redPitaLock

Two-channel PID intensity stabilizer for the Red Pitaya STEMlab 125-14, with a Qt-based remote control GUI.

The system reads photodiode signals on the Red Pitaya's fast analog inputs (IN1, IN2) and outputs corrective voltages on the fast analog outputs (OUT1, OUT2) to stabilize laser intensity through an AOM or similar actuator.

flowchart TB
    subgraph RP["Red Pitaya STEMlab 125-14  (rp-XXXX.local)"]
        direction LR
        IN1["SMA IN1"] --> PID1["PID Ch 1\n(RT thread)"] --> OUT1["SMA OUT1"]
        IN2["SMA IN2"] --> PID2["PID Ch 2\n(RT thread)"] --> OUT2["SMA OUT2"]
    end

    RP <-->|"TCP :5000\ntelemetry + commands"| PC["PC – Qt GUI\nPID on/off · gains · loop rate · plots"]

    style RP fill:#1565c0,color:#fff
    style PC fill:#607d8b,color:#fff

Loading

System diagram (per channel)

flowchart LR
    LASER([Laser]) -->|beam| AOM[AOM]
    AOM -->|beam| PICK{Pickoff}
    PICK -->|main beam| EXP([To experiment])
    PICK -->|sample| PD[Photodiode]

    PD -->|voltage| DIV["Voltage divider\n3.3 → ±1 V"]
    DIV --> RP["Red Pitaya\nADC → PID → DAC"]
    RP --> COND["Op-amp\noffset + gain"]

    RF["RF source\n80 MHz"] --> TTL["TTL switch"]
    TTL -->|RF| MIX["Mixer"]
    COND -->|"IF (DC)"| MIX
    MIX --> BPF1["Bandpass\n63–77 MHz"]
    BPF1 --> PRE["Pre-amp\n+20 dB"]
    PRE --> AMP["Amplifier\n+25 dB"]
    AMP --> BPF2["Bandpass\n63–77 MHz"]
    BPF2 -->|RF drive| AOM

    style LASER fill:#d32f2f,color:#fff
    style AOM fill:#ff9800,color:#fff
    style PD fill:#4caf50,color:#fff
    style RP fill:#1565c0,color:#fff
    style RF fill:#7b1fa2,color:#fff
    style EXP fill:#607d8b,color:#fff

Loading

Optical loop: The laser passes through the AOM; a pickoff sends a sample to the photodiode. A resistive voltage divider scales the photodiode signal to the Red Pitaya's ±1 V input range. The Red Pitaya's PID loop closes the feedback by outputting a corrective voltage through an op-amp conditioning stage.

RF drive chain: An 80 MHz source enters a TTL switch, then a mixer whose IF port receives the Red Pitaya's DC control voltage. The mixed signal passes through a bandpass filter (63–77 MHz), a pre-amplifier (+20 dB), a power amplifier (+25 dB), a second bandpass filter, and finally drives the AOM.

Features

• Two independent PID channels using 14-bit, 125 MS/s fast analog I/O
• Both channels disabled by default -- enable from the GUI
• Adjustable PID gains (Kp, Ki, Kd), setpoint, loop rate, and error sign
• One-click PID autotune using relay-feedback (Astrom-Hagglund) method
• Derivative on measurement (not error) to avoid noise amplification
• First-order low-pass filter on ADC input before PID
• Optional AOM linearization lookup table per channel
• Configurable PID output limits (out_min / out_max)
• Anti-windup integral clamping
• Real-time plots of input voltage, target output, and actual output (pyqtgraph, ~100 Hz)
• Simple text-based TCP protocol (debuggable with telnet)

Repository Structure

redPitaLock/
├── firmware/
│   ├── Makefile              # Build on the Red Pitaya
│   └── src/
│       ├── main.c            # Daemon: two RT PID threads + TCP server
│       ├── config.h          # Default parameters (gains, rates, scaling)
│       ├── pid.c / pid.h     # PID algorithm with anti-windup
│       ├── autotune.c / .h   # Relay-feedback autotune (Astrom-Hagglund)
│       ├── aom_lut.c / .h    # AOM drive linearization lookup table
│       ├── analog_io.c / .h  # librp fast analog I/O abstraction
│       └── tcp_server.c / .h # Telemetry streaming + command parser
├── scripts/
│   ├── monitor_rp.py         # PySide6 Qt control GUI
│   └── requirements.txt      # Python dependencies
├── .gitignore
└── README.md

Hardware Setup

Voltage Ranges

The Red Pitaya fast I/O operates at +/-1 V (LV jumper setting), which differs from the original 0-3.3 V Pico setup. External signal conditioning is required:

Signal	Pico Range	RP Fast I/O Range	Recommended Conditioning
Photodiode input	0-3.3 V	+/-1 V (LV)	3.3:1 resistive voltage divider
AOM drive output	0.18-1.25 V	+/-1 V	Op-amp offset + gain stage

The firmware has per-channel input_scale, input_offset, output_scale, and output_offset parameters adjustable at runtime via the GUI or TCP commands.

Connections

Red Pitaya Port	Signal
SMA IN1	Photodiode channel 1 (through voltage divider)
SMA OUT1	AOM drive channel 1 (through conditioning circuit)
SMA IN2	Photodiode channel 2 (through voltage divider)
SMA OUT2	AOM drive channel 2 (through conditioning circuit)
Ethernet	Network connection to PC

Building the Firmware

The firmware is compiled on the Red Pitaya itself (ARM Linux with gcc and librp pre-installed).

