My take on the HH2

[waltr]

This is my take on the HH PI detector. But first an introduction.
After watching the British TV show "The Detectorists" it seemed that the detecting parts were very accurate so started doing research on how these metal detectors work. Lance did have a CTX3030. Found the Geotech1.com site of Metal Detector Projects. Got out my proto-boards and started building and experimenting. After playing with BFO detector tried the basic PI detector and liked how this worked.
I have been playing with electronics since the mid-1960s and working in electronics since the mid 1980s. I obtained a BS in Physics in 1997 so the basics of PI is interesting.
On proto-boards I built up the HH detector with whatever parts I had on hand making a few substitutes. However, when first studying the HH schematics I went to a PIC processor instead of using all those timing chips. I've been using PICs for projects for a few decades so writing timing code is pretty easy. I had made PIC project PCBs so years ago for 18 and 28 pin PICs so used one to do the timing.
I then found this Geotech forum and started reading all the threads on PI detectors and studying how each worked. In the process of prototyping came up with a few changes to the HH. See the Schematic and PCB files.
1st change was to just use JFETs as sampling switches since I had these but no 4066s.
Next was to decrease the Pre-amp gain and add gain after the Sampling switches.
I did allow for options such as:
dual coils (TX/RX)
Jumper to select 7660 Change pump Clocking as internal, from TX pulse or from PIC.
Jumper to Audio out from 555 or PIC.
Resistor in Series with TX coil to adjust the coil's Tau.
Optional MOSFET to short the TX coil series resistor for experimenting with PIC controlled Coil current.
Option resistor divider off the TX/RX coil to directly measure or sample the coil Voltage.
Option op-amp circuits to level shift signals to the PIC's ADCs.
WR-PIdet-sh1-.pdf
WR-PIdet-sh2.pdf
WR-PIdet-sh3.pdf
WR-PIdet-sh4.pdf
WR-PIdet-PCB.pdf

[6666]

Thanks for this information.

[waltr]

That is part 1 of posting this project.

Here are photos:
Prototype on bread board
PCB stuffed and working
Side view showing the PIC board attached
12x12 inch DD coil
10inch Mono coil in progress

[waltr]

Timing and the PIC

I used a PIC16F88 since I have these and they have ADC input pins. Running a 20MHz ceramic resonator for a 200nsec instruction cycle.
TMR0 interrupts every 426usec and in the ISR the pulse/sampling sequence takes place. This is just like Carl's C code example. There is a TMR0 'tick flag' set.
In the MAIN() code section after all initialization is done there is a ' main _loop'. This watches the 'tick flag' and increments a counter. The counter is used for a crude 'state machine' to sequence reading the ADCs. This action obtain all ADCs 29 times a second which is plenty fast for manually turning a pot.
The five ADC inputs are connected to 5k Ohm 15 turn trim pots and allow adjustment of the following:
ADC0 - not used in this code but open for GEB or for direct sampling.
ADC1- first sample Delay time. This is a panel mounted pot
ADC2- TX pulse width
ADC3- Sample time
ADC4- second sample delay time
The ADC values are scaled and then added to the minimum time for TRM2.
TMR2 is setup with these 'values' for each of the times in the sequence.
This version of code is simply: TX pulse, delay1, sample1, delay2, sample2.
There is optional code here that I put in almost all PIC projects. These are an interrupt driven TX & RX UARTs with FIFO (ring buffers). This is handy for 'real time' debugging and was used to output ADC reading and the calculated times. There are a couple of "#define " (debug1 & serial) statements used to enable/disable building code.
There is the possibility of output info on the UART to an LCD display on future versions.
As Carl and others have stated- there are lots of possibilities for experimentation when using a processor. I setup trim pots to easily adjust all the timing so could try nd measure what works.
Currently with the 12" DD coil the timings are:
100us TX (with a 10 Ohm series R to coil), 8-15us delay1, 15us samples & 34us delay2.
I may just hard code the TX pulse, Sample time and delay2 time in the future to open up ADCs for other uses.



Schematic and the PIC assembler code (unzip and change extension from .txt to .asm)

PIC- 16F88 sch.pdf
16F88_detector.zip

[waltr]

Building and adjusting
Most of the Build was done in steps and tested as per Carl's instructions in the original HH article which is simply just good building practice.

