We are trying to connect a GF Signet 3-2537-2C-P1 paddlewheel flowmeter to an NCD IoT Long Range Wireless Production Counter Transmitter using the counter’s 5-pin M12 connection:
https://store.ncd.io/product/iot-long-range-wireless-production-counter-transmitter/
The GF 2537 is the three-terminal SSR/pulse-output version, with:
–power- Centre
PW/OUTpulse signal +power
We are powering both devices from the same regulated 12 VDC supply.
NCD M12 wiring
- Pin 1 / brown: counter input negative
- Pin 2 / white: unused
- Pin 3 / blue: counter input positive/signal
- Pin 4 / black: external power positive
- Pin 5 / gray: external power negative
Current wiring
0 V terminal block:
- Power-supply negative
- GF 2537
– - NCD gray, power negative
- NCD brown, counter-input negative
+12 V terminal block:
- Power-supply positive
- GF 2537
+ - NCD black, power positive
- One side of the pull-up resistor
Signal terminal block:
- GF 2537 centre
PW/OUT - NCD blue, counter-input positive
- Other side of the pull-up resistor
We initially used the GF-recommended 10 kΩ pull-up resistor between +12 V and the signal terminal. We have also tested a 4.7 kΩ resistor.
Measurements
With the 10 kΩ resistor:
- Supply voltage: approximately 12 V
- Signal-to-ground with NCD blue disconnected: approximately 12 V
- Signal-to-ground with NCD blue connected: approximately 6 V
With the 4.7 kΩ resistor:
- Signal-to-ground with NCD connected: approximately 8.7 V
This indicates that the NCD input is loading the pull-up circuit, but increasing the available signal voltage did not resolve the counting issue.
GF configuration tested
We have tested the 2537 in both:
TOTALmode, including 100 L per output pulseDIVIDERmode with divider/P-factor settings of 20 and 1
In Divider mode with the factor set to 1, spinning the paddlewheel causes the red output indicator on the GF 2537 to flash repeatedly, indicating that the meter believes it is producing output pulses.
However, the NCD counter does not increment.
NCD configuration tested
We tested NCD debounce settings of:
- 10 ms
- 20 ms
- 30 ms
- 50 ms
None produced counts from the GF output.
The GF documentation specifies a maximum pulse width of approximately 50 ms, while the NCD product page lists a 100 ms minimum detectable pulse width. We found previous NCD forum discussions suggesting that shorter pulses may be detectable by reducing the firmware debounce setting, but that has not worked with our unit.
Manual input test
When we disconnect the NCD blue wire from the signal terminal and manually touch it back onto the energized signal terminal, the NCD registers counts correctly.
We can repeat this several times per second and receive multiple counts. This appears to confirm that:
- The NCD counter is operational
- The M12 cable and blue input are working
- The wireless reporting is working
- The input voltage is sufficient when manually applied
We also tested an alternate NPN-style arrangement by swapping the blue and brown input functions and manually switching the brown side, but that did not register counts, so we returned to the documented NCD pinout.
Questions
- Is our wiring correct for connecting the GF Signet 2537’s sinking SSR pulse output to the NCD production-counter input?
- Does the NCD count a brief active-low interruption when its blue input is normally energized, or does it require a positive-going pulse from 0 V to a higher voltage?
- Is the published 100 ms minimum pulse width a hard hardware/optocoupler limitation, or can the current firmware reliably detect a 50 ms pulse through the debounce configuration?
- What voltage and current are required across the blue/brown counter input to guarantee detection? Is a signal sitting at 6–9 V through a 10 kΩ or 4.7 kΩ pull-up sufficient?
- Is there an input-polarity, edge-selection, counter mode, or firmware setting we may be missing?
- If the 50 ms GF output is fundamentally too short, does NCD recommend a particular pulse stretcher, one-shot module, optocoupler interface, or signal conditioner to convert it into a clean 150–250 ms positive pulse?
At this stage, our leading theory is that the GF output pulses are electrically present but are too short—or are the wrong polarity/edge type—for the NCD input circuitry to register.