Frequency climbing in Colpitts oscillator?

[frenky]

I wanted to make a simple metal detector with this scheme:


I made a coil with inductance approx 160uH, paired with 11nF should give 120kHz frequency.
I get 125kHz which is close enough.

Problem is that frequency keeps on rising.
It starts to rise with 2Hz per second and after minute slows down to 1Hz per second. But it "newer" stops (at least in 20min it doesen't).

What could be the problem?
I have made another circuit on the breadboard and it behaves exactly the same?

Please help. 

[BrianHG]

Measure your 5v with at least a 4.5 digit DVM and verify the osculation peak voltage on pin 5 with a 4.5 digit DVM as well over the 20 minute time.

[Twistedsnail]

Are your 22nF caps supposed to be between C and E, and E and GND? Like this example:

[amspire]

What sort of capacitors are you using?

Many types of ceramic capacitor are not stable. You need either NPO/COG ceramic or something like polyester capacitors.

[Twistedsnail]

Are your 22nF caps supposed to be between C and E, and E and GND? Like this example:

Actually now that I think about it, both have the crucial feedback capacitor from the emitter resistor.
Can someone explain the difference between having the second capacitor couple to V+ vs GND?

[frenky]

I am using this 22nF capacitors from a local electronic components store:


MULTILAYER
Brand: Millenium
Mac Valtage: 50V
Izolator   X7R
Tolerance   ±10%
Type   THT

Some aditional info...
I tried the circuit without arduino, so that I can rule that out.
And tried with different voltage sources: 5V from arduino voltage regulator, 9V battery 20V battery...

[amspire]

XR7's are not the most stable of capacitors. You are getting a drift of just under 2 percent per second down to one percent after a while.

The XR7 can have a temperature coefficient of over 2%/C, so it only takes a 1 degC change to cause the drift. You would be better with a  more stable capacitor, but make sure the capacitors are nowhere near the transistor since the transistor may be warming the capacitors.

[frenky]

Thank you! 

I'll try:
- replace caps with different type
- move it away from transistor
- thermally insulate capacitors

[frenky]

So I replaced all caps with this one:

https://www.digikey.co.uk/product-detail/en/epcos-tdk/B37987F1223K058/B37987F1223K058-ND/1596648

After a few minutes frequncy becomes much more stable. Now it climbs at a rate 1Hz every 30sec which is A LOT better. And the climb slows down with time significantly.

Tnx to all.

[tecman]

Those are X7R.

You really should put NPO caps in for stability

paul

[frenky]

I will try to get some NP0 caps, though it seems that that type mostly exist in pF range.

[amspire]

Film capacitors (polyester or polystyrene) are readily available and are much better then the XR7 ceramic capacitors. The film capacitor temperature coefficient is smaller then XR7 by at least an order of magnitude.

[frenky]

Tnx.
It seems that I managed to find all the wrong capacitors for this project. 

[frenky]

So have found some 47nF NP0 localy. The frequency went down to 85kHz but it remains withing 1Hz for more than 10min.
I could't be happier. 

Tnx to all.