Engineering 102 with fpv kpis

Tl;DR

Putting together an iot-dashboard v2 that now also controls a BLDC waterpump

Intro

• Why Im writting this post: bc I want to consolidate the 0-6-0 oss mbsd fwk notes, the electronics x water and these fpv kpi learnings
• What Ive learnt with it: Ive ended up sending for manufacturing STLs for the first time

Coming from this post.

npm install -g @openai/codex@latest

I found https://github.com/qunabu/Gravity

which is statically deployed to https://qunabu.github.io/Gravity/ with narration and smooth transitions powered by TypeScript + Three.js + Vite. It reminded my to some nice physics animations

• https://ciechanow.ski/moon/

Thought around escape velocity and optionality, anyone?

Now tell me thats not inspiring and superseeding my poincare lab: https://the-poincare-lab.pages.dev or my solar trajectory tracker

#git clone /ThreeBodies
cd ./ThreeBodies/ThePoincareLab

make deploy #https://the-poincare-lab.pages.dev/

MBSD

This year I tinkered within this repo: https://github.com/JAlcocerT/mbsd

First migrating matlab to python for mbsd, later with some NVH, 0D thermodynamics and fluid mechanics applied to a ICE.

To the initial renders with cadquery, i upgraded:

cd ./mbsd/integrated-engine-lab
make render
#cd ./build/renders

Recently made a recap with some improvements on the flow:

scp "jalcocert@192.168.1.2:/home/jalcocert/mbsd/integrated-engine-lab/build/renders/*.mp4" .
#ffmpeg -stream_loop 4 -i "engine_lab.mp4" -c copy "engine_lab_looped_5x.mp4"
scp "jalcocert@192.168.1.2:/home/jalcocert/mbsd/integrated-engine-lab/extensions/build/renders/*.mp4" .
#ffmpeg -stream_loop 10 -i "layout_ar_atlas.mp4" -c copy "layout_ar_atlas_5x.mp4"

How could i not output some videos

This was a good oportunity to use Typst > Latex to recap what I did.

Then, I wanted to make this available:

• https://multibodysystemsdynamics.com/whitepaper

Have been progressing with the roadmap: https://multibodysystemsdynamics.com/roadmap

• https://github.com/JAlcocerT/mbsd-core
• https://github.com/JAlcocerT/mbsd-examples

Last week I release v0.5.0 here, and now, its time for v0-6-0

#scp jalcocert@192.168.1.2:/home/jalcocert/multibody-tests/*.md . #v-0-5-0-concerns.md
scp jalcocert@192.168.1.2:/home/jalcocert/multibody-tests/v-0-5-0-concerns.md . 

cd /home/jalcocert/Desktop/mbsd-framework/mbsd-core

git switch main
git merge --ff-only v0.6.0-dev
git tag -a v0.6.0 -m "MBSD Core v0.6.0"
git push origin main
git push origin v0.6.0

awk '
  /^## v0\.6\.0 / { found=1; next }
  /^## / && found { exit }
  found { print }
' CHANGELOG.md | gh release create v0.6.0 \
  --repo JAlcocerT/mbsd-core \
  --verify-tag \
  --title "MBSD Core v0.6.0 - Experimental 3D Vocabulary" \
  --notes-file - \
  --latest

cd /home/jalcocert/Desktop/mbsd-framework/mbsd-examples

git switch main
git merge --ff-only v0.6.0-dev
git tag -a v0.6.0 -m "MBSD Examples v0.6.0"
git push origin main
git push origin v0.6.0

awk '
  /^## v0\.6\.0 / { found=1; next }
  /^## / && found { exit }
  found { print }
' CHANGELOG.md | gh release create v0.6.0 \
  --repo JAlcocerT/mbsd-examples \
  --verify-tag \
  --title "MBSD Examples v0.6.0 - Diagnostics and Validation" \
  --notes-file - \
  --latest

• https://github.com/JAlcocerT/mbsd-core/releases/tag/v0.6.0
• https://github.com/JAlcocerT/mbsd-examples/releases/tag/v0.6.0

Put together also couple of free non-walled godies:

• https://ebooks.jalcocertech.com/books/mechanism-analytics/
• https://jalcocert.github.io/JAlcocerT/understanding-batteries/#w-engine-simulation

Energy

Recently a friend told me that skoda (superb) got a diesel under 3L/100km

Thats crazy when we compared it on how things were with cars not so long back

Not sure how we avoided to think in terms of energy so far

How come we dont know that a typical laptop for wfh will consume ~150wh (~18w avg) while 9-5’ing?

