IGSL-NG

[Davor]

Time to move on. But first a final touch.
I decided to give one more go to the classic IB design before trying something new. IGSL-NG should be an IGSL, in a way that someone observing the schematic would say "it looks like an IGSL," but with refinements. I also intend to make it a collage of circuitry, so that newcomers may learn a thin or two. A sort of IT(IGSL)MD. It will also give me opportunity to finish it in a few months, as I'm a lot busier now than at times I joined this forum.
About IGSL...
It is an ingenious idea by Ivconic to marry two TGSL-s into one, to obtain both non-ferrous, and ferrous responses, by 2 different tones. Because these responses may also overlap, there is a third tone. I use it for foil. TGSL? Tesoro Golden Sabre Light, a classic Tesoro detector stripped down to essentials.

I'll start a series with an improved Tx. Feel free to suggest functions you miss/wish in a new design.

[dbanner]

First, from the brilliant mind of dbanner, I am fascinated by the term " remanence"
Second, by addition of channels, one can create two tone notch.
Many improvisations, little innovation.

[dbanner]

In think that it is more a matter of construction than any fundamental breakthrough. Physics don't change, except for string theory.

[Gwil]

The same physics operate a steam engine and a jet engine....

I would very much like to see any ideas Davor comes up with!

[dbanner]

I too would like to see any improvements, but I suspect that such improvements would require radical departure from original circuit, such that it would be like complete redesign.
I would also like to see some more work on PLL based designs. There are some circuits out there, but they don't seem to take the fancy of too many builders.

[rickb]

Nothing like another ill considered rant from dbanner to quell Davor's interest in returning to the forum.

dbanner: You are free to shut up and design your own state of the art detector (or not).

Like Gwil, I welcome his contributions and look forward to his improvements, state of the art or not.

[Davor]

No radical departures in this thread, only radical refinements. I said this will be a tribute to a classic IB analogue design, and that's what you'll find here. There are reports of IGSL, as is, making little babies to rigs like a CTX, so please pay some respect where respect is due.
As announced, here is a radically refined single ended Tx. As I intend to power it directly from batteries, it is necessary for such Tx to operate consistently in a wide range of supply voltages. Well, this one does that. It will give envelope equal to supply voltage from just above 4V to over 18V, with all harmonics at about -40dBc in a whole span. It is self-regulated to a scary level.

Important notice: it will use a F75 kind of coil. There are many reasons for that, but let it just sit for now.

There are many goodies in this circuit, some indicated in a schematic, but as I'm about to hit the hay, pictures and explanations will come soon. For now just the basics, a basic picture and LTspice files.

[ivconic]

The same physics operate a steam engine and a jet engine....
I would very much like to see any ideas Davor comes up with!

I like that! Nicely put!


Reason i was silent on IGSL subject for quite a while is because of the objectivity and fairness.
I stopped with any further works on it due some subjective and objective reasons forced by life occurrences at the time.
Lot of water under the bridge since than.
In meantime Davor proposed some significant changes and improvements which simply wins over most of my own ideas i had at the time.
Since than i am only a follower with great interest and attention.
Indeed i am looking forward to see this thing evolving in much better design. Even if it ends up totally redesigned and "departed" from the initial idea.
So this is having my full support and thumbs up, of course.
Welcome! Looking forward!

[ivconic]

... Important notice: it will use a F75 kind of coil. There are many reasons for that, but let it just sit for now...

That's the only part that worries me; design to be strictly tied to one specific type of coil.
Although F75 coil is indeed great coil, had chance to work with it and F70 for several months.
But from the diyer perspective i would be much happier if i could pick between wider range of coils.

[Davor]

You may. T2 is another such coil. When I said F75, the idea was a working frequency around 13kHz, and a Tx coil with inductance below 1mH. Tesoro Tiger Shark at 12.5kHz is OK. Tesoro Compadre at 12kHz as well. Whites MX5 and Matrix at 14kHz also qualify. Makro Racer at 14kHz as well. But I wouldn't go beyond 15kHz or below 11. Tesoro Outlaw at 10.6kHz is marginally there.
So, there are many options. When browsing ebay, F70/F75 coils make a significant presence, so It was a simple call: F75.
(or similar)

[Davor]

About the oscillator circuitry...

