Skip to content Where Reefkeeping Begins on the Internet

Personal tools
You are here: Home » Library » Aquarium.Net Article Index » 1298 » Voltage in the Aquarium By Gerry Parker

Voltage in the Aquarium By Gerry Parker

Gerry parker records some tests for voltage in an aquarium, Aquarium Net has numerous articles written by the leading authors for the advanced aquarist

Voltage in the Aquarium

By Gerry Parker

A recent thread in the marine aquaria newsgroups caught my eye. It referred to voltage measurements folks were making in their aquariums, and whether the voltages they measured might be a problem for the tank inhabitants. Since my professional background is electrical engineering, and I have quite a bit measurement equipment at my disposal, I decided to take a look at my tank.

What I've done in this paper is to make some basic measurements with laboratory grade test equipment, and provide the recorded data for public consumption with some commentary and opinions of my own. Individuals can now evaluate the quality of that information, and determine what they feel is the best course of action in their own tank. In the end, I'm not qualified to comment on, nor does this piece attempt to address the effects on the inhabitants of the tank. Still, I think the average aquarist might not have the first idea what's really happening inside his tank- electrically speaking, and I think these results are probably more typical than not.

I've provided the physical data for the tank I studied, followed by the data itself, and finally some general electrical information concerning seawater at 60 Hertz. In the last section, are included some parameters and reasonable conclusions that can be drawn from those.

Physical Configuration for the Tests

Oceanic 110 Gallon tank L = 60 inches W = 18 inches D = 24 inches

The tank is placed between to rooms in a pass-through. The sump is in the built-in cabinet beneath the tank. The tank bottom is 48 inches above the floor. There is no metallic connection between the sump and the tank. All plumbing is PCV.

Light Hood

Formica covered composition board with aluminum tape reflector inside. (3) 2-Tube, normal output, 48 inch fluorescent fixtures 5 inch distance between tubes and water surface

Bulbs used

  • (2) Marine Glow (Hagen)
  • (1) Power Glo (Hagen)
  • (1) URI Daylight
  • (1) Champion Daylight
  • (1) Philips F40DX Daylight


Power Lighting Products, Val-Miser #8G1024W Rapid Start, 2 lamp ballasts


18 inches below the tank, plumbed with 1.5 inch PVC drain and 1.0 inch PVC return


  • ETS style downdraft skimmer
  • (1) Rio 3100 pump (skimmer)
  • (1) Mag-Drive Model 7 pump (sump return)
  • (1) Penn Plax Therma Flow "PC" 100 watt heater

The Rio pump has only a two wire plug (no ground wire). The Mag-Drive uses a three wire plug, but in my normal configuration, I don't use the ground connection.

Water Composition

RO/DI with Instant Ocean to a specific gravity of 1.023.

Additives Used

  • B-Ionic (2 part)
  • Kalkwasser

Approximately 140 lbs Fiji live rock 2 inch Aragonite sand base over plenum

Test Equipment

The data was recorded using the following lab test equipment:

  • HP Infinium Oscilloscope
  • HP 34401A Multimeter (True RMS voltage and current)
  • Fluke 75 Series II Handheld Multimeter

Measurement Method

Voltage Data

The voltage waveforms were measured and stored using the HP Infinium Oscilloscope and HP1106A probes. The probes have a 10 MegOhm, 9 pF input impedance, and were connected to the tank water via a titanium grounding probe. The titanium probe is the standard type used for grounding an aquarium.

The titanium probe was inserted at a top corner of the tank using the supplied silicon suction cup. Approximately 2 inches of exposed titanium is available to be submerged, and all of it was submerged for the tests. The oscilloscope probe was connected to the opposite end of the probe wire; the end that would normally be plugged into a wall outlet. The probe was moved to different locations in the tank, and removed from the tank entirely and placed in the sump to verify that the probe placement had no measurable affect on the recorded waveforms. The titanium probe connection wire, as well as the measurement scope probe wire, was moved and twisted to verify that the wire placement had no measurable affect on the recorded waveforms.

