New Hammerhead Project

[KingJL]

Carl,
As you know, I am starting a new Hammerhead project. I will put the project outlines and initial system design to this forum shortly (when I nail down a couple of key decisions). One of the key requirements is to optimize low noise capabilities of the front end. Since I will be using multiple HH boards in the implementation, I wish to break the front-end into two stages. It will be a low noise preamp running a relatively low gain, followed by a a seconday amp (on the second board) to make up a total front end gain of around 1000. The gain in the second stage will be controlable (range to be determined).

To that end I am looking at a AD797 runing non-inveting with a gain of 10 for the 1st stage. Also the input resistor (currently R12 in the single board) needs to be no more than 100 ohm to take advantage of the AD797 low noise(ref. AD797 spec. sheet). The second stage of the preamp is currently planned to be a AD8031 inverting with a gain of about 100.

Enough for now as I want to focus on this issue of the pre-amp. The question is what are the issues (other that the obvious physical interconnection issue) that I have not forseen?

J. L. King

[Carl-NC]

Your boards were mailed today!

[KingJL]

Your boards were mailed today!

Again... thanks!
J. L. King

[bbsailor]

Other dual front end potential issues

The question is what are the issues (other that the obvious physical interconnection issue) that I have not forseen?

J. L. King

If you plan on making use of a DD coil, where the back emf pulse is reduced compared to a mono coil, the 100 ohm input resistor may not be an issue. But it could be an issue on a mono coil. Here is my thinking about this. The flyback voltage decays in three stages.
1. The MOSFET clamping stage. If the flyback voltage exceeds the MOSFET voltage rating the MOSFET will clamp the voltage at the MOSFET voltage rating and extend the fall time slightly.
2. The combined value of the damping resistor (Rd) is effectivly in parallel with the input resistor (Rin) while the decay voltage is above the clamping diode voltage. In a mono coil this would be a higher voltage than in a DD coil with a surpressed or somewhat balanced out flyback. You may want to try a few ways to minimize the effect of this input resistor by using two series clamping diode sets or using other types of diodes to minimize the effect of Rin affecting the voltage decay time.
3. The value of the damping resistor defines the final stage of the voltage decay and damping, and ultimately the potential earliest sampling time.

If you want to take advantage of a dual front end, consider that the front end will saturate while the any voltage is above the gain plus rail to rail IC supply voltage. With a gain of 10 you have more room to improve the recovery speed than having a gain of 1000 in a single stage.

You might want to consider making a small IC module that can fit off the main board with the following components.
1. Clamping diodes
2. Input and feedback resistors
3. IC, first amplifier stage
4. Power supply filtering
5. RX damping resistor (for a DD design)
6. Method to encase this module in a copper foil shield to minimize noise.
7. A quick way to experiment with a gain selection of 20X by using a small dip switch.

Point #6 can be expanded some. If you plan on remotely placing the first amplifier stage about 1 foot above a DD coil, you can eliminate about 2 feet to 6 feet of cable capacitance (depending on whether the control box is shaft mounted or body mounted) which could speed up the coil a little more. For test purposes, you could power this module with two 9V batteries.

One thought I have had about DD coils for PI machines is doing what Minelab does in their Sovereign IB coils. They embed the preamp module right in the coil housing. If this is tried, components need to be selected that don't allow eddy currents to form on the components themselves and become a target. The DD coil should be designed to be nulled when the module is placed in the coil to minimize any effects of it being very close to the coil. The circuit board would need to use small traces (surface mount) with no large loops that would allow eddy currents to circulate.

These are just some thoughts I have had over the years that seem appropriate for your dual amplifier experimentation.

bbsailor

[KingJL]

The question is what are the issues (other that the obvious physical interconnection issue) that I have not forseen?

