a tool for rocket test days

I built Redline so my test numbers would stop surprising me.

It's a free ground station for rocket static fires. It reads the stand live, warns you when a channel crosses a limit you set, then goes back through the data hunting for the things that quietly wreck an impulse number: a clipped load cell, a drifting zero, a burn that got cut off. If it can't stand behind the result, it says so instead of printing a clean-looking number anyway.

Download Redline Read how it started 20 Hz over USB, Chrome or Edge, no install

Fig 1. Thrust vs time. The shaded region is the burn window Redline integrates, bounded at 5% of peak.

01on the stand

You open one file and watch the fire.

Plug the ESP32 into a laptop and thrust, chamber pressure, and temperature start streaming in at 20 Hz. You set a redline on each channel; when you're armed and something crosses it, you get a sound and a line in the log, not a popup buried behind a window you'll never look at during a fire.

Your pre-fire checklist lives in the same screen. Every item you tick, every arm, every alarm gets a timestamp, so the mission log writes itself while you're busy watching the engine and not a spreadsheet.

When you stop recording, it works through the data and gives you burn time, total impulse, average and peak thrust, and a motor class, with the thrust curve and the exact burn window it used to get there. No hardware to test against? There's a sim stand built in. Sitting on data from an old fire? Drop the CSV in and skip straight to the analysis.

datasheet rev 0.4

rate20 Hz, USB serial

firmwareESP32, included

channelsthrust, press, temp

load cellssingle or summed

importCSV, US or EU

replay1x / 4x / 16x

outputreport, JSON, PDF

runsoffline, one tab

costfree

02checks its work

Before it classifies anything, it checks its own work.

A clean-looking curve can still be lying to you. The analysis runs a set of checks that come straight out of ways I've watched real data go wrong, then labels the result honestly. It also keeps an assumption ledger, so you can see the force unit it used, the baseline it subtracted, and the burn threshold it picked, along with where each of those came from.

verifiedevery check passed. the number is defensible.

needs reviewsomething's off enough that an engineer should look before you quote it.

not defensiblethe data can't support a class, so it won't invent one.

example flags, pulled from the real checks

load cells disagree. peaks 4,200 / 1,600 N, 2.6x spread. mount imbalance or a bad sensor.

ends at 78% of peak. almost certainly truncated. burn time and impulse are lower bounds.

zero drifted 0 N to 90 N across the record. impulse includes false area.

flat top held 11 samples at the rail. possible clipping, peak is understated.

03how it started

I was off by 2,000 N·s and didn't know why.

Early on, a team at Maryland sent me a CSV from one of their static fires so I could test the importer. I ran it, got a total impulse, felt good about it, then compared it to their own analysis and found I was high by about 2,000 N·s. Consistently. Same direction every time.

Fig 2. Vertical stand. The cell carries thrust plus the motor's weight, which drops as propellant burns.

The cell reads thrust + weight. Integrate it raw and you bank force that never left the nozzle. That was my 2,000 N·s.

Their stand was vertical. The motor sits on top of the load cell, which means before ignition the cell is already carrying the motor's full weight, and I was integrating that weight as if it were thrust. Every second of the burn I was adding force that never came out of the nozzle.

The naive fix, subtracting a constant, is also wrong. The motor gets lighter as it burns, so the weight on the cell drops the whole time. A flat subtraction over-tares the end of the burn and under-reports impulse.

So Redline now measures the weight at both ends of the burn and tapers it in proportion to the impulse delivered so far. Mass flow tracks thrust, so the weight comes off the cell in the same shape the thrust goes on. After that, my number landed within a couple percent of theirs.

That one email changed how I think about the whole tool. The math is the easy part. Figuring out which of your assumptions is quietly wrong is the actual job, so now the tool writes down what it assumed and shows its work, instead of handing you a confident number and hoping.

04dev log

It changes when a real test breaks it.

v0.4

Vertical-stand tare

Rebuilt the baseline after the Maryland fire above. Motor weight now tapers with delivered impulse instead of a flat subtraction.

v0.4

Forensic hash on every report

Each report carries a SHA-256 of the exact file it came from, so two people can confirm they're arguing about the same data.

v0.3

US and EU number formats

A file with European decimals (1.234,5) was silently misparsing. The importer now detects the decimal mark and delimiter on its own.

v0.3

Clipping detection

Added a check for load cells that rail out flat at the top. It was hiding the true peak and quietly lowering the impulse.

If you send me a file that makes it do something dumb, that's the next entry. That's mostly how this thing has grown.

05who

NT

Nico Turienzo Otero

16, high school junior, San Rafael, California

I taught myself to code and I'm teaching myself aerospace, and Redline is where the two ran into each other. There's no lab or company behind this. I built it in my room, tested it against real data from real teams, and fixed it every single time it got something wrong.

It started because I kept reading about student teams pouring a semester into an engine, firing it once, and then rebuilding the numbers by hand from a serial log at one in the morning. You get one fire. The data is all you keep. It felt wrong that the last mile, turning that data into something you can actually trust, was the part nobody had a decent free tool for.

I'm not pretending this replaces the DAQ systems the pros run. It's for student and amateur teams who have a load cell, a laptop, and no budget. It's free, it'll stay free, and it gets better every time someone hands me a file that breaks it.

Nico · nikolai.turienzo.edu@gmail.com

06get it

See it work before you trust it.

Download the tool. It's one HTML file, opens in Chrome or Edge, and runs with no internet. Or if you'd rather watch it work on data you already understand, send me a CSV from a past static fire and I'll send back a full report of your own numbers. If it flags something weird, we'll figure out together whether it's your stand or my code.

Download Redline Email me a CSV free, offline, no account, still improving