Call it ~350wh-400wh a day if for any reason you have 2 laptops while WFH.

A laguna has a a 660Ah 12V battery while a bluetti its 288wh (or 18ah) a sper the conversion

Aka, with one bluetti you will survive couple of days or one while moonlighting

Or that a 5inch dron betaflight energy analysis when properly scaled tells that hoovers at ~90W, with session avg 130W and peaks of 1KW+?

A x300 homelab + my CM needed ~600Wh a month, or 0.6kwh, a ~1% of your utility bill

Isnt energy the currency of the universe to make things happen?

Or was it enthropy?

pie title Monthly Electricity Expenses (PLN) ~105 PLN Total
    "Kitchen & Major Appliances (Fridge, Cooking, Dishwasher)" : 37.80
    "Work Laptops & Monitors" : 20.48
    "Homelab & Cable Modem / Wi-Fi" : 18.90
    "Washing Machine & Laundry" : 11.03
    "Lighting & Electronics Standby" : 9.45
    "Other / Occasional Loads" : 6.30

In copper wires, the difference between signal and power is not the wave itself, but its purpose, current, and impedance.

Both power and electrical signals travel along wires as electromagnetic waves through the copper’s electrical field, but they are handled completely differently:

While the wire itself has a characteristic wave impedance, power circuits prioritize near-zero impedance to push current without heat, whereas logic signal inputs use massive impedance so they can read a voltage without drawing any current at all.

They are fundamentally different in how they travel, how they represent data, and how you must design the circuit for them.

Logic signals are basic DC voltages switched on and off through wires to talk across a board, while Wi-Fi, ELRS, and Zigbee are high-frequency radio waves engineered to detach from conductors and fly through the air.

Termodinamics

Wrote about thermodynamics some time back here.

Whats new?

Boyle, exergy, Petela-Landsberg radiation exergy equation…

What can this help with?

How about sizing your house AC system? :)

Heat Transfer

You can do very cool projects around this.

Want to transfer some heat from pipes?

Not a problem, get it simulated:

Electronics

If electrical engineering is about power, we can say that electronics is about logic.

Both are energy anyways :)

Notice that as soon as you step into this field, you start to find links.

Like:

1. All MOSFETs are transistors
2. Behind every DC-DC buck there is a MOSFET
3. Every buck performs PWM, just like some cheaper solar panels setup do instead of MPPT

See MOSFET Id vs Vds and https://github.com/JAlcocerT/poc/tree/main/physics-electronics

Remember to counter-clockwise CCW to decrease the output voltage.

The XL4015 is a higher-power step-down (buck) converter.

LM2596 and MP1584EN modules are normally adjustable constant-voltage power supplies: set an output voltage and they try to hold it for electronics.

You can know in advance which one you’ll need by simulating your circuit:

docker run --rm --network host eclipse-mosquitto:2 \
  mosquitto_pub -h 192.168.1.2 \
  -t esp32/pump/cmd \
  -m 'pulse:3000'

To recreate the wifi password file safely:

#read -rsp "Wi-Fi password: " WIFI_SECRET
printf '%s' "$WIFI_SECRET" > /tmp/iot-water-wifi-pass
unset WIFI_SECRET
chmod 600 /tmp/iot-water-wifi-pass
  # - Wi-Fi password expected at /tmp/iot-water-wifi-pass
  # - SSID is automatically detected with iwgetid

Then connect the ESP32, keep the pump-positive switch OFF, and run:

#cd /home/jalcocert/Desktop/poc/iot-esp-water/esp32-bms-prepwork/esp32-bms-mosfet
make board-list
make upload PORT=/dev/ttyUSB0
make clean

Use the port reported by make board-list if it differs.

Make sure that Arduino CLI is already installed.

Created and uploaded the WiFiManager demo successfully.

Files:

• esp32-wifi/Makefile
• esp32-wifi/wifi_manager_demo/wifi_manager_demo.ino

Test it:

1. Keep the pump-positive switch OFF.
2. On your phone, join the open Wi-Fi network:

WaterPump-Setup-E342A8

3. Accept the “no Internet” warning.
4. The portal should open automatically. Otherwise visit http://192.168.4.1.
5. Select your home Wi-Fi and enter its password.
6. The setup network should disappear and the ESP32 should reconnect to your Wi-Fi.
7. Press the ESP32 reset button. It should reconnect automatically without reopening the portal.

