Add drop-test protocol, Edison synthesis, and first-data analysis#67
Conversation
|
@copilot send to edison scientific (high effort literature). Fetch this session. Commit all artifacts. Summarize in your comment reply and provide direct link to md file. |
|
@copilot consider comment from Jeff in #36 (comment); I had forgotten/didn't realize it was bungee assisted |
Done. Edison Scientific LITERATURE_HIGH task 📄 Report (direct link to .md): Headline recommendations (from §"Summary of Actionable Recommendations"):
Standards stack surfaced: ASTM D5276 / D7136 / D3332, ISO 6603 / 1683 / 5347, MIL-STD-810 method 516, SAE J211. Idempotent driver at |
Good catch — folded Jeff's bungee-assisted note into
The Edison literature report's recommendations still apply: bungee-assisted (and pneumatically-assisted) drop towers are standard in the cited drop-weight literature (Rajput 2018, Anas 2024) and the magnet-release-jerk / fixture-resonance gotchas there extend to the bungee preload too. |
|
@copilot have at it with #36 (comment) |
Plotted and analyzed all five runs. Raw data, a reproducible script, figures, and a writeup are committed:
Format: 4-channel accelerometer, 125 kHz (8 µs), 0.2 s window. CH1 is the impact channel; CH2/CH3 are low-level; CH4 carries a fixed ~1.4 kG spike at t≈4.2 ms in every run including the control — a trigger/magnet-release artifact, not impact (please confirm the channel map). Peak acceleration (raw peaks are ringing-dominated — PSD energy out to ~20 kHz, structural peak ~550 Hz — so the SAE J211 CFC-180 column is the meaningful structural number):
Headlines:
Caveats: channel map unconfirmed; CH4 artifact; only the 200 ms shock window (no ~10 s ringdown); n=1 for control/PETG; no Δv/SEA quoted yet (needs the input-vs-transmitted channel pair + clean baseline + full-event integration). Next-capture recommendations are in §6 of the analysis doc. |
|
@ctrhjk could you run tests using three of the T3 prisms (each with distinct parameters), doing the following:
Noting of course that we'll be using the lower sensitivity single-axis accelerometer per #71 and #74 @achris0520 will print these 6 structures for you per #35 (comment) In the meantime though, if you could grab a couple existing T3 prisms (noting the IDs) and run a vertex test and an acrylic test without cyclic testing, just a single drop. It's up to you whether you do the vertex or the acrylic test first. Please get videos if you can as well. |
|
We have a new printing whose ID is '7xadt6' I ran 200 drops with this specimen. Test settings are the same as last time, but I switched the trigger channel from ch4 to ch5 again because when I ran the practice drop test, ch4 trigger didn't engage at a drop height of 5in. I switched the trigger channel to ch5 and dropped at 5 in height, but it didn't work. So this time I changed the drop height to 10 in. Here are the results of drops. Please analyze these data using OLS regression and report the results. |
|
Claude finished @ctrhjk's task in 11m 46s —— View job 200-drop campaign on
|
| series | mean | CV | slope (%/drop) | p | R² |
|---|---|---|---|---|---|
| TOP output (CH2–4) | 244.6 G | 1.98 % | −0.028 | 1.2e-37 | 0.58 |
| CH5 plate (trigger) | 238.7 G | 3.40 % | −0.036 | 2.9e-18 | 0.33 |
| T = TOP/CH5 | 1.025 | 2.37 % | +0.009 | 3.6e-03 | 0.04 |
| BOT (alive drops, n = 77) | 175.8 G | 5.62 % | +0.055 | 1.9e-14 | qualitative |
| T* = TOP/BOT | 1.403 | 6.54 % | −0.073 | 1.6e-21 | qualitative |
TOP and CH5 decline together (−5.6 % / −7 % accumulated) with plate Δv falling in lockstep — the familiar rig-level drift (the strike softened over ~46 min; mirror image of the 5-in run's +9 % hardening), back-loaded into the second half (split-half: +0.000 %/drop n.s. → −0.023 %/drop). T cancels it again — mean 1.025, CV 2.37 %, residual slope ≈ +1.7 % accumulated (R² = 0.04). That's the fourth consecutive campaign where a rig-level drift cancels in T. Reliability: DW 0.37–0.70 (positive autocorrelation → OLS over-eager, so the tiny T slope is if anything overstated); start-drop sweep stable.
