TX Experiments

[Silver Dollar]

I made this PID transmitter just to see what it would do.


Using it on my Minelab TX coil at 1.18 uh (2.5 ohms) I got the following results;

Using 9V ps

1 uf cap
4.6 khz
22.2v P/P
19 ma!

0.47 uf
6.7 khz
25v P/P

0.22 uf
10.1 khz
26.4v P/P

0.1 uf
14.5 khz
28.9v P/P
11.3 ma

I'm not sure it was working correct (I just made it!) as the Amplitude adjustment
did not do anything. I could adjust the duty cycle and get crazy voltages (46v!)
but the coil started buzzing and current shot up higher than my meter could measure
(24 ma).

It might be worth a try on TGSL to see if any depth is gained but it does use a lot
of current. The H Bridge or AB/C Minelab oscillator might be a better choice...

I might try this one next. Aziz designed it. It's simple and supposedly puts a lot of
current into the coil but doesn't use much juice. You do need a center tapped TX
coil though so it's not for everyone...

[Silver Dollar]

So I wired up the AzizTX oscillator using 7.5v zeners and +-9V supplies and
with the Deep Search Tesoro coil (6.1mh CT) I get;
14.2 Khz
27V P/P
+- 4.8ma

That's close to 10 ma! I thought he said 2ma. Oh well the search goes on.

I'm going to try his magic inductor oscillator. I found a 2.6mh inductor today
to make the magic. Here's the schematic;




Here's the thread where I found it;


http://www.geotech1.com/forums/showt...ht=balanced+tx

I couldn't find 51v zeners to protect the transistors but found 33v ones and some high voltage
transistors, I'm not sure if they are high beta though. They are FMMT597 and FMMT497 (hfe 100 to 300)

[Carl-NC]

Efficiency can depend on coil resistance and cap ESR.

[Silver Dollar]

So I made this version of the magic oscillator;


Some preliminary results using the centertapped RX of my Deep Search coil
(6.1mh) with R1 + R2 = 22k

and C3 = 47n

10 khz
33v P/P
9.8 ma input current
Not sure how much current is in the coil

With C3 = 22n

14.2 khz
33v P/P
6.6 ma input current

These were all using 9v input power

I'd like to get current draw down to 4 ma
maybe try 47k for R1 + R2?

And I want to try it on my Fisher 440 coil
as that's the one I will need a CT oscillator for.
The sims show 14.5khz and 4 - 5 ma in
for +_27v P/P and 50 ma in the coil.

[Davor]

I was not able to find a way to stabilise this oscillator and I can't recommend it for any serious development. In most "normal" oscillators a cyclostationary operation is achieved when oscillation hits the transistors' saturation regime. Such oscillators' amplitudes are locked to the Vcc and thus very stable. Noise performance of such oscillators is ~ equal to the power supply noise, and that means you should keep Tx supply rails as quiet as possible.
There are also oscillators with active amplitude control with various PID regulators. I'm not fond of such solutions as they are much more complicated, and if any kind of integration is introduced in a PID regulator, your envelope loses stability. Problem is with a tank Q factor that introduces group delay in the envelope response (integrator) and two integrators in a same loop = oscillator. If you lose integrator in a regulation loop, what you get is equivalent of a rail limited oscillator, only running at some peak voltage that is lower than the rail. What goes around ...
Amplitude stability is of great importance with VLF detectors as the air signal creeps into the Rx, and its amplitude modulation component translates to chatters. For that reason it is also good to limit 1/f noise of the very Tx transistors, and the only way to do it properly is to introduce some small emitter degeneration resistor. 10ohms is fine.

IMHO rising amplitude by "magic choke" is not a very good idea.

A solution with Colpitts configuration is a bit more sensible, as long as the Q factor is high and the oscillation tethers to Vcc.

The approach that makes sense to me would be going way below 1mH with total Tx inductance while keeping a tank Q high, and some AB->C class mechanism to further reduce power consumption and harmonics. A proper way to do it would be using a stranded or Litz wire to reduce eddy currents in a cross section of a Tx coil, as it becomes quite thick. You can achieve some serious ampereturns this way, while supply current remains low. The remaining problem could be current that flows through a connecting cable, so a meaningful coil embodiment would include a tank capacitor as well.

[Silver Dollar]

Hi Davor, Thanks for the input. I'm looking for an oscillator to drive my new Fisher 440
coil. It uses a 12.6 mh center tapped TX. Is there a way to use this oscillator with a
centertapped coil (maybe by grounding the center?)?


