Colpitts values for NXP's SA605?

I'm trying to put together a simple superhet radio, receiving at 161.75MHz (this is for AIS reception.) I've been trying to design around NXP's SA605 chip (see SA605) but I'm having a hard time figuring out what values to use for some components. My IF is 10.7MHz & the LO is 161.75 - 10.7 = 151.05MHz. I'd like to avoid using a crystal. From the docs, it seems like a Colpitts oscillator should do the trick, with L = 82uH and C = 27pF (I know this is off a little bit, but I assume I can put a trimming cap in parallel with the capacitor across pins 3 & 4 to get it right.)

From one of the application notes, it looks like I can do this fairly simply:

...but the question becomes, what values do I use for the DC blocking cap, and the one on ground? I assume the value varies based on the oscillation frequency, but I don't know where to go to calculate that.

I've also tried to simulate this in the spice simulator now embedded in KiCAD:

...using the equivalent circuit from the application note:

...but I haven't had any luck getting it to actually oscillate. Do you have any hints on getting this going?

• Your schematic shows 10uF in parallel with the inductor. Jan 22, 2019 at 2:35

It seems that the local oscillator transistor has marginal gain at 150 MHz...achieving stable oscillation may be difficult. App-note suggests boosting DC operating current by adding an external resistor from pin 3 to ground: a value between 10K - 22K is recommended for oscillator frequency above 80 MHz. Boosting oscillator DC current will improve its gain-bandwidth product.
In addition, RF oscillator amplitude at pin 3 should be 220mV (RMS) to get full mixer gain. Relying on a SPICE simulation is very risky, because this transistor is not well-specified in the data sheet - what transistor should you substitute?

Oscillator frequency will drift with temperature and supply voltage. Expect to adjust frequency often (either with variable L or a variable C). At these frequencies, a voltage-controlled-oscillator would be used, as part of a phase-locked-loop to achieve stable frequency. Doing so with this feeble on-board transistor makes oscillator design even more risky.
An alternative might use a digital oscillator: something like SiliconLabs Si5351. Its top frequency end goes at least to 160 MHz (some claim 200 MHz). A tiny microcontroller would be required to set its frequency on power-up, and a fixed-frequency ~25MHz crystal is required as well. You can program its output frequency very close to 151.05 MHz. Its 3.3V CMOS output square wave can be reduced in amplitude with a resistor divider, and injected via a small capacitor to SA605 pin 4.

Colpitts oscillator: In the schematic below, R3 is an added external resistor to pin 3. R1 & R2 are internal bias resistors inside the chip. These might serve as a starting point for choosing values:

Oscillating frequency is roughly 150 MHz. C3 could include a very small variable capacitor, or L1 might be wound on a form with an iron powder slug or brass slug. R4 is not a real resistor; it simulates finite Q of the coil L1.

This simulation kick-starts the oscillator by injecting a current pulse with an initial condition: .IC I(L1)=10n

• Amazing work -- thank you! A note, and a question: it seems like the best way to figure out component values is to actually build something rather than play with the simulator. And, I can use a square wave generator for the LO? I didn't think that was an option -- shouldn't it be sinusoidal? Jan 22, 2019 at 9:20
• I start by choosing the resonator L reactance about 100 ohms @ 150 MHz., then see if capacitor values are reasonable. Yes, most mixers accept square waves but you should include a frequency-selective filter (162 MHz) between antenna and mixer RF port. Jan 22, 2019 at 15:39

You need to inject a couple of cycles at the required frequency into the circuit between the inductor and Gnd, to get the simulated oscillator going. In a real-world circuit there is always some noise present that has the same effect.