Design of an Hartley oscillator

[ppTRN]

Hi everyone,
I have a metal detector coil that I need to oscillate at a frequency of around 12kHz. This coil is center plugged with an inductance of 125uH per side.
Previous oscillators that I used are Colpitts oscillator, thus they do not use the coil central plug. I wish to use it because it is supposed to screen the coil if connected to ground. Also, the oscillator in image A does not work with such low inductance value. Or at least, I think. It works perfectly with coils of 5mH or so, but when I try this new one only starts to oscillate for frequency too high for my needs (>35kHz). Connecting a capacitor larger than 50nF stops the oscillations.

I think that it could work if properly modified, but I have no idea where to start.

I found many schematics on the internet but there are really no explenation on how to choose the working point of the amplifier transistor, and I do not have many experience on that.

The sine wave does not need to be extra pure, I just need it to have a peak voltage equal, or almos equal, to the supply voltage.

Can you guys point me to the right direction or provide some schematics?

Thank you!

photos attached:

A: oscillator I have used so far for coils with L > 4mH
B: anlother oscillator that worked perfectly with L > 4mH
C: inspo from "The Art of Electronics", page 440. It could be perfect for my goal. Unfortunately, no design reference on the book.
D: the most common configuration I found online. The choke on the collector could be replaced by a resistor given the low frequency required. Still no guidelines on how to design it.

[jwet]

The problem you have is that with 250 uH of L, the cap required to resonate at 12 KHz is less than 1 pF.  The coil itself likely has this much winding C- you can get there from here.  You need to get to a reasonable sized C values of say 33 pF so you're not just in the parasitics.  It easy to pick up a few pF with just wiring, etc.  This sets a min value of L at about 5 mH.  You can do the math or use an online calculator.  Your L is just too little.  If you get values less that a few pF, its not really "realizable".

Hope this helps.

>>>> Sorry I botched this- this would be for 12M. I think others have you on the right track.  Sorry for the distraction.

[moffy]

At 12kHz, 250uH is just 25 reactive ohms of impedance, to drive that you will need a unity gain current buffer of some sort. My calc shows that for 250uH you need about 680nF of capacitance to get 12kHz.

[PCB.Wiz]

I found many schematics on the internet but there are really no explenation on how to choose the working point of the amplifier transistor, and I do not have many experience on that.
The sine wave does not need to be extra pure, I just need it to have a peak voltage equal, or almos equal, to the supply voltage.

The emitter coupled oscillator is simple and forgiving of skewed or lossy LC circuits.

It can be NPN or PNP, this example is PNP allowing a grounded coil. A single resistor sets the injection current.

https://commons.wikimedia.org/wiki/File:Emitter_coupled_osc.svg

If the DC resistance of the inductor is significant, you may need to add a R//C to the base of T2 so the collector currents nominally balance.

At 12kHz, 250uH is just 25 reactive ohms of impedance, to drive that you will need a unity gain current buffer of some sort. My calc shows that for 250uH you need about 680nF of capacitance to get 12kHz.

Yes,  680n is close to 12kHz with 250uH.
With collector current balancing, a cross coupled oscillator starts ok with even 12 ohms of series loss resistance in the inductor.
Note however, at those high losses, the Q is lousy, and the oscillator frequency is less Fc aligned.

[ppTRN]

The problem you have is that with 250 uH of L, the cap required to resonate at 12 KHz is less than 1 pF.

I do not know how you came up with that value, but the actual value is far from that. I think you are using the wrong formula.

At 12kHz, 250uH is just 25 reactive ohms of impedance, to drive that you will need a unity gain current buffer of some sort. My calc shows that for 250uH you need about 680nF of capacitance to get 12kHz.

But 125uH is just half of the inductor, shouldn't I consider the whole inductance?

The emitter coupled oscillator is simple and forgiving of skewed or lossy LC circuits.

It can be NPN or PNP, this example is PNP allowing a grounded coil. A single resistor sets the injection current.

https://commons.wikimedia.org/wiki/File:Emitter_coupled_osc.svg

If the DC resistance of the inductor is significant, you may need to add a R//C to the base of T2 so the collector currents nominally balance.

Never used (or heard of to be honest) of such oscillator. I may try but still it does not uses the center plug of the inductor.

[moffy]

At 12kHz, 250uH is just 25 reactive ohms of impedance, to drive that you will need a unity gain current buffer of some sort. My calc shows that for 250uH you need about 680nF of capacitance to get 12kHz.

But 125uH is just half of the inductor, shouldn't I consider the whole inductance?

It's just to point out that because of its low impedance some current buffering is needed, the actual impedance seen will be dependent on the total reactance and how close to resonance it is.