IGSL

[o.mag]

Simpler - the better.
I guess any better PIC should do.

Hi Ivconi
My analysis for the choice of the peak:
At frequency 8.66kz period T takes 115us
The sampling time period lasts half 57us
For geb there a setting range of 9us to start sampling
For ferrous there a setting range of 24us to start sampling
if the PIC has a 20MHz quartz (16F) its internal frequency is 5MHz therefore between 0.2 us instructions
it gives 45 position for Geb (enough?) and 120 to the other setting
a 48MHz (18F2550) so it gives 108 position adjustment for Geb and 288 for other settings

opinions and comments are welcome

[ivconic]

More than enough!
4-10Mhz clock will provide sufficient timings and resolution.
Whenever we talk about uPc implementation ; i am having PIC16F877 on my mind, don't know why!

[o.mag]

More than enough!
4-10Mhz clock will provide sufficient timings and resolution.
Whenever we talk about uPc implementation ; i am having PIC16F877 on my mind, don't know why!

same family : PIC16F877 (40 pin) is the big brother of PIC16F873 (28 pin)

[ivconic]

same family : PIC16F877 (40 pin) is the big brother of PIC16F873 (28 pin)

[Davor]

I think the PIC would be a much better signal source for Tx as well. PIC provides phase granularity, and that's OK, but phase reference is still prone to PWM and phase noise. If Pic provides Tx excitation AND phase references, the phase noise is kaputt.

[golfnut]

Fundamantally this machine will have relatively poor sens, like all of the vlf IB kits on here Half of the Rx signal is discarded - like a 1/2 wave rectifier.



All of the commercial designs that get results in the feild have a full wave demod, so you get better signal to noise.

There was a design called a verator on here that has a good demod.


Its like saying Im going to design this new car and leaving the rear wheels off - yes it may go, but not as good as one with 4 wheels.

S

[Davor]

true, but it is fun anyway. You can get a pretty decent rig from any concept there is, yet some solutions are more elegant in solving some small troubles.

Maybe it is a perfect moment to join forces and make some new VLF. It will not be from the scratch, but from the best concepts there are. You can't beat the discrimination ability and energy efficiency of a VLF, so it is by no means outdated.

[simonbaker]

Fundamantally this machine will have relatively poor sens, like all of the vlf IB kits on here Half of the Rx signal is discarded - like a 1/2 wave rectifier.



All of the commercial designs that get results in the feild have a full wave demod, so you get better signal to noise.

There was a design called a verator on here that has a good demod.


Its like saying Im going to design this new car and leaving the rear wheels off - yes it may go, but not as good as one with 4 wheels.

S

Design competition --

1. "full wave" SD with minimum parts / least complexity.

2. Full wave SD with lowest noise.

[o.mag]

ok for a prototype with "PIC16F877A PCB CEPARK minimum core of the system board " I ordered.
I start to think to manage the crossing of two parts.

[Davor]

In case you are into optimising an existing design AND wish to introduce full wave switching, you can freely discard one of the GEB channels and port a single GEB to two comparators stages. Unless you really wish to have two separate GEBs this mod would enable you to use a single 4053 for full wave switching of 3 phases: GEB, Fe Disc, and Cu Disc. I joined two GEBs in my mods, and it works perfectly. Both GEBs on my board do exactly the same thing, so one is redundant.
On the plus, you can as well use a LM833 in a cross coupled configuration to supply a perfectly balanced input for the switchers, and simultaneously perfect load for balanced and common mode on the coil.

[ivconic]

I see things are getting more serious here!
So, in that spirit; i suggest you: Davor and O.Mag to join forces and design common project.
Davor, you should input your suggestions in format of readable schematic, preferably done in one of the frequent software we are using here (my favorite is ExpressSCH, but it is not problem if you use any other).
And O.Mag you should than take Davor's suggestions in consideration and possibly include them in your final design.
I can contribute and help as much as it is in my capabilities.
Yes, i don't like too much lot of IC's there too. Rising power consumption.
Also i think too that fullwave demodulator should be used.
PIC16F877 can do all the tasks needed.
Join forces and make "mo' better blues" than IGSL is now.
Cheers!

