TGSL Experiments

[Jerry]

Hi Simon:

Did you add bypass caps on the power supply pins of each IC with this board? I seem to remember that your last board did not have them. Just wondering.

I still have some noise issues in my workshop. For some reason I was only getting about 28, cm with a 2 euro coin. A few weeks ago I was getting more.

Here is a few pic's of what I have built so far. The metal case is for testing purposes.

Jerry

[Jerry]

Signal Strength Question

A few weeks ago I did some testing and recorded some phase delay and null readings. After looking at the data for a while, I decided to put it in graph form to see how it looked. The results were somewhat surprising in that the -20 degree point on this particular coils is pretty far removed from the deepest part of the null. What was interesting is the point of lowest null voltage was not point of the best air test distance.

I would really like to add another curve to this graph showing relative signal strength at some point in the circuit but cannot think of a place to measure that will give good results.

Signal strength should correlate to air test distance. I hesitate to use air testing since it is somewhat subjective at how the ticks and beeps are heard.

I am thinking about removing the limiting diodes from u103 and 104/5 for testing purposes and measuring the outputs of u104/5.

The coil I made is: tx 5.79 mh, rx 6.23 mh. Tx freq is 14.4 Khz.

Anyway, here is the chart I made of my TGSL coil. It would be interesting to see what others get. Also I plan on repeating this with date for both inside and outside of the null point.

I hope you find it of interest.

Jerry

[Qiaozhi]

That looks correct. Approximately 12mV at 20 degrees.
Try monitoring the pre-amp output to get the relative signal strength.

[simonbaker]

Hi Simon:

Did you add bypass caps on the power supply pins of each IC with this board? I seem to remember that your last board did not have them. Just wondering.

I still have some noise issues in my workshop. For some reason I was only getting about 28, cm with a 2 euro coin. A few weeks ago I was getting more.

Here is a few pic's of what I have built so far. The metal case is for testing purposes.

Jerry

Hi Jerry:

No, I do not have the bypass caps. It is worth trying.

A year or two ago I was extensively testing such bypass caps using a solderless breadboard circuit of the TGSL, which really showed huge voltage spikes everywhere generated by the 4024 chip and the -5V supply circuit. Even with bypass caps, it was hard to eliminate the spikes. But the odd thing was that the bypass caps did not change the operation of the circuit at all that I could observe. So I get the feeling that the spikes, being perfectly synchronous, are not a big problem -- although you would think they would be. The spikes are also extremely high frequency, and so maybe don't affect much of the detection circuitry. But I should definitely try the caps anyway.

The noise I am most aware of definitely seems to come from the RX coil picking it up like an antenna. You can test this by shorting out, or disconnecting, the RX coil and looking at the output of the LM308s with a scope set to about .1 Second/division. The noise becomes very small.

Also, I'm pretty sure the noise does not come from oscillator jitter or amplitude modulation coupled between the TX and RX coil, because I substituted a choke coil for the TX coil (not near the RX coil), and the RX noise was still there.

Also, when I go to the mountains, the noise seems to drop way down and the detection depth goes up.

This RX noise reduces the detection depth from about 30 cm to about 15 cm in my workshop. However, there may be other types of noise, such as circuit noise, that are significant when you are in an EMI-free area and want to get 30 cm depth. I would need a mobile oscilloscope to study that.

Many people have reported that mysteriously some days are more noisy than others in their workshop, and the depth measures less.

---------------

Hey, really nice PCB and case! Another fine piece of workmanship. So far it seems that workmanship is an important key to making a really good TGSL -- at least the people that make the prettiest machines also seem to make the best performing. I'm trying to defy that rule, but not successful yet.

I have a feeling you have a nice MD in the making.

Regards,

-SB

[Jerry]

I agree completely that the receive coil picks up most of the noise. I found that unplugging the power for my travel trailer makes a big difference. It is parked next to my workshop and has an inverter for charging the battery. The inverter is at least 50 feet away from the metal detector and my workshop is inside a metal building but taking the inverter off line makes for about 5 - 6 cm improvement in air test distance easy. So you never know.

I was going to add the bypass caps since it is pretty standard fair in most other circuits I have built. Will let you know the results.

Jerry

p.s. I managed to break of a few chunks of ferrite about 1 cm in diameter. Will glue them to a stick next week. Right now I have to get packing for the weekend.

