Moodz' Awesome Gold Pulse Induction Version 3 - MAGPI V3 Project

[moodz]

Hi Moodz,
Adjusting the timing to minimise the clipping the best I could get is pictured below green being the ouput from the buffer and red the input to the LM6171.
Timing is 1.3us and 100ns sample does this look ok to you or is it still too deep..?

Hmmm ... the negative going error looks better but the waveform is now clipping at the 5 volts .... the only clipping during the RX cycle should be during the flyback .. which is the first thing that happens in the RX period.
Should look like the CRO shot posted earlier ...
A large target can cause clipping ?? Remove all targets from your sim if you have them.

moodz.

[Dean Sarelius]

The best way to resolve clipping is to set the sample pulse to midway of the clipping "pulse". In the V3 the flyback is about 200 - 250 negative volts and will then swing positive if not damped ... the flyback damping error is therefore positive at the input of the preamp which amplifies and inverts this error voltage.
If the error sample is too early you will be sampling the actual flyback and if the sample is too late then in either case ( too early or too late ) the magnitude of the damping error voltage will cause clipping.
For a 30 microsecond TX time the sample pulse might be 80 to 85 counts ( corresponding to 1.6 microseconds ) however I have mentioned before that this timing depends on the SRF of the coil when it is attached to the detector ( so to take account of the capacitance that the front end as well ).
This 1.6 microseconds was determined for a REAL circuit .. not a simulation. The simulation timing might be different as its harder to account for all the stray capacitances in a simulation. ( could be more or less than a real circuit ).

Short answer is set it to the middle of the clipping.

In the pic below the error signal does not exceed -5 volts ( thus not clipping ).



moodz

Hi Moodz,
Adjusting the timing to minimise the clipping the best I could get is pictured below green being the ouput from the buffer and red the input to the LM6171.
Timing is 1.3us and 100ns sample does this look ok to you or is it still too deep..?

[Rick S]

Moodz,
Building your MD I want to stay as close as possible the components used in the schematic. Build the MAGPI V3 first, then do mods, up grades etc... Where did you purchase the crystal for the PIC. Having
hard time finding a thru-hole 10 MHZ 10pf oscillator. SMD 10MHZ 10pf oscillators find them in most electronic warehouses. All the Geotech projects I have built, one of them for example TGSL, become
a headache with all the changes to the first build. For me I want to KISS. I enjoy your cool head approach to this project.

[Emersonpaz]

what is the meaning of this resistor R21 and GND on both sides

[moodz]

Thanks Moodz,
Regarding the clipping, have set the timing at 1.6us with 100ns sample.
To avoid the clipping in your experience should I set the timing to be closer to the return pulse ie shorter or further away say 1.8us...?

The best way to resolve clipping is to set the sample pulse to midway of the clipping "pulse". In the V3 the flyback is about 200 - 250 negative volts and will then swing positive if not damped ... the flyback damping error is therefore positive at the input of the preamp which amplifies and inverts this error voltage.
If the error sample is too early you will be sampling the actual flyback and if the sample is too late then in either case ( too early or too late ) the magnitude of the damping error voltage will cause clipping.
For a 30 microsecond TX time the sample pulse might be 80 to 85 counts ( corresponding to 1.6 microseconds ) however I have mentioned before that this timing depends on the SRF of the coil when it is attached to the detector ( so to take account of the capacitance that the front end as well ).
This 1.6 microseconds was determined for a REAL circuit .. not a simulation. The simulation timing might be different as its harder to account for all the stray capacitances in a simulation. ( could be more or less than a real circuit ).

Short answer is set it to the middle of the clipping.

In the pic below the error signal does not exceed -5 volts ( thus not clipping ).



moodz

[Dean Sarelius]

The idea sprung from a voltaqe controlled resistive damper using mosfets ( see patents by Nelson ) when I realised that constant current sink utilising the I/V curves of a suitable mosfet was more effective with a simple feedback loop driven by the damping "error".
This was patented ...A DAMPING CIRCUIT FOR A COIL - AU2013101058.

Looking at the loop distortion you are trying to avoid by adding a buffer .. analysis will show that the frequency response of the loop is not above 1 Mhz but the primary distortion components start at around 10 Mhz upwards ( high speed switching edges ) and will not make it around the loop thus not impacting the operation of the loop.
The only time I would consider a buffer would be where I might be using a high speed ADC to implement the loop ( but then I would not be using the sampling switches .... so .... hmmmm ).

So the only criteria for including a buffer would be a demonstrable improvement in performance of the overall signal chain.

The green trace is seems to be clipping at about -8 volts ... this means the damping error sample is not correct in timing.

moodz

Thanks Moodz,
Regarding the clipping, have set the timing at 1.6us with 100ns sample.
To avoid the clipping in your experience should I set the timing to be closer to the return pulse ie shorter or further away say 1.8us...?

[IBGold]

Hi Ian ... glad you got the unit. You dont need my permission its yours to do what you want.

That unit was made to be (almost ) as light as possible with no power on/off or LCD etc. I dont have any power on/off circuitry ... using the old school air-gap technique.

Please also note the default timings are really for "beach mode" ... maximum sensitvity and no mineralisation.

There is a previous post about the encoder wiring ... it works either way.

moodz.

