Building a Servo Valve for Cryogenic Rocket Propellants

Designing for cryogenic temperatures means everything you know about standard components stops working. Seals shatter, metals contract, sensors fail. Now try solving all those problems in a 4" x 4" package. That was my challenge designing Phoenix's valve system at UCSD's rocket lab.

This size constraint shaped everything. Most cryogenic valves look like plumbing nightmares - they need long stems to handle thermal gradients, vented balls to prevent pressure locks, and bulky seals that won't shatter in the cold. The commercial ones we looked at were massive, some bigger than our entire rocket diameter. However, most of our rocket was already designed with limited room for valves, so we needed to do something clever to make this work.

A cryogenic valve needs to:

• Open and close reliably under high pressure (600 psi)

• Not leak explosive propellants

• Tell us exactly where it is at all times

• Not shatter when hit with -297°F LOX

• Fit in a 4" x 4" box

Previous teams tried solving this by designing a case that went around the valve, adding a magnet to the valve stem, and integrating external Hall effect position sensors near the magnet to track rotation. This seems reasonable, until you consider thermal contraction. Magnetic field strength drops with the cube of distance. Even after adding heaters, this design would have likely failed since the case was a box which would have deformed when exposed to cryogenic propellants.

I realized that we needed a more robust system before launch. To start with, I found a Sharpe 50M76 stainless ball valve that was almost small enough. It wasn't rated for cryo, but its simple construction gave us something to work with. The real challenge was figuring out how to actuate and track its position reliably.

The valve specs showed we needed 75 kg-cm of torque:

• Base torque 50 kg-cm from pressure

• Cryo multiplier 1.5x

• Safety factor 1.0x for pre-launch cycling

Unfortunately, we couldn't just grab a bigger motor due to the size constraint. We had to be smarter. I found an unlikely solution - a RC servo meant for robot arms. The DS3235 35KG had built-in position feedback through a potentiometer, 35 kg-cm of torque in a tiny package, and 270° rotation range.

35 kg-cm wasn't enough, but it could be with gears. However, because the servo couldn't spin continuously, we needed to be wary about limiting the range of rotation.

• Needed 90° valve rotation

• Had 270° servo travel

• Required >75 kg-cm torque

• Everything had to stay compact

The solution was a 58:20 gear ratio. This would amplify torque to 101.5 kg-cm (35 * 58/20), allow for 93.1° (270 * 20/58) of valve rotation, keep the footprint minimal, and leave room for adjustment.

The gearing materials mattered. At -297°F, everything shrinks and warps. For the 20-tooth gear, I used an existing brass pinion. I picked Delrin for the 58-tooth gear because its thermal properties work better in the cold (it gets harder and more wear-resistant) and it was easy to laser cut for this prototype. However, future versions may need to be machined out of steel to increase thermal resistance.

I CADed and press-fit a two-piece case that would lock the valve geometry exactly, align gears on pitch diameter, account for thermal shrinkage, and stay rigid under load.

Next, I used an Arduino with a simple script to turn the servo to its maximum and minimum positions on command. I also added a sweep mode, where it would automatically alternate between the two.

#include <Servo.h>

Servo myServo; // Create a Servo object to control the servo
int pos = 0; // Variable to store the servo position
int direction = 1; // Variable to store the direction of the servo sweep

void setup() {
  myServo.attach(A5); // Attach the servo to pin A5
}

void loop() {
  // Sweep the servo from 0 to 270 degrees
  for (pos = 0; pos <= 270; pos += 1) {
    myServo.write(pos);
    delay(15); // Adjust the delay time to change the speed of the servo sweep
  }
  // Wait for 1 second before sweeping the servo in the opposite direction
  delay(1000);
  // Sweep the servo from 270 to 0 degrees
  for (pos = 270; pos >= 0; pos -= 1) {
    myServo.write(pos);
    delay(15); // Adjust the delay time to change the speed of the servo sweep
  }
  // Wait for 1 second before sweeping the servo in the original direction
  delay(1000);
}

We built a water flow test stand to validate the mechanics. Started at low pressure, worked up to 250 psi, monitored for leaks, checked position accuracy, tested emergency shutoff. Our pump failed before hitting the target 600 psi, but even at 250 psi the servo worked well below its torque limit. This suggested the system would easily handle full pressure.

Nevertheless, the design needs several upgrades before flight:

• Drill pressure relief vent in ball

• Replace standard seals with PTFE for cryo

• Consider aluminum casing for thermal matching

• Test with LN2 to verify cryo performance

• Validate at full operating pressure

What made this project interesting was that every decision cascaded:

• Compact valve → needed higher torque

• Higher torque → required gearing

• Gearing → thermal concerns

• Thermal concerns → material choices

• Material choices → new failure modes

The real test will come at 30,000 feet. Rockets have a way of finding any weakness in your design. But that's exactly why I love this kind of engineering. Sometimes the best solutions are just well-executed fundamentals. I'm looking forward to seeing these launch to 30k feet!