[simonbaker]

A few weeks ago I did some testing and recorded some phase delay and null readings. After looking at the data for a while, I decided to put it in graph form to see how it looked. The results were somewhat surprising in that the -20 degree point on this particular coils is pretty far removed from the deepest part of the null. What was interesting is the point of lowest null voltage was not point of the best air test distance.

I would really like to add another curve to this graph showing relative signal strength at some point in the circuit but cannot think of a place to measure that will give good results.

Signal strength should correlate to air test distance. I hesitate to use air testing since it is somewhat subjective at how the ticks and beeps are heard.

I am thinking about removing the limiting diodes from u103 and 104/5 for testing purposes and measuring the outputs of u104/5.

The coil I made is: tx 5.79 mh, rx 6.23 mh. Tx freq is 14.4 Khz.

Anyway, here is the chart I made of my TGSL coil. It would be interesting to see what others get. Also I plan on repeating this with date for both inside and outside of the null point.

I hope you find it of interest.

Jerry

Would you describe more how you conducted the test and what you measured to make the graph?

Because of the noise in my work area, I can't do a sensitive test of air depth vs. null point. But I have never yet seen the fine particulars of the null point make a difference in the depth. And theoretically, I have not seen a convincing argument why it should. So I'm interested in your observation.

Your "merry-go-round" is a valuable tool for testing this. I recommend monitoring the output of the LM308's (DISC and GB channels) to see the target pulse, with scope sweep at about .1 S/div (or something like that, whatever is easiest). With your targets rotating (I would use a single target), you should be able to measure a pretty consistent pulse amplitude.

You can remove those diodes if you want -- it would help extend the range of measurements and make the waveforms more informative, but not necessary if you test with small signals.

Then adjust the null slightly around the minimum and see how it affects the pulses at the outputs of the LM308s.

-----------------

I don't know what to make of your graph yet because I am thinking these days about a model for the null signal, which I'll maybe discuss again in another thread. But basically I'm speculating that it may be useful to view the null as the sum of three basic components:

Vair + Vcon + Vcap

Where:

Vair is the pure electromagnetic coupling of the RX and TX coil. It's phase should not change with shifting the coil positions except that the polarity reverses suddenly as you cross the point where the magnetic field balance changes. It's phase is mainly determined by the RX resonant circuit. However, it's amplitude does change continously as you shift the positions of the coils while nulling.

Vcon is a signal due to any conductive target, including the ground and our foil shields, that electromagnetically couple the RX and TX coils. I don't think its phase should change much as you shift the coils, but not sure. It's amplitude probably doesn't change much. It may be useful to divide this into Vshield and Vground, as distant targets couplings shouldn't change significantly as you shift the coils.

Vcap is a capacitive coupling due to things like cable capacitance, shield capacitance, etc. It's amplitude and phase may not change significantly as you adjust coil positions.

The key concept is that the phase shift we see in the null signal is really due to an amplitude and polarity change in Vair added to some relatively small constant signals. There is no single signal whose phase is actually being shifted. It is the changing amplitude of one of the components of the sum that makes the resultant null signal shift phase.

Another important conclusion is that ferrite should only affect Vair and Vcon, not Vcap. This means that you only really have two choices of null phase for ground balancing corresponding to the two polarities of the null signal. The intermediate null phases are due to the Vcap signal probably, but you cannot ground balance to that intermediate phase.

I believe this explains why you can adjust the null signal phase over a wide range but always end up with the ground balance pot in the same position -- because the Vair and Vcon signal phase is fixed by the RX resonant circuit and doesn't really vary as you adjust the coils, except for a polarity reversal.

If you really need to shift the "magnetic null" phase for ground balance reasons, the only practical way I know is to put a piece of metal on your coil, like mikebg once mentioned.

Sorry for all the words, but I've been pondering this, that's my current thinking, subject to change of course.

-SB

[Jerry]

Hi Simon:

I ran this phase/nulling test a while before you proposed your test with the ferrite. My thoughts at that time were to see if there was a solid relationship between the nulled voltage and phase. I am confident that at least on the test coil, I can set the null voltage and be the null phase delay will follow the chart.

It was a sample of one which statistically means nothing. However I was hoping to run the same thing on other coils or better yet have others who have these nice scopes that read out all the parameters including phase perform the same test.

