The Mighty "Fisher CZ5 Quicksilver" and it's secrets

[Altra]

Neat work and nice drawn schematic, easy to follow.
Two questions:
1) Why U7 at TX? I understand the role and what it is doing, yet i don't understand why is not used some more simple inverter solution? Cmos.. or two small mosfets... or similar?
2) U12? PIC is supplied with 15v than? Why? Since TX is supplied with 10v.

Thanks, U7 is used as a level shifter. The pic +Vdd is connected to +10v and pic -vss is connected to the output 79l05, which is 5v below +10. It's a little confusing, but allows using the cd4053's without level shifting. If you think in terms of a +/-5 volt supply, the micro and logic are 0 to 5 volts

[ivconic]

Thanks, U7 is used as a level shifter. The pic +Vdd is connected to +10v and pic -vss is connected to the output 79l05, which is 5v below +10. It's a little confusing, but allows using the cd4053's without level shifting. If you think in terms of a +/-5 volt supply, the micro and logic are 0 to 5 volts

Thanks for reply.
Yes that's confusing to me.
Usually at Atmegas(Arduinos) i get pwm signal within voltage range supplied to chip. So if chip is supplied with 5v, pwm is in range 0-5v.
In this case i would expect signal coming from PIC to swing between -5v and +10v... that's confusing?!
If so; than 4053 is wired correctly, "accepts" full swing of 15v and provides swing of 10v to match TX supplied also with 10v.
Only part that confuses me is why 15v at PIC? Why not 10v, and tie Vee,Inh and GND all to B-?
If PIC (supplied with 10v provides signal swing -B to +10v; than just tie Vee, Inh and GND all to -B, and all sits to the proper places then.
Then no need for U12 and bit more complicated setup.
Point is to extract U12 from the schematic and make things simpler.

[Altra]

There is no 15V, the input to U12 is 10 volts, it drops 10v to 5v which is the minus supply for the micro. You could delete U12, but I like to decouple the digital supply from the analog. I posted this as an example not a "build it". There are many ways to skin a cat, so I hope people experiment and create their own circuits.

B+ = input LM7810
10V = +5v analog and +5V micro/digital
5v = NE5532 Analog Ground 0v and output LM79l05 0V micro/digital
0V = B- = analog -5v

[ivconic]

There is no 15V, the input to U12 is 10 volts, it drops 10v to 5v which is the minus supply for the micro. You could delete U12, but I like to decouple the digital supply from the analog. I posted this as an example not a "build it". There are many ways to skin a cat, so I hope people experiment and create their own circuits.

B+ = input LM7810
10V = +5v analog and +5V micro/digital
5v = NE5532 Analog Ground 0v and output LM79l05 0V micro/digital
0V = B- = analog -5v

You are absolutely right!
I completely overlooked some things, sorry!
Yes, +5v then is ground, correct!
Cheers!
...
I am interested in this, do you have any video of how it is working?

[Riss]

I can't see the notch filter that removes 15kHz signal in channel 5kHz ?

[Riss]

This was my attempt at multi-frequency. I lost interest when the Equinox was introduced and later the Vanquish. But now may revisit the project after a long rest.

It worked pretty good, but was weak on the low conductors. The gain in the harmonic channel needs to be increased to equalize the gain between high and low conductors.

I used the CZ timing as a model, 5khz & 15khz generated by a cheap pic18f1220. It uses a full bridge tx driver to gain the extra coil current. The tx driver is straight from the ML Sovereign. The the rest of the circuit is standard analog. It ignores salt water and has manual ground balance.

I can't see the notch filter which removes 15kHz signal in channel 5kHz ?

[Riss]

Based on the reverse-engineering of the 4.7 kHz / 14.2 kHz mode used on the Deus2 TX signal, I see that this waveform could be applied to the CZ, to create a viable hybrid.
The problems of generating the waveform for the Transmit, and the creation of I and Q demod triggers at both 4.7k and 14.2k ( or whatever) can all be solved neatly with more modern digital techniques.
Either dedicate a microcontroller to doing the signal generation ( lookup table, possibly 120 bytes, using 6 bits of data per byte ?)

