"Clipped sine wave" TCXO as Propeller clock source

Christoph_H

I've discovered only recently[*] that temperature-compensated crystal oscillators are available from the usual distributors for quite affordable prices. Most offer a so called clipped sine wave output with a guaranteed 0.8 Volt peak-to-peak amplitude connected to a 10kOhm // 10pF load.
The one I've ordered is from the Taitien TX series (second from top on http://www.taitien.com.tw/en/products_vctcxo.aspx).

With only the 10M//15p oscilloscope probe as load I observe a 0 to 1.1V signal on the TCXO output. So I guess the options for connecting this signal to the Propellers XIN pin are:

1. Direct connection.
   Won't work - 1.1V is most certainly too low for the logic threshold.
2. Coupling capacitor to remove the existing DC-Bias, followed by a voltage divider to set a new bias voltage near the Propellers Vcc/2.
   Might work with careful bias point selection.
3. Coupling capacitor to remove the existing DC-Bias, use XTAL3 clock mode instead of XINPUT.
   Could this work? Would only need one additional capacitor.
4. Use an inverter with a feedback resistor (self biasing) between the TCXO and propeller to provide a full amplitude clock signal with sharp edges.
   Will work, needs an additional IC (single gate).

The manufacturers appnote on this topic is available here
Thank you in advance for your opinion/thoughts/experiences on this.


[*]Last year I bought a few small OCXOs via eBay. Compared to the TCXOs in questions those are designed for 5V and want approx. 65mA at room temperature. They need a much larger footprint on the pcb (not to mention the height..). But they are quite a bit more stable, both with temperature and aging if they datasheet is correct.

Beau Schwabe

Can you use 1 diode drop on the TCXO to raise the output to within threshold of the Propeller? (see attached)

Phil Pilgrim (PhiPi)

I like #3. At least the XIN pin will be seeing the kind of input it would be getting from a crystal feedback.

-Phil

jmg

Christoph_H wrote: »

The one I've ordered is from the Taitien TX series (second from top on http://www.taitien.com.tw/en/products_vctcxo.aspx).

What price are these ?

Christoph_H wrote: »

With only the 10M//15p oscilloscope probe as load I observe a 0 to 1.1V signal on the TCXO output. So I guess the options for connecting this signal to the Propellers XIN pin are:
3. Coupling capacitor to remove the existing DC-Bias, use XTAL3 clock mode instead of XINPUT.
Could this work? Would only need one additional capacitor.

I'd use that one, ie configure as a Xtal osc (biased inverter) and AC couple. Much more tolerant than resistor biasing, and Diode kludges, and will give a close to 50% clock input.

Another alternative would be a Dual-Supply Logic Translator, where you can adjust the IP side Vcc to set Duty cycle.

Or, you could look at a device like a 74HC6323 (SO8), which would allow a more common 10MHz TXCO, to feed a 5MHz Prop CLK out
.

Christoph_H

Beau Schwabe (Parallax) wrote: »

Can you use 1 diode drop on the TCXO to raise the output to within threshold of the Propeller?

This goes into the direction of #2 with fewer parts but without the ability to fine tune the bias point. I don't want to rely on the clock signal having 1.1Vpp. So the question is if the 0.8Vpp mentioned in the TCXO datasheet plus the diode voltage drop will be enough to reliably trigger the threshold when going low -> high.

jmg wrote: »

What price are these ?

Here (Austria) I can get them for about €2.5 in single piece quantities. In the US DigiKey offers similar parts.

jmg wrote: »

you could look at a device like a 74HC6323

Yes, that would be a good choice if I needed the input clock to be in the 4-8MHz range for the PLL. At the moment I don't need the Propeller to run at maximum speed, so I'll skip the PLL.

I will try #3 on a spare board and see how it works out.

EDIT: Works just fine with #3 from first post and a 26MHz TCXO, PLL disabled of course.