J. L. King

If you plan on making use of a DD coil, where the back emf pulse is reduced compared to a mono coil, the 100 ohm input resistor may not be an issue. But it could be an issue on a mono coil. Here is my thinking about this. The flyback voltage decays in three stages.
1. The MOSFET clamping stage. If the flyback voltage exceeds the MOSFET voltage rating the MOSFET will clamp the voltage at the MOSFET voltage rating and extend the fall time slightly.
2. The combined value of the damping resistor (Rd) is effectivly in parallel with the input resistor (Rin) while the decay voltage is above the clamping diode voltage. In a mono coil this would be a higher voltage than in a DD coil with a surpressed or somewhat balanced out flyback. You may want to try a few ways to minimize the effect of this input resistor by using two series clamping diode sets or using other types of diodes to minimize the effect of Rin affecting the voltage decay time.
3. The value of the damping resistor defines the final stage of the voltage decay and damping, and ultimately the potential earliest sampling time.

If you want to take advantage of a dual front end, consider that the front end will saturate while the any voltage is above the gain plus rail to rail IC supply voltage. With a gain of 10 you have more room to improve the recovery speed than having a gain of 1000 in a single stage.

You might want to consider making a small IC module that can fit off the main board with the following components.
1. Clamping diodes
2. Input and feedback resistors
3. IC, first amplifier stage
4. Power supply filtering
5. RX damping resistor (for a DD design)
6. Method to encase this module in a copper foil shield to minimize noise.
7. A quick way to experiment with a gain selection of 20X by using a small dip switch.

Point #6 can be expanded some. If you plan on remotely placing the first amplifier stage about 1 foot above a DD coil, you can eliminate about 2 feet to 6 feet of cable capacitance (depending on whether the control box is shaft mounted or body mounted) which could speed up the coil a little more. For test purposes, you could power this module with two 9V batteries.

One thought I have had about DD coils for PI machines is doing what Minelab does in their Sovereign IB coils. They embed the preamp module right in the coil housing. If this is tried, components need to be selected that don't allow eddy currents to form on the components themselves and become a target. The DD coil should be designed to be nulled when the module is placed in the coil to minimize any effects of it being very close to the coil. The circuit board would need to use small traces (surface mount) with no large loops that would allow eddy currents to circulate.

These are just some thoughts I have had over the years that seem appropriate for your dual amplifier experimentation.

bbsailor

bbsailor

I thought going to 100 ohms for Rin might be a little aggressive, and I did consider the fact that it was in parallel with the damping resistor while the pulse level was above the diode clamp level. I was wondering how low I could take it. 1000 ohms is the upper limit for this particular amplifier. After that the low noise advantages are overwelmed by resistor noise from Rin. I was wondering if a little over damped in the middle stage is really a bad thing. In reading Corbyn, he recommendend slightly overdampening for the transmit coil and critical dampening in the receive coil in his 2 coil design. I was wondering if a little over dampening in the middle stage might effectively accomplish the same in a monocoil.

As to your recommendations, I am going to put the in my notebook. I actually like them. I think putting the initial stage pre-amp in a dd coil is a project that I want to pursue.

But right now, my intention is to build a PI consisting of 2 (and then 3) HH boards to explore the advantages of a second channel (e.g. Goldscan IV), and for a third channel dedicated to ground balance. I plan on stacking these boards and only populating them with what id needed for that channel. In doing this, I also have the opertunity to easily break the high gain pre-amp into 2 stages with very little work to the boards. Hence my original question.

I plan to focus first on the main channel (which will be pretty much complete HH) and the ground channel. The ground channel board will have the second pre-amp section, sample pulse timing section, and the integraor sections populated (no audio, no oscillator, and no transmit). The sampling pulses for the ground channel wiil be delayed far enough so that the only thing in the integrators should be the ground signal. The integrators in the ground channel will have the integration time lengthed significantly. The output of the ground channel integrator should only be the ground error. This is the fed back into the ground channel integrator and the main channel integrator.

After this is working satifactorily, I plan on adding the 3rd board to implement a iron channel with the timing delayed so that only iron targets are in this channel. This will open up opportunities for some discrimination.

The reason that I am going to try to use 3 HH boards is: (1) the board is alreay designed and it works, (2) the HH is so tunable, all the variables are adjustable, (3) it seems to be the fastest way to prototype all the features I want. After every thing is working and tuned, then it may be time to design one board with everything that I have learned.

Somewhere in all of that a try at a DD coil should be in the mix (probably towards the end after I have sorted out the primary functions).

