chance pi build

[baum7154]

I do see your point that the feedback diodes do have better control so I think I'll try them on both stages. This should help to preserve the 6us decay time of my current 8" coil.

Thanks

Dan

[green]

baum
I see in your reply in gold simulation you are using the 2 stage amp. Is it working better than the single amp? I use a two stage amp for faster response on the bench, but I'm not sure that it would be better in a field metal detector due to some oscillation and more noise from higher frequency response.

[baum7154]

My purpose in doing the 2 stage amp is to get/keep the amp out of saturation. With the new amp my 6us coil decay at the input is increased by only 1 us to 7us decay at the output. In the process of doing this though I noticed the improvement in detection of the glass vail of small gold flakes (about the volume of a book match head) that in the past has only given the slightest notice of detection. With the 2 stage amp it is solid at 1" distance. Also in this process I am not supplying the A/D exactly what it wants in order to give good VDI discrimination and audible tone discrimination of different metals than gold.

Right now I am trying out some tricks on the original 1 stage amp and also trying to characterize exactly how different levels and biases at the A/D affect the discrimination performance.

Regards,

Dan

[Waikiki_Sweep]

My purpose in doing the 2 stage amp is to get/keep the amp out of saturation. With the new amp my 6us coil decay at the input is increased by only 1 us to 7us decay at the output. In the process of doing this though I noticed the improvement in detection of the glass vail of small gold flakes (about the volume of a book match head) that in the past has only given the slightest notice of detection. With the 2 stage amp it is solid at 1" distance. Also in this process I am not supplying the A/D exactly what it wants in order to give good VDI discrimination and audible tone discrimination of different metals than gold.

Right now I am trying out some tricks on the original 1 stage amp and also trying to characterize exactly how different levels and biases at the A/D affect the discrimination performance.

Regards,

Dan

It is very interesting results. Too bad there is no chance to change software to get it serve your updates of schematic better.

Experimenting with coils I have noticed that speed of the decay is very much depends on inductance of coil.
If you like you can try some coils with lower inductance. They have to be much faster.
There will be a problem with more current in lower inductance coils because TX impulse is fixed. Probably you can add a resistor of few ohms in series with powering MOSFET.
Fastest way to build test coils for Chance is cut cardboard disk and make non-odd number of cuts 4 cm. long from edge to the center of the disk than wind that basket coil.
With cardboard it takes 5 minutes to make a test coil instead of plastic which is brittle and you need a lot of time to cut it.

[baum7154]

For the past month I have been building and testing a 2 stage input amplifier based on the original OP37 amp. The problem to be overcome is amplifier saturation and consequently the extended coil decay time presented to the A/D. My fast spider coil decay time is 6us but the 470 gain of the original amp caused saturation and extension of the coil decay to 16us. This extended decay encroached on the A/D sample window that opens 8us after coil decay begins thereby masking short target responses from small gold.

The present 2 stage amp modification uses 2 OP37 amps with the first stage inverting amp set at a gain of 18 and the second stage non-inverting amp set to a gain of 28. Positive voltage is limited in the first stage to 1.4 volts by placing two diodes in series across the feedback resistor to be in forward conductance on the positive going voltage. The original negative voltage limiter diode is also used in the feedback loop of the first amp keeping the negative to 0.7 volt . These measures keep the first amp out of saturation and the low gain keeps amplification of noise down.

The second amp (non-inverting --28 gain) also has positive voltage limiter diodes in the feedback loop across the feedback resistor to limit positive output voltage to about 11 volts, still below the 12.5 rail voltage and the minimum operating 11 volt set point of the battery supply. I may reduce this further if testing shows no benefit of the higher voltage. Once again the negative voltage is limited to 0.7 volt with a single diode in the feedback loop.

Gain of the two stages is now 504 instead of the original design of 470. Once this amp was put into operation greater noise became apparent. In order to mitigate the noise, R/C filtering was implemented in each of the gain stage feedback loops and in the interstage coupling and the coupling to the A/D. I designed the low pass filtering for an Fc of about 680 kHz in order to pass signals in the range of 1.5 to 2us and longer. The 4 levels of filtering reduced noise from about 1 volt down to under .02 volt. Filtering was accomplished by adding proper capacitance across the feedback resistor in the feedback loops of the two amps and between the interstage coupling resistors and ground, carefully taking into consideration the capacitance of the diodes present in the feedback loops.

