Well it has taken a week but its done. I tried good advice from a Guru but it did not help so then I got stuck in. I designed all kinds of circuits to fix it including a timer driving a small ww resistor placed near the oscillator (well why not..
), but nothing really worked. But then, it all became clear. This circuit is very much the same as you can find all over the internet, and it works, well sort of. The real problem for me was three issues in the end.
As designed originally it runs at about 200KHz and indeed, with careful adjustment it gives an acceptable range. the clue is 'careful adjustment'. Note the tuning cap(s) is 1-3nf. The circuit itself has a very high Q but we are not too worried about freq, just amplitude, but as this thing goes a little off tune, the amp changes too. So for me, C1 and C2 must be X7R types to cope with temp change, and I also made C3 the same type. Better, but no prize
.
So then I changed C1/2/3 to 10nf and deleted C4. Now the freq is 167KHz. I had to make a small change in the 'sense' R chain. R6 is now 390R - the fact I used 1% metal oxide is probably irrelevent, it was first out of the box. Now, a small temp drift is much easier to cope with because the Q has dropped a lot so a poss drift of +-10% over 20nf does not cause much loss of amplitude.
Finally, I used a 10 turn 100R pot. With this I could get an incredibly fine adjustment - and this was very bad because, as I realised, turning on and off again, that adjustment to an ohm or two is bound to be wrong. I replaced it with a 250R standard pot. Sorted
. Nice hot weather (over here, up to the 90s
pinpointer.pdf.
Left it in the garden, left it in the garage overnight. A very minor touch (I mean the smallest movement) was all I needed to do. Also, the range is really good; I have picked up 10p at 4", and I can detect an American penny at about 1", a 5c coin at 2".
My work here is done. Hi ho Silver, away....
Forgot to mention, lots of noise on Vcc rail so used 78L05 and good decoupling. Less then 47uf did not kill the noise from the osc.

As designed originally it runs at about 200KHz and indeed, with careful adjustment it gives an acceptable range. the clue is 'careful adjustment'. Note the tuning cap(s) is 1-3nf. The circuit itself has a very high Q but we are not too worried about freq, just amplitude, but as this thing goes a little off tune, the amp changes too. So for me, C1 and C2 must be X7R types to cope with temp change, and I also made C3 the same type. Better, but no prize

So then I changed C1/2/3 to 10nf and deleted C4. Now the freq is 167KHz. I had to make a small change in the 'sense' R chain. R6 is now 390R - the fact I used 1% metal oxide is probably irrelevent, it was first out of the box. Now, a small temp drift is much easier to cope with because the Q has dropped a lot so a poss drift of +-10% over 20nf does not cause much loss of amplitude.
Finally, I used a 10 turn 100R pot. With this I could get an incredibly fine adjustment - and this was very bad because, as I realised, turning on and off again, that adjustment to an ohm or two is bound to be wrong. I replaced it with a 250R standard pot. Sorted


Left it in the garden, left it in the garage overnight. A very minor touch (I mean the smallest movement) was all I needed to do. Also, the range is really good; I have picked up 10p at 4", and I can detect an American penny at about 1", a 5c coin at 2".
My work here is done. Hi ho Silver, away....
Forgot to mention, lots of noise on Vcc rail so used 78L05 and good decoupling. Less then 47uf did not kill the noise from the osc.