# Capacitance based drip logger

I want to know if the following is possible. I need to log the interval between drips at a fieldwork site in a cave without physically interacting with the drip (ie the drip can't strike any surface). I need to do this with the minimum possible current draw since I need to power this rig with a battery/batteries for 6 months between battery changes.

I see that there are IR beam drip sensors out there, but these typically drain 30-50 mA which would drain my battery (6V lantern batteries) within weeks at best. I was wondering if there might be a lower-power way of doing this. One thought that sprung to mind was to try to use the transient change in capacitance as a drip fell through capacitor plates, but I'm not sure how to start, or how to data-log this.

Has anyone heard of something like this?

-
 WHat is missing here is a more accurate physical description of the site restrictions ( if any ), proximity of the drip, interval, rate, duration of data collection, budget for sensor, for logger, for power unit, size and any other requirements relevant. These are ALL trade-offs. Cheapest $100 but short batt life with USB interface. – Tony Stewart Jul 13 '12 at 1:32 From the end of my answer, questions so far unanswered: Max drop speed = ? Max 1D (horizontal) or 2D (area) of drop dispersion at sensor plane. General environment? dark/light, windy/calm. (Assumed dark and calm in cave, but maybe not) Mimimum interdrop period? How big a battery is acceptable? Logging specs (see answer for details ) Do you want/need 1/10/100/1000000 or these (Call me if 1000000 :-) ). How many would the international cave water drop logging fraternity buy in a typical year and what would they be prepared to pay? Other ...? – Russell McMahon Jul 13 '12 at 4:14 ## 4 Answers I need to log the interval between drips at a fieldwork site in a cave without physically interacting with the drip (1) A capacitance based sensor is definitely possible. IC's that allow you to measure capacitance changes far smaller than you require and with automatic self trimming are available at reasonable prices. Example only. AD7150 - datasheet here. As little as 1 femtoFarad detection. Response time is 10 ms but a falling drop may be able to be detected as it falls through the sensor (window). Would need to be investigated. Can be operated "stand alone" without microcontroller support and MAY be able to be made to provide input to logger without other 'conditioning'. (2) IR! However, you can very easily [tm] achieve the time scales you want using IR techniques, and may find this easier. The enabling "trick" is to modulate the IR beam at a low duty cycle at a fast enough rate to not miss any drops. Battery mAh capacity and drop speed are relevant. Assume for now a 5 m/S drop speed and 2000 mAh battery capacity. I'll also assume that the drop path does not diverge horizontally "too much" and that it can be persuaded to fall though a small-ish area circle such that an IR beama can be interrupted noticeably without too much effort. Falling through an IR dark space will help. These assumptions can be modified as desired. For example a 3600 mAh 3.7V mean LiPo battery would very happily exceed the above capacity with enough voltage to run an IR LED based system. . 6 months ~= 4000 hours so average battery drain = 2000 mAh/4000 hours = 0.5 mA. This is 0.5 mA with DC IR drive 5 mA at 10% IR on 50 mA at 1% IR on 5 mA may be marginal in some cases. 50 mA should be more than adequate. At 5 m/s a water drop falls 5mm in 1 millisecond. You can probably detect the beam interuption of a drop over a physical 5mm distance, but if not, the distance and thus the time can be reduced. The IR source has to be on for at least one "transmit period" during the time the water drop is present in the sensor "aperture". At 1 mS and 10% modulation you need a 100 uS transmit period and 900 uS off period. At 1mS and 1% modulation you need 10 uS transmit and 990 uS off. If you want to reduce the 'aperture to say 1 mm you need a frame time of only 200 uS (for a water drop at 5 m/s) 10% on = 20 uS, off = 190 uS. 1% 0n = 2 uA, off = 198 uS. The 2 uS transmit time is starting to become mildly challenging but still very achievable with basic off the shelf componentry. The above requirements can be reduced to the following formula. Tcycle = A/V milliseconds (= pulse repetition period) Ton = D x A / V milliseconds. • D = duty cycle ( 0 <= D <= 1) A = aperture distance (fall distance ) in mm V = drop velocity in m/s If there was a known minimum inter-drop time this could be used to allow the detector to be powered down between drops BUT this should not be necessary. Data logging is "just a matter of doing it". The logger may be integrated with the detecttor in some implementations but is conceptually a different subsystem. You need to specify what you want to log. eg drop to drop time - all data recorded. and/or drop to drop time, saved as a distribution. And or drops per minute / hour/day ... - with what resolution. etc Modern memory capacities are such that logging every even is doable if necessary.As an extreme example, 1 drop per second for 6 months = 4000 hours x 3600 s/hr = 14.4 million drops. 128 MB of memory would allow about 9 bytes per drop. Logging time absolutely to 1 second accuracy over 6 months requires about 3 bytes of data per drop. Doing it incrementally requires less. A 32 kB static RAM or FRAM or Flash memory would handle the task with ease with minimal mean current for the whole logger compared to IR sensor needs with careful design. An off the shelf logger may be able to be used or a custom design could be implemented. Actual concept is very simple. As ever, practical implementation will usually have a few traps along the way but the logger is a simple project. Questions: Max drop speed = ? Max 1D (horizontal) or 2D (area) of drop dispersion at sensor plane. General environment? dark/light, windy/calm. (Assumjed dark and calm in cave, but maybe not) Mimimum interdrop period? How big a battery is acceptable? Logging specs (as above) Do you want/need 1/10/100/1000000 or these (Call me if 1000000 :-) ). How many would the internation cave ater drop logging fraternity buy in a typical year and what would they be prepared to pay? Other ...? - What do you have for a data logger now? You can condition a mic to pickup drops and count each one with filters to prevent double counting. Data loggers are diverse with low power and large storage but a wide range of prices. Sensors are extra. Is there ambient light? I think pulsed IR would work for detecting path interference ( reflection) with 0.1% duty cycle or less. IR detectors do not consume much power. What is the proximity of drip? inches feet? 10ft? – Tony Stewart Jul 12 '12 at 0:56 Also, do you need to record every drop or could you window your sampling (eg sample continuously for 2 minutes every 15 minutes) – spearson Jul 12 '12 at 0:59 YOu are best to treat this as two separate requirements. A Micropower Data Logger say with USB interface and Li-ion rechargeable battery. then (2) a drip sensor for the counter interface in a logger. – Tony Stewart Jul 12 '12 at 1:08 While I'll admit I haven't looked at the numbers, in my mind "micropower" and "usb" do not go together. I like the sampling rather-than-constant-running idea though. – Chris Stratton Jul 12 '12 at 15:43 The USB is only for data extraction and recharging. otherwise it is a micropower data logger that handles all your future needs and will not be cheap.. But if you want cheap, they exist but not low power, so if need low power and cheap and flexible UX, I/O and long operation U need$ more accurate specs to tell us. – Tony Stewart Jul 13 '12 at 1:24