The effect of this modification is better detection of small gold. I have been using a small glass vial of placer flakes equivalent in volume to a book match head and the detector now gives a better response to this target. The largest of the flakes is about the size of 4 stickpin heads combined. It does not go into 'tone lockup' but does give definite beeps correlated to the passing of this target that approach tone lockup at times. Detection distance is now 1" to 2" greater for most targets.

As far as implementing this amp on the original CHANCE PI board, it is 'Plug and Play'. It is designed to plug directly into the original single stage amp IC socket using the original power, ground, amp input, and output to the A/D. I did add another soldered ground lead to the ground plane on the new PCB.

I'll post some pictures of the board and O'scope screen shots soon. I plan to take this out to the my prospect area this coming Monday and hope to bring some better gold targets home!

Regards,

Dan

[greylourie]

This is great work you have done to improve your machine. Thanks for taking the time to describe your work so intimately, and share with us. I'm sure plenty of us will be coming back to re-read this and let it sink in.

[baum7154]

Thanks for your kind comments Greylourie. I'm unsure if my approach to this amp design is the conventional way things are done but it seemed logical to me. I'm open to suggestions to further improve the performance of this amp. There are many on this forum more knowledgeable of this topic than me and I have learned a great deal from them since joining this group. That said, I did not have much luck in searching the forum for a 2 stage amp design that I could easily implement in CHANCE PI. It would be great if we had a library of designs that are easily found. Maybe we already do and I have just missed it in this case.

I don't have the CAD software to produce the schematic and PC board layout but I will soon post a photo of the hand drafted versions I did. I'm more than happy to contribute details to this group.

Regards,

Dan

[6666]

Dan if you are looking for some free simple cad software just to produce a schematic have a look at Tinycad, its free and simple to use.

[baum7154]

Thanks for the tip! I actually used tinycad about 5 years ago on some ham radio designs and it was pretty good but a little short on some IC selections. I'm sure the selection is better now.

Thanks,

Dan

[6666]

Interesting , on a little sim I have, I uped the preamp voltage from +-5v to +-8v and the preamp does settle quicker.

[baum7154]

Very good!

[bbsailor]

Thanks for your kind comments Greylourie. I'm unsure if my approach to this amp design is the conventional way things are done but it seemed logical to me. I'm open to suggestions to further improve the performance of this amp. There are many on this forum more knowledgeable of this topic than me and I have learned a great deal from them since joining this group. That said, I did not have much luck in searching the forum for a 2 stage amp design that I could easily implement in CHANCE PI. It would be great if we had a library of designs that are easily found. Maybe we already do and I have just missed it in this case.

I don't have the CAD software to produce the schematic and PC board layout but I will soon post a photo of the hand drafted versions I did. I'm more than happy to contribute details to this group.

Regards,

Dan

Dan,

See this link where Eric Foster describes his PI machine that can sample at 5 uS. http://www.findmall.com/read.php?34,...171#msg-137171

The TX pulse width that he uses is only 20us long but his pulse rate is 20,000 PPS.

This should give you some more ideas.

bbsailor

[6666]

What a pity the pictures have gone from that thread.

[baum7154]

Dan,

See this link where Eric Foster describes his PI machine that can sample at 5 uS. http://www.findmall.com/read.php?34,...171#msg-137171

The TX pulse width that he uses is only 20us long but his pulse rate is 20,000 PPS.
This should give you some more ideas.

bbsailor

Thanks for the thread Joe…very interesting reading and something to shoot for…someday.

Dan

[green]

[The present 2 stage amp modification uses 2 OP37 amps with the first stage inverting amp set at a gain of 18 and the second stage non-inverting amp set to a gain of 28. Positive voltage is limited in the first stage to 1.4 volts by placing two diodes in series across the feedback resistor to be in forward conductance on the positive going voltage. The original negative voltage limiter diode is also used in the feedback loop of the first amp keeping the negative to 0.7 volt . These measures keep the first amp out of saturation and the low gain keeps amplification of noise down]

Some thoughts, maybe correct maybe not. With the two stage amp the bandwidth might be high enough that the input resistor acts as a damping resistor. The damping resistor and input resistor are in parallel giving you a damping resistance closer to 600 ohms. I doesn't matter in the circuit just the thinking on how it works. The circuit I've been playing with uses the input resistor only, allowing a lower value input resistor (less resistor noise).