Voltage Waveform File Listing

Image Number

Test Condition


Everything ON (lights, pumps, heater)


Nothing ON


Rio 3100 ON


Mag-Drive Model 7 ON without ground wire connected


Mag-Drive Model 7 ON with ground wire connected


Heater ON


Heater OFF (Thermostat)


All 6 Fluorescent Tubes ON (cleaned)


Outer 4 Fluorescent Tubes ON (cleaned)


Center 2 Fluorescent Tubes ON (cleaned)


Outer 4 Fluorescent Tubes ON (uncleaned)


Center 2 Fluorescent Tubes ON (uncleaned)


Outer 2 Ballasts ON only (no tubes in fixture)


Center 1 Ballast ON only (no tubes in fixture)


Everything ON, with line voltage for phase reference

Current Data

Using the same titanium probe setup as the voltage measurements, current measurements were attempted with two different devices, a Fluke 75 Series II multimeter, and a lab quality HP 34401A Multimeter. The initial measurements attempted were AC current measurements from the tank to ground (house wired). The AC currents induced in the tank proved to be so small that neither the Fluke 75, nor the HP 34401A were capable of measuring them directly.

Next, I connected a 110k ohm resistor (outside the tank) in series with the titanium probe (with the oscilloscope still connected) and connected the other end of the resistor to ground. Using the HP 34401A multimeter, I measured and recorded the AC RMS (true) voltage drop across the resistor, and used that measurement to calculate the AC RMS current from the tank to ground.

Please note that only the devices listed are ON during the measurement. If a device is not listed, it is not connected in any way to AC mains while data is recorded. A value recorded as 35 E -9 is Thirty-five times ten to the minus nine power, or 35 nano-amps.

Current Calculations

Test Condition

AC Voltage (RMS)

Calculated AC current (RMS Amps)

Everything ON

3.90 mV

35 E -9

Nothing ON

0.15 mV

1.4 E -9

Rio 3100 ON

0.20 mV

1.8 E -9

Mag-Drive ON (no gnd)

0.18 mV 1.6

E -9

Mag-Drive ON (w/gnd)

0.20 mV 1.8

E -9

Rio + Mag-Drive On

0.29 mV 2.7

E -9

Heater On

0.14 mV 1.3

E -9

Heater Off (thermostat)

0.14 mV

1.3 E -9


6 Tubes ON

8.0 mV

73 E -9

Outer 4 Tubes ON

8.8 mV

80 E -9

Center 2 Tubes ON

7.2 mV

65 E -9

Ballasts On (No Tubes)

All 3 ballasts ON

0.67 mV

6.1 E -9

Outer 2 ballasts ON

0.71 mV

6.5 E -9

Center ballast ON

0.45 mV

4.1 E -9

Old Power Head

Totally Submerged

29.0 mV

264 E -9

Dry, case touching water

29.0 mV

264 E -9


Measured Induced Voltage Phase

The phase of all voltages induced in the tank were 180 degrees from the supply line phase. The radiated voltage phase from the fluorescent tubes (in air) was 90 degrees shifted from the supply line. The voltage phase at the bulb contacts was in phase with the supply line.

Voltage and Current Surprises

The measured data have caught me by surprise all the way around. I expected much lower voltages, and much higher currents. I even thought the pumps would produce higher currents than anything else, and in fact the opposite is true (with the pumps I tested). By far, the worst offender for tank current (excluding the old powerhead) is the fluorescent tubes, and the least are the pumps. Equally surprising is that utilizing the ground plug on the Mag-Drive reduces the induced voltage by half, without really altering the induced current.

Not really a surprise, but it can be noted from the table of currents that the individual currents don't add up correctly if compared to "everything on". The lights draw current in a very non-sinusoidal way, and the individual measurements exclude phase, so measuring them individually vs. all together really isn't the same, even using True RMS. The lights also interact quite a bit with each other due to proximity.

Hot Wire Connection Only

It has been reported by two separate sources that if only the "hot" side of a pump is connected to power mains, then the measured induced voltage in the tank doubles. I have verified that in the case of the pumps this is true, in that the peak-to-peak voltage recorded by the oscilloscope doubles if only the hot side of the supply voltage is connected. I have also verified that the same thing occurs with the fluorescent light fixtures. Thus, in Voltage Waveform File Listing , Image 9 , the peak-to peak waveform doubles in amplitude to approximately 160 VAC under those test conditions described above. In the case of the lights, the waveform becomes nearly purely sinusoidal, rather than the somewhat distorted waveform in Image 9.

Bizarre Power Head Measurements

Utilizing an old powerhead I had lying around, I did some experimentation to attempt to determine if the pump need be submerged in the tank to induce a heavy voltage, or if simply being near to the tank was sufficient. In the course of these observations, I found that it was the same whether the pump was completely submerged, or if any small part of the plastic case touched the surface of the tank water.

