DS3860HSAA | GE Factory Tested & ISO Certified

Description

Product Introduction

Standard analog inputs sample at 50-100 Hz—fine for temperature and pressure, but useless for vibration or surge dynamics. The DS3860HSAA is the high-speed alternative, capturing rapid transients at 100 kHz per channel. This board is typically found in turbine condition monitoring systems, compressor surge detection, and high-speed bearing vibration analysis where you need to see the waveform, not just the average.

Compared to the DS3860HSA (non-A revision), the “A” revision doubled the onboard FIFO buffer from 4k samples to 8k samples and added a hardware trigger input. The older unit could only stream data continuously—you had to catch the transient in software, which was hit-or-miss. The new trigger input lets you start capturing on an external event (e.g., a trip signal or a speed threshold), ensuring you don’t miss the important 100ms window before and after the event. Terminal assignment is identical, but the trigger pin on the backplane changed—verify your wiring if you’re using the external trigger feature.

Key Technical Specifications

Compatible Replacement Models

Frequently Asked Questions (FAQ)

Q: What’s the actual use case for this board?
Anything that moves faster than 100Hz. Typical applications: (1) Bearing vibration monitoring (up to 10 kHz, but you’ll see 100 Hz to 10 kHz harmonic content), (2) Compressor surge detection (pressure transients in the 50-200 Hz range), (3) Fast temperature transients (e.g., thermocouple response to a fuel trip—we’ve seen 100°C/sec on some hot gas paths), (4) Actuator response testing (measuring the delay and overshoot of a control valve). It’s overkill for standard process monitoring (flow, pressure, temperature)—use the HAFA for that.

Q: The sample rate is 100 kHz per channel. Is that simultaneous or multiplexed?
Simultaneous. Each of the 8 channels has its own ADC (delta-sigma type), so you get a true snapshot of all 8 signals at the same instant. This is critical for vibration analysis where the phase relationship between sensors matters. If the sampling were multiplexed, you’d have time skew between channels—up to 10µs per channel for 8 channels, which would mess up orbit plots and shaft centerline tracking. GE designed this board for simultaneous capture, and we’ve verified it in our lab.

Q: What’s the trigger latency from hardware trigger to the start of data capture?
About 5µs. The hardware trigger input (TTL rising edge) directly gates the FIFO write enable—it doesn’t go through the CPU. This is fast enough to capture transients within a few microseconds. The board also supports pre-trigger capture: you can program the FIFO to store data for X samples before the trigger, so you see what happened leading up to the event. For a 100kHz sample rate, 8k samples is 80ms of data—enough to capture most turbine transients. Post-trigger capture is also configurable.

Q: Can I capture data continuously, or only in burst mode?
You can do both. In burst mode, the board fills the FIFO, triggers a DMA transfer to the CPU, and stops (or re-arms). In continuous mode, it streams data to the CPU at the sample rate—but at 100kHz, that’s 1.6 MB/s (8 channels × 100 kHz × 2 bytes each). The Mark VI’s CPU can handle that over the VME backplane, but it will eat up 15-20% of the CPU’s bandwidth. If you need continuous high-speed data, you’re better off using a separate condition monitoring system (e.g., Bently Nevada) and sending the analog signals to both the Mark VI and the CMS. Don’t try to run 8 channels at 100kHz continuously on a Mark VI—you’ll overload the CPU.

Q: The measured value has a 0.5% offset at 100kHz vs. DC. Is that normal?
Yes. At 100 kHz, the ADC’s settling time and the alias filter introduce a small gain and offset error. GE’s spec is ±0.15% of reading at 100 kHz, so 0.5% is outside spec—you may have a calibration issue or a bad channel. Check your alias filter setting: if you have it set to 50kHz, you’ll get more roll-off at high frequencies. Set it to “off” for DC accuracy, but then you’ll alias high-frequency noise (anything above 50kHz folds back down). For vibration monitoring, we recommend the 10kHz filter—it gives a flat response up to 10kHz and rejects noise above that. At 10kHz, the accuracy is within ±0.05%.

Q: What’s the maximum input voltage before damage?
±30V DC or AC peak. The board has a clamp that limits the input to ±30V, but sustained voltages above 30V will damage the input protection diodes and possibly the ADC. We’ve seen plants accidentally connect 48V DC from a faulty transmitter—the board survived for about 10 seconds before the protection diode shorted, taking the channel out. If you’re connecting to sensors that might produce higher voltages (e.g., piezo-electric accelerometers with built-in charge amps), add an external voltage limiter or a zener clamp at ±12V.

Q: The Trigger LED is flashing amber, but I’m not getting any data. What’s the likely issue?
Flashing amber means the board is armed and waiting for a trigger event. If it doesn’t get the trigger, it’ll sit there forever (or until a timeout if configured). Check your trigger source: (1) If you’re using software trigger, make sure the CPU has issued the “Start” command—it’s not automatic. (2) If you’re using the hardware trigger input, verify the TTL signal is reaching the backplane connector—pin 20 on the auxiliary connector. We’ve seen plants forget to connect the trigger wire. (3) If you’re using a level trigger (e.g., start when channel 1 crosses 5V), ensure the signal actually crosses that threshold. The LED will stop flashing and turn green when the trigger fires.

Q: Can I use this board for vibration analysis without an external antialiasing filter?
The board’s onboard 2-pole filter (50kHz cutoff) is not enough for vibration analysis above 10kHz. For IEC 60034-14 compliant vibration monitoring, you need a 4-pole or higher filter to reject alias frequencies. GE’s standard vibration monitoring uses a separate CMS with its own filters—the HSAA is typically used for transient capture, not continuous vibration analysis. If you’re using it for vibration, add an external 4-pole Bessel filter at 10kHz to prevent aliasing. Without it, any vibration components above 50kHz will fold back into the 0-50kHz range, corrupting your data.

Q: What’s the drift over temperature at 100kHz?
GE specifies 100 ppm/°C for gain drift and 50 µV/°C for offset drift. Over a 40°C range (20°C to 60°C), gain could shift by 0.4% of reading—a noticeable error at full scale. If your plant cabinet varies in temperature, we recommend calibrating the board annually at the actual operating temperature. For critical vibration monitoring, consider placing the board in a temperature-controlled enclosure. We’ve seen plants with 55°C cabinets get 0.3% drift after 6 months.

Q: Is there a limit on how many HSAA boards I can install in one rack?
Yes. Each HSAA board draws about 2.5W from the backplane. The Mark VI’s 5V backplane supply is rated at 50W total. A fully populated rack (10 boards) would draw 25W, well within spec. However, the HSAA’s DMA transfer uses the VME backplane bus heavily—two HSAA boards running at 100kHz simultaneously will use about 30% of the backplane bandwidth. We recommend a maximum of 3 HSAA boards per rack to avoid bus contention. GE’s system design guide advises this—you’ll start getting DMA errors if you exceed it.

Q: What’s the lead time for a surplus HSAA?
We keep 2-3 units in stock. U.S. domestic: 2-3 business days ground, overnight if ordered by 2 PM EST. International: 5-7 days via DHL. This board is subject to ITAR regulations—it’s used in gas turbine control systems, so we need a valid end-user certificate for international shipments. We’ll send you the form—allow an extra 24 hours for the paperwork if you’re outside the US. For domestic orders, no export license is required. We ship with anti-static protection and a detailed test report showing the board’s calibration at 100kHz.

DELTA TAU 602804-109
A-B 2094-BC02-M02-S
PROSOFT MVI56-PDPMV1
ABB DSAI130A