I used the ICL7660 (10V part) at first but put a 3.3V Zener diode in series to the battery. This dropped the Voltage to the 7660 to 9.6V which it can handle. Output was good so got the +5V above the ground (VB+). I left the resistor and cap un-installed for initial testing even though the 7660 noise may be an issue. Once the other circuits were tested came back and install the RC to feed the 7660's clock from the TX pulse. This did not want to work all the time. Sometimes the +5V would just drop-out. I did get an order out to DigiKey and included some ICL7660A's which can take 12V. Swapped to the 7660A and the TX pulse clocked the 7660A just fine with good +5V output.
I first adjusted the balance pot to match the integrators without the NPN JFET drivers as per Carl's instructions. This works but the HH 1 or 2 doesn't easily allow to adjust later without un-soldering parts. With a PIC on a separate PCB, not having the PIC connections allows jumpering the sample switch NPN base's to ground turning ON the JFETs to re-adjust integrator balance.

I noticed TX pulsing in the audio and could see these bumps at the TX pps rate after the Integrator and SAT stages. Further digging with the scope I found that the two gated samples both were positive on the Integrator output. This was being caused by the Pre-amp output offset. I re- adjusted the Pre-amp offset while watching the integrator out on the scope. Adjust for minimum steps when the samples are taken. Now the TX pulses are not heard in the audio.

I have read in the forums here about NOT mounting the Headphone jack on a 'grounded' (VB+) enclose but forgot about that until I was assembling everything into the aluminum box. The headphone 'ring' is connected to VB- so this will cause a short across the battery. I would have re-designed the transistor output drives to reference 'ground (VB+) if I remember this earlier. After thinking about this for a bit I dig out a old miniature audio transformer, 1k CT input, 8 Ohm output. A quick test with jumper leads showed this worked just fine and isolated the head jack from the battery. The x-former got mounted with some double sided foam tape into a corner of the box. Big problem solved.

[waltr]

Coils- TX pulsing and RX recovery
The very first coil I tried when prototyping was just a bundle of 30ga wire-wrap wire of 32 turns about 7 inch diameter. I hung this coil above the work bench to keep it away from metal. This rough coil measured 400uH, 6.7 Ohm and 67pF. It actually worked quite well. Also tried a few other rough coils of wire and all worked ok showing that the coil is not critical for basic PI detector operation.

There has been lots of discussion about not using thin wire in PI coils- all the reasons are to get higher coil current. However, a low resistance raises the coils Tau ( Tau = L/R) which then requires longer TX pulses, higher duty cycles and less battery life. Does all this really help?

I did some experiments and measured the Coil current (note the 1 Ohm resistor between the TX OFFSET and VB- in the schematic). The voltage across this 1 Ohm resistor was measured with the scope (no 1000uF cap). with the original loose coil (24t, WW, 7in- 400uH, 6.7 Ohm) required a pulse of 150us and reached a peak current of 1.5 A. Tau = 400uH/ (6.7 + 1 + 0.55) = 48us, 3Tau = 145us. With a 60us pulse the current reached 0.6A but was still raising at shut-off. I then add a 12 Ohm resistor in series with the Coil/MOSFET. Tau = 400uH/(6.7 + 1 + .55 +12) = 19us, 3Tau = 59us.

Sensitivity to my gold wedding ring and other test targets was actually better with lower peak current. Therefore, using a larger diameter wire for the coil is not needed. The added resistance and shorter Tau also reduces the Average current and gives a longer battery life. I used a 1W, 10 Ohm resistor in the final circuit. The current setup draws and average current of 180mA.

Since the first coils were loosely wound coils the inter-winding capacitance is quite low. Next coil was wound at 10 inch diameter with 24 turns of #28 enamel wire wrapped with 1/16 inch ID PE spiral wrap (see pictures in post #3). This coil required a lower value damping resistor therefore the capacitance is higher as expected. Haven't measured the L, R of this coil but coil calculator says 474uH, 4 Ohm and guessing C = 150pF from the damping R value.

The Coil delay is longer and also so is the Pre-amp recovery. With the loose coil sampling could be done at 8us delay but now needs 15us. No real surprise there and this would be perfectly fine for Beach and general coil hunting. Still getting good response to my gold ring (thick men's ring).
So try another coil. This one is same diameter of 10 inch but 21 turns 30ga WW with 1/16" ID spiral wrap. Calculated L = 341uH but this needed a slightly lower value damping R so C must be slightly higher. Testing showed this coil is very close to the previous coil.