[Davor]

I kind of left my PCB routing skills back in p-cad times, so I'll gladly leave that part to, well, anybody else. In lieu of PSpice I switched to LTspice and became fond of that program. I'll gladly share all that I learned while building my IGSL.

[o.mag]

I see things are getting more serious here!
So, in that spirit; i suggest you: Davor and O.Mag to join forces and design common project.
Davor, you should input your suggestions in format of readable schematic, preferably done in one of the frequent software we are using here (my favorite is ExpressSCH, but it is not problem if you use any other).
And O.Mag you should than take Davor's suggestions in consideration and possibly include them in your final design.
I can contribute and help as much as it is in my capabilities.
Yes, i don't like too much lot of IC's there too. Rising power consumption.
Also i think too that fullwave demodulator should be used.
PIC16F877 can do all the tasks needed.
Join forces and make "mo' better blues" than IGSL is now.
Cheers!

I'm willing to consider suggestions Davor, but I do not know the timing used for a full wave detector!
IGSL on the GEB adjustment it is the same on both channels?
PS: I prefer an independent oscillator which I synchronize the PIC.
Torque LC has its own frequency, and also much less harmonic

[Davor]

I'm willing to consider suggestions Davor, but I do not know the timing used for a full wave detector!

It is exactly the same, except that the signal from the preamp is supplied in counter phase as well, and a switcher toggles between phase and counterphase. From the PIC point of view it is exactly the same. The advantages include lower losses, less offset troubles, better common mode handling, and it is by all means more sexy.

IGSL on the GEB adjustment it is the same on both channels?

Not in the original because Ivica had an otherwise good idea to sense earth by slightly underadjusting the GEB in a Fe channel. Trouble with such idea is that it requires an additional control, and in most cases you'd not use this feature anyway. In case GEB is perfectly adjusted in both channels (no underadjusting in Fe), they'd be switching at the same angle, so it can be joined in a single GEB, and make a whole shebang less complicated.
Besides, a handy 4053 has 3 switchers inside - just enough for GEB + two Disc channels.

PS: I prefer an independent oscillator which I synchronize the PIC.

There is nothing wrong with independent oscillator as long as it supplies phases correctly. In fact, controlling delay of the phase is more exact means of supplying vectors for discrimination, because the eddy currents do the very same thing - they delay. There is a potential problem of rising phase noise in 1/f region, but I guess it also depends upon the granularity of such delay. With high clock it might be just fine. Having a Tx fed by an amplified fraction of a system clock would produce no phase noise at all.
In such cases analogue is often better than digital.

Torque LC has its own frequency, and also much less harmonic

You may keep the LC tank at Tx even with forced oscillation at the clock fraction frequency. In such case it works as a filter and provides energy conservation. In this configuration it can provide even cleaner signal than the self-oscillating Tx. Just a thought.
Basically I have nothing against keeping the most analogue parts and just improving the binary ones, such as phase switchers. It would enable forking the project into a pure analogue one, and a PIC meld.

[ivconic]

I kind of left my PCB routing skills back in p-cad times, so I'll gladly leave that part to, well, anybody else. In lieu of PSpice I switched to LTspice and became fond of that program. I'll gladly share all that I learned while building my IGSL.

No, i don't mean you to draw new pcb, just to draw your corrections on existing schematic.
I posted schematic made in ExpressSCH here.
All you have to do is to install ExpressPCB software (including SCH part too) and open my file with it. Than to input your corrections and repost file again here.
That's why i almost always do post all my working files here : so to others can easily manipulate with them, input own corrections and share with others.
I hope it is easy job for you and others.
I am pushing ExpressPCB (SCH part actually) here on these forums, for long time ago, simple because i think it is most easiest software to work with.
It has everything that hobbyist may need. Inputing symbols and whole schematic is indeed very easy.
After all; that software is completely FREE and downloadable from it's official site.
...
So what i want to suggest again is; you to input your corrections in existing .sch file and repost it here.
Cheers!