Crystal Clock Soruce - Tuning for Low Voltage

Crystal Clock Source - Tuning for Low Voltage

So I've constructed the circuit shown in Fig. 8 from this Linear Technology Clock Source App Note on a bit of Veroboard, but I don't want to run it at 24V.
Now... when I say "Fig. 8", at the moment I don't want variable frequency, so I've replaced the varactor diode with a fixed capacitor, and omitted the variable capacitor and 100kΩ resistor.

I was a little lost with the "tuning process", but I managed to keep it oscillating all the way down to the desired 3V supply by connecting four trim-pots in-place of the 10kΩ, 2k7Ω, 12kΩ and 4k7Ω resistors and adjusting one at a time, purely trial and error.
The default circuit would only oscillate down to about 12Vdc.

In order to adjust for the supply, what are the goals that will keep it oscillating?
Not overloading the crystal? Keeping the transistor biased in the linear region?
Are there some simple rules to follow for calculating the values rather than simply trialling them?

I've never had much luck at XOs but critical is keeping the
driver output from being lossy. You can't have the driver
looking linear resistive when "on" or "off". Weak gate drive
can do that. You also want the receiving end to be right
at the peak gain point (consider a dummy autobiased
inverter and cap coupling, or a single resistor shunted
stage with subsequent gain).

In fig 8, the transistor has an IC ~ .5 mA, has yours? In this circuit it has a 12 K collector load - lots of gain here, with low Vcc, you will have a problem with 12K, try using an RF choke in series with a lower value resistor.
The circuit needs the crystal to be inductive, so the tuned bit is like a capacitivley tapped tuned circuit, with the 100 PF+ 25PF variable being the Hi Z side and the 330 + 220 PF being the Lo Z side. What value cap has you used in place of the vari-cap diode?
Frank

chuckey said:

In fig 8, the transistor has an IC ~ .5 mA, has yours?

Click to expand...

How is this calculated? I was thinking more around 1mA based on Vb and Vbe, but maybe I've made some incorrect assumptions.
Infineon's BFP405 is the transistor I used since I had it on hand.

chuckey said:

In this circuit it has a 12 K collector load - lots of gain here, with low Vcc, you will have a problem with 12K, try using an RF choke in series with a lower value resistor.

Click to expand...

I didn't use a choke (there may be some intrinsic inductance), but I simply wound down the resistances one at a time, noticing when the output started to decrease, and then decreased another resitor to increase the output as far as it would before changing to decreasing another resistor. I seemed to go by:

1. Decrease Vcc to the oscillator's lower limit.
2. Decrease Rb-gnd (I think this increased the output for a while, then decreased it if I didn't stop).
3. Decrease Rb-vcc (from memory I kept doing this until just before it started to decrease again).
4. Decrease Rc.
5. Decrease Re.
6. Repeat until working at 3V.

chuckey said:

The circuit needs the crystal to be inductive, so the tuned bit is like a capacitivley tapped tuned circuit, with the 100 PF+ 25PF variable being the Hi Z side and the 330 + 220 PF being the Lo Z side. What value cap has you used in place of the vari-cap diode?

Click to expand...

I didn't look at the 220pF as being part of the tank... I figured that was part of an RC filter on the output. How would I use it in calculations with the 100Ω resistor between it and ground? Can I ignore the resistor?
I placed a 100pF cap in place of the varactor, just since it was a value that fell within the varactor's range.

Vb = 2.7/14.7 X 24 ~ .2 X 24 = 4.8V, -.7 = 4.1 (Sorry I took it as 2V). Ie = 4.1/4.7 ~ .8 mA = IC. Volts drop across Rl = 12 K X .8 mA = 10V, giving a Vc of 24- 10 = 14V, slightlu greater then your 3V!.
Frank

chuckey said:

Vb = 2.7/14.7 X 24 ~ .2 X 24 = 4.8V, -.7 = 4.1 (Sorry I took it as 2V). Ie = 4.1/4.7 ~ .8 mA = IC. Volts drop across Rl = 12 K X .8 mA = 10V, giving a Vc of 24- 10 = 14V, slightlu greater then your 3V!.

Click to expand...

Ah, thanks a lot for that.
That's the current I'd calculated, but I hadn't really thought about the 12kΩ voltage drop exceeding the rail. Makes me feel a little silly.
I'll see how I go with calculations and testing. I've build and tested about five different oscillator topologies so far; mods to this one should't be too much hassle. :grin: