LT Spice tx lines

Td={expr} Z0=50 will only work where "expr" is a fixed amount during the simulation.

If you're trying this inside of a loop, I'm sure it won't work. ...Jim Thompson

Doesn't seem to. I have to build variable delays other ways.

I'm simulating a system that has to run for almost a second, and I'd like the max time step to be 5 ps, something like that. It does crazy things in auto mode, or at 50 ps max step size. At 5 ps, I might learn something useful in a year or two.

In real life causality is imposed consistently, though, and charge is conserved. With a general expression for the delay, you could do all sorts of acausal things that would cause the simulation to diverge.

Also, in real life changing the delay rapidly will cause a large change in signal amplitude, e.g. if you reduce the capacitance per unit length, the voltage will go up to keep CV constant.

Probably a more physicsy approach would work.

Cheers

Phil Hobbs

You might try an all-pass with a multiplier... works, at least for a single-tone sinusoid.

Digital, make an ideal one-shot.

Sometimes a breadboard is faster ;-) ...Jim Thompson

"physicsy" ?>:-) ...Jim Thompson

Yes, like an N-section variable LC, or possibly a big mux and binary-weighted tx lines.

Cheers

Phil Hobbs

I'm using a simple analog ramp (diode, cap, current source) and a tanh() comparator.

Looks that way. Roughly 1e9 faster. I'm laying out a test board.

I can't post the Top Secret overall circuit, but I'll try to isolate one of the weird-behaving sections, for general amusement.

The HP8133A pulse generator did that. It's full of inductor-varicap delay lines.

You can buy adjustable delay lines, sort of like wirewound pots with a slider wiper thing. The speed of moving the slider is of course well under the speed of light.

I've used trombone lines a fair amount, back in me yoot.

Cheers

Phil Hobbs

That works.

OK ;-) ...Jim Thompson