Files

T

etwenandClaude Opus 5 9cc41f6fbd fix(margin): Reference VNOM to what the rails run at, and split profiles per CONN

86 of the 144 margin channels moved up +0.8% .. +2.5% to the measured
median for this platform (3.3 -> 3.38, 0.9 -> 0.915, 0.8 -> 0.82, ...).

That is not a display change. VNOM is the denominator of margin_status'
deviation %, and it is also the basis margin_set derives high/low/OV/UV
from - so both the number you read and the voltage you apply move with
it. A rail steady at 0.756 V used to read +0.80% against a nominal 0.75
and now reads 0.00%. Margin logs from before and after cannot be
compared: the reference moved, the hardware did not.

comboA/comboB are replaced by combo_SWB_CONN13..16_{high,low}. The
per-channel percentages are positional, so they follow the rails of one
specific board - applying a CONN13 profile to a CONN16 margins the wrong
rails and nothing detects it. SWB0 and SWB1 share a profile per CONN
because their rails are identical.

Fixes two things the deletion broke:
  - margin_apply_profile_all.sh still named comboA for all nine boards,
    which would now stop at "Profile not found" nine times over. It names
    the matching per-CONN profile instead, with CB on the generic
    combo_high3, and a note that the low pass is _high -> _low.
  - docs/LTC2980_channel_map.csv is generated from settings/*.conf, so
    all 86 changed rails were stale in it. Regenerated.

The eight new profiles also still carried comboA's header comment
("alternating high/low"), which describes neither what they contain nor
what they are for.

Script A -> V1.1.3.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014RetWKFZFG1ZHcitQAyhwM

2026-08-30 22:13:43 +08:00

4.7 KiB

Raw Blame History

v1.1.3 — Margin now measures against what this platform actually runs at

✨ New features

Per-CONN margin profiles for the switch boards

combo_SWB_CONN13_high / _low through combo_SWB_CONN16_high / _low replace the old comboA / comboB pair. Each file's per-channel percentages follow the rails on that board, so a CONN13 profile is no longer a rough fit for CONN16.
SWB0 and SWB1 share a profile per CONN — the two boards carry identical rails and differ only by I2C bus.
A high pass and a low pass are now the same run with _high swapped for _low.
A profile applied to the wrong CONN margins the wrong rails and says nothing. The percentages are positional, so nothing detects the mismatch. margin_apply_profile_all.sh now names the right profile per board instead of one name for all nine.

🐛 Bug fixes

VNOM was the spec value, not what the rails actually sit at

86 of the 144 channels moved up between +0.8% and +2.5% to the measured median for this platform — PVDD3_3V_SYNTH_LDO_* 3.3 → 3.38, P0V9_AVDD_* 0.9 → 0.915, PVDD0_8V_T3 0.8 → 0.82, and so on.
This is not a display change. VNOM is the denominator of margin_status's deviation %, and the basis margin_set derives high / low / OV / UV from. A rail sitting steady at 0.756 V used to read +0.80% against a nominal 0.75 and now reads 0.00%; and asking for +3% now lands on a different absolute voltage than it did last week.
Margin logs from before and after this build cannot be compared directly. The numbers moved because the reference moved, not because the hardware did.

The batch profile script pointed at files that no longer exist

margin_apply_profile_all.sh still named comboA for all nine boards. With comboA.conf removed, every board would have stopped at Profile not found — a nine-line wall of errors on a script that is normally left to run.

The channel map CSV was stale

docs/LTC2980_channel_map.csv is generated from settings/*.conf, so all 86 changed rails were wrong in it. Regenerated with tools/gen_channel_map.sh. It is output, not source — regenerate it, never hand-edit it.

📦 Downloads

File	Contents
`Script_ABC_Blanton_V1.1.3.zip`	The full Tera Term working directory, including `Blanton_Script/` to copy onto the DUT

After copying to the DUT:

chmod +x ~/Blanton_Script/*.sh      # see below
grep -rlU $'\r' ~/Blanton_Script    # expect no output

blanton_ber.sh and nfc_polling.sh are invoked by the macros through ./ or ~/, so they need the execute bit along with bmc_monitor.sh and usb_target.sh — neither git nor a Windows/USB copy carries it reliably.

⚠️ Before you run

Re-baseline your margin expectations. Any pass/fail limits or golden logs derived from the old VNOM values need redoing against this build.
margin_save in any form writes NVM permanently and now writes margin windows derived from the new reference. Run margin_status_all.sh first and never call it during a test.
SWB profiles are per-CONN. Applying combo_SWB_CONN13_high to a CONN16 board is silently wrong.
These three tests cannot share the switch: blanton_ber.sh, blanton_tr518.sh and blanton_traffic_linespeed.
PRBS takes the links down while blanton_ber.sh is armed.
ce0 keeps running until you stop it.
Two numbers must stay in step: Script C's sleep 30 and blanton_ber.sh's BER_INTERVAL.
FAN_SPEED is 100 — the fans run flat out unless you change config.ttl.
nfc_polling.sh start clears the log, so Script B wipes Script A's NFC baseline file (the numbers are already in the console log).
The NFC tool is written for the customer's tag. It polls Type A / B / F only and parses Type A strictly; a tag of another technology is never discovered and fails as no RF interface activation notification received. Without --expected-id it does not check identity, so a different tag of the same type still passes.
The three items listed as unverified in v1.1.0 are still unverified: the ce0 ingress mirror, the 120s settle in blanton_tr518.sh, and test mode nr=no on stop.

🔗 Links

ARCHITECTURE.md — test flow, data models, constraints
CLAUDE.md — commands and the bench gotchas
LTC2980_channel_map.csv — all 144 margin channels with the new VNOM

Full changelog: V1.1.2...V1.1.3

4.7 KiB Raw Blame History