Tinker's threads were very interesting reading and in particular the use of IB coils. So I took the two 10 inch coils and bent them into 'D' shapes. Pulled the jumper to separate the TX from the Pre-amp. Connect and check the RX coil then added a resistor to Dampen the RX coil.
With the scope on the Pre-amp out then slide the RX coil over the TX coil to find the Null. As for the detector response and sensitivity the Coils Null is not critical. Best thing is that the Pre-amp now recovers in under 8us.
See photos attached. Scope is 20usec/div.

What about mineralized ground.
Did lots of reading of GEB, etc and found suggestions of using a Red Brick, Ferrite and permanent magnet to test Ground response. see photo.
With the Mono coil all of these test object caused a strong response. With the DD coil there is No response to Ferrite, a very, very slight response to the brick if within an inch of the coil but still responses to the magnet. Second good reason for DD coil on a PI.

This does sound promising and can't wait to get this detector outdoors for Real testing.
The third coil I will be trying is an 8 inch concentric, 4 inch RX & buck.
I also will be trying a method of GEB that has been discussed where the processor adjusts the second sample width and adds a third sample to null ground effects.
Last is a couple of photos of the circuit and enclose. Note enclosure is built from junk I had laying around. The one Pot needs it's shaft shortened but works. The control from Left to Right: GEB (not installed yet), Gain, Delay, Threshold. Then the Audio volume and headphone jack.

[waltr]

It is Together
Finally got the detector put together. A little bit crude but seems to work.
Old broom handle for the pole.

This is the DD coil from photo in above post. The "coil housing" is foam board, corrugated card board and mat board held together with nylon screws/nuts.
mount is 1/4 plywood and a plastic screw/nut used for toilet seats (yes, this is a new one).
I did this so easy to disassemble if Shielding or other mods are needed. Once I have coil configuration working properly then I'll go with a "real" coil housing.
The coils are not shielded and uses the coil wires inside 1/16 spiral wrap to the box. This means Minimal Capacitance and this can sample with 8 usec delay.

It does detect all the nails in the wood floors of my house.

Took it outside for a few minutes- it is currently 6°F (-14°C) and windy with 3 inches of snow so didn't stay long.
There was no response when lowering and raising (pumping) the coil to the ground/snow so electro-static/ground C seems ok.

Sweeping next to my walk way did detect something but not attempting to dig since the ground a frozen like rock.

Next is to clean-up the construction, shorten pot shafts, apply adjustment markings, etc.
Then start building another PCB. My wife does want to go detecting on the Beach so a water-proof version will be coming.

[6666]

Took it outside for a few minutes- it is currently 6°F (-14°C) and windy with 3 inches of snow so didn't stay long.

good work, it was only plus 42' C here yesterday

[waltr]

good work, it was only plus 42' C here yesterday

Thanks.
I'll think of that temperature just to stay warm.

[Tim]

Great stuff waltr. I am yet to start the HH I RevD, open my new scope or work out my pickit. But I shall return here later. DD, GB and 8us or less should make a good Oz gold detector.
A very good and comprehensive description thank you.
This is my go-to design for my detectors. Box from KitstopAU, brackets and armrest made from PVC sewer pipe bent with heatgun. Upper shaft 700mmx 25mm Ali from the hardware. Lower shaft (900mm) and handle laminated bamboo (leaf rake handle). Very strong. Very light. Very cheap. 10x5 coil made from pvc downpipe formed in wooden former in oven@100C . Clamped and glued with white neutral cure silicone. 12v1w solar panel, 3x18650s with internal charge regulator.
I hope this helps.

47C here yesterday.

[waltr]

Great stuff waltr. I am yet to start the HH I RevD, open my new scope or work out my pickit. But I shall return here later. DD, GB and 8us or less should make a good Oz gold detector.
A very good and comprehensive description thank you.
This is my go-to design for my detectors. Box from KitstopAU, brackets and armrest made from PVC sewer pipe bent with heatgun. Upper shaft 700mmx 25mm Ali from the hardware. Lower shaft (900mm) and handle laminated bamboo (leaf rake handle). Very strong. Very light. Very cheap. 10x5 coil made from pvc downpipe formed in wooden former in oven@100C . Clamped and glued with white neutral cure silicone. 12v1w solar panel, 3x18650s with internal charge regulator.
I hope this helps.

