Close

PLL Debug

A project log for Photo-acoustic Spectroscopy

Creating an at home photo-acoustic spectroscopy system

JasMoHJasMoH 01/09/2018 at 20:430 Comments

Back from the holidays, and back into the swing of things. This post is rather rough because I don't feel like cleaning up my personal notes, but here's some raw, unfiltered black tar debugging. as always, the TL;DR is I'm an idiot who can't read letters...

Double check your footprints kiddos!

Pll testing notes:

Initial checkout:

Input midpoint tracker works at a variety of frequency down to the 10HZ range - coo.

Comparator is not comparatoring - uncool.

I’m thinking that this is a hysterisis problem. Pull up r88 - consider new values for r87

and r88 - or remove entirely and use on board hysterisis (dpending on low freq. performance?)

Pulled off r88 (feedback resistor). In process noticed issues with solder bridges on comp. Pins.

It has worked intermittently, but appears to have issues w/ manufacturing (wrong footprint).

That was an ordering problem. I ordered sc70 (.65 pitch) package - i laid out sot-23 (.95)

Next time: include nicer grounding studs for scope probes!

1/8/18

Replaced comparator with correctly packeged part. Always fun to do smt rework with my heated screwdriver soldering iron...

Midpoint tracking issues -

Seem to have issues with the input midponit tracking, even on 1KHz. Unsure why that was working before but not now.

measured 3v3 at non-inverting input (about 10x), and railing at output. Colud this be due to tolarance issues w/ rc?

Quick’n’dirty spice-y sims say that it shouldn’t be that sensativ eat 1k (20% mismatch between the input and output)

Op amp damage?

Loading effects from comp?

5000 rA input bias current, 10^13 input resistance. Should be good fine.

Op amp lifted off due to incorrect footprint sizing. Likely under stress… fix that and try agin...

1/9/18

Repaired input midpoint tracking opamp (u7) and it works again. So that’s cool.

Still no PLL lock though. Comparator is just railing out...

Quick midpoint characterization. “Decent” at 10 Hz, falls appart in the single Hz range.

This should be good enough for a first pass, it’s unlikely to ever be used bellow 20Hz due to 1/f noise.

Sceenshots:

 I bought the wrong footprint. Fucking again. I am not a clever man

Discussions