TGSL Tuning & troubleshooting

[Davor]

Try bobbing the coil over a real ground when fine tuning the ground balance. You'll be surprised how good it works.

So where is the catch?!

Easy. Ferrite has no conductivity, so a ground balance setting achieved using ferrite would be 90° sharp. Real ground has some conductivity, so it's vector is not exactly at 90° but a bit towards the real side, or more ohms on your GEB potentiometer. The biggest catch of all is that this vector angle will not change with elevation of a coil above ground, so bouncing the coil up and down will induce some voltage in a GEB channel if your ground balance is not zeroed against that particular soil. When you hit the bulls eye, your toy will be silent, and your targets just perfect.

[Jerry]

Got it figured out! The GEB setting was too aggressive. Backing down the ground balance setting a bit and it went away completely. By backing down I mean having the tendency to accept more of a ferrite signal. I had it set beyond the cutoff point for ferrite. So I was not dreaming things but at the same time, the detector was not set up correctly.

This just re-enforces the importance of a good ground balance for me.

Jerry

[Jerry]

I have to add a footnote to my last post. I did some more testing this morning and the audio loudness vs target conductivity was no where near as pronounced as it was yesterday. It was there to some extent but if not looking for it, could easily be passed over. I have to make sure that it isn't bias creeping into my testing.

We have company coming later today but will do some more testing to verify if what I posted yesterday was a false alarm or not.

I would like to be able to duplicate it but if I cannot, then it was a false alarm. Either way, I like Davor's mod.

Jerry

[Jerry]

What you can do is replacing R44 with a real potentiometer. For my IGSL I bought a real stereo pot (2 channels) because variable R44 provides a real audio level control. OK, my values will be somewhat different, and I did some other changes as well. Anyway, if you want to experiment, replace R44 with 100k lin pot, and replace C26 with 22nF. Leave everything else as is. Or you may use my other values and replace R44 with 10k lin pot, replace R42 with 10k, R43 with 47k and leave everything else as is (don't replace C26). In case R44 is turned down to 0ohm, LM358 turns into a comparator and produces pure binary response.

When I tested this earlier, I was not completely happy with the results but not for any of the changes suggested above. I changed the value of the filter feedback resistors R29 and R31 on U 103 back to 470k which is what Ivconic had on the original TGSL schematic.

Finally got the detector mounted on a new set of shafts and tested it for a while this afternoon.

The proportional audio based on target depth is the same as my last test with one big exception. With the pot that is installed to replace R44 turned to about the middle of it's rotation, I am getting a very noticeable change in audio volume in relation to the conductivity of the target. Nickels, and pull tabs (both ring and figure eight type) sound off much louder than a U.S. Quarter dollar at the same depth. This behavior may have been there last time but I did not notice it.

When I turn R44 (25k pot) all the way up so there is saturated audio this effect goes away. Also I checked the same targets with my original TGSL and the beep tones sound pretty much the same for all of the mentioned targets.

I have always thought that the TGSL 14.5 khz frequency made it hit hard on nickels and pull tabs but this is the most dramatic evidence I have seen for that. This change is a keeper. I think there is still more to be gained right in the area of U107 so will keep experimenting and trying a few more or your suggestions on my test bench TGSL which is almost ready to smoke test.

I like the way it is looking right now.

Jerry

[Jerry]

Tweek to U107, R44

I had a few minutes this morning so I changed out the 100K pot and replaced it with a 22 log taper and a 470 ohm in series with it. This combination replaces R44.

Took it out and tried it for about a half hour and it appears to be the combination what I was looking for. With a minimum resistance of 470 the beep tone gets much louder but it is still pleasant to listen too. Turning the pot to max resistance quiets the beep tone down a lot. This combo is usable over the entire rotation of the pot. I used a 22k Log taper because that is what I had that fits into the space available. Works well enough that that it not worth the bother of changing it for a linear one.