47C here yesterday.

Nice build. I'd like to see more details on how you made the coil housing. Maybe start a thread on this in the Coils forum.

So far I found the DD coil rejects Ferrite and 'brick' but not a magnet so not sure how it will do with highly minimized soil like you have out in the gold fields.

I will be experimenting with Software to do GEB and will post here results and code.

[SVEN1]

Building and adjusting

I have read in the forums here about NOT mounting the Headphone jack on a 'grounded' (VB+) enclose but forgot about that until I was assembling everything into the aluminum box. The headphone 'ring' is connected to VB- so this will cause a short across the battery. I would have re-designed the transistor output drives to reference 'ground (VB+) if I remember this earlier. After thinking about this for a bit I dig out a old miniature audio transformer, 1k CT input, 8 Ohm output. A quick test with jumper leads showed this worked just fine and isolated the head jack from the battery. The x-former got mounted with some double sided foam tape into a corner of the box. Big problem solved.

Its very hard these days to find the standard 1/4" switching stereo jacks like these with the threaded portion being plastic that will not ground to the metal enclosures.



So I now use these plastic threaded types of headphone jacks as they do not ground out on the metal enclosures https://www.ebay.ca/sch/sis.html?_nkw=20PCS%2C1%2F4%22+6.35mm+STEREO+PCB+P ANEL+MOUNT+HEADPHONE+JACK+4+pin+socket&_id=1214211 71150&& _trksid=p2057872.m2749.l2658

[waltr]

I need to get some insulated phone jacks.
Those look pretty good.

[waltr]

New PIC code
A few changes to the PIC code include:
16F88_detector-GB.zip

Added a variable delay to the PPS time to avoid interference. Used ADC 2 and made the TX pulse length fixed at 100usec. I was having low frequency wobbles at the SAT output at times. Did notice that when I changes the PPS slightly it could be better or worse. Since the original HH and the TDI have PPS adjust I added this to the code. Then setup and went through the adjustment range (in my basement workshop so lots of AC mains @ 60Hz). As I adjusted the Wobbles were much worse but would nearly go away at some points in the adjustment. So this is a worthwhile code addition. If it is only the AC mains 60Hz interfering then I'll just leave this adjustment as a trim pot inside the enclosure.
The current PPS is 1503 Hz with a minimum 665usec period. The Adjust can add up to 12usec for a minimum PPS rate of 1490Hz.

Ground Balance
Added GEB adjustment, ADC4, then applied the techniques in these threads:
http://www.geotech1.com/forums/showt...-Balance/page3
http://www.geotech1.com/forums/showt...Ground-Balance
I use the Three Sample Method with variable Sample times instead of Variable GB Gain. This can be done on any PI detector with a single differential integrator and a processor for timing. The sequence is:

EFE sample : delay3 : TX pulse : delay1 : Target sample : delay2 : GEB sample

TX pulse fixed at 100usec. Coil Tau = 400uH/ (4 + 10 + 1) = 26usec. 4Tau = 108usec. Coil peak = 0.6Amp.
D1: Delay1 is Pot and variable from 6usec to 35usec.
S1: Target sample set at 15usec (Trim pot adjustable)
D2: Delay2 set at 20usec (Trim pot adjustable)
S-GEB: GEB sample time is adjustable from Target sample time (15us min) to 226usec. This sample time is Target + GEB.
S-EFE: EFE sample time is GEB sample time - Target sample time (0 to 204usec)
D3:delay3 is fixed at 12usec

When GEB control is Zero then the EFE sample time is zero and the GEB sample time = the target sample time. This is the same timing as the original HH sample sequence and allows to turn OFF GEB.
Did some experiments using a 25cm Mono coil and a 33cm DD coil on various targets listed below.
With the GEB time = Zero, all targets respond. As the GEB time is increased then some targets drop into the "GB Hole" and not respond and then respond with a low going signal instead of a high. This change to Low going has been described when GEB is used (White's TDI) so seems like this 3-sample variable time method will work. Big test is when I find some mineralized ground.

