6
\$\begingroup\$

I want to use the signal from a GPS disciplined oscillator (GPSDO) as a clock signal for a microcontroller.

The GPSDO outputs a 10 MHz square wave between 0 V and 3.3 V designed to drive into a 50 Ω characteristic impedance.

This is delivered to my circuit over a short (500 mm or so) length of coaxial cable. My circuit terminates it with 50 Ω to GND (which also reduces the peak voltage of the square wave to about 1.5 V).

I'm using a Schottky diode to clamp the signal (through a 2200 Ω resistor) to the 3.3 V rail to protect a microcontroller from being powered by the square wave when the 3.3 V rail is turned off.

As a result of this, the final clock signal is reduced to a slightly curved triangular wave between 1.3 V and 1.78 V (in other words, a 480 mV pk-pk with a +1.55 V DC offset). The waveform is very clean and appears free from any distortions that might make it unsuitable as a clock signal. However the amplitude is very small and it is currently unbuffered.

schematic

simulate this circuit – Schematic created using CircuitLab

What could I use to amplify this signal into the microcontroller? The microcontroller expects a 0 - 3.3 V signal, but I think it's fairly tolerant of anything better than 2 V pk-pk centred on 1.65 V.

So far I have tried to use an NPN BJT transistor (2N2222) as a simple voltage amplifier but I couldn't get that to work. I wondered about using something like an LM741 op-amp but the 1.5 MHz bandwidth is insufficient.

\$\endgroup\$
1
  • \$\begingroup\$ A triangle wave is not right. How much capacitance does the diode have? Make sure you use one with small capacitance like 1N5711 whose datasheet lists 2pF. \$\endgroup\$ Commented Oct 24, 2022 at 1:02

1 Answer 1

6
\$\begingroup\$

I think the problem here is that you haven't heard of series terminations applied at the source end of the clock signal. If you put a 50 Ω series resistor in-line with a signal output just before it feeds the coax then, the output from the other end of the coax will be 3.3 volts peak and no buffering is needed to put that right.

However, by the sound of it, the GPS output already has this internal output impedance built in so, no need to add anything; just feed the GPS output onto the coax and you should see a clean 3.3 volt clock signal at the far end. You can easily check this of course.

Then if you want to protect the MCU input from surges, you should be able to look at the data sheet of your device and add a series current limiting resistor to your input.

AN-903 A Comparison of Differential Termination Techniques <-- read the section on series termination on page 4.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you. Yes, you're right about the source-end termination. The GPSDO's documentation doesn't mention it at all. I removed the 50 Ohm termination on my board and the situation did improve. I also reduced the 2200 ohm resistor to a 1000 Ohm, which boosted the signal further. Regarding the diode, it wasn't really surges I was worried about, but rather parasitically powering the micro from the clock signal, since the clock is always present but the 3.3 V rail would be periodically turned off. Problem solved, cheers! \$\endgroup\$ Commented Oct 23, 2022 at 14:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.