Or use a PLD ( programmable logic device ), programmed to behave as a 120-state sequential counter, with 6 outputs used to drive TX and demods, and one master clock input. The clock is likely to be in the 500 kHz - 1.5 MHz range, so a simple common quartz crystal oscillator with 2, 3, 4 stages of binary division beforehand is all that is needed to make all the timing functions. 120 states means 7 bits, so I expect that the timing functions should fit inside a common ATF22V10 ( or equivalent ), or failing that the 'super-22V10' , the ATF750. ( this was a great device when Atmel created it as the ATV750, but it got obsoleted by massive CPLD's. When Microchip took over Atmel, they did a smart thing, and reintroduced it, in modern F Flash format ). these PLD's are 2 or 3 US dollars (unprogrammed ).

If a 5k / 15k TX is used, the Fisher CZ coils should be capable, they may work a bit higher. If you want a higher-freq operation ( 13k / 39 k for example, a good choice ) ,the Equinox coil is an obvious solution. And having the pre-amp internally means the first stage amp inside the 'control box' can be a modest low-noise differential amp, no need for expensive ultra-low noise parts, or discrete/opamp hybrids like in the CZ.

I would use the synchronous counter CD4520 and diode decoder with diodes 1N4148 and one resistor to ( + ) - scheme AND, or to (-) -scheme OR . Takes much less time when the project is raw - temporarily- permanent spider-like montage (assembly ) - spatial network of wires, integrated circuits and other elements

[Carl-NC]

It worked pretty good, but was weak on the low conductors. The gain in the harmonic channel needs to be increased to equalize the gain between high and low conductors.

The best way to equalize the gains is at the TX. The DFX did a really good job with this waveform:



That's a 5:1 ratio which is better than a 3:1 ratio. If you want to keep the 3:1 ratio then one of these waveforms will work:



Any of these can be bit-banged in a cheap 8-bit PIC or if you use a better processor with DMA-based timers you can do complex waveforms in the background. Demod waveforms stay the same.

[ivconic]

[Altra]

I can't see the notch filter which removes 15kHz signal in channel 5kHz ?

You do not want a notch filter. The circuit has two bandpass filters. At 5khz and 15khz

[Altra]

[ATTACH]56575[/ATTACH]

Nice work! With my old eyes, I split the schematics up so that I read them.

I don't have any videos yet. I'll pull the circuit out and see how it working.

[Altra]

The best way to equalize the gains is at the TX. The DFX did a really good job with this waveform:



That's a 5:1 ratio which is better than a 3:1 ratio. If you want to keep the 3:1 ratio then one of these waveforms will work:



Any of these can be bit-banged in a cheap 8-bit PIC or if you use a better processor with DMA-based timers you can do complex waveforms in the background. Demod waveforms stay the same.

Thanks for your insight and drawings. My pcb has all the digital signals lines on a header. I can take the pic18 out and use a dspic to do the more complex timing.

Also attached is a recent mf patent the may interest some.

[ivconic]

Nice work! With my old eyes, I split the schematics up so that I read them.
I don't have any videos yet. I'll pull the circuit out and see how it working.

My eyes are not much better i suppose.
I do this regularly with all shematics in which i find something interesting, part or whole of it.
To be able later to edit, replace, cut, add... try different approach etc.

[Taktyk]

Altra, you have a production file for PCB?

[Riss]

You do not want a notch filter. The circuit has two bandpass filters. At 5khz and 15khz

However, why is there a notch filter in the circuit on CZ 5, CZ 6 ? There is nothing accidental in this world . There is mathematics everywhere - more or less . No one can afford unnecessary expenses ,even for one resistor... there are two stages of active filters in channels 5 and 15 kHz . Let's see which of you will answer why the damn notch filter? And once this elementary question has been clarified, we can move on to Fourier analysis, and the easy-to-understand sine-cosine function . Because the slightly more complex current waveform of DFX - there is nothing accidental in it . To find out ( to decompose ) you need it :

1. a sheet of checkered paper
2. pencil
3. mathematical reference book
4. Chivas Royal Salute 21 Year Old - one bottle (it doesn't give me a headache , but my headache starts immediately when I open mathematical reference book