I am attaching a pdf of the schematic for the implementation of the ground channel. The IC's do not reflect what is used on the HH. They were only changed so that I could implement it on the only interactive simulator that I have available. But the resistors, capacitors, U numbers correspond to the HH schematic. The ground channel has 100 added to all component numbers.

Thanks you for you thoughtful response and suggestions. As I go forward I will appreciate any and all feedback and suggestions.

Question: How far out should I sample to just obtain predominantly ground excited signal. From reading everything that I can get my hands on, I have concluded that at about 150 usec, all metal has decayed. Is that a valid assumption?

Thanks,
J. L. King

[bbsailor]

Question: How far out should I sample to just obtain predominantly ground excited signal. From reading everything that I can get my hands on, I have concluded that at about 150 usec, all metal has decayed. Is that a valid assumption?

Thanks,
J. L. King



J. L. King

Read all the posts by Bugwhiskers. He has done some testing on very mineralized soil in Autralia and can give you some upper limit ground delay numbers. http://thunting.com/geotech/forums/f...splay.php?f=38

Try using the inductor discharge curve of the MiscEl program. My quick analysis of a 300 uH coil with 1A of current, pulsed for 54us is that it takes 14.88us for the voltage to fall to .7V using a 100 ohm load resistor. This is what a 100 ohm input resistor might look like when the clamping diodes are conducting. If you went to 1000 ohms the time drops to 1.48us and if you use 500 ohms it is 3.9us.

I hope Bugwhiskers jumps in here to answer some of your questions as he has played with a digital version of what you are trying to do with multiple Hammerhead boards. Read his posts in the above topic thread.

I believe that your design should start with identifying the maximum target decay for the type of targets that you primarely seek. Some common US coins can go out past 200us till the eddy currents totally decay. You probably want to focus on the different decays near the knee of the decay curve rather than at the end of the decay curve.

bbsailor

[KingJL]

I,m convinced!

Question: How far out should I sample to just obtain predominantly ground excited signal. From reading everything that I can get my hands on, I have concluded that at about 150 usec, all metal has decayed. Is that a valid assumption?

Thanks,
J. L. King



J. L. King

Read all the posts by Bugwhiskers. He has done some testing on very mineralized soil in Autralia and can give you some upper limit ground delay numbers. http://thunting.com/geotech/forums/f...splay.php?f=38

Try using the inductor discharge curve of the MiscEl program. My quick analysis of a 300 uH coil with 1A of current, pulsed for 54us is that it takes 14.88us for the voltage to fall to .7V using a 100 ohm load resistor. This is what a 100 ohm input resistor might look like when the clamping diodes are conducting. If you went to 1000 ohms the time drops to 1.48us and if you use 500 ohms it is 3.9us.

I hope Bugwhiskers jumps in here to answer some of your questions as he has played with a digital version of what you are trying to do with multiple Hammerhead boards. Read his posts in the above topic thread.

I believe that your design should start with identifying the maximum target decay for the type of targets that you primarely seek. Some common US coins can go out past 200us till the eddy currents totally decay. You probably want to focus on the different decays near the knee of the decay curve rather than at the end of the decay curve.

bbsailor

bbsailor,
I'm convinced... Rin will remain 1K. One question answered.

For the ground channel, the idea is to have the integration TC set long enough, that hopefully a coin shouldn't integrate enough to make much of a signal gain in that channel with normal MD sweep motion. My intention is to have the differential integration set such that only a slow and constant (or fairly constant) ground signal change will cause a resultant integrated output change. On this ground channel, I want to sample far enough out, that there is very little meaningful target integration, but close enough in that I get an increase in integrated signal for mineralizied soil which should last over a most of the sweep. The ground channels only function is to determine ground signal and provide a signal that can be used to adjust it out. In doing so it will also subtract it out of the primary (normal) channel also. That is the plan anyway.

J L King

[KingJL]

Your boards were mailed today!

Boards arrived today!!

Carl,
Do you have any further guidance on parts for the SMT board? What wattage resistors did you use on your SMT build?

