Oscillator circuit design

[Chryseus]

Hey guys, I've been trying to build various LC oscillators (Armstrong, hartley, etc) with very little luck, I can maybe get 5-10 cycles out but no sustained oscillation, I could use some suggestions for reading material and the best way to wind inductors. 
Specifically what are the design differences from a BJT oscillator and one using a JFET, I do have some JFETs but I'd like to experiment with both.

[kg4arn]

What frequency range?

[PA4TIM]

How do you build them , dead bug or  breadboard ? (For RF (more as 500kHz or so) you can forget the breadboard)
For +200 you can forget veroboard
for > 1 GHz you best use microstrip

[Chryseus]

In the 1-30MHz range, I've tried both breadboard and copper clad board, the problem is I can't get any of my designs to oscillate for more than 20 cycles, I'll post some of my designs in the morning, I'm probably doing it wrong.

[Chryseus]

Here are the two designs I've tried so far:

Hartley:


Armstrong:


No matter what I try they do not oscillate.
I'm using an inductor with about 100 turns for L2 and 15 turns for L3 on a ferrite core.

[Andy Watson]

Chase the losses.
1. Have you decoupled the power supply rails close to the oscillator?
2. Are the phases of L2 and L3 correct?
3. You want L2 and C1 to dominate the circuit. C2 is probably putting too higher load on them (via the base-emiiter resistance of Q1).
4. Ditto the Armstrong circuit, except R1 and R2 are also a substantial load.
5. I would consider bypassing R4 with a capacitor - or a capacitor and resistor if you really want sinewaves.

I can maybe get 5-10 cycles out but no sustained oscillation

So at some point the biasing is correct as C2 charges up, but then it re-enters a state of low gain. You only need enough gain to replace the losses, but with BJTs it's not always easy to see where the losses are.

[PA4TIM]

Use seperate inductors and not a transformer.

[ElectroIrradiator]

You probably want the L/C ratio at resonance to be at least 10 times larger than it is, and probably higher still. 100nF + 10uH is a pretty nasty combination to use for 150 KHz. The transistor cannot couple to the very low impedances you have created, making it impossible for it to amplify the RF signal, thus no oscillation.

To give an example: For the oscillator in an AM receiver, the inductor is typically 100uH for a frequency around 1 MHz (off the top of my head, not exact numbers). At 1 MHz and L = 100 uH, the reactance will be roughly 630 ohm. You need a capacitor of about 250 pF (not nF) to make this resonate at 1 MHz.

For 150 KHz you want an inductor of about 670 uH plus a capacitor of about 1.7 nF. So your current values are a bit far off, to be similar to the oscillator in an AM receiver.

[vk6zgo]

Use seperate inductors and not a transformer.

An Armstrong Oscillator relies on magnetic coupling between the collector & base circuits to provide feedback,so a transformer is necessary,or at least,some coupling between the coils/

OP:-Starting with the Armstrong circuit, try making L3 the same value as L2,& put C3 with the same value as C1 in parallel with it.(separate inductors,not a transformer

This will create a BJT version of the old "Tuned plate,Tuned grid" (TPTG) oscillator,which oscillates if the "grid" (base) circuit and the "plate" (collector) tuned circuits are tuned slightly differently.

[PA4TIM]

Sorry, I was only talking about the Hartley, my fault. I do not think a transformer used this way will work there but I have never tried.
I do not know the Amstrong oscillator. Looks interesting and I agree about the values you give.

[Andy Watson]

Sorry, I was only talking about the Hartley, my fault. I do not think a transformer used this way will work there

It does work, but you have to consider the mutual inductances as well to find the frequency of oscillation. Think about it as one inductor with a tap.

[PA4TIM]

Yes, mutual inductance was my consideration. I have done some measurements on that subject last week and after that done some calculations to see if measurent and theory were in line (and they did, pffff  ) 

I was also thinking if it matters how you connect the two windings, thus if the secundair is in or out phase, but I do not know that for sure.
I have build a lot of oscillators (from 1 kHz to 4 GHz) but never used a transformer.
Interesting subject.

[Chryseus]

Thanks guys this is going to keep me busy experimenting, I'm going to pick up a LC meter so I can build with some level of accuracy, as for the transformer that's just a symbol in NI Multisim to indicate inductive coupling, the inductors are wound on a ferrite rod and I have tried reversing phase.
I'll give the TPTG oscillator a spin tomorrow. 

[Chryseus]

Finally got it working using a TPTG oscillator, turns out using a resistor instead of a RF choke is a very very bad idea. 



Base tank waveform:


Collector waveform looks awful as I expected:


I'm rather pleased with how well it works despite the simplistic design, I'll probably try do the same or a similar circuit using a JFET and a triode and compare them, my money is on the JFET. 

[vk6zgo]

TPTG circuits really like to oscillate,as the first people trying to design Tuned Radio Frequency amplifiers found out!

Triode vacuum tubes have quite a lot of grid to anode capacitance even in a non operating condition,& to make things worse,when they are operating,this capacitance is effectively multiplied by the device gain (Google Miller Effect).

In this respect,BJTs are "triodes" too!