I further tried rinsing and soaking the powerhead overnight in RO/DI water, and then allowing it to dry thoroughly before repeating the experiment. The results were the same.

The implication is that the plastic housing is sufficiently (or has become sufficiently) conductive to transfer the impressed voltage to the tank whether fully or partially submerged.

Salt on the Bulbs

This is apparently not a factor, as the data shows no real difference between Really Encrusted Bulbs, and Very Clean Bulbs.

Some NSW Electrical Facts and Conclusions

Natural Sea Water has a number of interesting electrical characteristics.

It is a conductor, not an insulator. Good insulators (glass and rubber) have conductivities in the range of 1E -12 to 1E -15 [S/m]. Distilled water has a conductivity in the range of 2E -4 [S/m], copper has a conductivity of 5.8 E 7 [S/m], and NSW has a conductivity of 4 E 0 [S/m]. So it can be seen that even distilled water is a weak conductor, compared to "good" insulators like rubber or glass. NSW is even a better conductor, (but is still seven orders of magnitude poorer than copper) nevertheless, we will not expect to measure any real voltage drop from one end of a tank to the other with any type of voltmeter. Another way of saying that is to say that any low frequency induced charge in the tank will be very evenly distributed throughout the tank regardless of the location of the source or any ground connection.

Seawater at low frequencies is not a "good" low-loss dielectric. Only above about 2 GHz does NSW pass the test for a good, low-loss dielectric. One major reason for this is due to the conductivity of the water. The electrons in good, low-loss dielectrics are tightly bound to their atoms. In NSW, electrons can be induced to flow very easily. Therefore, no two parts or components of the tank should behave like capacitor plates with the NSW as the dielectric. This does not preclude the water behaving as a single capacitor plate.

The skin depth of a conductor is defined as the outer layer of a conductor where most of the current flows. This depth is related to frequency and conductivity, and must be recalculated for each material and frequency. The "skin depth" of NSW for 60 cycle energy is about 32.5 meters. In other words, 60 cycle current induced from outside the NSW container will mostly be confined to the outside 32.5 meters of the tank. You can see that this means that all "small" tanks have uniform current density throughout the tank from 60 cycle sources. If you own Seaworld, well, that's a different thing altogether.

The wavelength of a 60 Hz signal in air is about 5,000,000 meters (over 2500 miles). The wavelength of a 60 Hz signal in NSW is about 203 meters. From this, we can conclude that there will be no significant voltage difference due to electrical phase measured across the typical SW tank.

The Intrinsic Impedance of NSW at 60 Hz is

(1 + j) (7.7 E -3) ohms

The Intrinsic Impedance of Free Space (air) is 377 ohms [Real].

The natural conclusion here is that the impedance mismatch at 60 Hz between seawater and air is very high. In normal conditions of signal propagation (TEM, far field), no significant 60 cycle energy would penetrate the air/water interface because of this extreme mismatch (less than 1/10th of 1%). It should be noted that the definition of far field is 3 to 5 wavelengths distance between the radiator and the receiver (8000 miles minimum for 60 Hz). So, in all cases for the SW tank, 60 cycle sources are in the extreme near field. In the near field, radiated energy does not follow the normal conditions for transmission between media. Thus, when we see fluorescent tubes 5 inches away from the water surface impressing an 80 V signal into the tank, we already know that the transmission method is not related to the impedance mismatch between the media.

Some Concluding Thoughts

Regardless of the method by which the energy is being coupled into the tank (and we can debate this further) the voltages are quite real, and quite high for a sufficiently isolated tank. The currents, however, are so small that they very nearly defy measurement with laboratory grade equipment. In all cases tested for this paper, no voltage was detectable in the tank when the titanium ground probe was properly connected to (earth).

So if you now accept this as the normal case for SW tanks, you begin asking yourself if this is a problem. I have discussed this with several educated folks, and we have concluded that the minute currents in the tank, due to grounding with titanium probes, are much less significant than unterminated voltages on the order of tens or hundreds of volts. We could not conceive of a case in nature where the volume of water would consistently be charged to 80 volts, but could very easily see situations where nano-amp currents would flow throughout the media. It was the consensus, then, that it would be better to ground the tank, in that it would more accurately simulate the electric environment found in nature.

This is not to say that when equipment fails (insulation on power cords, for example) that fatal currents cannot be impressed in SW tanks.

/ Articles / 1298

Created by liquid
Last modified 2006-11-23 01:39