Watch logs with:

cd /home/jalcocert/Desktop/poc/iot-esp-water/esp32-wifi
make monitor

The combined firmware is uploaded and working:

• Wi-Fi connected: three quick flashes
• EMQX connected: two longer flashes
• Ready: short heartbeat every two seconds
• Pump active: rapid blinking
• Pulse completes: GPIO23 returns LOW

The safe MQTT status test succeeded; ESP32 is 192.168.1.17.

When supervised and ready, enable the physical pump switch and run:

cd /home/jalcocert/Desktop/poc/iot-esp-water/esp32-wifi
make pub-pulse PULSE_MS=3000

Or use your original Docker command. The pump should run for three seconds and stop locally.

Files:

• Combined firmware (esp32-wifi/wifi_manager_pump/wifi_manager_pump.ino)
• Testing instructions (esp32-wifi/wifi_manager_pump/readme.md)
• esp32-wifi/Makefile

The provisioning-only sketch remains untouched.

MQTT is currently unauthenticated and unencrypted, suitable only for this local prototype.

WiFiManager stores the Wi-Fi SSID and password in the ESP32’s persistent NVS flash—not inside the .ino file.

Uploading a new sketch replaces the application firmware but normally preserves NVS. Therefore, the new sketch found the previously saved credentials and connected automatically.

Credentials are removed only when you:

• Type r in this sketch’s serial monitor.
• Call WiFiManager.resetSettings().
• Erase the ESP32’s entire flash.
• Upload with an operation that explicitly clears NVS.

This is exactly what you want for future firmware/OTA updates: firmware changes without forcing clients to configure Wi-Fi again.

An unlimited command could leave the pump running if the off message is lost.

The firmware now accepts 1–60000 ms and compiles successfully.

It is not uploaded yet because the physical switch state is unknown.

With the pump switch OFF:

cd /home/jalcocert/Desktop/poc/iot-esp-water/esp32-wifi
make pump-upload

Then test progressively:

make pub-pulse PULSE_MS=10000
make pub-pulse PULSE_MS=30000
make pub-pulse PULSE_MS=60000

For water, a 60-second test gives approximately:

weight gained in grams ≈ flow in mL/min

Keep the pump wet and check its duty-cycle rating.

Made some updated firmware: esp32-wifi/wifi_manager_pump/wifi_manager_pump.ino. so that it asks your for WIFI pwd instead of having to bake it with the esp32 firmware.

docker run --rm --network host eclipse-mosquitto:2 \
  mosquitto_pub -h 192.168.1.2 \
  -t esp32/pump/cmd \
  -m 'pulse:3000'

docker run --rm --network host eclipse-mosquitto:2 \
  mosquitto_pub -h 192.168.1.2 \
  -t esp32/pump/cmd \
  -m 'off'

#Or from esp32-wifi:
# make pub-off

Hydraulic output is and always was for any pump: ρgQH

cd ./electronics-101

On top of all the above, the DHTs keep feeding data to a sqlite in the homelab:

#cd ./poc/iot-rpi-dht
sqlite3 /home/jalcocert/poc/iot-rpi-dht-insulation/ingester/data/readings.sqlite "WITH ranked AS (SELECT device, metric, topic, value, received_at, received_ms, strftime('%Y-%m-%d %H:00:00',received_at) AS hour_bucket, ROW_NUMBER() OVER (PARTITION BY device, metric, strftime('%Y-%m-%d %H', received_at) ORDER BY received_ms ASC) AS rn FROM readings WHERE received_ms >=(strftime('%s','now') - 24*60*60)*1000) SELECT device, metric, hour_bucket, topic, value, received_at FROM ranked WHERE rn = 1 ORDER BY hour_bucket, device, metric;"

To get the latest temperature and humidity from ESP32 and Pico W, excluding Pico’s internal temperature:

sqlite3 -header -column \
  /home/jalcocert/poc/iot-rpi-dht-insulation/ingester/data/readings.sqlite \
  "WITH ranked AS (
    SELECT *,
      ROW_NUMBER() OVER (
        PARTITION BY device, metric
        ORDER BY received_ms DESC
      ) AS rn
    FROM readings
    WHERE device IN ('esp32', 'pico')
      AND metric IN ('temperature', 'humidity')
      AND topic NOT LIKE '%/internal'
  )
  SELECT device, metric, value, topic, received_at
  FROM ranked
  WHERE rn = 1
  ORDER BY device, metric;"