5. ⚠ Problem 3 — a CH5 excursion at drops ~140–175
CH5 sags to a 210–230 G shelf across drops ~140–175 (min 210 G at drop 173), then partially recovers; T spikes to 1.09–1.16 at drops 170–176, and the campaign's only two impact-timing outliers land at drops 170/177. The window overlaps the BOT recovery (drop 172), so something disturbed the rig there — most plausibly the plate sensor's tape coupling momentarily degrading (TOP shows no matching feature, so it's not the specimen). Since CH5 is both the trigger and the T denominator: refresh the tape (or move to the stud/cement mount) before the next campaign. Silver lining — the per-drop T series caught this live; it doubles as the rig-health monitor.
6. Specimen 7xadt6 — no damage over 200 drops, one genuine watch item
Pulse width 1.48 ms (CV 0.69 %, −0.8 % total, stiffer direction), dominant ringdown mode ~550 Hz with no trend (p = 0.20), noise floors healthy. But: the dominant mode flips to a ~122 Hz component for the final 9 drops (192–200) consecutively (sporadic at 172/179 earlier). Level/width/centroid are unchanged, so it reads as mode-trading between comparable PSD peaks — yet a low-frequency mode winning 9 straight drops at the end of the campaign is what tendon-relaxation onset would look like. Please photograph/inspect 7xadt6's tendons and run a short 5-drop re-check — if ~122 Hz now dominates from drop 1, it's a real specimen change (and our first observed one on an intact print).
7. About the 5-in no-trigger problem you hit
The numbers say √h scaling can't explain it: at 10 in this specimen puts 2.9–5.4 kG on CH4 and 3.6–6.2 kG on CH5, which would naively predict ~2–4 kG at 5 in — far above 1000 G. So either 7xadt6 is strongly nonlinear in drop energy (a compliant fresh print can absorb a 5-in impact without the sharp bottoming-out spike — RW5F61, stiffer and much-exercised, only reached 1.4–1.9 kG at 5 in), or the practice-run setup differed. Practical fix either way: at 5 in, lower the trigger level to ~300–500 G instead of raising the height — pre-impact activity is ≤ 9 G, so even 100 G would have ≥ 10× clearance over the floor. For this campaign, 10 in + 1000 G worked flawlessly.
Caveats: n = 1 specimen; 200 ms window; Δv partial-pulse; tri-axis orientations unverified; BOT saturation-biased and absent for 113 captures; T ≈ 1.03 at 10 in is not comparable to RW5F61's 0.96 @ 13 in / 0.945 @ 5 in (different specimen and height, so this is not yet a geometry-discrimination result).
- Gather context (prior 5-in 100-drop analysis script, channel map)
- Download & extract the 8 zips, commit the 200 raw TP4 exports
- Classify captures (200/200 real; BOT dropout forensics; saturation audit)
- Burn-in + OLS drift regression + reliability checks
- Damage indicators over 200 drops (first fresh specimen — ~122 Hz watch item)
- Figures + writeup + README
- Commit & push (
a136ab6) - Report results + errors/problems observed
· branchcopilot/get-video-drop-test-data
…(Signals 61-173), CH7/CH8 over FS at 10 in, common-mode rig softening cancelled by T, ~122 Hz end-of-campaign mode watch item Co-authored-by: Jinkwan Han <282969777+ctrhjk@users.noreply.github.com> Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Here are another 30 drops run after your report. Diagnose those three problems you mentioned in the earlier report from this data, and report back to me check_Signal203.csv |
|
Claude Code is working… I'll analyze this and get back to you. |
|
Looks like you got stuck. Can you start working on this comment again?
