Multi-Channel Time Series Generation
Multi-channel generation allows you to create datasets with multiple independent or correlated signals sharing the same time axis. This is essential for simulating multi-sensor systems common in industrial environments.
What Are Multi-Channel Time Series?
Multi-channel time series have: - Single time axis - All channels share the same timestamps - Multiple data columns - Each channel has its own values - Independent or correlated - Channels can be statistically related
Single-Channel vs. Multi-Channel
Single-Channel:
Timestamp Value
2024-01-01 00:00:00 100.5
2024-01-01 00:00:01 101.2
2024-01-01 00:00:02 99.8
Multi-Channel (3 channels):
Timestamp X-Axis Y-Axis Z-Axis
2024-01-01 00:00:00 0.5 0.3 -0.2
2024-01-01 00:00:01 0.6 0.4 -0.1
2024-01-01 00:00:02 0.4 0.2 -0.3
When to Use Multi-Channel
Common Use Cases
Vibration Monitoring - 3 accelerometers (X, Y, Z axes) - Partially correlated (mechanical coupling) - Different vibration characteristics per axis
Temperature Arrays - Multiple temperature sensors in same environment - High positive correlation (similar ambient conditions) - Small variations due to local effects
Process Monitoring - Multiple process variables (temperature, pressure, flow) - Complex correlation patterns - Different units and scales per variable
Environmental Monitoring - Weather station (temp, humidity, pressure, wind speed) - Some correlation (temperature affects humidity) - Independent and dependent variables
Creating Your First Multi-Channel Time Series
Step 1: Add Channels
By default, Phoenix starts in single-channel mode. To enable multi-channel:
- Find the "Channels" section in the sidebar
- Click "Add Channel" button
- Phoenix creates "Channel 1" (your first channel)
- Click "Add Channel" again for more channels
You can add up to 10 channels total.
[Screenshot Required: Channel Tabs Interface] 1. After adding 3 channels 2. Capture: Channel tabs showing "Channel 1", "Channel 2", "Channel 3" with delete buttons 3. Purpose: Show the channel management interface
Step 2: Configure Channel Properties
Click on a channel tab to configure it. Each channel has:
Channel Identification
Channel Name (required) - Default: "Channel 1", "Channel 2", etc. - Customize to match your scenario: "X-Axis", "Temperature", "Pressure" - Appears in chart legend and data export
Unit (optional) - Measurement unit for this channel - Examples: "°C", "kPa", "m/s²", "V" - Used for documentation, not affecting calculations
Base Signal Configuration
Each channel independently has:
Mean Value - Baseline level for this channel - Can be different for each channel - Example: Temp1=20°C, Temp2=22°C, Temp3=19°C
Noise Amplitude - Random variation for this channel - Independent noise per channel (unless correlated) - Example: Clean sensor (noise=0.5), noisy sensor (noise=2)
Trend Slope (optional) - Linear drift for this channel - Independent trends per channel - Example: Sensor 1 drifting up, Sensor 2 stable
Oscillations
Each channel can have multiple oscillations:
- Click "Add Oscillation" within the active channel
- Configure frequency/period, amplitude, phase
- Channels can have different oscillation characteristics
- See Oscillations Guide for details
Step 3: Configure Multiple Channels
Switch between channel tabs to configure each one.
Example: 3-Axis Accelerometer
Channel 1: X-Axis - Name: "X-Axis" - Unit: "m/s²" - Mean: 0 - Noise: 0.1 - Oscillation: Frequency=30 Hz, Amplitude=2
Channel 2: Y-Axis - Name: "Y-Axis" - Unit: "m/s²" - Mean: 0 - Noise: 0.1 - Oscillation: Frequency=30 Hz, Amplitude=1.5
Channel 3: Z-Axis - Name: "Z-Axis" - Unit: "m/s²" - Mean: -9.81 (gravity) - Noise: 0.1 - Oscillation: Frequency=30 Hz, Amplitude=0.5
[Screenshot Required: Channel Configuration] 1. Configure 3 channels as above 2. Capture: Configuration panel for one channel showing all fields 3. Purpose: Show channel-specific settings
Step 4: Time Configuration
Time settings are shared across all channels:
- All channels have the same duration
- All channels use the same sampling frequency
- All channels share timestamps
This ensures synchronized multi-channel data.