The test conditions were pretty basic and simple:

1. Coil connected to TGSL.
2. Channel one of the scope connected to Tx J1-1
3. Channel two of the scope connected across Rx terminals 1 & 2 of j2.
4. Digital voltmeter connected also to J2 terminals.
5. The scope is triggered off of channel one. (Tx)
6. With TGSL turned off, both scope channels were centered on the baseline so the zero crossing on the retical was as close to 50% of max/min as I could make it.
7. I spread out the time base so that one half of one Tx cycle nearly filled the scope screen. Then counted the units for the half cycle and doubled it for a full cycle.
8. Next I counted the scope units between the two zero crossings and divided by the full cycle number and the multiplied X 360.

I think I am right on determining the phase. If not I would appreciate some help.

BTW I agree that the null phase does not affect depth all that much. I was getting good air tests with much larger than 20 degree delay.

I keep forgetting: I have mentioned the specs on my coil except for one. It is 8 inch (20 cm) in diameter. Once I get this potted and working, I plan on trying another 10 inch.

Jerry

[raff33]

Hi frends,
can modified TGSL for detect only gold not other metals ?
REGARDS

[simonbaker]

Hi Simon:

I ran this phase/nulling test a while before you proposed your test with the ferrite. My thoughts at that time were to see if there was a solid relationship between the nulled voltage and phase. I am confident that at least on the test coil, I can set the null voltage and be the null phase delay will follow the chart.

It was a sample of one which statistically means nothing. However I was hoping to run the same thing on other coils or better yet have others who have these nice scopes that read out all the parameters including phase perform the same test.

The test conditions were pretty basic and simple:

1. Coil connected to TGSL.
2. Channel one of the scope connected to Tx J1-1
3. Channel two of the scope connected across Rx terminals 1 & 2 of j2.
4. Digital voltmeter connected also to J2 terminals.
5. The scope is triggered off of channel one. (Tx)
6. With TGSL turned off, both scope channels were centered on the baseline so the zero crossing on the retical was as close to 50% of max/min as I could make it.
7. I spread out the time base so that one half of one Tx cycle nearly filled the scope screen. Then counted the units for the half cycle and doubled it for a full cycle.
8. Next I counted the scope units between the two zero crossings and divided by the full cycle number and the multiplied X 360.

I think I am right on determining the phase. If not I would appreciate some help.

BTW I agree that the null phase does not affect depth all that much. I was getting good air tests with much larger than 20 degree delay.

I keep forgetting: I have mentioned the specs on my coil except for one. It is 8 inch (20 cm) in diameter. Once I get this potted and working, I plan on trying another 10 inch.

Jerry

Thanks, good description.

What do you mean by "Channel two of the scope connected across Rx terminals 1 & 2 of j2"? I can't visualize that with a single probe.

I believe in your data for phase vs. voltage, I just don't know what the significance is. It could be useful for us to compare such graphs and see how they correlate with our perceived performance of our coils. However, I'm still feeling that a particular coil's null phase can be deceptive to make judgments on, so I'm cautious about thinking of the graph as an important design criteria. But still interesting and possibly important if we can correlate it to some benefit or problem.

Right now my guess is that adjusting coils to achieve a particular absolute phase is misleading. I'm thinking that what we really want to do is to pick a phase polarity (of the underlying magnetic phase signal), and then pick an offset from minimum (of that same signal) to settle on. The trick is that we can't directly see that underlying magnetic phase component, and I think people chase the overall null phase desperately without good effect. For some coil contructions it works fine, others it is frustrating because it can't be achieved as desired. But both types of coils may be fine.

I'd like a better understanding of which null polarity is best, and why. And what offset is best, and why. For air tests, I have not seen a significant difference. I'm suspecting that for searching over real ground and for ground balancing, maybe that is where the choice matters. But I don't know. I'd like to hear a theory and then see it born out in real tests (which doesn't prove the theory of course).

So still lots of fun to be had with the good old TGSL.

Regards,

-SB

[simonbaker]

Hi frends,
can modified TGSL for detect only gold not other metals ?
REGARDS

Use TGS circuit with notch for best chance at that. See first post http://www.geotech1.com/forums/showthread.php?t=12692

-SB

[simonbaker]

more null thoughts

not that anyone cares...

I have to revise/expand on what I said above on nulling (as usual).

I said practically we just have two choices of null phase as far as ground balance is concerned, because a component of the null is capacitively coupled and not affected by ground/ferrite, and the electromagnetically coupled component only has two polarities and fixed phase for each polarity.