|
|
Claude finished @ctrhjk's task in 6m 23s —— View job 30-drop check run — diagnosis of the three 200-drop problems (
|
| channel | full scale | median peak | max | > FS |
|---|---|---|---|---|
| CH6 | 1,002.0 G | 70.0 % FS | 90.5 % | 0/30 |
| CH7 | 991.1 G | 102.3 % FS | 107.6 % | 26/30 |
| CH8 | 989.1 G | 107.1 % FS | 109.8 % | 29/30 |
Slightly worse than the campaign (CH8 50/87, CH7 31/87) — plausibly because the electrical path now delivers the full signal. The downstream damage is visible: BOT CFC-180 scatters 85–200 G (CV 23.8 %) and T* spans 1.19–2.80 — unusable. Verdict unchanged: at 10 in, BOT is qualitative only; quantitative bottom-vertex data needs ~5 in or a ≥ 3 kG sensor. (Everything else has ample headroom: CH4 max 34.7 % FS, CH5 max 49.0 % FS.)
Problem 3 — CH5 excursion: ✗ the sensitivity loss persisted — re-mount CH5
| quantity | campaign healthy | campaign excursion | check run |
|---|---|---|---|
| CH5 CFC-180 | 242.2 G | 224.7 G | 219.7 G (CV 1.14 %) |
| TOP CFC-180 | 244.6 G | — | 239.1 G (CV 1.13 %) |
| T = TOP/CH5 | 1.025 | 1.09–1.16 | 1.088 (CV 0.99 %) |
TOP is within ~2 % of its campaign level, but CH5 sits ~9 % below its healthy level — below even the excursion shelf — so T is biased up to 1.088, inside the excursion band. The differential (TOP unchanged, CH5 down) rules out a softer strike — every prior rig-level drift moved both channels together — so this is the tape coupling durably degraded, not recovered. The degraded state is at least stable (CH5 flat within the run, p = 0.44; T CV 0.99 % is the tightest of any 7xadt6 series), i.e. the coupling settled at a new lower sensitivity rather than continuing to slide.
Action: re-mount CH5 (fresh tape at minimum; stud/cement preferred) before the next campaign, then re-baseline T — the "compare T within a mount only" rule applies to the re-mount.
Bonus — the ~122 Hz watch item: ✅ cleared, no specimen damage
This run doubled as the 5-drop re-check I asked for, and it's good news: the dominant ringdown mode is back at 519–580 Hz on all 30/30 drops from drop 1; the ~122 Hz component is present but subdominant throughout (band-power ratio vs ~550 Hz: median 0.26, never ≥ 1). Pulse width 1.482 ms (CV 0.29 %, flat, matching the campaign's 1.48 ms). The final-9-drop low-mode dominance was transient mode-trading, not tendon-relaxation onset — 7xadt6 shows no damage signature at ~230 cumulative drops. (Minor: the 122/550 ratio creeps up mildly within the run, 0.13 → 0.43, p = 0.013 — the per-drop ratio is now in the metrics JSON so future campaigns can watch it directly.)
Bottom line
| item | verdict | action |
|---|---|---|
| 1. BOT dropout | recovered (30/30) but cause untouched; Signal 203 says still marginal | wiggle-test CH6–8 cable + connectors |
| 2. BOT saturation | confirmed, CH7 and CH8 over FS at 10 in | keep BOT qualitative, or 5 in / ≥ 3 kG sensor |
| 3. CH5 excursion | persisted — coupling down ~9 %, T biased to 1.088 | re-mount CH5, re-baseline T |
| ~122 Hz mode | cleared — transient mode-trading, no damage | none; ratio now tracked in metrics |
Caveats: n = 1 specimen; 30 drops is short for catching an intermittent electrical fault; 200 ms window; tri-axis orientations unverified; BOT saturation-biased throughout; the CH5 verdict is inferred from the TOP/CH5 differential, not an independent coupling measurement.