Step 5: Add Channel Correlations (Optional)
If your channels should be statistically related, add correlations.
See Channel Correlations Guide for comprehensive coverage.
Quick Setup: 1. Scroll to "Channel Correlations" section (appears when 2+ channels exist) 2. Click "Add Correlation" 3. Select two channels 4. Enter correlation coefficient (-1.0 to 1.0) 5. Repeat for other channel pairs as needed
[Screenshot Required: Channel Correlations Section] 1. Add correlation between Channel 1 and Channel 2 with value 0.7 2. Capture: Channel Correlations section showing correlation list 3. Purpose: Show correlation configuration interface
Step 6: Preview Multi-Channel Data
Click "Preview" to generate and visualize.
Multi-Channel Chart Features: - Multiple traces - One colored line per channel - Legend - Channel names with colors - Toggle visibility - Click legend items to hide/show channels - Independent scaling - Y-axis auto-scales to fit all channels
Multi-Channel Statistics: - Statistics shown per channel - Min, max, mean calculated independently - Total point count = points per channel × number of channels
[Screenshot Required: Multi-Channel Chart] 1. Generate 3-channel example from above 2. Capture: Chart showing 3 overlaid signals with legend 3. Purpose: Show what multi-channel visualization looks like
Step 7: Iterate and Refine
- Switch between channel tabs to adjust individual channels
- Preview after each change
- Compare channel behaviors
- Adjust correlations if needed
Step 8: Save or Export
Multi-channel data can be saved or exported just like single-channel:
CSV Export:
Timestamp,X-Axis,Y-Axis,Z-Axis
2024-01-01 00:00:00,0.5,0.3,-9.9
2024-01-01 00:00:01,0.6,0.4,-9.8
Excel/JSON: Includes all channels in structured format
See Export and Save Guide for details.
Channel Management
Adding Channels
- Click "Add Channel" button
- Channel appears as new tab
- Configure channel settings
- Maximum 10 channels
Removing Channels
- Click delete button (×) on channel tab
- Channel is immediately removed
- Correlations involving that channel are automatically deleted
- Cannot undo - re-add if needed
Renaming Channels
- Click on channel tab to activate it
- Find "Channel Name" field
- Enter new name
- Name appears in tab, chart legend, and exports
Reordering Channels
Channels appear in the order created. To reorder: 1. Delete unwanted channels 2. Add channels in desired order 3. Reconfigure settings
(Note: Future versions may add drag-to-reorder functionality)
Point Count Limits with Multi-Channel
Critical: Point limit applies to total points across all channels.
Calculation
Total Points = Points Per Channel × Number of Channels
Where:
Points Per Channel = Duration (seconds) × Sampling Frequency (Hz)
Examples
Example 1: 3 Channels at 1 Hz for 1 Hour
Duration: 1 hour = 3,600 seconds
Sampling: 1 Hz
Channels: 3
Points Per Channel: 3,600
Total Points: 3,600 × 3 = 10,800 ✗ EXCEEDS LIMIT (10,000)
Fix: Reduce sampling to 0.9 Hz
Points Per Channel: 3,600 × 0.9 = 3,240
Total Points: 3,240 × 3 = 9,720 ✓ Within limit
Example 2: 10 Channels at 0.1 Hz for 10 Minutes
Duration: 10 minutes = 600 seconds
Sampling: 0.1 Hz
Channels: 10
Points Per Channel: 600 × 0.1 = 60
Total Points: 60 × 10 = 600 ✓ Well within limit
Strategies for Staying Under Limit
- Reduce sampling frequency - Lower Hz value
- Reduce duration - Shorter time span
- Reduce channels - Fewer channels
- Optimize for use case:
- Fast events: High frequency, short duration, few channels
- Slow events: Low frequency, long duration, many channels
Common Multi-Channel Patterns
Temperature Sensor Array
Scenario: 4 temperature sensors in the same room
