LVDS line receiver with an unencrypted SPICE model?

Something sort of like this might work.

Fun. I like it in general--it should have a nice clean Tx edge, for one thing, and the ON resistance of the pHEMT is a lot better than a Schottky diode for the same capacitance. That might eliminate the need to use multiple sampling pulses to get the sampling loop to converge. (Which would be good, because you can't easily stack samples with a track/hold architecture.)

ESD protection is one possible issue, and the hold capacitor loads down the line throughout the pulse and return, so it has to be very small.

So far I'm envisioning a two-diode series sampler, with a fast sampling pulse made by taking a fast edge from a BFP840 and differentiating it with a shorted Tx line, maybe with a bit of capacitive help.

As long as the final voltages across the diodes don't change from sample to sample, the nonlinear capacitance funnies pretty well go away. There might be some schmutz due to reflections off the bridge in the OFF state and other discontinuities rattling round. It'll be hard to get rid of the reflection from where the tines hit the soil surface, for instance.

As I say, fun.

Cheers

Phil Hobbs

How about this?

This slow sampler drives C1 hard, unlike a faster feedback sampler that can only poke a bit of incremental charge into the cap each shot.

Filed away. Good to have one that I can share with the customers.

Thanks again!

Phil Hobbs

That one's nice too. How hard it drives the sampling cap depends on the video resistance of the diodes. BAT1502s seem to work okay.

At the moment I'm trying to do it with just BFP840s and diodes, to make sure I understand all the details.

I'm someplace between you and Joerg on the ideal functionality-vs-BOM cost continuum--some years ago ChesterW and I did a 100-ps class TDR sampler for a fuel gauge that sells for $25. My bit was well under $2, including the fast pulse generator, 1-diode first stage sampler, fast buffer amp, second stage sampler, and slow output buffer.

It worked great, for a gas gauge--in fact CW and Co. Muntzed it a bit and got a few cents of cost out, in exchange for turning it into a

The present one needs good waveform fidelity because we're basically measuring the shape of the drooly return pulse and converting that to the water and conductivity measurement.

The spherical cows think that a FIN1002 driving a BFP840 can easily make a < 20 ps edge. I may disagree with that, but at least it's encouraging. ;)

Cheers

Phil Hobbs

A BAT1502 should be well below 10 ohms, which into a few pF will be plenty fast.

Sounds ambitious, but if you try it I'd love to know what happens.