Current result:

esp32 humidity 78.0 esp32/humidity/dht11 esp32 temperature 16.1 esp32/temperature/dht11 pico humidity 60.3 pico/humidity/dht22 pico temperature 20.6 pico/temperature/dht22

And ive combined all that:

#cd ./poc/iot-dashboard
cd ./poc/iot-dashboard-v2 #that takes prior version (shown at the tech talk) 
#and adds the watering capabilities via mqtt ./poc/iot-esp-water

More about DRONS

Before going to FPVs, Ive been doing some tinkering around the DJI tello lately.

The Tello communicates through raw UDP:

A browser can use HTTP, WebSocket, WebRTC, and WebTransport, but none exposes an arbitrary UDP socket that can communicate directly with these Tello endpoints.

Thats why I put together a desktop app that works for the Tello:

#ssh -T git@gitlab.com #Edit profile → Access → SSH keys
git clone git@gitlab.com:fossengineer1/dron.git
cd dron
make run-pyqt

Now, for FPVs:

• https://app.betaflight.com/
• https://blackbox.betaflight.com
• https://fpv-logs.pages.dev/
• When you fry a VTX module, you cna actually learn stg https://github.com/JAlcocerT/poc/tree/main/physics-stationary-wave

cd ./poc/fpv-build

Future post ideas in the same drone/firmware lane:

• INAV: Betaflight-adjacent, but more navigation/GPS/fixed-wing/long-range focused.
• ArduPilot: huge OSS autopilot stack, missions, rovers, planes, copters, boats, simulation.
• PX4: professional/research autopilot stack, MAVLink-heavy, SITL/dev workflow.
• Bluejay / BLHeli_S / AM32 ESC firmware: the motor-control layer after Betaflight.
• MAVLink: protocol post tying ArduPilot/PX4/ground stations together.
• QGroundControl / Mission Planner: UI/configuration layer for autopilot stacks.
• INAV Configurator: good comparison with Betaflight App.
• Multi-Protocol TX Module firmware: relevant to EdgeTX/OpenTX radios and non-ELRS receivers.
• OpenHD / RubyFPV: open digital video/telemetry/control stack for FPV.
• Cleanflight / Baseflight history: useful “where Betaflight came from” context.

https://www.youtube.com/watch?v=1lfVKcKQ5BI https://www.youtube.com/watch?v=C0KBu2ihp-s&t=19s https://www.youtube.com/watch?v=lvcPUcZ9wAA https://www.youtube.com/watch?v=BoLeQCqns5A

FPV KPIs

Made recently a battery KPI checks for the drons based on betaflight logs:

Drop a Blackbox log. Get the electrical story: https://fpv-logs.pages.dev/

• https://github.com/JAlcocerT/meteor-dron/blob/main/betaflight-energy-check/z-hovering-learnings.md
• https://github.com/JAlcocerT/meteor-dron/blob/main/betaflight-energy-check/z-motors-learnings.md
• https://github.com/JAlcocerT/meteor-dron/tree/main/betaflight-energy-check/cf-pages

#cd ./poc/fpv-kpis

can you put a brd for an agent to put an interactive html explaining all the concepts we have seen in this conversation?

Done with hyperframe btw and OSS TTS model KoKoRo

After talking about energy/thermodynamics, batteries and aerospace/propulsion/aerodynamics, it was time to land some KPIs for a 5inch dron:

My 5inch dron has

CAD x Design

Design for manufacturing was an interesting book.

But before manufacturing, you have to design:

git clone https://github.com/JAlcocerT/3Design
#cd ./poc/blender/dron-standing-v102

I got my first couple of 3d prints orders done as a code here and here with validations at blender.

Just recently I got to know about: https://sketchforge3d.com which is oss https://github.com/Formsmith746/SketchForge-3D and ive forked here for further exploration at some point.

