1

I have a mechanical water flowmeter with reed switch output and I connect it to my low power, battery powered device to count pulses.

The reed switch is connected to my device via a 15 m long cable and it shorts the pulled up input (1M) of a comparator with hysteresis in my device - detection works okay.

The problem is that once in a year the flowmeter needs to be disconnected and inspected - while the long cable combined with large impedance, humid and electrically noisy environment causes the comparator to trigger and count a lot of false pulses. Even touching the cables triggers counting. Changing the pull up value to 47k works great, but also increases the battery current a lot when the switch is shorted.

Currently the device is powered from 3 V coin battery, the comparator window is set between 1.4 to 1.7 V. I thought about moving the window into a lower voltage range, ie. 0.2 - 0.5 V - is it a good idea? Maybe there's a better solution? Battery life is the top key priority.

I'll only add that it's not my design, I'm just trying to fix someone's work.

Adam
  • 135
  • 2
  • 10
  • A circuit diagram of ALL the components involved is always a good idea. If you have told us about ALL the components the diagram will be simple and easy. Often we find there is "stuff" not mentioned.|| Try adding "some" capacitance to the comparator input. IF there is already capacitance there TELL US SO. | Try adding hysteresis to the comparator. (positive feedback) IF thgere is already hysteresis ... !!!! | Show us a circuit !!! – Russell McMahon Jun 14 '19 at 13:15
  • Is your cable shielded? Touching/flexing the cable suggest a possible triboelectric cause of phantom pulses as well. – glen_geek Jun 14 '19 at 13:44
  • No, unfortunately the cable is not shielded. And I'll only add that the pulse counting device is currently mounted in about 2,500 locations and not a single shielded cable is used, just a typical, two wire cable. – Adam Jun 15 '19 at 12:03

2 Answers2

2

A circuit diagram of ALL the components involved is always a good idea.
If you have told us about ALL the components the diagram will be simple and easy.
However, OFTEN we find there is "stuff" not mentioned.

The term "comparator window" suggests hysteresis but this may not be what you mean.
Just moving the window is unlikely to be a vast help - except if the open circuit input is biased in some manner. The improvement due to lower pullup value MAY be due to harder 'pull' against the input signal (ie lower impedance vs input signal impedance) but this is not certain.

'A LARGER hysteresis window should help - but will probably not fix the problem totally as an open circuit line MAY swing rail to rail. .

Try adding "some" capacitance to the comparator input.
IF there is already capacitance there TELL US SO.

Try adding hysteresis to the comparator. (positive feedback)
IF there is already hysteresis there TELL US SO !!!!

Show us a circuit !!!

Why not disable the circuit and/or short circuit the line when / while the sender is disconnected?

Russell McMahon
  • 147,325
  • 18
  • 210
  • 386
  • Thank you for your comment. Unfortunately I can't share the circuit diagram directly because of signed NDA. The circuit is a typical window comparator with positive feedback and window between 1.4 and 1.7 Volts. The counting frontend consists of a 1M pullup and then a series 1k resistor directly to the comparator 'counting' input. Short circuiting the line causes the current to increase - I would like to avoid that. – Adam Jun 15 '19 at 12:16
1

The best solution is a current limited snubber with a Schmitt Trigger CMOS input, to attenuate line frequency but well below a typical Reed relay current yet well above the pulse rate. Such as 100 Ohm to 1k Ohm pullup instead of 1M

This can be designed for optimum tradeoff of switching current and attenuation of interference, once you characterize all the variables for impedance ratios, current limit and pulse rate. The pullup must be less than the stray impedance from your finger at line frequency which may be a few hundred pF. A LPF may be included.

There is more than one snubber solution, it merely attenuates noise well below signal so that the hysteresis ignores stray high impedance noise. If possible, use STP cable for shielding of Common mode noise.

Tony Stewart EE75
  • 1
  • 3
  • 54
  • 182