Fine Paul but on my encoder the centre green wire and link to the P/B and the red wire are transposed and OK I will fit a switch I have some genuine new SD power switches in my spares box then we will have to talk about timings.

Regards, Ian.

[moodz]

Hi Moodz,
Succesfully simulated your circuit with and without the buffer, I must say this is a very interesting circuit I would really like to know where you got the ZPD concept from..?
When comparing the results there doesnt appear to be much of a difference in the performance both circuits perform exactly as expected providing the much needed negative bias voltage for Q2.
The bias voltage settles at around -4V and it takes approximatley 90ms to reach this value with and without the buffer.
Below are screen shots results with the addition of a buffer using the NE5532 A and B one for the buffer the other for the TX feedback replacing the NE5534.

The idea sprung from a voltaqe controlled resistive damper using mosfets ( see patents by Nelson ) when I realised that constant current sink utilising the I/V curves of a suitable mosfet was more effective with a simple feedback loop driven by the damping "error".
This was patented ...A DAMPING CIRCUIT FOR A COIL - AU2013101058.

Looking at the loop distortion you are trying to avoid by adding a buffer .. analysis will show that the frequency response of the loop is not above 1 Mhz but the primary distortion components start at around 10 Mhz upwards ( high speed switching edges ) and will not make it around the loop thus not impacting the operation of the loop.
The only time I would consider a buffer would be where I might be using a high speed ADC to implement the loop ( but then I would not be using the sampling switches .... so .... hmmmm ).

So the only criteria for including a buffer would be a demonstrable improvement in performance of the overall signal chain.

The green trace is seems to be clipping at about -8 volts ... this means the damping error sample is not correct in timing.

moodz

[moodz]

Moodzs is the schematic still current or has any improvements been made ?

Hi David,

The most up to date schematic is post #23 https://www.geotech1.com/forums/foru...820#post409820

moodz

[moodz]

Moodz,
Found monolithic multilayered 1uf caps. Sorry for the previous question.

Hi Rick ... I used 1 uF for maybe improved noise reduction ... 0.1uF will work also.

moodz

[moodz]

Hi Paul,

Package arrived today with your permission I want to make a few changes and I have some questions the changes first,
.1. I want to remove the DC power socket and replace with a LTW that I use reason found on the other device the DC plugs were very noisy .
.2. Remove the 3.5 mono socket and replace with a 3.5 stereo one wired as a mono as that is how the Bluetooth TX that goes to my hearing aids is configured.
.3. After initial testing I want to remove the encoder and fit it into an enclosure which will also house the screen and fit on the handle I have the screen and the enclosure is almost complete.
Now for the questions .
.1.I intend running the Magpie V3 using my existing 2S Li-Ion 7200mAh is that OK?
.2. I notice that the encoder is wired differently to the pictures you posted earlier two wires on this encoder are transposed to the ones in the pictures the plugs on the boards are all the same as each other so which is correct.
.3. I intend to run the screen and encoder Via a shielded 9 core cable via an 8 pin plug and socket on your box and hard wired to the screen box does this sound feasible?
.4. I notice that their is no On/Off switch should I add one between power socket + positive (my LTWW socket) and the + wire to the board unless of course you have something already enabled to turn the power On/Off?
That should do for now.

Regards, Ian.

Hi Ian ... glad you got the unit. You dont need my permission its yours to do what you want.

That unit was made to be (almost ) as light as possible with no power on/off or LCD etc. I dont have any power on/off circuitry ... using the old school air-gap technique.

Please also note the default timings are really for "beach mode" ... maximum sensitvity and no mineralisation.

There is a previous post about the encoder wiring ... it works either way.

moodz.

[David_1]

Moodzs is the schematic still current or has any improvements been made ?

[Rick S]

Moodz,
Found monolithic multilayered 1uf caps. Sorry for the previous question.

[Rick S]

Moodz,
Looking at this project and the build seems really good. New to programming. Use this project to learn more. Looking at the schematic and pictures posted, all the caps used for decoupling are 1uf? My limited experience they are usually 0.1uf. The pictures are so that
I can't read the yellow caps on the pcb. I don't see any 1uf caps on your pcb electrolytic or film. What are the caps polyester? The reddish looking ones look like metallized polypropylene. Also, where did you purchase the 10 Mhz 10pf ocscillator? I have found
10 Mhz load capacitance 18pf or higher. BOM would be nice. Thanks.

RickS

[Dean Sarelius]

Thanks Dean .. dont get me wrong I am not trying to dismiss your ideas for improvement .. I just want to lock down the V3 hardware as there are some built up units already operating by various people and whereas we can change the software because that only involves re-programming those existing units ... if we change the hardware then those existing units and ones being built up might become obsolete and those people would not be happy.

For your info ... I have V4 in simulation already with NMOS and significantly higher power and V5 which is bipolar ... not sure if target sensistivity will increase that much though we hopefully will start to see a discrimination function.

moodz

Hi Moodz,
Thinking about adding another opamp ex the buffer to provide an edge triggered pulse as auto sat mechanism to position the damping point which the MCU can read on the fly.
In your experience with this circuit do you find that for minealised soil the timing of the delay for sampling the ZPD can vary, for example with high mineralisation does the timing need to be adjusted further out for example.?
So instead of 1.6us it could be required that the sampling point needs to be adjusted to 1.9us...?
Cheers,
Dean