- Gather context (200-drop analysis script + findings)
- Download & commit the 30 check-drop CSVs
- Problem 1: BOT electrical dropout — alive 30/30, Signal 203 marginal
- Problem 2: BOT saturation at 10 in — CH8 29/30 and CH7 26/30 over FS
- Problem 3: CH5 excursion — persisted; re-mount recommended
- Watch item: ~122 Hz mode — cleared, back at ~550 Hz from drop 1
- Figures + writeup + README
- Commit & push (
475937b) - Report results
· branchcopilot/get-video-drop-test-data
…: BOT recovered but marginal (Signal 203), CH7/CH8 still over FS at 10 in, CH5 tape coupling durably degraded (T biased to 1.088), ~122 Hz watch item cleared Co-authored-by: Jinkwan Han <282969777+ctrhjk@users.noreply.github.com> Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Here are another 30 drops run after the previous one. Analyze these data and diagnose whether the same problem exists or not. |
|
Looks like you got stuck again. Can you start working on this again?
|
|
Claude Code is working… I'll analyze this and get back to you. |
|
@claude try again |
|
Claude Code is working… I'll analyze this and get back to you. |
|
@copilot pick up from where claude gh actions was at, since claude seems to be running into some issues @ctrhjk I made a new issue and corresponding PR, and claude pings worked on both, so there might be something weird/wrong about this one.. @copilot also make sure that all the raw data files (CSV in particular) across all comments in this PR are appropriately committed and organized within the branch, especially since I might need to create a new PR associated with this thread per the issues that are happening above with claude. |
|
Carrying on over here: #82 (comment) |
…e-level (SHA-256) completeness audit of every PR #67 attachment Co-authored-by: Sterling G. Baird <45469701+sgbaird@users.noreply.github.com> Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Move @sgbaird's root-level video uploads to data/drop-tests/60in-5felts-validation/video/{7xadt6,9GMQYQ}_slomo.mp4 and run the frame-by-frame pass that was previously blocked by YouTube's bot gate. fps recovered from the repo record + PR #67 camera spec: Sony RX100 IV HFR at 960 fps; the script deduplicates the 24->30 pulldown so real time is exactly unique-frame/960 (19.8%/14.1% duplicates found). Findings: impact pulse <= 2 capture frames (~1-2 ms) corroborating the DAQ's ~1.6 ms CFC width; anti-rebound brake catch at ~2.1-2.4 g holding the carriage 130-150 mm above the felt from ~86-89 ms after impact (no secondary hits); top-vertex elastic snap-back at ~0.7x impact speed with visible strut flexure fully recovered; cross-framing scale consistency (82 vs 78 mm specimen extent) under the free-fall anchor (DAQ dv 5.53/5.69 m/s corroborates). Compression itself falls between frames - >=5000 fps DIC still needed. Co-authored-by: Sterling G. Baird <45469701+sgbaird@users.noreply.github.com> Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Issue is operational (coordinating the first crush/drop tests on Jeff Hill's tower) and this repo is the LaTeX MRG proposal with no code or existing test-protocol surface. This PR consolidates the moving parts from the issue thread, adds a literature synthesis, and analyzes the recorded accelerometer data.