Right now I am really happy with the results. Will use it for two days of detecting this weekend so will get a chance for more complete evaluation.

I am out of time right now or I would have drawn this up. I can if anyone is interested.

Jerry

[Jerry]

Hey Jerry,

I like your prototyping work. Good idea. Thanks for sharing this.
Cheers,
Aziz

Thanks, but I cannot take credit for it. I got the idea from this guys article: http://www.k7qo.net/manart.pdf

I have used it for building radio receivers so there is no reason why it will not work for a metal detector.

Jerry

[Aziz]

Hey Jerry,

I like your prototyping work. Good idea. Thanks for sharing this.
Cheers,
Aziz

[Jerry]

Rather than keep on changing parts on the PCB, I am going to call this one complete for now and continue the experimentation with my modular TGSL which is about 75% built. As luck would have it, I am right at the sections for U-107 and U- 107.

Jerry

Should have proofread a little closer....... that is U-106 and U-107

[Jerry]

You seem to be happy with binary output

Anyway, I have a solution for you, and I know you'll like it: replace the diode D12 with 68k resistor and enjoy - it will give you that extra boost for low level signals, however, It will ruin their level response a bit, so it will become binary-like for small signals. If you find it too much of a boost, just replace R54 (1M) with larger value, say 2M2. Anti chatter functionality remains untouched.

Actually I prefer proportionate audio. The original 2.2K for R44 was a good choice but by installing the pot in it's place, I was able to find a sweet spot where the proportionate audio is much more pronounced. I did take it out and play around a bit this afternoon and found that full binary (zero ohms) gives short cracks of audio which are not pleasant to listen to. Right now I am thinking that a 25k linear pot with maybe 560 ohms between it and ground will be a good compromise. The way it is now, much of the rotation of the pot causes little change. I am going to take it out this coming weekend for an overnight metal detecting trip and I am sure that will give me a lot better idea on where to go with it.

As it is right now, I would rate this TGSL equal to my At Pro for sheer depth ability and somewhat better than the Ace 250, which is saying a lot because both Garrett's are good detectors. Of course depth isn't everything and the Garrett's have much to offer with their audio.

Rather than keep on changing parts on the PCB, I am going to call this one complete for now and continue the experimentation with my modular TGSL which is about 75% built. As luck would have it, I am right at the sections for U-107 and U- 107. Here are a few pictures of my approach. This one is intended for pure experimentation.

I like your ideas and it is interesting to see how they work out.

Jerry

[Davor]

You seem to be happy with binary output

Anyway, I have a solution for you, and I know you'll like it: replace the diode D12 with 68k resistor and enjoy - it will give you that extra boost for low level signals, however, It will ruin their level response a bit, so it will become binary-like for small signals. If you find it too much of a boost, just replace R54 (1M) with larger value, say 2M2. Anti chatter functionality remains untouched.

[Jerry]

Well I tried the first suggestion and changed R44 to a 100K pot and C26 to 22nf. This did make an improvement in that I can now get an audible beep with a coin waved about 32 cm above the coil. As suggested, turning the pot all the way to zero ohms makes the response very binary. For now I am going to leave it that way but it still isn't exactly what I was looking for and I probably did not explain it properly.

I ran a little test by unplugging U101 and connecting a 100K pot between +12 vdc and ground with the wiper connected to the junction of R45 and the base of TR2. I found that the actual threshold point to get diodes D3, 4, 5 to conduct is +.44 volts. Above that the audio get louder and below that it is completely silent. Much like the squelch control on a two way radio.

With U107 installed again, Pin 7 measures a -6.13 to -6.14 volts. Pin 7 must go from -6.14 to at least +.44 volts before any audio will be passed through the audio circuit. What I have been trying to accomplish with a threshold is to have a control where I can move the voltage on pin 7 closer to the .44 volt threshold. Exactly where it should be set depends on the amount of noise that is present at a given location.

