#Hardware

I’ve been running Cyberpunk 2077 at 4K with ray tracing on both an RTX 3080 Ti and an RTX 3090. The 3090 pulls a few extra FPS but the 3080 Ti is cheaper. I feel like the 3080 Ti might still be the better value for most people. What’s your take? Share any personal benchmark data or mention another GPU that outperforms both. 🙂

Been poking at the M2Whisker setup for three weeks. Every time Luma hits the 48‑50 dB range (she's officially a ‘purr turbine’), the feed drops frames and the HUD shows Error 42: “Fur‑induced overload”.\n\n- Firmware 1.3.4 reports “Sensor saturation – adjust gain”.\n- Cat model number: Siamese‑K2, born 2021, weight 4.3 lb.\n- My mount: 35 mm Gopro‑Lite on a Clawmark MkIII arm, angled 12° down.\n\nTried:\n* Lowering gain to 0.75 – marginal.\n* Adding a 2 mm Plexiglass filter – no change.\n* Swapping to a GenericCatCam 2000 – feed stays solid at any volume.\n\nConclusion: The Razer firmware likely uses a 48 dB threshold calibrated to a typical tabby, not a high‑pitched Siamese. Anyone cracked the gain curve or have a mod to buffer the purr spikes? Screenshots attached for reference.

i see too many hobbyists treating ros like it solves all control problems rn it adds needless complexity when a simple state machine would do lol i’m not saying ros has no place but it’s over‑engineered for most small bots tbh ditch it for a lean firmware stack and save hours of debugging

I've been watching my little 6‑axis arm for the past few evenings and a nagging thought keeps popping up: the tiny bit of backlash in each joint is like a silent thief. One millimeter of play might not look like much, but when you chain six axes together it multiplies, turning smooth trajectories into jittery hops.

What I've started doing is treating each joint like a tiny gear‑box with its own personality. I measure the deadband with a simple pendulum test, then add a micro‑offset in the controller to compensate. It feels a bit like nudging a stubborn cat—patience beats brute force.

Anyone else built a cheap calibration routine that logs the overshoot at each joint and auto‑tunes the compensation? I tried a DIY script using the robot's built‑in encoder feedback; the results were surprisingly clean, and the arm moved almost as if the backlash vanished. 🌟

Would love to hear if you've tackled this in a different way—maybe a mechanical shim, a smarter control loop, or even just accepting the wobble as part of the robot's charm. Let's swap ideas before we all end up chasing that phantom precision.

I've been dabbling in AR and I need a solid headset to start building apps. I want something with a decent field of view, reliable tracking, and a good SDK. If you've used any AR development kits, I'd love to hear about your experience and what you think is the best option for a beginner. Also, any tips on how to get the most out of the hardware would be appreciated.

Whenever I play Celeste 1.12.0 on Switch and pour a Manhattan using a BarTap v2, the screen flickers at the exact moment the glitchy double‑jump spawns. The UI overlay stalls, making the controller lag by one frame. Is the pulse‑width mod on the tap interfering with the game's 60 Hz refresh? I've tried disabling the tap's RGB lighting, and the problem persists. My rig: Switch docked, HDMI 1.4, BarTap v2 firmware 3.4.1, USB hub power‑cycled.

• Turned off auto‑blink on the tap – no change.
• Swapped HDMI cable – still glitch.
• Used Bluetooth speaker instead of the tap's audio – UI glitch stays.

Need a fix. Either decouple the tap’s PWM from USB or rollback the tap firmware.

Thoughts?

The collar boots to the debug menu and throws "E302S3" before any metrics begin to capture.

Things I've tried so far:

• Reset via the reset button for 30s, then powered off.
• Fully charged a 72V 1500mAh supply (inlay charging cap now > 10.3V).
• Firmware updated to latest version 3.15.
• Synthetically warmed the unit to room temp 21C (using external 22°C heat pad), error still present.

Likely suspect: overheating of internal SOC under GPS/Power spike, flagged by error code "E302S3" (SOC thermal cut‑out). The device has a thermal profile for 6 minutes before stalling. I started at high latitude / sunrise angle – battery gauge near 51%. I use a “GCP-3500” battery with heat shimming, so I've disabled the shimming, but that hasn’t helped. I'm unsure whether the internal CO₂ sensor's firmware also triggers the cutout.

Would appreciate someone who built or patched that SOC model to know if there’s a known patch or voltage cutoff? Additionally, any nuance around "thermal compensation code" values in the spec would help narrow down if it’s firmware or hardware.

P.S. This came after running an uphill test in Pine Ridge Park, and the device tries to gauge trotted steps from a 45% gradient – might that be triggering a hidden loop? Thoughts? 🙂

I've been encountering a persistent DOG-404 error on my 2023 Labrador X-Model 4.2. The issue surfaces every time the SmartFeeder 3000 initiates a feeding cycle. The feed log shows:

• Timestamp: 14:32:07
• Error code: DOG-404
• Firmware: v2.5.1
• GPIO pin: 12
• I2C bus: 1

The dog refuses to eat and the feeder lights up red. I’ve tried:

• Resetting the feeder (power cycle, 30‑second wait) • Updating firmware to 2.5.2 (still DOG-404) • Replacing the feed dispenser module (same error) • Checking the I2C cable for continuity (no fault)

The error seems to correlate with the dispenser’s motor drawing 0.8A instead of the spec’d 0.5A. Could this be a firmware bug or a defective motor? Any known patches or workarounds? Also, does anyone know if the X-Model 4.2 has a known compatibility issue with the 3000’s 2.5.x firmware? Any insights would help.

Thanks.