Regards,
J. L. King

[bugwhiskers]

Soils ain't soils

Hi Guys,

I am no expert on "ground" as seen by a PI. What I do know is that the classic 1/T ground response is probably just an average. I daresay those making that assumption haven't tested some of the worst ground in Australia. There is a location called Beggary Hill and the "soil" there is so ferrous that if a magnet is put into it and withdrawn everything sticks to the magnet and nothing falls away.
In doing some testing for Dave Emery's Pulse Devil/Nemesis Dave concluded that the ground was the equivalent of holding a pair of long nosed pliers above the coil. Clearly the 1/T would not apply to this situation nor would the assumption that the ground response is linear.

regards
bugwhiskers

[KingJL]

Parts Parts Parts!

Well,
All of the parts (enough for 3 SMT boards) that I ordered arrived today. The first phase of this project is to split the pre-amp to two sections and also allow for a DD coil configuration using the SMT boards. The first pre-amp will be a non-inverting AD797 with a gain of 10. This will be mounted on the second board which has been modified for non-inverting operation. Also the receive coil will attach to the second board. The output of the AD797 is the piped down to the main board with a normal inverting AD8033 running with a gain of 150. The output of the AD8033 is passed normally to the integrator circuit of the main board as well as fed back up to the second board integrator which is going to be the ground balance channel.

Progress today: Modified the second board for non-inverting operation. Populated the power supply section of the main board and tested. All voltages OK. Installed IC5, the 555. Tested the 555 timing...all OK. Went to install 74HC221s... Uh-Oh! Cant seem to find any 74HC221's. Dang, they didn't get ordered!! Went to Mouser and ordered 6 74HC221's.

Pictures in attached zip.

[KingJL]

Error in above post

In the above post I identified the second stage of the pre-amp as a AD8033. This was wrong! The second stage of the pre-amp is a AD8031.

[KingJL]

Progress!!!

Well the first iteration of my HH project is assembled on the bench and has undergone bench testing. I am happy with the results so far. The second board (one on right in pics) has the initial stage AD797 pre-amp, non-inverting with a 10X gain. Also on the second board is the ground channel differential integrator section. The main baord (left in pics) has the second stage of pre-amp, inverting AD8031 with 150X gain. Also on the main board is the transmitter, main channel receiver backend, and audio. I set it up to use a DD coil (300uH tx, 410uH rx). System works great on bench. My gold wedding band at 23cm, a nickel at 21cm in a noisy environment. I have about 100mv noise/interference measured at the output of the second stage pre-amp. If I move coil to vertical plane noise reduces to about 50 mv normal noise. Detection was done in horizontal plane with interference.

Ground channel integration gain set a 1/47 of normal (C17,C18 4.7uF). Output of ground channel fed thru a 100k pot to pin 2 of IC8a of main channel (in final hookup, I believe I will make this a multi-turn pot).

Had a noise issue while running 7660 synced with main oscillator. Now running non-synced (R37 removed) with no issues. Blew Q4 3 times, until I put a 50ohm 1 watt resistance in series with R41 to limit max current.

Settings:
PW 60 usec, Freq 670 Hz, Primary Delay 9 usec, Secondary delay 150 usec Sample width 15 usec, battery voltage 10.8 volts.

Next steps, mount boards together and mount them in a shielded metal box; mount box in final enclosure with all controls.

[KingJL]

Coming Together

Latest pics of project progress. Next to mount case to rod!

[KingJL]

Schematic

Attached is a schematic that shows how the two boards were used to implement a split pre-amp and a ground balance channel. Eventually I may add a 3rd board to implement an "Iron" channel. The schematic detail was not carried out to show the nulling circuits as they are effectively unchanged. In the schematic the original components are prefaced with the board ID (e.g. IC6 is B1-IC6 for board 1 and B2-IC6 for board 2). New components are in the 100 range (i.e. R101 - 106). Both boards had the power and receiver and receiver timing sections populated. In addition, the main board had the master oscilator (IC5), transmitter, and audio section. The master oscilator signal (N1 in the original schematic) is fed to the same point on the second board. The main board had the trace cut from the coil to R12. The Rx coil is connected to the added resitor (R101) on the second board.

[bugwhiskers]

Hi KingJL,

The AD797 as shown has a gain of 2, is the ground resistor at 10k a typo ?

regards
bugwhiskers