You can import such STLs and it renders in the browser

Get inspired:

• Weather Vane
• Windturbine
• Interesting box and holder
• meshtastic box or 2- wtf is meshtastic https://meshmap.net/
• solar tracker 1
• solar furnace
• OpenSCad box
• birdcamera
• rc car

FEM x FEA

Yes—the local FEA workflow is operational on this PC: got it done here, after the fpv stand 3d printed

• FreeCAD → BRep reconstruction
• Gmsh tetrahedral meshing
• CalculiX solving
• JSON summaries and VTU visualization output
• Coarse/fine screening cases

FPV Standing x FEA

FEM/FEA example

Local Deep Researcher Repo

Local Deep Researcher with langGraph - Source Code on Github

Upgrading my Desk

I got an IKEA MITTZON desk, with DELL U2518D upgraded to U2724DE with KVM included and ERGOTRON MXV gas desk mount for the VESA 100x100.

How about simulating the tensions?

DIY Solar Tracker

To better be sure that winds dont take away my solar panel :)

Some time back https://solar-trajectory.pages.dev/

#cd ./poc/go-solar-trajectory
#npm run build
#npx wrangler pages deploy dist --project-name=solar-trajectory

CFD

1. Can we predict lift?
2. Prop Wash science
3. Robotics meets Fluid Dynamics: A Characterization of the Airflow below a Drone as a Turbulent Jet
4. Torsional Rigidity test on Formula Student Chassis/Frame using Solidworks FEA | FSAE | DP DESIGN

Generative Design

You can get started with Calculix x Beso

And you have them both in FreeCAD!

Conclusions

The automatic watering setup worked with the 3sx18650 and custom power stage pcb autonomously with these components 3W and 20W with 3 diodes in paralel

But there was a piece missing: adding solar with a CN3791 IP2326 and a voltage divider

For which Im considering IP2326 or CP

So I took this DIY IoT Dashboard and made:

cd ./poc/iot-dashboard-v2

It includes all the goodies, from the sonoff/zigbee integration, the DHT with deepsleep and mqtt for esp32 and also picoW support

Implemented and running at http://192.168.1.2:3038.

• Project: poc/iot-dashboard-v2
• SQLite: poc/iot-dashboard-v2/data/readings.sqlite
• MQTT mappings: poc/iot-dashboard-v2/config/ingest.json
• Documentation: poc/iot-dashboard-v2/readme.md

Key results:

• Dashboard and ingester run together in iot-dashboard-v2.
• SQLite is a bind-mounted file, not a Docker volume.
• Migrated 395,932 historical rows after deduplication.
• Current count is already 395,953 and growing from live MQTT.
• Supports ESP32, Pico W, Zigbee2MQTT sensors, battery/LQI and smart-plug power fields.
• Unknown Zigbee devices are automatically discovered and graphed.
• The old volume contained only ESP/Pico topics—no Zigbee or smart-plug history was available there.
• Lint, TypeScript, production Docker build, SQLite integrity and dependency audit all pass.
• Existing dashboards and ingesters remain untouched and running.

Power data is ingested automatically, but its dashboard panel requires FEATURE_POWER=true.

Starting with this code with such makefile:

cd /home/jalcocert/poc/iot-dashboard-v2
make help
#docker stop iot-ingester iot-dashboard

http://192.168.1.2:3038/?range=90d

Useful examples:

make db-latest
make db-frequency DAYS=7
make db-hourly-presence DAYS=7
make db-hourly-diff DAYS=7
make db-events
make mqtt-live
make health
make ingest-status
make db-query SQL="SELECT COUNT(*) FROM readings"

All SQLite/help targets were tested successfully.

I didn’t modify or restart any containers.

Added poc/iot-dashboard-v2/docker-compose-zigbee.yml. It:

• Runs only Zigbee2MQTT—no Mosquitto or port 1883 conflict.
• Publishes directly to EMQX at 192.168.1.2:1883.
• Maps the Sonoff dongle using its stable device path.
• Uses the correct ember adapter.
• Exposes the Zigbee2MQTT UI on port 8081.
• Persists pairing/network state under data/zigbee2mqtt/.

