Lumped-element oscillator

[Agronski]

lumped elements

I'm new to this world (microwaves) so please be gentle

I've worked out a basic design for an oscillator:

-fc = 2.4GHz
-Colpitts, single BJT transistor (BFP520W, probably)
-Common-base, Icq ≈ 40mA
-5-10mW output power (more would be nice)

...now a couple of questions:

Is 10mW optimistic for a single transistor at this frequency?

All the capacitors and inductor values are available as surface-mount components, so can I build the entire oscillator on PCB using only those lumped elements, or will I have to use a microstip-match at the output?

If lumped elements are fine, what precautions should I take when laying out the tracks? (Or are there automatic routers that will do it for me?)


Thanks in advance,
_______
Agronski.

 

[vfone]

+oscillator+single transistor-crystal-patent

**broken link removed**

 

[Agronski]

simple 2.4 ghz oszillator

Thanks for that - it's a helpful document, but not completely what I need. I've only just started using ADS, so simulating circuits is still a bit tricky. My PSPICE model of the oscillator runs fine, as long as I use a specific transistor - most of the other models for NXP BJTs don't seem to want to oscillate at 2.4GHz

What I really need to know is what's the best way to lay out a PCB? Can I just assemble it with large ground-planes and a 50-ohm transmission line output? If it's all surface-mount, I think I could get the whole thing down to about 15mm squared...

_______
Agronski.

 

[vfone]

Are few rules to follow for PCB design of the oscillator at this frequency, but the main rule is to use RF traces shorter as possible. Also use decent distance between output components and LC tank circuit, don’t place inductors close together or place them perpendicular to one another, don’t use too many vias to minimize breaks in the ground plane, but if use vias place few in parallel to minimize inductance, control and bias lines could be longer but routed away from RF lines, use high-Q components in LC tank to improve phase noise, etc.

 

[Agronski]

Thanks for the great tips! I'm only using surface-mount stuff, so short tracks shouldn't be a problem.

One final thing - a bit embarrassing to ask, because it seems so simple....but I've seen a lot of circuitboards with solder blobs all over the ground-track traces on the comonent side - are these actually connected (with pins, or wires) to the _bottom_ ground-plane? Is that a what a 'via' is, that we hear so much about? If so, can I just use little bits of wire, or is there some magic technique for doing them?

Once again, many thanks,

________
Agronski.

 

[vfone]

Yes they are vias, or plated holes.
http://en.wikipedia.org/wiki/Via_(electronics)

Is not something special, and soldered wires should be fine if you do not have other possibility. Just watch for parasitic inductance and check how to integrate it into the circuit.

 

[Agronski]

Re: Oscillator power supply...

Going to start laying out my PCB now - nice and small, with the output at least 15mm away from the tank...with a 50-ohm coplanar output strip....

I'm drawing 5mW from this circuit, so I guess I'll need some bypass capacitors. Is it alright just to use a 10pF chip capacitor near the transistor, with a choke in the DC line? Is C_bypass necessary, or are both a good idea? What sort of value should I chose? 10nH? 100nH? I'm not sure, but I guess any inductor with a large XL 2.4GHz...

_______
Agronski.

 

[vfone]

C_bypass capacitor is necessarily, and if you use 0402 SMD a 10pF as a decoupling cap at 2.4GHz is fine.
The RF choke shall have XL>500ohms at 2.4GHz (>47nH)