Configuration: - Channels: 4 - Names: "Sensor 1", "Sensor 2", "Sensor 3", "Sensor 4" - Unit: "°C" - Means: 20, 20.5, 19.5, 21 (slightly different locations) - Noise: All 0.3 (similar sensor quality) - Oscillation: All have Period=86400 (daily cycle), Amplitude=3 - Correlations: All pairs at 0.85-0.95 (high correlation)
[Screenshot Instructions: Temperature Array] 1. Configure 4 channels as above 2. Add correlations between all pairs (6 total): values 0.85-0.95 3. Duration: 2 days, Sampling: 0.001 Hz (1 sample per 16.67 minutes) 4. Preview 5. Capture: Chart showing 4 highly correlated temperature traces 6. Purpose: Demonstrate realistic multi-sensor scenario
3-Axis Vibration Monitor
Scenario: Accelerometer on rotating machinery
Configuration: - Channels: 3 - Names: "X-Axis", "Y-Axis", "Z-Axis" - Unit: "m/s²" - X-Axis: Mean=0, Noise=0.1, Oscillation=30 Hz/Amp=2 - Y-Axis: Mean=0, Noise=0.1, Oscillation=30 Hz/Amp=2 - Z-Axis: Mean=-9.81, Noise=0.1, Oscillation=30 Hz/Amp=0.5 (gravity + small vibration) - Correlations: X-Y at 0.3 (some mechanical coupling), others at 0.1 - Time: 10 seconds at 100 Hz
[Screenshot Instructions: Vibration Monitor] 1. Configure 3 channels as above 2. Duration: 10 seconds, Sampling: 100 Hz 3. Preview 4. Capture: Chart showing 3 axes with dominant 30 Hz vibrations 5. Purpose: Show high-frequency multi-axis data
Process Variables
Scenario: Chemical reactor with temperature, pressure, flow
Configuration: - Channel 1: Temperature - Name: "Reactor Temperature", Unit: "°C" - Mean: 85, Noise: 1, Trend: 0.01 (gradual increase)
- Channel 2: Pressure
- Name: "Reactor Pressure", Unit: "kPa"
-
Mean: 150, Noise: 2, Trend: 0.02 (increases with temperature)
-
Channel 3: Flow Rate
- Name: "Inlet Flow", Unit: "L/min"
-
Mean: 50, Noise: 3, Trend: 0 (controlled)
-
Correlations:
- Temp-Pressure: 0.8 (strong physical coupling)
- Temp-Flow: -0.2 (slight inverse)
- Pressure-Flow: -0.3 (inverse relationship)
Time: 1 hour, 0.1 Hz (1 sample every 10 seconds)
[Screenshot Instructions: Process Variables] 1. Configure 3 channels as above 2. Duration: 1 hour, Sampling: 0.1 Hz 3. Add correlations 4. Preview 5. Capture: Chart showing 3 different-scale variables trending upward 6. Purpose: Demonstrate process monitoring scenario
Tips for Multi-Channel Success
Channel Naming
Good Names: - Descriptive: "Front Left Temperature", "Motor Vibration X" - Include location: "Room 1 Temp", "Bearing 2 Accel" - Include axis/direction: "X-Axis", "Vertical"
Avoid: - Generic: "Channel 1", "Sensor A" - Abbreviations: "T1", "CH_A" (unless standard) - Too long: "Temperature Sensor Located in the Northwest Corner"
Setting Different Means
When channels measure the same phenomenon in the same conditions: - Use similar mean values (temp sensors in same room: 20°C, 20.5°C, 19.8°C)
When channels measure different phenomena: - Use appropriate scales (temperature: 20°C, pressure: 101 kPa)
Noise Levels
Same Sensor Type: Use similar noise amplitudes - 3 identical temperature sensors → all noise=0.3
Different Sensor Quality: Vary noise by quality - High-end sensor: noise=0.1 - Standard sensor: noise=0.5 - Low-quality sensor: noise=2.0
Oscillation Configuration
Same Source: Give channels similar or related frequencies - All axes vibrating at motor speed: 30 Hz
Different Sources: Use independent frequencies - Daily temp cycle: Period=86400 - Pump pulsation: Frequency=5 Hz
Using Correlations
When to Use: - Sensors measure related phenomena - Physical coupling exists - Shared environmental factors
When Not to Use: - Completely independent measurements - Different physical processes - Intentionally uncorrelated test data
See Channel Correlations Guide for detailed guidance.
