Super TGSL With H Bridge TX Diff RX Quad Sampling

[Silver Dollar]

Maybe no one else is still working on this one but I just got interested and bought a board
from Silverdog. Thanks to Davor and others I came up with some improvements. I added the
H bridge TX and a differential RX with discrete preamp (note for use with the Minelab Coils).

With a TX of 8.7 Khz the negative power supply only gets 4 khz so I changed the 4024 to
an inverter (CD40106 should be used ((not in library))). The audio will come from a simple
inverter osc. This might create less noise or maybe not (as is closer to filter bandwidth).

I just got my board running in stock form so these will be incorporated soon. I coludn't
get the Minelab osc to work. I built two different ones too. But have to wait for my BC337's
tp come on the slow boat!

This is my first post with an attachment I hope it shows up!

[joop]

What is your goal? Better discrimination or more depth? Or just improving the design? Thanks.

[Silver Dollar]

I just want to make the best detector I can. The H bridge has less second harmonics. The preamp is quiet.
Adding the two phases reduces offsets. I'm still not sure on the negative supply. I'd have preferred a virtual
ground but I'm working with a already laid out PCB.

There were a couple of mistakes in the original schematic so here's version 2.0;

[ashefr10]

Hello, your work is very interesting, regarding the improvement that you think you can improve with these changes? better discrimination, best sensitivity?
personally mine lacks depth and is unstable in the wet grass.
Detection limit in the detector emits an unpleasant sound swing.
So if your version of the scheme solves some problems that I encounter that interested me greatly.

[ashefr10]

I do not see the floor for TX in your schematic I see two coils RX1 and RX2 but not TX. why do you use two chokes RX?

[joop]

I just want to make the best detector I can.

Will follow this thread with great interest.... So far I am satisfied with my TGSL, will be interesting to see what the outcome is...

[Silver Dollar]

The TX is not grounded on one end as it is balanced using an H - Bridge oscillator. The RX is center tapped.

You may get away without shielding using a center tapped RX. These mods should end up with better sensitivity...

[Davor]

Yes, they should. I like the way you combined the balanced oscillator with the rest of the rig. It has a potential for a perfect low budget device, but without performance defects.
It does, however, require a coil that is very simple to make, but non-existent on the market. You won't have to make a shield for it, but you surely won't be able to buy a ready made coil for this device.
Are you going to make this device? I'd like to see it in action and I have big expectations on it's performance.

[Silver Dollar]

I have it working as per the standard TGSL and have the parts to upgrade it on order.
Over the next several weeks I will be modifying it to the new schematic.

Yes unfortunately you would have to make your own coil for it but I hope the results
will be worth it!

I found a possible audio mod over the weekend. It places a Fet between the last GEB
gainblock and the 358 for audio.

Checkout version 3.0;

[Silver Dollar]

Oops there was a mistake on that last schematic with biasing on the Fet.
Here is a corect one. Also note the caps on the H bridge were wrong as I
had just copied some and forgot to change the values....

[Davor]

I like it. FET may prove to be even faster than my resistor mod in IGSL. Nice catch.

There may be a better solution to the phase shifters. I'll try to come up with a solution tomorrow (unlikely), or some time soon (more likely). It is quite possible to obtain quadrature switching signal from the oscillator. With some clever phase play, from there you can obtain rock-solid phase shifts that would enhance discrimination.

[Silver Dollar]

Sounds good. I'm looking forward to see what you come up with!

[Gwil]

???

Should there be a 47k resistor from U101a pin 5 to ground (like R15)? Also, doesn't C23 short the signal to 0V (via the 8V rail)?

Gwil

[Davor]

Yes, there should be a 47k resistor to signal ground. It should work without it, but CMMR and PSRR are reduced. C23 is supposed to be there, as it improves high frequency behaviour and linearity.

[Gwil]

OK, but if 4u7 means four point seven microfarads that's pretty close to a short circuit at the operating frequency. Maybe I'm missing something.

G