TESORO GOLDEN SABRE

[ivconic]

Huh! Again error. Now i am more sure it has bugs....Look;
I took previous oscillator schematic with 8v supply and othe zener...
Just changed 8v to 9v and look what i got!?
Now, on this, when i repeat it changes scope amplitude to 16v...???
OK no big deal. I'll remember this thing and each time i need accurate scope i'll repeat those steps twice to be sure...until find out what is the problem?

...

[ivconic]

1N750 is actually 4V7 so zener is not problem here...

[Max]

1N750 is actually 4V7 so zener is not problem here...

Hi,
yes it's ok. I've repeated your test using +8 and then changed to +9... and it goes well.

You have to close the diagram window and make again Simulation->Run so it will recalculate the points of transient a. again with right params.

Otherwise it will be not so happy of changes.

Another thing to keep in mind is that labels mean nothing here and could be confusing also.

Best regards,
Max

[Max]

1N750 is actually 4V7 so zener is not problem here...

Hi,
yes is the feedback circuit of TGS osc that makes output be almost not dependant from supply variations up the a certain level (depending on zener voltage and diode forward drop).

That's cool cause you can program e.g. double voltage accross coil just changing the zener and the regulator!

I think there's enough to test with some spare zener and look for results.

Kind regards,
Max

[Qiaozhi]

One important thing you have missed out in your schematic - the DC resistance of the coil. At the moment you are using an ideal inductor. You need to include a resistor in series with the inductor, as it will make a difference to the results.

[ivconic]

Good point....i'll try later...
Now i am in the middle of something else....be back..

[porkluvr]

Don't forget R parallel, too!

Qiaozhi mentioned the coil's series resistance. I have also started adding parallel resistance to help simulate the unknown quantity of cable loss. Adding parallel resistance smooths out the collector voltage hump and also helps tame the FET gate waveform. The AVS (Automatic Voltage Stabilization?) circuit, or whatever it's called, will not work smoothly when the circuit Q is too high - or when too much power is being produced, or even if the frequency fo is raised significantly. I can't fully explain this last statement and don't want to try and elaborate, but you will see what I mean whan you start playing with the supply voltage. (Or maybe it's just ME whose screwed up!) Look for jagged wave on the FET's gate.
When I first started working with the TGS I changed the FET gate's RC components, or else the transmit signal would not stabilize. Now, I leave the original RC circuit values, but keep lowering and lowering the value of parallel coil resistance (simulating circuit parasitic LOSS) until the amplitude stabilizes. I usually must have a value of around 8kΩ or sometimes even less, with fo=14kHz.Trial and error. Hey, I don't know what else to do.... . I never suggested LTspice is perfect, but it's all I got because I do not have a lab!

Max, Ivconic: I see you have discovered that the simulations show a voltage "hump" on the collector. My simulations also show that the collector waveform does not have 50-50 symmetry. You have scope, I believe so, please tell me: how bad is it in real life? It may be an anomally of the software and models used.

Adding parallel resistance on the coil will smooth the hump, but there is still an apparent failure to attain a symmetrical waveform. How bad it is in real life, is what I would like to know. Please measure and tell me.
??
Max I still owe you some reference material. It's coming. Gotta go.

[Max]

One important thing you have missed out in your schematic - the DC resistance of the coil. At the moment you are using an ideal inductor. You need to include a resistor in series with the inductor, as it will make a difference to the results.

Hi,
actually not. I've used the porkluvr's model that is:

5.8mH inductance
22 ohm series resistance (actually we have a bit more)
25Kohm parallel
950pF capacitance

So we are about ok using it.

Kind regards,
Max

[Max]

Qiaozhi mentioned the coil's series resistance. I have also started adding parallel resistance to help simulate the unknown quantity of cable loss. Adding parallel resistance smooths out the collector voltage hump and also helps tame the FET gate waveform. The AVS (Automatic Voltage Stabilization?) circuit, or whatever it's called, will not work smoothly when the circuit Q is too high - or when too much power is being produced, or even if the frequency fo is raised significantly. I can't fully explain this last statement and don't want to try and elaborate, but you will see what I mean whan you start playing with the supply voltage. (Or maybe it's just ME whose screwed up!) Look for jagged wave on the FET's gate.
When I first started working with the TGS I changed the FET gate's RC components, or else the transmit signal would not stabilize. Now, I leave the original RC circuit values, but keep lowering and lowering the value of parallel coil resistance (simulating circuit parasitic LOSS) until the amplitude stabilizes. I usually must have a value of around 8kΩ or sometimes even less, with fo=14kHz.Trial and error. Hey, I don't know what else to do.... . I never suggested LTspice is perfect, but it's all I got because I do not have a lab!

Max, Ivconic: I see you have discovered that the simulations show a voltage "hump" on the collector. My simulations also show that the collector waveform does not have 50-50 symmetry. You have scope, I believe so, please tell me: how bad is it in real life? It may be an anomally of the software and models used.