Not capacitive, but here is a paper measuring kinetic energy of water drops with a piezo.

-
A fun paper - but he says that drop impact is not an option. – Russell McMahon Jul 12 '12 at 1:26

You say the drip can't strike any surface, but it's going to have to strike something eventually. And when it does, it will have to make a noise. If you can mount a microphone close to the impact point, then you should be able to make a very sensitive detector that uses very little power.

Using a low power op-amp and comparator, you can make a circuit which wakes up your microcontroller from deep sleep at each drip. The MCU then records the drip and sleeps again until the next drip lands. This will be a much lower power solution than anything involving IR, and you should be able to keep this going for at least a year.

Regarding IR solutions. Is this a completely dark cave?

If so, then 0.5mA should be plenty of current to run the LED. You will probably even be able to make it run on a lower current. No need even to pulse it. To get maximum efficiency use several IR LEDs in series. If the power supply is 6v, and the IR LEDs have a forward voltage drop of less than 1.5v, then you can probably get away with using 3 in series, plus a small series resistor. This way, more of the power goes to the LEDs and less to the resistor. Also, use a lens to focus the light onto the sensor.

If the cave is exposed to the daylight, then you might find that the daylight interferes with your sensor. In that case, it's well worth pulsing the LED so that you can use a high pass filter to detect the difference between the LED light and the daylight.

How tightly spaced are the drips? An IR beam won't be very wide, and so scattered drips (perhaps caused by changing air currents) will miss the beam.

One solution is to use one IR LED and several detectors. That way you essentially have several beams. However, you wil need to use a cylindrical lens to create a flat sheet of light, rather than a beam.

Another solution is to use a beam and bounce it back and forth between two mirrors, creating a web of light that should be able to catch drips over a larger area. This is harder to set up, and would be easier if you were using visible light, so you can see where to place the sensor.

-

Another completely different approach is to use an ultra sensitive air pressure sensor to feel the drip whizzing past.

These sensors have two ports can measure a pressure difference between them. This is convenient because the drip has a higher pressure in front of it, and a slightly lower pressure behind it.

Let the drip fall through a tube to help concentrate the pressure wave. Mount the sensor so that you are measuring the difference in pressure between the top and bottom of the tube (not right at the top and bottom though).

Connect the output to a high pass filter, and a comparator, and you should get a pulse for every drip that comes through.

The down side is that the power consumption is about 2mA. However, you could save power by pulsing these things. If the drip rate is reasonably constant, and doesn't change rapidly, then measure the drip rate for the first couple of drips. Now you can predict roughly when the next drip will arrive. Switch the sensor off until a moment before the next drip arrives, then off again as soon as you detect the drip.

-