In the field of condition monitoring and predictive maintenance of industrial rotating machinery, the accuracy of data acquisition, the reliability of the system, and the flexibility of deployment directly determine the effectiveness of fault diagnosis. The MCM-100 and MCM-102 edge computing platforms launched by ADLINK are specially designed for 24/7 continuous vibration signal acquisition. They integrate high-precision Δ - ∑ ADC, flexible trigger mechanism and rich I/O interfaces, which can help equipment maintenance engineers easily cope with the challenges of data fragmentation, poor real-time performance and algorithm deployment in traditional patrol inspection. This article is based on the technical characteristics of the platform, providing a complete deployment and optimization guide from hardware installation, BIOS tuning, simulated front-end configuration, trigger strategy, calibration process to programmable I/O applications, aiming to provide practical technical references for on-site engineers.
Key points of hardware installation and power supply design
1.1 Mechanical installation specifications
MCM-100/102 adopts a compact metal chassis (183 × 110 × 54 mm), supporting both DIN rail and wall mounted installation methods. Special attention should be paid to two points during installation:
Wall mounting direction restriction: Due to the antenna port being located on a specific panel, the document clearly requires that the device can only be wall mounted with the front and rear panels on both sides. It is strictly prohibited to install the top or bottom facing outward to prevent the antenna interface from being subjected to force or the heat dissipation air duct from being obstructed.
Heat dissipation gap: Heating elements such as CPU and PCH are concentrated on the left side and passively dissipate heat through heat sinks. To ensure air convection, at least 5 cm (2 inches) of clearance should be left at the top of the equipment to avoid stacking and installing in a closed cabinet.
1.2 DC power supply and surge protection
The device is equipped with a built-in 6-36 VDC wide voltage input (3-pin pluggable terminal with locking buckle), suitable for car or industrial power supply. But it must be noted that:
Polarity confirmation: The manual emphasizes that "incorrect voltage or polarity may cause equipment damage", and it is necessary to use a multimeter to verify V+, V -, and chassis ground before powering on.
Selection of power adapter: When operating at full load (CPU 100%+graphics load+USB full load), the power consumption can reach 38W or even exceed 40W. It is strongly recommended to use an adapter with 90W or more (can be ordered from ADLINK) to avoid voltage drop and restart in high temperature environments.
BIOS Advanced Settings Optimization
BIOS is the cornerstone of underlying system stability. MCM-100/102 adopts Aptio UEFI firmware, and the following key items directly affect on-site applications:
2.1 Serial port mode configuration (COM1/COM2)
Two DB-9 serial ports support RS-232/422/485 software switching, path: Advanced → Onboard Device Configuration → COM1/COM2 Control. For long-distance sensor communication, RS-485 differential mode is recommended, and it is important to enable HSUART (High Speed UART) in the BIOS to support higher baud rates.
2.2 Power Management and Wake up Strategy
State After G3: Set the power on behavior after power failure recovery (normally on/keep off), suitable for unmanned stations.
LAN Wake: Supports Wake On LAN for LAN # 1/# 2, which can be used for remote wake-up and data transmission.
RTC Wake: It can be set for timed wake-up (fixed time or dynamic delay) to achieve low-power intermittent sampling in conjunction with acquisition tasks.
2.3 Watchdog and System Monitoring
The System Management menu in BIOS allows real-time viewing of CPU temperature, motherboard temperature, input current/voltage, and power consumption statistics. More importantly, the SEMA (Smart Embedded Management Agent) firmware provides hardware watchdog (WDT) functionality that automatically resets when the system or application hangs. It is recommended to set WDT timeout (e.g. 60 seconds) in the production environment through SEMA API and feed the dog periodically in the application main loop.
2.4 Clear CMOS Recovery
If the wrong settings result in the inability to start, the CMOS can be cleared by shorting the CN24 jumper (Pin1-Pin2) on the motherboard to restore the factory default values. This operation is particularly useful during the debugging phase.

Simulation input front-end configuration strategy
MCM-100 provides 4 synchronous sampling BNC inputs, MCM-102 provides 2 channels, ADC resolution is 24 bit Δ - ∑ type, and sampling rate is programmable from 1 to 128 kHz. Front end configuration directly affects signal integrity.
3.1 Differential vs. Pseudo Differential Input
Grounding signal source (such as accelerometer conditioner with low output impedance): Differential mode should be selected to eliminate ground loop noise using BNC's positive and negative common mode rejection capability.
Floating ground signal source (such as IEPE sensor or ungrounded equipment): Pseudo differential mode should be selected, where the negative terminal is connected to ground through a 20 k Ω resistor to provide a reference potential and avoid common mode voltage exceeding the limit.
3.2 Coupling Method and IEPE Incentive
DC coupling: retaining the DC component in the signal, suitable for low-frequency vibration or static strain measurement.
AC coupling: Remove DC bias through a 0.4 Hz high pass filter (-3 dB), suitable for accelerometer signals to avoid amplifier saturation.
IEPE current source: When IEPE (2 mA constant current) is enabled, the system automatically switches to AC coupling and provides 24V compliance voltage, which can directly drive standard IEPE accelerometers. Attention: If the sensor does not require excitation, the IEPE current must be turned off by software, otherwise it will cause measurement errors.
3.3 Anti aliasing filtering and frequency response
The Δ - ∑ ADC is equipped with multi-stage FIR digital filters, providing linear phase and 100 dB stopband attenuation. But the front-end is still equipped with a simple RC low-pass to suppress out of band high-frequency noise. The document provides two modes:
High resolution mode (sampling rate ≤ 52.734 kHz): -3 dB bandwidth of approximately 0.49 × fs, passband flatness ± 0.01 dB (20 Hz~20 kHz).
