then my coil is a 52 inch octagon.
JC
A
Anonymous
Guest
Wanted to add, I won't necessarily believe that the number of short pulses, the sampling times, or anything else for that matter, are the way the currently manufactured product is vs the patent. These things change and Minelab may at least make you buy one if you want to figure out the lastest revision. Anyway new products every year tells you they ain't got it quite right yet. The coil is not optimum for short and long pulses for example. The ground signal contains goodie signal in it too, so you get to make some guesses about how this detector is going to be used and set up the look up table to reflect the kinds of targets the user will be going after. If you get the wrong mix of targets, you may not get a response at all, even though you just put the stuff on the ground.
Great forum as usual
JC
Great forum as usual
JC
A
Anonymous
Guest
Thought about transformers, no improvement, some problems.
I use litz wire, but regular wire works fine, too. Plenty of other places to screw up and lose any advantage of litz wire.
Eric knows what he is talking about, listen to him.
Good forum
JC
I use litz wire, but regular wire works fine, too. Plenty of other places to screw up and lose any advantage of litz wire.
Eric knows what he is talking about, listen to him.
Good forum
JC