Facktor & Vendan
Facktor
Have you ever considered how a single millisecond delay in a turret’s servo could translate into missed shots over a kilometer? I’ve been crunching the data.
Vendan
Yeah, a millisecond on a servo is a whole lot of lag at a kilometer. What does your data say—did the target drift that far? Let’s tighten the loop and shave that lag off.
Facktor
The data shows the target’s velocity was about 0.3 m/s, so a one‑millisecond lag moves it roughly 0.3 mm. That’s well within the error margin, but if you want to shave the lag, tighten the PID gains: increase the proportional term a few percent and add a small derivative component to damp the oscillation. Then recalc the error after each cycle. That should keep the loop within sub‑millisecond response time.
Vendan
Sounds good, just tweak the gains and keep a watch on the error curve. If the loop still bounces, we’ll need to bump the derivative further or add a feed‑forward term. Keep it tight and you’ll stay in that sub‑millisecond band.
Facktor
I'll log the current PID values, apply the suggested increase, and plot the error over the next 100 iterations. If we see oscillation beyond 0.5 ms, I'll increment the derivative by 0.05 and re‑evaluate. Keep the data tidy.
Vendan
Log the numbers, run the 100 cycles, and keep the plots neat. If the error jumps over 0.5 ms, bump the derivative by 0.05, re‑run, and see if the oscillation dampens. Straightforward, no fluff.