0
\$\begingroup\$

I've got a microcontroller I want to clock from an external 10 MHz reference. I'd like the reference input to be as versatile as possible. At the moment, my reference is an LVCMOS square wave, but other folks have various levels of sine wave references.

I'm going to use a 74LVC1G17 to condition the input, fed with 3.3 volts (the microcontroller runs at 5v, but I've already tested that it will accept a 3.3v clock).

The remaining question is how best to terminate and/or condition the input, and what range of input signals it will accept (assuming regular, stable, 10 MHz, 50% duty cycle input).

I've googled around and learned about AC parallel termination and Thevenin termination, but in simulating them in circuitlab, I don't see the transitions lining up the way I'd expect.

In particular, a series cap feeding a voltage divider on the input doesn't give me a square wave centered on Vcc/2 like I'd expect.

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
4
  • \$\begingroup\$ RE "In particular, a series cap feeding a voltage divider on the input doesn't give me a square wave centered on Vcc/2 like I'd expect." If you're using a CMOS source, it's probably not able to drive a 50-ohm load. Try with maybe R1 = R2 = 10 kohms. \$\endgroup\$
    – The Photon
    Commented Aug 7, 2015 at 20:37
  • \$\begingroup\$ I mean that in circuitlab I'm not getting that square wave centered on Vcc/2. Circuitlab's voltage and function sources are supposed to be ideal. \$\endgroup\$
    – nsayer
    Commented Aug 7, 2015 at 21:07
  • \$\begingroup\$ Are you not getting a square wave, or not getting it centered on VCC/2? How long did you run the simulation? You should run it for at least 15 us to give enough time for V2 to charge up C1. \$\endgroup\$
    – The Photon
    Commented Aug 7, 2015 at 21:23
  • \$\begingroup\$ @ThePhoton is absolutely right. I was starting the simulation at T=0. Skipping forward a few µs fixed it. If you write that as an answer, I will happily accept it. \$\endgroup\$
    – nsayer
    Commented Aug 8, 2015 at 4:36

1 Answer 1

1
\$\begingroup\$

Transient simulators typically start all capacitors with 0 V across them. That will put the buffer input initially at 0 V. You will need to run the simulation for a few times the time constant (100 nF * 50 ohms, or about 5 us) to get C1 charged up and put the buffer input voltage due to V2 near the steady state value.

\$\endgroup\$

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.