How can the temperature drift in this unusual absorption detector be reduced?

I am using this simple circuit in a sensitive hand-held probe to detect metals in the ground. It works very well except for the temperature drift.

The drift means having to often re-adjust P1+P2, which are external multiturn pots. This is very pronounced when clouds are moving overhead and the sun's warmth on the probe is fluctuating.

I measured the temperature drift by the necessity to change resistance of P1+P2 to reach the equilibrium again after temperature change. This point is where the circuit has the greatest detecting range.

I measured around +480Ω per degree Celsius thru 0 to +50 degrees. It looks linear to me. Base value of P1(1k fine tune) +P2(100k rough tune) at 20 degrees C is approx. 80kΩ.

My first idea was to implement a PTC resistor (Tempco) in series with P1+P2 to compensate for this. But I just couldn't find any with such high value changes. I could put lots of them in series, but together they would reach over the base value of 80kΩ. So no solution here.

From what I have read it is a Hartley oscillator in conjunction with a relaxation oscillator formed by the CD4024 (Ripple-Carry Binary Counter) IC.

The Hartley oscillator windings are 330uH & 1600uH on a ferrite rod from an am-radio. The main causes of temperature drift as far as I understand are the transistor and coil.

Are there other ways of using PTC's or NTC's here to reduce the drift? Is there perhaps a better biasing scheme that could be implemented here?

The circuit is from the following forum: http://www.geotech1.com/forums/showthread.php?21373-Can-pinpointer-detecting-thin-chain&p=224345#post224345

Yes, I did ask Nupi if it's OK to post this in an open forum. It is.

P1+P2 are multiturn pots and are found on the left side of the schematic in the middle.

Coils are wound on the same ferrite rod, the coupling factor k is close to 1.

I assume you mean why Q5 (pin 5) of the 4024 is connected to clock. I'm not sure. Someone mentioned it may be a type of regenerative feedback on the forum I have this from. It's quite arbitrary I think. I've played around with the different outputs coupled with clock, also with different capacitors. The output tone will change and thermal stability can be slightly enhanced/weakened when doing this.

Do you mean 'no gold' as an extra indication literally (that would be ground breaking) or do you just mean 'no metal'? This circuit does actually have a background whine (high pitch as you call it) which I find annoying.

Yes, the power absorption raises the current. I only have to adjust the pots when temperature has changed.

I will describe how this detector works out in the field, as I think you may have a different picture of what goes on:

Adjust P1+P2 to the most sensitive point, just where the beeping stops.
When metal gets close you get a slow beep-beep.
The closer the metal gets the faster it will beep.
The same happens when the temperature goes up (not as fast though).

Here's a more technical quote about this probe I found on the forum I got the schematic from:

'Tr1 forms a Hartley oscillator and it is biassed so that it operates in linear regime that is prone to amplitude changes. When a metal object is introduced to the coil, the amplitude of oscillation is reduced because of the eddy current losses, but more importantly, the transistor power consumption is enlarged, and thus the emitter voltage gets shifted. This arrangement is not new, and most pinpointers work at this or similar principle. The change in the oscillator working conditions is translated into some sort of indication.

What is new and very good here is the way this working point gets detected. It is the role of 4024 that forms a relaxation oscillator with narrow strip of input signal levels that brings it from non-operation to saturation, and in a process it provides a very sensitive indication. It operates on the oscillator's envelope rather than the DC level, pulse durations are set by pulse counting method, and in a process it provides pulsed audio.'

• L's are NTC and C's with PTC are possible for frequency stability , but for gain control, use R2 acros Vcb instead for negative feedback with less gain than open loop gain. (ie. bypass Re) – Sunnyskyguy EE75 Apr 10 '17 at 23:10
• is this running at 30Hz? why? – Sunnyskyguy EE75 Apr 11 '17 at 1:12
• R1? R2? Drift? What drift are you talking about? – Andy aka Apr 11 '17 at 8:33
• Get a thermal control circuit, insulate it and temperature regulate it – Voltage Spike Apr 11 '17 at 15:22
• @ Tony Stewart: I will try this and post what happens. – Curious Kitty Apr 12 '17 at 0:35