Experimented with Delay and sample times as per the second GeoTech link above. This has statements of the timing relationships: S1/D1 = S-GEB /(S1 + D1 + D2) Eq-1
One thing this says is that if D2 (delay between target sample and GEB sample) is longer, then the GEB sample time must be longer. I verify that this is true by adjusting the GEB sample time to find a target's pivot point. Longer D2 delays required longer GEB sample times. This can cause very long GEB & EFE samples which can force slower PPS rates. By keeping the D2 time short (some say that this delay can or should be zero) then the GEB & EFE sample times allow faster PPS rates. Faster PPS rates allow for fast sweeping of the coil across the ground with the same integrator TC.
This showed up in testing when finding the GEB sample time that caused a 'pivot point' on targets. I settled on a D1 time of 20usec which puts the S-GEB time at 67.5usec to satisfy Eq-1.
The pivot S-GEB times for various targets are:
25cm Mono 33cm DD
silver Dime: 23 20
silver 1/2 dollar: 29 18
copper penny: 25 24
Clad quarter: 26 21
sm piece of alum foil: 25 26
1954 Nickel: 170 100
Gold ring: >220 >220
Steel eye screw: 26 negative with no GEB added
Brass key: 45 40
lg alum foil: 122 >220


Therefore, all targets that are close to what the S-GEB is set to will NOT respond and targets that are less then what the S-GEB is set to will produce a Low going SAT output. If S-GEB is set to 67.5us then all of the above targets except Nickel, gold ring and lg alum foil will respond with a Low tone.
This does require an Audio circuit that can produce a Low or High tone like the TDI. I have prototyped a circuit with a full-wave rectified and 4046 VCO chip that does this (schematics posted in the forums here). Am working on replacing the three op-amps and VCO chip with an 8-pin PIC (I'll post details once have this working).

Interesting is that the Pivot points are different for the DD coil verse the Mono coil. The silver and clad coins are about the same but the Steel eye screw is very different.
Played with the different targets again and watched the Pre-amp output on the scope. With a Mono coil all targets increase the Decay time (inverted pre-amp output lowers). However, with the DD coil most targets decrease the delay at the output when at the non-over lapped coils and increase where the coils over lap. Also, during TX ON the different targets cause a very different shape. Tinker showed this in his threads and proposed sampling during TX on as a possible way to obtain discrimination info. Planning on trying this. The steel eye screw produced a response reversed from the other targets.
The a DD coil seems to add more information if there is a High/Low audio tone circuit than a mono coil. An IB concentric coil should have a similar response.

DD coil Pre-amp output scope pictures:
20usec/div.
Top trace is Pre-amp out & 2V/div, shows the full TX cycle and coil ring/decay.
Bottom trace is Integrator output. Note the Voltage level without and with targets.

No target - base line for comparing
Gold ring -
half Dollar -
Eye screw -
Screw driver - Note the Integrator out is nearly saturated and the 1st sample pulse shows.
Screw driver plus Gold ring with GEB added - The GEB lifted the integrator output Voltage off the bottom rail and allows detection of the gold ring.






Decay Voltage during S1 is higher than no target & the zero crossing time during TX on is sooner with the two Steel targets.

Coils
Measured the 25cm mono coil. It is 21 turns, #28 enamel wire, PE Spiral wrap with #28 twisted leads to circuit. 366uH, 3.4 Ohm, 56pF. Sampling can start at 8usec but finding 10-12usec is better.
I am still running Unshielded coils and cables and have not seen noise issues except the Wobbles described in the beginning of this post which is fixed with a PPS rate adjustment. I did try shielding to fix the Wobbles but the shield did nothing. None of the unshielded coil respond to my hand or moving closer to the ground.
This has made me wonder if I have something wrong or is it all the other PI detector circuits/builds have something wrong. Since I getting close to putting coils into shells with epoxy this is making me nervous.
Anyone have thoughts or experience with this?

Also received some coil shells from Don Bowers today.
https://sites.google.com/site/dbcoilshells/home
He lives only 113 miles (182 km) from me. Shells in picture are for: 25cm mono, 25cm DD, 21cm Concentric.

[Tim]

Great stuff waltr. Very comprehensive, and all too technical for me (yet). Small natural gold and heavy mineralisation will be my main concerns for fiddling with the HH. Yes, that damned mains electricity is a problem. My last residence had only minimal interference in close proximity to the fuse box, but where I am now (single story unit block) all the wires run through my roof, and there is a distributor across the road.
Awaiting my Crazy-tech coil molds for the IDXpro and IGSL Musky, but still pre-occupied with a simple BFO and an IB....
Keep up the good work.