Adding parallel resistance on the coil will smooth the hump, but there is still an apparent failure to attain a symmetrical waveform. How bad it is in real life, is what I would like to know. Please measure and tell me.
??
Max I still owe you some reference material. It's coming. Gotta go.

Hi prokluvr,
and really thanks for all the stuff you provided and references to LTspice too.
Me personally found the software really good for our (limited) purposes. I think that it's made good enough to give some answers we await from little mods on original schematic without directly putting on hands on pcb everytime.

I've tested voltage at collector a number of times during tuning of TGS and from what I remember the software give a good approximation of the effects there. And yes there could be instability at osc too depending on many factors. Most important is the feedback circuit, what you call AVS.

The very hi-gain it has could introduce oscillations at low frequency cause of overdamping effect that occours sometimes... you can see it on the diagrams of LTspice too... are like output bursts followed by zero voltage signal... till, after a while, output stabilize. That's cause of the gain... but I guess that also temperature on the real world could give similar effects: fets are well know for changing dinamic params really easy with temp. variations due to the structure they have internally...respect e.g. BJT configuration.

Is it cause of bleeping we are looking for ?

Maybe ! Or maybe, more realistic, is just one cause of erratic bleeping that sometimes occours.

Anyway, thanks again for your help and interest in TGS circuit.

Kind regards,
Max

[Max]

Hi,
actually not. I've used the porkluvr's model that is:

5.8mH inductance
22 ohm series resistance (actually we have a bit more)
25Kohm parallel
950pF capacitance

So we are about ok using it.

Kind regards,
Max

Ops... parallel is 12Kohm not 25....

[Max]

Hi,
now I've made a little test to simulate the "moisture" effects on PCB... I mean soldering flux not removed... look

Suppose that instead of 100K at R1 the circuit see 50K... the pulses suffer of an heavy intermodulation and you can see that peaks are not all at same level anymore -> as expected this would damage SNR

Then I've tested with a 470K at R1 and you can notice a very good sinus...

Other interesting thing is the capacitor that filter signal on the gate... if you reduce too much (e.g. 22nF) you will see the signal fluctuate in a funny way (underdamping) ...but if you increase to 1uF signal will be more stable.

The alteration of gate components change the feedback behaviour like when in a proportional control you add a delay block... if delay is too short you could experience underdamping...

Conclusion is that feeback system is a good thing... but you have to keep clean circuit and use right values... otherwise... it will give you just problems.



Kind regards,
Max

[Leto]

feedback testing..

Hi...
Max.
If you have the time; could you please check alternate values for C3.
(100nF and 470nF)
Would this change the output amplitude?

[Max]

Hi...
Max.
If you have the time; could you please check alternate values for C3.
(100nF and 470nF)
Would this change the output amplitude?

Hi,
the amplitude is stabilized by the feedback effect...so you have always that limit imposed by zener + fet etc.

Then is the damping effect that could change... if you use 100nF you have underdamping than with 220nF; 470nF is much better cause you critically damp or overdamp oscillations (so have less fluctuations from one sinus to another)...that's what we need first: perfect sinus.

Kind regards,
Max

[porkluvr]

4.7 zener?

Max, Ivconic, one of the things you might be overlooking in trying to get more power from the oscillator:

Zener diodes with voltages below 4.7V have a less abrupt breakdown than 4.7V zeners.

I have a feeling that the 1N749 was chosen as much for this soft-knee characteristic as it was for the breakdown voltage.

Going to a different FET (with a more negative turn-off) may be a more desirable alternative? Just keep that in mind.

I not sure about the exact curve that this knee effect vs zener voltage effect follows.... who's got the exact data, anybody?

[ivconic]

One thing is still buging me;
again i went outdoor with TGSL. It is still producing false signals from time to time on the ground? I checked everything. Cant find what is the problem? So i have made TGSL on 2 different pcb's. One (bigger) from first page of this thread and second, smaller.....I removed all components included in notch section from larger pcb. Comparing those i saw that device with larger pcb is acting much better than device with smaller? First one is not producing to much false signals at all. Smaller one is producing to much false signals.
What is interesting; i can not provoke false signals to be produced by bending cable,shaking coil,punching it....remains silent. Not sensitive on any kind of "torture"...But when small item moved in coil vicinity, just about to provoke weak detection....device detect it weak and when item removed, device continue to "beep,beep,beep...."-ing all the time. Outdoor, when coil is in the air, i get simillar behavior, but when lower to the ground it stops beeping. Moving forward with coil 3-5cm of the ground, touching grass it is producing false signals oftenly?
Device with larger pcb is acting simillar but noticeable less false signals....very rare...on some fields almost none?
I collected those 254 coins in not time, with larger pcb.
I tried everything possible to find source of problem....without success.
I guess my huge problem is cose i dont have good osci. to accuratelly measure all suspectable "spots" here.
I am posting smaller pcb with component layouts, maybe some of you will spot some mistake on it. I got very tired of looking in it, and havent find anything bad yet....