I feel that not hearing a beep when pin 7 is not being driven above the threshold is leaving something on the table depth wise. Probably not a lot but that is what I am working towards. From what I have seen on my scope, getting an air test of about 40 cm on a 2 Euro coin should be possible.

I am thinking that the answer may be in the comparator circuit as porkluvr suggested.

Anyways, I am learning a lot in the process.

Jerry

[Jerry]

Thank you for the suggestions. I will give them a try tomorrow and see what happens.

This detector is working very well as it is right now but I can see where overcoming the threshold overhead, can make it a lot better. It would be nice to hear the signals that I can see on the scope.

I will see if I can download spice and play with it a bit. Tried it a couple of years ago but gave it up. Thanks for the files.

Jerry

[Davor]

R40 is there to prevent chatter.

I actually analysed this circuit and posted my results at the IGSL page. There are some misconceptions regarding the circuitry here, so here are my two cents.
Chatter-wise most of the job is done by R42 C26. C26 is by my opinion ~4 times too large, and if left as is it just produces odd delays in audio response at weak target responses.
R43 sets the output threshold level to ~1.4V to linearise audio (darlington) and it is best if left as is.
"Sens." potentiometer is actually setting a threshold at which comparators will trigger, and subjectively it does not do much, because the same targets will sound the same at various different settings of this pot. If you set it too high you'll miss lots of deep targets, but the shallow ones will sound the same.

What you can do is replacing R44 with a real potentiometer. For my IGSL I bought a real stereo pot (2 channels) because variable R44 provides a real audio level control. OK, my values will be somewhat different, and I did some other changes as well. Anyway, if you want to experiment, replace R44 with 100k lin pot, and replace C26 with 22nF. Leave everything else as is. Or you may use my other values and replace R44 with 10k lin pot, replace R42 with 10k, R43 with 47k and leave everything else as is (don't replace C26). In case R44 is turned down to 0ohm, LM358 turns into a comparator and produces pure binary response.

Here is a LTspice file of this circuit for your pleasure (400nF at the rightmost end is a piezo stand-in)

[Jerry]

Have you measured the DC voltage at the output of the opamp? Try adjusting the threshold pot over its range, and see how it affects the output voltage.

I did measure the output at pin 7 of U107. It remains at 6.14 VDC regardless of where the threshold pot is to Max or Min.

Pin 6 of U107 varies from +.036 with Threshold at Min to -.26 at Max.

Pin 5 of U107 averages right around -5.7 volts with some noise spikes pushing it in the positive direction. Whenever it becomes more positive than pin 6, there is an audio beep.

I tried changing R43 from 100K to 47k and also R44 from 2.2k to 4.7k. This give a swing on pin 6 U107 of +.72 at Min to -.99 at Max.

This made a little improvement but there is still a lot of space between pin 5 & 6. I can see the pulse of a 2 Euro coin on pin 5 when waved at about 40 cm over the coil. However I have to get down to about 28 to 30 cm before it is strong enough to break the audio threshold.

R40 is there to prevent chatter. If you get the value too low what you get is clamping from D12, and NO targets getting through.

Why don't you try increasing the value incrementally until you can get a threshold tone with your favorite sens setting?

Actually this detector is pretty quiet chatter wise. I am trying to push the limits a bit by adding the external sensitivity and threshold controls so I crank it up beyond the point where it is stable and then back it off until it is.

I tried taking a picture of the scope screen showing pins 5 and 6 with their relative voltages and pulses on pin 5 but the battery died before I could get a decent one worth showing. I will try that again tomorrow.

I just had another thought. I had always looked at R42 on U107 as being a feedback resistor to establish gain. I wonder if your comments regarding R40 would also apply to U107?

Thanks for your help

Jerry

[porkluvr]

R40 is there to prevent chatter. If you get the value too low what you get is clamping from D12, and NO targets getting through.

Why don't you try increasing the value incrementally until you can get a threshold tone with your favorite sens setting?