When ready:

cd /home/jalcocert/poc/iot-dashboard-v2
make zigbee-init
make zigbee-up
make zigbee-logs

Then open: http://192.168.1.2:8081

sudo docker stats iot-dashboard-v2

• SSH client: installed
• SSH server: not installed
• Port 22: not listening
• LAN address: 192.168.1.10

To enable it:

sudo apt install openssh-server
sudo systemctl enable --now ssh
sudo ufw allow OpenSSH

Then connect from another device with: ssh jalcocert@192.168.1.10

Run this on the new server (192.168.1.2):

cd /home/jalcocert/poc/iot-dashboard-v2

mkdir -p data/zigbee2mqtt

rsync -av --progress \
  jalcocert@192.168.1.10:/home/jalcocert/Desktop/Home-Lab/zigbee2mqtt/data/ \
  data/zigbee2mqtt/

Then verify and start:

make zigbee-state-check
make zigbee-config
make zigbee-up #sudo docker network rm da0100acf963
make zigbee-logs

And zigbee UI flows to http://192.168.1.2:8081 but im not using the mqtt function not to collide with my emqx

Ensure Zigbee2MQTT is stopped on the old PC before starting the new instance. The frontend will be at:

http://192.168.1.2:8081

does any of the makefile sqlite clis shows whats the last readings per sensor?

make db-latest

Latest values for one sensor:

make db-latest DEVICE=zigbee.snzb_02d_laptop
make db-latest DEVICE=zigbee.forta_igz_k_plug
make db-latest DEVICE=esp32-dht11-01
make db-latest DEVICE=picow-dht22-01

For the newest raw rows across all sensors:

make db-tail LIMIT=20 #db-tail refreshes every two seconds

https://github.com/JAlcocerT/poc/blob/main/iot-dashboard-v2/z-learnings-migration.md

https://github.com/JAlcocerT/poc/blob/main/iot-dashboard-v2/docker-compose-zigbee.yml

FAQ

This (\sigma(z) = \frac{1}{1 + e^{-z}}) is an inline expression.

then

$$F(\omega) = \int_{-\infty}^{\infty} f(t), e^{-j \omega t} , dt$$

more

$$ \begin{aligned} \nabla \cdot \mathbf{E} &= \frac{\rho}{\varepsilon_0} \ \nabla \cdot \mathbf{B} &= 0 \ \nabla \times \mathbf{E} &= -\frac{\partial \mathbf{B}}{\partial t} \ \nabla \times \mathbf{B} &= \mu_0 \left( \mathbf{J} + \varepsilon_0 \frac{\partial \mathbf{E}}{\partial t} \right) \end{aligned} $$

$$ \begin{aligned} 2x + 3 &= 7 \ 2x &= 4 \ x &= 2 \end{aligned} $$

The logistic function: $$\sigma(z) = \frac{1}{1 + e^{-z}}$$

The Fourier Transform: $$F(\omega) = \int_{-\infty}^{\infty} f(t), e^{-j \omega t} , dt$$

Analogies

Heat and Electr can be compared, but so can fluids

If you can visualize water flowing through a pipe, you can understand a circuit or a heat exchanger.

In all three systems, Power is calculated by multiplying the “Push” (Across variable) by the “Flow” (Through variable):

• Electric Power: $P = V \cdot I$
• Thermal Power: $P = \Delta T \cdot q$ (Though $q$ is already a power unit, Watts)
• Fluid Power: $P = \Delta P \cdot Q$

Prep ULM x PPL

After watching some motivating videos about ppl who can already fly

I decided its time to get LLMs to help me with the prep of the ULM exam

The static version: https://ulm-ppl-test.pages.dev/

cd ./poc/ulm-ppl

And I got a PWA ready: https://jalcocertech.com/ulm-ppl

#git init
claude --dangerously-skip-permissions -p """i want to prepare for the ULM and PPL flight exams and have thought to make a simple web app that sources qna from a markdown
with the options and will help me prepare for the exam, if i fail, it should show an explanation in the end of the reasoning""" #yolo
#.\serve.ps1

Verified on the built dist/ (served locally, headless Edge):

• /ulm-ppl/ renders: 103 questions, 9 banks, 3 generators, Ready to start, no load error
• 200 on /ulm-ppl/sw.js, manifest.webmanifest, questions/manifest.json, assets/icon-192.png
• _headers and _redirects present at dist/ root where Cloudflare reads them
• Site root unaffected — 200, title JAlcocer|Tech Made Simple

Not verifiable locally: the _headers/_redirects rules themselves — the Python test server ignores them; only Cloudflare applies them.

Check after deploy that jalcocertech.com/ulm-ppl (no slash) 301s to the slash form.

Remaining on your side: commit and push, and confirm Rocket Loader is off on the zone.

C:\Users\j--e-\Desktop\ulm-ppl untouched — still the source of truth. When you edit questions there, re-copy questions/ into public/ulm-ppl/questions/.

http://localhost:8080/ oh I mean jalcocertech.com/ulm-ppl