TGSL Tuning & troubleshooting

[Jerry]

Thanks Simon, that is the one I was thinking of. Will try and get a picture of U101b-1 and post it later today. Right now I have a huge lawn that needs cutting

Jerry

[simonbaker]

Simon,

I remember you posted a voltage divider mod for the signal from the TX to the Disc and GEB circuits a while ago. Other than cleaning up the Disc pulse when set at minimum, did it have any other benefits for discrimination? As long as I have the covers off, I just may do that and see.

Jerry

Hi Jerry:

I did not see any other benefits that I could measure. I actually expected to possibly get some depth improvement because it eliminates the "double pulse" (as seen in your top scope photo) that can happen, which diminishes the integration interval of the Synchronous Detector JFet. However, I believe that the ground-balance channel always has a weaker signal than the DISC channel, and so is the limiting factor, and as a result degradation of the DISC channel is not noticed.

In my circuit however there is a capacitor (I called it Cdiv) that compensates for a slight phase shift that the circuit introduces -- this gives you some control over the "phase bias" of the DISC control, if you want to play with values of the cap.

Another way to change the phase bias is with a small capacitor (like 10pf to 50pf) in parallel with the feedback resistor R17 of U101b (LF353). Someone else talks about it, I forget who. I don't know if such a cap would shift the DISC phase bias in the right direction for you, but it is worth a try, and easier to install than my mod.

I would like to see your scope picture of the output of U101b (pin 1) that drives the U102a LM393 DISC pulse shaper -- it would be good to see this output for both cases, where you adjust the oscillator for no bump, and with maximum bump. Or am I mixed up -- are you using pin 7 to drive the LM393???

I wonder what changed in your circuit? I agree with golfnut that the trigger point of the pulse generator (and thus phase bias) can be sensitive to the "oscillator bump", because that bump becomes a real blip at U101b pin1, and typically that blip is near the zero voltage trigger point. However, the puzzle is why when you restore the bump you don't go back to your previous operating point?

Here is a link to the "voltage divider" mod I did for reference: http://www.geotech1.com/forums/showp...9&postcount=73

-SB

[Jerry]

Hi Jerry, Adjusting the R value has biassed the device on less - so its more linear.
Spectral output is cleaner.

The amplitude may have decreased a little too.


If the amplitude is a little lower the trigger point at the detector is later in time (phase) - so you see the shift in desc setting.

Steve

Hi Steve,

I missed your post until I read my email. You are right, the amplitude increases until the oscillator is saturated and they starts clipping the top peaks. I backed it down just enough to clean it up.

For some reason, setting it back to where the bump at the top reappears does not fix the discrimination problem. I caused a change somewhere but have not found it yet.

Jerry

[Jerry]

I just noticed something about the series of pictures that I posted. By coincidence, the center vertical graticule is very near the dimple at the top of the square wave. It starts out where the rising edge of the right hand double bump is and the square wave seems to pass in front of this point and it finally falls off to the left or rising edge in the 5th picture.

It is amazing the details you can see once you put a picture on a big computer screen

Do not understand the significance of this but it is just an observation.

Jerry

[Jerry]

Simon,

I remember you posted a voltage divider mod for the signal from the TX to the Disc and GEB circuits a while ago. Other than cleaning up the Disc pulse when set at minimum, did it have any other benefits for discrimination? As long as I have the covers off, I just may do that and see.

Jerry

[Jerry]

I just spent a couple hours checking to see if a shift in TX frequency or voltage is causing the problem and it is not. I was surprised to see that the TX frequency was 14.21 Khz. Fortunately all of the transistors and IC's on the TGSL are socketed so it was a simple matter of removing the TX transistor and connecting a sine wave signal generator.

With the signal generator, I checked ground balance and discrimination settings while moving the frequency from 14.2 to 14.8 Khz and changing the TX voltage from 8 to 18 volts.

After each change in frequency or TX voltage, I did reset the ground balance. Sometimes by a lot. However, the point at which nickels would discriminate out changed very little and seemed to be in the 2 o'clock position of the Disc knob.

I am satisfied that frequency and/or voltage shifts are not the cause of the problem in this particular machine. I agree that they probably should so there is a problem somewhere on my board. I even put my old 8 inch coil with the heavy foil shielding and tried it. It discriminates the same way. At least things are consistent.

Once again, the machine is very usable the way it is and gets good depth on coins. My issue is with the ability to ID coins or trash with the Disc control. I should be able to discriminate out all targets up to and including zinc pennies.

I did take a series of pictures of the signals from U-101, pin 7 and the Gate of TR-4.

No. 1 shows the Disc at minimum. (counter clockwise)
No. 2 is with Disc at ten o'clock where the dip in the square wave just goes away. It seem to hit a very critical point where a very small movement of the disc control makes it blip out completely. It does however leave a small dimple in it's place at the top.
No. 3 is Disc at 12 o'clock, mid rotation.
No. 4 is at 1:30 where the trailing edge just stops moving to the right as the disc is advance clockwise. The rising edge does keep moving as the disc is turned further.
No. 5 is with the disc full clockwise. Width of the top of square wave is no as wide as it was and the bottom is wider.