I spoke too hastily about the electromagnetically coupled component. I previously described it as being possibly modeled by two components, Vair and Vcon (pure mutual inductance of the coils (Vair) and a conductive coupling due to presence of some target, specifically our shields).

Adjusting the coil overlap adjusts the magnitude of these two signals. We're familiar with Vair -- it dominates and we can see the effect clearly. Vcon is smaller and hard to distinguish from Vcap.

Let's suppose that Vair and Vcon both go to zero at some overlap of the coils. Then the statement that we have only two null phases for ground balance seems true.

But probably they don't zero at the same overlap. So the combined signal will never be zero, rather probably shift continuously from one phase extreme to the other.

Because Vair and Vcon are both due to electromagnetic coupling, it is reasonable to assume they both will be affected by ground/ferrite.

Therefore, we may have some choice of choosing a null phase for ground balance purposes. But it depends on the size and phase of these signals how practical that is vs. really a choice of two dominant null phase-polarities.

As to what null phase to choose, if we have the opportunity -- another story...


-SB

[Jerry]

Hi Simon:

I dont think the phase vs voltage graph is of any real significance other than an interesting chart unless it is taken further and another curve is added to show either relative signal strength or detection depth. I was just messing around a couple of weeks ago and for once decided to actually write something down

You are correct, it is only meaningfull for this particular coil unless or until other coils are charted and compared.

On connecting the scope to the receive terminals, I should have stated it the same was as for the tx. Signal and ground on terminals 1 & 2 respectively.

Just got back from a long weekend so have not got settled in yet.

Jerry

[simonbaker]

Hi Simon:

I dont think the phase vs voltage graph is of any real significance other than an interesting chart unless it is taken further and another curve is added to show either relative signal strength or detection depth. I was just messing around a couple of weeks ago and for once decided to actually write something down

You are correct, it is only meaningfull for this particular coil unless or until other coils are charted and compared.

On connecting the scope to the receive terminals, I should have stated it the same was as for the tx. Signal and ground on terminals 1 & 2 respectively.

Just got back from a long weekend so have not got settled in yet.

Jerry

Hi Jerry!

Thanks for answers. Just to be clear, do you have a ground wire on each probe, and are both ground wires connected to circuit at same time?

I'm wondering, because I have trouble trying to measure both those voltages without grounding together two points that aren't directly grounded on the PCB. I need more probes so I can use differential probes on the RX coil. So instead I usually measure the voltage of the LF353 pin 7 instead relative to single ground.

Just interested in how others do it.

Regards,

-SB

[Jerry]

Hi Simon:

Yes I do. I have an adapter with which I connect the voltmeter in parallel with the oscilloscope. I use a BNC to Banna Jack adapter. Dont know how else to describe it. I can take a picture of it tomorrow.

All that is connected directly to the detector at J2 is the scope probe. (J2-1) Now that I think about it, the scope ground is chassis ground on the detector but it could be connected to J2-2 . The digital volt meter is connected at the input to the scope but is measuring the same voltage.

I will set it up and snap a picture. That will do more than I can describe I think.

Jerry

I just re-read your post. Right now I am grounding the rx in the coil head like Ivconic and dfbowers describe it. Have tried it pretty much all ways and this seems to work well so I have left it that way.

[simonbaker]

Hi Simon:

Yes I do. I have an adapter with which I connect the voltmeter in parallel with the oscilloscope. I use a BNC to Banna Jack adapter. Dont know how else to describe it. I can take a picture of it tomorrow.

All that is connected directly to the detector at J2 is the scope probe. (J2-1) Now that I think about it, the scope ground is chassis ground on the detector but it could be connected to J2-2 . The digital volt meter is connected at the input to the scope but is measuring the same voltage.

I will set it up and snap a picture. That will do more than I can describe I think.

Jerry

I just re-read your post. Right now I am grounding the rx in the coil head like Ivconic and dfbowers describe it. Have tried it pretty much all ways and this seems to work well so I have left it that way.

I'll look at your pics because I agree its hard to describe exactly.

My questions involve the meaning of RX signal phase -- measuring one input of the LF353 is not necessarily the whole story because it is differential amp -- unless one input is grounded. But maybe matters where it is grounded (coil shield vs. PCB ground) -- and some people use ungrounded RX coil too. And two probe grounds of course could connect together two points of circuit directly.

Probably not a big issue but something I wonder about when taking phase measurements.

Regards,

-SB