I also found the Viking Oscillator that works with center tap;



Your AB/C mod to the Minelab Oscillator works great but not with higher restance coils.
Any idea on how to mod it to work with 3.3mh 15 ohm CT TX coil?

[Davor]

Any asymmetric oscillator can drive a centre tapped coil. The only difference is that you connect the tank capacitor on the coil ends and drive only one end with the oscillator and the tap is connected to ground (provided the coil is driven by a PNP). That way you get double the voltage. Symmetry is not perfect, but you benefit from doubling the voltage.
You can also drive such coils with half bridge with better symmetry and thus better harmonic suppression.
Full bridge oscillators are a solution for balanced drive of coils without centre tap. You can use such oscillator with a centre tapped coil, but in that case you'd connect coil ends to the oscillator, and centre tap remains unconnected. You also lose voltage doubling.

Free running oscillators depend upon a coil Q factor, which is XL/R for a given frequency. Coils driven by a separate oscillator do not benefit much from high Q, and thus often have poor Q. Such coils take less copper, and hence may be lighter than high Q coils.
As the tank efficiency is proportional to Q, ultimately the power consumption will be lower with higher Q for a free running oscillator. Some oscillators require a bit tweaking when operating with lower Q tanks, as the loop requires some additional gain. AB->C mod reduces gain by turning the oscillator into class C, so operating a poor Q tank would require increasing loop gain a bit. However, poor Q tanks have poor power efficiency on their own, so using such coils with AB->C kind of defeats the purpose of the exercise.

[Silver Dollar]

The sims were not that good for that Viking Oscillator. It was from a Pinpointer anyway I think.
So I tried the oscillator from the Fisher 440 as the coil was meant for that MD.

Here's the Schematic;


The sim said 14 khz with 4ma current draw. I read;
10 khz
4.5 ma
17.8v P/P

Using 9v for power.

I read this from one side to ground (-7v in schematic)
The problem with this oscillator is it runs from the minus
supply to get the center tap to ground. I'm not sure my minus
supply can give 4.5 ma to run it...

[Davor]

You can reverse it by simply replacing all NPN transistors with PNP and vice versa, and simply turn the supply opposite. Easy peasy.

[Silver Dollar]

Well the 440 Oscillator tuned to 14.9khz via a 6.8n cap gives
14.9khz
33v P/P
Input 3.8 ma at 9V
24ma per side in the coil (50ma total)

I went back and tried the Magic oscillator
with 9V in and 18n
15khz
32v per side (64V P/P!)
6.5 ma in
98 ma in coil (1 side!)

I adjusted the input to get around 4 ma and got

5.5V in
4.1 ma in
61 ma per side in coil
20v per side
15khz

Picture of current config;


So this one looks very efficient. Not sure how much
amplitude jitter there is. The waveform in sim looks poor
but on the scope it looks fine. I have some MAX667's so
I will use one in variable mode for this TX so I can set the
amount of power for future experiments.

This is destine for an experimental rig so no biggy if it doesn't perform
well I'll just rip it out and try something else...

[Silver Dollar]

Well the magic oscillator puts out a lot of power but it has a lot of second harmonic.
I think I need a more regulated solution so today I flipped the sex of all the transistors
in the 440 oscillator so I can run it off positive voltage. Here's the schematic;



For some reason I just swapped the transistors (though from BC337's to 2N4403's)
and the frequency changed from 14.9khz to 21 khz. Here's the data;

6.8n
21khz
34v P/P
4.3 ma in

8.2n
20.6 khz
34v P/P
4.3 ma in

15n
15.4 khz
33V P/P
4.8 ma in

18n
14.3 khz
33V P/P
5ma in

22n
13.6 khz
32V P/P
5.1 ma in

I kept the 15n and tried varying the power supply aiming to get 4ma draw.

with 7.48V in
15.5 khz
26V P/P
3.9 ma in

With 7V in
15.4 khz
24.8V P/P
3.7 ma
74 ma in the coil! (37ma per side)

I'm thinking of having 2 settings for power say 5v and 7v?
At 5V I estimate around 2.5 ma input current and 40 ma in the coil
so you'd have high power of 74 ma and low power of 40...

[Davor]

There are some odd connections in a swapped transistors version.

[Silver Dollar]

? Should something change when used in this configuration?

Here's the schematic for the power selectable version. I'm not sure if the Max667
is readily available, I found some languishing in an old parts bin...

[Davor]

I think I found what I was missing. Silly me. Do you have this circuit in spice?

[Silver Dollar]

Here you go;


And a picture;