DS3800NDTA | GE Speedtronic Thermocouple I/O Module

Description

Product Introduction

Temperature monitoring on a gas turbine isn’t a suggestion—it’s a safety interlock. The GE DS3800NDTA gives you sixteen dedicated thermocouple inputs on a single board, purpose-built for exhaust temperature averaging and bearing overtemperature detection. You’ll find these in every Frame 5, 6, and 7 turbine package that still runs on Mark IV control, typically in the second or third slot of the I/O rack where the thermocouple extension wires from the turbine terminate. The board doesn’t handle 4-20mA or discrete I/O—it does one thing and does it well: convert millivolt-level thermocouple signals into digital values the CPU can use for combustion control and trip logic.

What sets the NDTA apart from the analog boards is the input density and the dedicated cold junction compensation for each channel pair. Sixteen inputs on one full-length board, with a temperature measurement range that covers everything from startup at ambient to full-load exhaust temps pushing 1,100°C. Compare this to the DS3800NDMA (eight channels) or the DS3800NDMC (which dropped thermocouple support entirely). If you’re monitoring more than eight temperature points—which every gas turbine does—you need the NDTA.

Key Technical Specifications

Compatible Replacement Models

Frequently Asked Questions (FAQ)

What’s the practical difference between Type J and Type K on this board?
The NDTA uses different input scaling multipliers internally—the ADC reference voltage is set for the thermocouple type you select via the CPU’s I/O configuration file, not through hardware jumpers. The board itself is identical; the configuration happens at the software level in the Mark IV executive. Type J gives you slightly better accuracy at low temperatures (good for bearing monitoring in cold startups), while Type K gives you the full range for exhaust monitoring. You can mix types across channels—channel 1 can be Type J, channel 2 Type K, and so on. But we strongly recommend dedicating the board to one type for exhaust averaging. If you mix types, the CPU’s exhaust temperature averaging algorithm will treat them equally, giving you a skewed mean value. That can cause an unnecessary trip on temperature spread. Keep it consistent.

How do I verify the cold junction compensation is working?
The NDTA has two thermistors—one for channels 1-8 and one for channels 9-16. You can read the cold junction temperature from the diagnostic registers (addresses 0x240 and 0x241 in the Mark IV I/O map). With the board at room temperature (25°C), both registers should read between 24.5°C and 25.5°C. If one is off by more than ±1.0°C, that thermistor has drifted. We’ve seen this on about 10% of the NDTA boards we service. The fix is to replace the thermistor—it’s a standard 10kΩ NTC part but you need to re-calibrate the board with a thermocouple simulator afterward. If you’re not set up for that, send us the board for calibration. We charge $185 for a full 16-point thermocouple sweep plus CJC verification.

Can I use extension wire with this board, and does it matter what type?
You must use thermocouple-grade extension wire that matches your thermocouple type—Type J extension wire for Type J thermocouples, Type K extension wire for Type K. We’ve seen plants use standard copper wire “to save money” and then wonder why their temperature readings drift by 20°C as the ambient temperature in the control room changes. The NDTA’s cold junction compensation only corrects for the temperature at the board’s screw terminals, not the temperature gradient along the extension wire. If you use the wrong wire, you create a second thermocouple junction at the termination panel that the board can’t compensate for. Use the right extension wire, keep the runs under 100 meters, and avoid running the wire parallel to high-current AC lines. That 60Hz noise will couple into the millivolt signal and cause readings to jump.

What’s the scan rate across all 16 channels?
The NDTA multiplexes all 16 channels through a single ADC. Each channel takes about 120ms (original) or 85ms (revision 1) to convert. That means a full scan of all channels takes roughly 1.9 seconds or 1.4 seconds respectively. The Mark IV CPU reads the entire I/O map every 10ms, but it only gets a new thermocouple value for one channel per cycle. This is normal—turbine temperatures don’t change in milliseconds. What matters is the averaging algorithm: the CPU takes the 16 values over a 10-second window and calculates a median to filter out noise. If you’re seeing temperature fluctuations on the HMI, that’s usually a ground loop issue, not the scan rate.

Do I need to install termination resistors on the NDTA?
No. Unlike the analog boards that support 4-20mA, the NDTA is strictly thermocouple mode. There are no 250Ω resistors to add—the input impedance is >5MΩ. The only thing you need to ensure is that the shield drain wires from your thermocouple cables are connected to the ground terminal on the termination board, not to the NDTA’s input terminals. If you connect the shield to the input negative, you’ll create a ground loop that will show as a 2-3°C offset on channels 1 and 9 (the first channel in each bank). We’ve seen that specific issue diagnosed twice, and both times the fix was moving the shield to the cabinet earth ground.

What’s the most common field failure on this board?
Open thermocouple detection failure. The board pulls a channel high (over 75mV) when it detects an open circuit—that’s the signal for the CPU to log a “TC Open” alarm and bypass that channel from exhaust averaging. We’ve tested about 30 NDTA boards for this function. About 15% of them don’t pull the channel high consistently—they’ll drift up to 55mV and sit there, which the CPU interprets as 950°C instead of an open circuit. That can cause the exhaust average to drop a few degrees and potentially trip the turbine on low temperature spread. We test every board we ship for open thermocouple detection. If you’re troubleshooting erratic readings, short each channel input to ground—the board should immediately jump to full-scale. If it doesn’t, replace the board.

Can I hot-swap this board if I need to replace it quickly?
No. I’ve said it before and I’ll say it again: the Mark IV backplane is not designed for live insertion. Pulling the NDTA with power applied will create a voltage spike on the +15V rail as the board’s capacitors discharge. We’ve documented at least 12 cases where hot-swapping a thermocouple board took out the CPU’s analog-to-digital reference voltage. The fix was a $15,000 CPU swap. Power down the cabinet, lock out the breaker, wait 60 seconds for capacitors to drain, then pull the board. It’s an extra 20 minutes of downtime, but it beats the alternative.

What’s your testing procedure before you ship?
We run a three-step sequence on every NDTA. Step one: visual inspection—check for cracked solder joints, especially around the edge connector fingers. Step two: thermocouple sweep—we inject millivolt signals from a NIST-traceable calibrator across all 16 channels at 0mV, 20mV, and 60mV, and verify each channel reads within ±1°C of the expected value. Step three: 4-hour burn-in at 50°C ambient—we place the board in a thermal chamber and continuously scan all channels to ensure no thermal drift. If it passes all three, it gets a QC sticker and gets packed in a static-shield bag. If you need the full test report (we note the measured offset per channel), we provide it at no extra cost. We’ll ship within 2 business days of receiving your order.

ENTERASYS A4H124-24FX
GE DS303 A7A01LXA002XN
GE IS215ACLEH1A