High speed mode (sampling rate 52.734~128 kHz): The bandwidth is correspondingly expanded, but the noise is slightly increased (65 μ Vrms vs 50 μ Vrms).
Engineers should select an appropriate sampling rate based on the target fault frequency (such as bearing characteristic frequency), follow the Nyquist criterion, and use anti aliasing filters to avoid false frequency component folding.
Trigger mechanism and collection mode selection
MCM-100/102 supports flexible trigger sources (software, external simulation, external digital) and multiple trigger modes to adapt to different working conditions.
4.1 Source of Contact
Software trigger: immediate execution, suitable for simple startup collection.
External simulation trigger: Select any AI channel as the source, and set "below low threshold" or "above high threshold" to generate the trigger. The resolution is 24 bits and the level range is ± 10V.
External digital trigger: Receive rising or falling edge through GPI/O pin, minimum pulse width of 20 ns, TTL level (3.3V). Suitable for external tachometers or proximity switches.
4.2 Trigger Mode Application Scenarios
Applicable scenarios of the mode
Record the complete waveform after the post trigger fault event is triggered for transient analysis.
Pre trigger collects N samples (N ≤ 8k/channel) before triggering, suitable for capturing precursors of sudden impacts.
Middle trigger simultaneously collects specified length data before and after triggering, balancing the information before and after.
Gated triggering is only collected during high (or low) trigger levels and is suitable for signals with varying duty cycles.
Retragger responds to multiple triggers during a single acquisition process and is suitable for continuous shock sequences.
Key constraint: The total number of samples (M+N) for pre trigger and mid trigger must not exceed 8k samples/channel, otherwise the trigger signal will be ignored.
Calibration process and accuracy maintenance
To ensure long-term measurement accuracy, the equipment provides factory calibration constants (stored in EEPROM) and user automatic calibration functions.
5.1 When calibration is needed
Environmental temperature changes exceeding ± 10 ℃
Regular maintenance every year
After replacing the sensor or extending the cable
5.2 Automatic calibration steps
Preheating: After powering on, run for at least 15 minutes to stabilize the internal temperature.
Run software calibration program (called by UD-DASK or MAPS Core API provided by ADLINK).
No external signal source required, onboard reference voltage and digital potentiometer automatically correct offset and gain errors.
Save constants: The calibrated new constants can be stored in the user specific EEPROM area, and the original factory constants are permanently retained and can be restored at any time.
Attention: The typical gain error after calibration is ± 0.15%, and the offset error is ± 0.15 mV, ensuring high-precision DC measurement.
Advanced applications of programmable GPI/O
In addition to AI, the platform provides 2 channels of 3.3V TTL programmable I/O (shared DGND), which can be configured through software to perform one of the following functions (select only one at a time):
6.1 Static Digital I/O
Software polling to read or output levels, suitable for controlling indicator lights, relays, or reading simple states.
6.2 Frequency/Event Counter
Measure the external pulse frequency, up to 4 MHz, and set the effective rising or falling edge. Suitable for speed calculation or flow meter pulse accumulation.
6.3 PWM output
Generate adjustable frequency and duty cycle square waves with a maximum frequency of 4 MHz by setting Pulse_initial-cnt and Pulse_length_cnt. Calculation formula:
text
F_PWM = F_Timebase / (Pulse_initial_cnt + Pulse_length_cnt)
Duty = Pulse_length_cnt / (Pulse_initial_cnt + Pulse_length_cnt)
Can be used to drive external actuators or simulate sensor signals.
6.4 External digital trigger input
As mentioned earlier, it can serve as a hardware trigger source for AI collection, reducing software latency.
6.5 Synchronous sampling clock input
When multiple devices need to synchronously collect data, GPI/O can be configured as an external clock input (frequency 1-128 kHz) to ensure multi-channel phase consistency and no delay error.
Drivers and Software Ecology
Proper installation of drivers is a prerequisite for maximizing hardware performance. For Windows 10 IoT Enterprise, it is recommended to install it in order:
Chipset driver (configured with SATA, USB, etc.)
Graphics card driver (Intel core graphics)
Network card driver (Intel I210)
USB 3.0 driver
I/O controller driver
SEMA kit (including WDT and hardware monitoring API)
AI and DI/O drivers (MAPS Core or DAQ-LabVIEW Plus)
Note: MAPS Core must use version 18.10 or later to fully support MCM-100/102. API calls require administrator privileges.
Common troubleshooting and maintenance suggestions
Based on the precautions in the manual and on-site experience, summarize the following quick troubleshooting checklist:
Possible causes and countermeasures for the phenomenon
Check V+, V -, and GND for polarity reversal or low voltage when there is no response during power on; Ensure 6~36V and within the rated current
Frequent system restarts due to insufficient power (full load>40W), replace with a 90W adapter; Check the power consumption of USB peripherals
AI signal distortion or no signal IEPE not turned on or coupling mode error checking software settings; Measure the BNC center to ground voltage (IEPE should be around 24V)
Invalid trigger source selection error or unreasonable level threshold confirmation trigger channel and voltage range; Digital trigger check pulse width>20ns
USB devices cannot recognize the use of USB 3.0 port to boot and install OS. Cold start is required to first plug in the device and then power it on, or cold restart can be used
CMOS cannot enter BIOS settings, chaotic short circuit CN24 Pin1-Pin2, clear CMOS