Vref is straightforward. Two diodes modestly biassed are a cheap and quiet solution. Function of Vref is assuring there is AB class bias at Q2 to start oscillation at any voltage. This bias is formed via D2/D3, R3, and R2. R2 is slightly larger than R3 for AB bias of Q2. Vref also promotes class C function once oscillation is established. I tried some better Vref sources, but there was no improvement, so two diodes are here to stay.

The AB->C circuit is now intrinsic to the oscillator. It is formed by C4, R4, and Q1 BE diode as a charge pump. When there is sufficient oscillation, C4 fills above the Vref-Vbe, and in turn Q1 goes from AB into class C. A measure of class C depth is a conduction angle. Conduction angle? from 360? conduction for class A (current flows for the whole period of a sinus), down to ~190? for class AB, down to 180? for class B, and everything below that is a class C. In this particular case, conduction angle is about 120?. Class C is good for frequency stability, and more importantly amplitude stability, and it also somewhat reduces harmonics, as lower conduction angle improves effective tank Q factor. It also reduces consumption, but that is of lesser importance.
Cut off frequency of R4C4 is ~480Hz, and because it also stabilises amplitude, it must be faster than useful spectrum (~1Hz to 20 Hz), but slower than the oscillation itself (13kHz), so there you have it, a geometric middle.

Miller killer is formed by C2 and C3. Its role is ~ 1/5 tank voltage divider, but because it is a capacitive divider, it overwhelms the Miller capacitance of a transistor. It therefore kills the Miller ill effects. Because of Miller capacitance, most of the classic oscillators have a small lag that effectively turns all classic oscillators into off resonance devices. It is bad for both frequency and amplitude stability. I found that C3 value should be about 2% of a tank capacitor. Miller killer results in a scarily symmetrical tank driving signal.

More about the rest of Tx circuitry later ...

[ivconic]

...

[Nandor]

Thank you guys for the good work, really interesting topic!I feel so lucky that I can witness the birth of this project.

[Davor]

Moving on...

A role of D1 and R10 may not be obvious. I tried a Schottky diode here, and while it prevented Q2 to saturate, while providing excess tank amplitude a higher impedance path in saturation, it also reduced amplitude for about half a volt. Silicon diode is a compromise that gives me half a volt back, and R10 makes sure the vast majority of saturation current still goes through D1. R10 may as well be omitted, but I left it there anyway. If there is no diode like D1, the driving transistor takes more time to switch back from saturation, and it is yet another source of lag in classic oscillators.

About the shaper...

I'm proud of this one. Whoever designed oscillators before might have noticed that when oscillation goes near the rail, it causes the driving transistor to saturate. During saturation the tank is exposed to low impedance which in one hand stabilises the amplitude (good), but on the other it spoils the tank Q factor (bad) and sinus purity (very bad). When driven heavily, the 2nd harmonic becomes very strong, and Tx signal loses symmetry.
It first came to my mind that it may be possible to extend the time the driving transistor acts as a current source, in order to reduce tank loading, but not too much to retain the amplitude limiting ability. So I've put a voltage limiter to a Q1 base. And it kind of worked, but not for a wide voltage span. Then I realised it must be possible to do it as a soft limiter, and also to make it adapt to supply voltage, and boom! it makes the working conditions consistent for a whole voltage span. I'll put a few pictures later on that indicate what does what.

In conclusion, this oscillator follows supply voltage verbatim. This means I can hook it to a battery, and no additional voltage regulation, and know that its amplitude noise skirt will resemble the voltage source noise. In case of batteries it is good, because those are the best voltage sources regarding the noise below 10Hz. Which means less chatters and phantom targets.

[Davor]

Attached are the FFT responses, first one at 15V straight, taken from 70ms to 160ms, and another one from the sloping amplitude, 15ms to 60ms. Windowing function is Flat top, as it is perhaps the best one to capture exact amplitude. If there is anything going on at the skirt, this windowing function would capture it. Anyway, it is clean as a whistle.