# From your PC -- copy the source to the Red Pitaya
scp -r firmware/ root@rp-XXXX.local:/root/redPitaLock/

# SSH into the Red Pitaya (default password: root)
ssh root@rp-XXXX.local

# Build
cd /root/redPitaLock/firmware
make

# Run (requires root for real-time thread priority)
./stabilizer_rp

If the Red Pitaya web apps are running and conflict with the FPGA acquisition/generation, stop them first:

systemctl stop redpitaya_nginx

Running as a Service

To start automatically on boot:

cat > /etc/systemd/system/stabilizer.service << 'EOF'
[Unit]
Description=redPitaLock PID Stabilizer
After=network.target

[Service]
ExecStart=/root/redPitaLock/firmware/stabilizer_rp
Restart=on-failure
Nice=-20

[Install]
WantedBy=multi-user.target
EOF

systemctl enable stabilizer
systemctl start stabilizer

Running the GUI

On your PC:

cd scripts/
pip install -r requirements.txt
python monitor_rp.py                    # default: rp-XXXX.local:5000
python monitor_rp.py 192.168.1.100     # explicit IP
python monitor_rp.py rp-XXXX.local 5000  # explicit host + port

GUI Controls (per channel)

Control	Description
PID ON/OFF	Toggle PID feedback (off by default)
Setpoint (V)	Target photodiode voltage
Kp, Ki, Kd	PID gains
Loop period (us)	Control loop update interval
Error sign	+1 (normal) or -1 (inverted feedback)
AOM linearization	Enable/disable the AOM drive lookup table
PID out min / max (V)	Output clamp range for the PID controller
Reset PID integrator	Zero the integral accumulator
Autotune amp (V)	Relay half-amplitude for autotune excitation
Autotune	Start/cancel relay-feedback autotune; shows live progress

TCP Protocol

The daemon listens on port 5000. Connect with telnet rp-XXXX.local 5000 to debug.

Telemetry (server -> client, ~100 Hz)

D <ch> <input_V> <target_out_V> <actual_out_V> <setpoint_V> <enabled>
AT <ch> <crossings> <measured_cycles> <elapsed_s>    # while autotune is running
A <ch> <Ku> <Tu> <Kp> <Ki>                           # autotune completed
AF <ch>                                               # autotune failed (timeout)

Example:

D 0 0.4832 0.7521 0.7521 0.5000 1
D 1 0.3210 0.0000 0.0000 0.5000 0
AT 0 4 1 2.3
A 0 12.3456 0.0234 5.5555 284.3210

Commands (client -> server)

SET <ch> enabled 1              # enable PID
SET <ch> enabled 0              # disable PID
SET <ch> setpoint 1.500         # set target voltage
SET <ch> kp 4.0                 # proportional gain
SET <ch> ki 2.0                 # integral gain
SET <ch> kd 0.04                # derivative gain
SET <ch> loop_rate 1000         # loop period in microseconds
SET <ch> error_sign 1.0         # feedback polarity (+1 or -1)
SET <ch> use_lut 1              # enable AOM linearization
SET <ch> in_scale 3.3           # input calibration scale
SET <ch> in_offset 0.0          # input calibration offset
SET <ch> out_scale 1.0          # output calibration scale
SET <ch> out_offset 0.0         # output calibration offset
SET <ch> out_min 0.0            # minimum PID output (V)
SET <ch> out_max 1.0            # maximum PID output (V)
SET <ch> reset 0                # reset PID integrator
SET <ch> autotune 1             # start relay-feedback autotune
SET <ch> autotune 0             # cancel autotune
SET <ch> autotune_amp 0.5       # relay half-amplitude (V)
SET <ch> autotune_hyst 0.005    # hysteresis band (V)
GET <ch> params                 # query all parameters

PID Algorithm

• P term: Kp * error
• I term: Ki * integral, with clamping to +/-integral_max and anti-windup (integral freezes when output saturates in the same direction as the error)
• D term: Kd * d(measurement)/dt (derivative on measurement, not error, to avoid kick on setpoint changes and reduce noise amplification)
• Input filter: first-order IIR low-pass on ADC before PID (INPUT_LPF_ALPHA)
• Output clamped to [out_min, out_max] (default 0-1 V)

When the AOM LUT is enabled, the normalized 0-1 PID output is mapped through a 13-point piecewise-linear curve to the physical AOM drive voltage (~0.18-1.25 V).

Autotune

The autotune uses the Astrom-Hagglund relay-feedback method:

1. Enable PID and set a valid setpoint
2. Click Autotune -- the PID is temporarily replaced by a bang-bang relay
3. The relay forces the output to oscillate around the setpoint, swinging between center + amp and center - amp (center = midpoint of out_min/out_max)
4. After discarding 2 settle cycles, the firmware measures the oscillation period (Tu) and peak-to-peak amplitude (a) over 5 full cycles
5. Ultimate gain and PI gains are computed:
   • Ku = 4d / (pi * a) where d = relay amplitude
   • Kp = 0.45 * Ku, Ki = 0.54 * Ku / Tu
6. Gains are written into the PID and normal control resumes
7. The GUI spinboxes update automatically with the computed gains

The autotune respects error_sign so it works with both normal and inverted plant polarity. A 10-second timeout aborts if no oscillation is detected.

Origin

Ported from stabilizerPi, a Raspberry Pi Pico-based single-channel laser intensity stabilizer.