Added
docs/drop-test-protocol.md— single source of truth for the drop-test setup, covering:g_max, SEA, full ~10 s ringdown (not just the 200 ms shock), reusability, slow-mo framing from t=0edison-trajectories/drop-test/— Edison Scientific LITERATURE_HIGH synthesis (task653d7d39) on drop-tower troubleshooting for small 3D-printed lattice/tensegrity specimens: ~57 KB report, full JSON dump, submission record, and README. Idempotent driver atscripts/edison/submit_drop_test.py. Surfaces a standards stack (ASTM D5276/D7136/D3332, ISO 6603/1683/5347, MIL-STD-810 method 516, SAE J211) and recommendations (linear sleeve bearings, magnetic/elastic top-plate hold-down, ≥10 s ring-buffer DAQ, SAE J211 CFC filtering, n ≥ 5 + CV, ≥5000 fps DIC; closest analogues Pajunen 2019, Dwyer 2023).First drop-test data analysis of the five TP4 accelerometer exports posted by @me-madsen (
Signal 10–14, 4-channel, 125 kHz, 0.2 s window):data/drop-tests/raw/— committed raw export filesscripts/analysis/drop_test_analysis.py— loader, SAE J211 CFC-1000 / CFC-180 filtering, peak/pulse/PSD metrics, and figure generationdata/drop-tests/figures/— full-window CH1 overlay, per-run impact zoom (raw vs filtered), peak-g bar chart, PSD, and CH4 trigger-artifact plotdocs/drop-test-analysis.md+data/drop-tests/README.md— findings: the "audrey" tensegrity specimen reduces CFC-180 peak acceleration ~74–79 % vs the no-specimen control (~370–463 G vs ~1,792 G), while the PETG run's raw peak is within ~1 % of the control (≈ direct plate-on-plate hit), strong evidence of the bungee-driven lift-off; CH4 carries a fixed ~1.4 kG trigger/release artifact at t≈4.2 ms in every run. Caveats noted: unconfirmed channel map, 200 ms window only, n = 1 for control/PETG, no Δv/SEA quoted yet.Vertex vs. acrylic-plate T3-prism drop-test data and analysis posted by @ctrhjk (PR Add drop-test protocol, Edison synthesis, and first-data analysis #67), under
data/drop-tests/vertex-acrylic/:raw/— eight TP4 exports{n0jdwk, m6cyoq, T3_0103, T3_0000}_Signal{1,2}.csv(Signal1= vertex-mounted,Signal2= acrylic-plate), single drop per configuration at 13 ft, 200 ms / 125 kHzREADME.md— channel map (CH1 removed; CH2–CH4 tri-axis; CH5 single-axis; CH4 = 1000 G trigger) with full-scale/sensitivity per channel, per-specimen file index, and @ctrhjk's observations (clip-height/no-trigger issue, hot-glue z-axis mount limitation,m6cyoqstrut andT3_0103TPU-tendon damage after the acrylic test, and the invalidT3_0000acrylic run where the accelerometer fell off)scripts/analysis/drop_test_vertex_acrylic_analysis.py— locates the impact via the triggered CH4 channel (windowed ±1.5 ms peak search in the first 10 ms, not a global max), baseline-corrects, and reports raw / SAE J211 CFC-1000 / CFC-180 peaks for the single-axis CH5 (primary go-forward sensor) and tri-axis CH4, auto-flagging invalid / no-clean-impact runsdata/drop-tests/vertex-acrylic/figures/— vertex-vs-acrylic CH5 impact windows, CFC-180 peak-g bar chart, and vertex CH5 PSDdocs/drop-test-vertex-acrylic-analysis.md— findings + an explicit SOP / test-method section: vertex mounting is repeatable (4/4 clean, CFC-180 229–284 G, CV ≈ 9 %) while the acrylic configuration is not (3/4 runs registered no clean impact — clips too low so the plate seats on the specimen, plus the fell-offT3_0000); the vertex peaks do not yet discriminate geometry so fresh intact distinct-geometry samples (vertex-only, n ≥ 5) are needed before peak-g is a trustworthy BO objective; replace the hot-glue mount with a z-axis-aligned seat; and the single-axis sensor's raw peaks reach 70–90 % of its 9,442.9 G full scale (near saturation on the m6cyoq-acrylic run). Caveats: n = 1 per (specimen, mount), 200 ms window only, partial-pulse Δv, unconfirmed CH4/CH5 axis correspondence.Clip-height sweep & base-plate accelerometer-check diagnostic posted by @ctrhjk, under