Also of interest is the slight bump on the U101/7 signal that moves with the square wave and then coincides with the spike on the trailing edge at maximum rotation.

I checked the value of R18, the 100K Disc pot and it measured 74K from end to end so I replaced it with on that measures 97K (closest I had) and it did not change anything.
I am using MPF102 JFET's for TR4 and TR5. Don't know if that is an issue or not but looking at this part of various Tesoro circuits I don't see where it would be.

So that is my progress, (or lack of) so far.

Looking for ideas.

Jerry

[golfnut]

Hi Jerry, Adjusting the R value has biassed the device on less - so its more linear.
Spectral output is cleaner.

The amplitude may have decreased a little too.


If the amplitude is a little lower the trigger point at the detector is later in time (phase) - so you see the shift in desc setting.

Steve

[simonbaker]

Today I changed R3 in the oscillator circuit which is the 47k from base to ground with a 100K PCB pot. This is the change that was noted by Ivconic when he wrote up the IGSL thread about a year ago.

The purpose was to fine tune the oscillator for a nice sine wave and get rid of the little bump at the top of the TX waveform that we are all used to seeing in oscilloscope pictures.

This cleaned up the TX waveform really nice. However it also changed where things discriminate. For example, nickels used to null out at mid rotation of the Disc pot or about the 12 o'clock position. After the changes, it moved clockwise to between the two and three o'clock position. Pull tabs are silenced further clockwise but Zinc pennies are not. Note...... I could not silence zinc's prior to the tx change so that is nothing new.

It was raining today so I did not take it outside to the test garden but the air tests are as good if not better than before the change. It is supposed to rain again tomorrow so it looks like this will be the project for the day.

I am confused as to why a cleaner TX signal would change the discrimination. Anyone have any ideas?

Jerry

I think Qiaozhi's idea sounds very plausible -- let us know what you find.

One problem with getting rid of the bump is that the oscillator amplitude may be less "stable" (more affected by ground and other effects) - and you may lose a little depth. But it should help clean up some sync pulse generation messes if you have them.

-SB

[Qiaozhi]

I am confused as to why a cleaner TX signal would change the discrimination. Anyone have any ideas?

Jerry

Try measuring the TX frequency before and after the change. I suspect you will find a difference. If so, you can compensate for this by adjusting the value of the tuning capacitor.

[freedan3d]

Hi to all

I maked TGSL with this information in attachment
And its very cool and it discriminator is nice too !

But that sense is low in my mind
It detect a 1 penny in about 20 cm in air and a gold coin (2.03 g) in 15 cm .is this correct ?
tx=6 mH and rx=6.5 mH

thanks

freedan3d

[Jerry]

TGSL Oscillator/Discrimination question.

Today I changed R3 in the oscillator circuit which is the 47k from base to ground with a 100K PCB pot. This is the change that was noted by Ivconic when he wrote up the IGSL thread about a year ago.

The purpose was to fine tune the oscillator for a nice sine wave and get rid of the little bump at the top of the TX waveform that we are all used to seeing in oscilloscope pictures.

This cleaned up the TX waveform really nice. However it also changed where things discriminate. For example, nickels used to null out at mid rotation of the Disc pot or about the 12 o'clock position. After the changes, it moved clockwise to between the two and three o'clock position. Pull tabs are silenced further clockwise but Zinc pennies are not. Note...... I could not silence zinc's prior to the tx change so that is nothing new.

It was raining today so I did not take it outside to the test garden but the air tests are as good if not better than before the change. It is supposed to rain again tomorrow so it looks like this will be the project for the day.

I am confused as to why a cleaner TX signal would change the discrimination. Anyone have any ideas?

Jerry

[porkluvr]

I already had them writ, so may as well post. (Drawn with Design Science MathType.)

[simonbaker]

values coils tgsl

hello , values coil tgsl are 5,9 and 7 mH , What value should I put c6 for it accords with my Rx 7 mH?
__________________
hello , Nico

Nico:

Use formula for resonant freq and choose C6 so resonant freq of RX is 1.11 times the Tx oscillator frequency.

You will find formula somewhere in this thread or other TGSL threads. http://www.geotech1.com/forums/showp...&postcount=494

And please learn how to post to this forum. Maybe Qiaozhi can send you another email (I hope).

Regards,

-SB

[Qiaozhi]

Post any other questions here. Look at Qiaozhi's message above for advice on getting permission to post messages.

-SB

I think he's already figured it out, as he's now posting in the Silver Sabre / Bandido / TGSL thread.

Have fun with the questions.

[simonbaker]

tgsl frequency 16 khz for RX COIL

hello, I want to know how to change the frequency RX, because I have 14.5 and 14.7 kHz for TX to RX and not 16. So I want to know how to change the frequency of RX 16 khz to do it. THANK YOU
__________________
hello , Nico

Hi Nico:

TX frequency of 14.5 or 14.7 kHz is good!

RX resonant frequency is determined by the RX coil inductance and the parallel capacitor (typically called C6, .015 uF), and is around 16 kHz. You can use a formula to adjust to a different value of C6 if your RX coil inductance is much different.

Look at this link for more information: http://www.geotech1.com/forums/showp...&postcount=494

Post any other questions here. Look at Qiaozhi's message above for advice on getting permission to post messages.

-SB