Previewing Multi-Channel Data
Check Each Channel: - Use legend to toggle channels on/off - Verify each channel looks correct individually - Check relative amplitudes make sense
Check Correlations: - Correlated channels should move together - Uncorrelated channels should be independent - Zoom in to verify correlation patterns
Troubleshooting
"Data point limit exceeded" with multi-channel
Cause: Total points = points per channel × number of channels
Solutions: 1. Calculate current total: Duration × Frequency × Channels 2. Reduce any of the three factors: - Shorter duration - Lower sampling frequency - Fewer channels
Example:
Problem: 5 channels × 30 min × 1 Hz = 9,000 points ✓
Add 2 more channels: 7 channels × 30 min × 1 Hz = 12,600 points ✗
Solution: Reduce to 0.75 Hz
7 channels × 30 min × 0.75 Hz = 9,450 points ✓
Channel correlations section doesn't appear
Cause: Need at least 2 channels for correlations
Solution: Add a second channel (click "Add Channel")
Can't tell channels apart in chart
Solutions: 1. Use descriptive channel names (show in legend) 2. Toggle channels on/off using legend 3. Ensure channels have different amplitudes or means 4. Export to CSV and analyze separately
Channels look identical despite different configurations
Check: 1. Verify you're configuring the correct channel (check active tab) 2. Confirm you clicked "Preview" after making changes 3. Check that means/amplitudes are sufficiently different 4. If highly correlated (0.9+), channels will look very similar (expected)
Correlation validation fails
Error Message: "Correlation matrix is not valid"
Causes: 1. Impossible correlation combinations 2. Correlation values outside -1.0 to 1.0 range
Solution: See Channel Correlations Guide for troubleshooting
Wrong channel gets deleted
Prevention: - Pay attention to which channel tab has the × button - Delete buttons appear on each tab - No undo - be careful!
If It Happens: 1. Click "Add Channel" to re-create it 2. Reconfigure the channel settings 3. Re-add correlations if needed
Advanced Multi-Channel Scenarios
Harmonic Analysis
Create channels with harmonic relationships:
- Fundamental: 10 Hz, Amplitude=5
- 2nd Harmonic: 20 Hz, Amplitude=2
- 3rd Harmonic: 30 Hz, Amplitude=1
Use high sampling (100 Hz) to capture all harmonics properly.
Phase-Shifted Channels
Use phase offset in oscillations to create phase-shifted signals:
- Channel 1: Frequency=1 Hz, Phase=0
- Channel 2: Frequency=1 Hz, Phase=π/2 (90° shift)
- Channel 3: Frequency=1 Hz, Phase=π (180° shift)
Useful for simulating multi-phase systems (3-phase power, etc.)
Mixed Time Scales
Combine slow and fast phenomena:
- Channel 1: Daily cycle (Period=86400, Amp=5)
- Channel 2: Hourly fluctuation (Period=3600, Amp=2)
- Channel 3: Fast vibration (Frequency=10 Hz, Amp=0.5)
Requires appropriate sampling frequency (2× fastest oscillation).
Next Steps
Master Correlations
Multi-channel data becomes powerful when channels are correlated: - Channel Correlations Guide - Learn to create realistic multi-sensor data
Understand Each Channel Component
- Oscillations - Add periodic behavior to channels
- Basic Usage - Review single-channel fundamentals
- Sampling and Aliasing - Proper sampling for multi-channel
Simulate Real-World Issues
- Data Degradation - Add quality issues affecting all channels
Export and Use
- Export and Save - Work with multi-channel data in external tools
Multi-channel generation opens up realistic simulations of complex multi-sensor systems. Practice with different channel configurations and correlation patterns to master this powerful Phoenix feature.