data/drop-tests/clip-height/— drilling into why the acrylic-plate configuration repeatedly fails to trigger:raw/Accelerometer_check_Signal1.csv+README.md— the one triggered base-plate CSV (tri-axis on the bottom plate, 13 in drop) plus a setup README documenting both experiments: the clip-height sweep (extra bungees cured fly-off; tri-axis on the acrylic plate; clips at 0.5/1/1.5/2 in, two drops each; 0/8 drops triggered, video only) and the base-plate accelerometer check, with the shared channel mapscripts/analysis/drop_test_clip_height_analysis.py+data/drop-tests/clip-height/figures/— windowed CH4 impact location, SAE J211 CFC-1000 / CFC-180 peak/pulse/Δv metrics, and figures (base-plate impact window, full-window CH4, PSD)docs/drop-test-clip-height-analysis.md— findings: the base-plate hit triggers cleanly (CH4 raw 3072 G ≈ 3.1× the 1000 G trigger, CFC-180 280 G, Δv ≈ 3.3 m/s; CH4 dominates the off-axis channels ~23–55×), so the acrylic-plate "no trigger" failure (0/8 across the clip sweep) is a load-path problem — the plate seats on / is damped by the bungee-restrained specimen — not the sensor, DAQ, or trigger level.Input-output (transmissibility) drop-test data and analysis posted by @ctrhjk (PR Add drop-test protocol, Edison synthesis, and first-data analysis #67), under
data/drop-tests/input-output/— @ctrhjk's input-output instrumentation design: a single-axis accelerometer on the bottom plate = input (now the triggered channel CH5), a tri-axis accelerometer hot-glued to the top vertex = output (CH2–CH4), bungees removed, four distinct-geometry specimens (practice,n0jdwk,yqpmx1,h8Lbev) each dropped five times at 13 in:raw/— 20 TP4 exports{practice,n0jdwk,yqpmx1,h8Lbev}_Signal{1..5}.csv(Signalindex = drop number) +README.mdwith the channel map (trigger moved to the single-axis input CH5) and @ctrhjk's setup notesscripts/analysis/drop_test_input_output_analysis.py— locates the impact on the triggered CH5 (windowed ±1.5 ms peak), baseline-corrects, and reports raw / SAE J211 CFC-1000 / CFC-180 peaks for the input (CH5) and the tri-axis output resultant, the transmissibilityT = output/input, pulse width and Δv, with per-specimen mean ± 1σ / CV aggregatesdata/drop-tests/input-output/figures/— input-vs-output impact windows (5 drops overlaid), transmissibility bar chart, input repeatability, output PSDdocs/drop-test-input-output-analysis.md— findings: the input-output design works — 20/20 drops triggered cleanly, removing the bungees makes the input nearly constant (235–248 G CFC-180, ≤1.7 % CV), and transmissibility now discriminates geometry (yqpmx1≈ 0.96 is the only attenuator,h8Lbev≈ 1.09,practice/n0jdwk≈ 1.17–1.19), makingT(or output-peak-at-fixed-input) a usable BO objective; a mild within-run drift across the five cyclic drops is flagged as most likely hot-glue-mount-driven. Caveats: n = 1 specimen per geometry (5 repeat drops), 200 ms window, unverified tri-axis orientation, IDs not yet tied back to design parametersedison-trajectories/input-output/— Edison Scientific ANALYSIS (taskfe044079) that independently reproduced the transmissibility values exactly to two decimals, confirmed the within-run drift is statistically real and mount-driven (pooled +0.015/drop, p = 0.0001), endorsedTas a first-pass screening objective (recommending FRF / SRS-band metrics and output-peak-at-fixed-input as it matures), and gave a prioritized SOP (rigid z-aligned keyed sensor seat, keep bungees removed, extend capture past 200 ms, n ≥ 5 distinct prints per geometry with randomized order, anchor in SAE J211 / ISO 5347 / ASTM D3332). Idempotent driverscripts/edison/submit_input_output.py+ fetchscripts/edison/fetch_input_output.py; a cross-check section is folded into the analysis doc.