DIY Auto Top Off Design Discussion:



Automation. A wonderful thing. It lessens our daily load of reef tank maintenence. It allows us to take long trips with some confidence that some aspect of the tank will be able handle itself. It can also provide a more stable environment for our tank inhabitants.

I've implemented several (ATO) Auto Top Off designs and have seen quite a few more. I've participating in hundreds of forum discussions, helped others design their own systems, and viewed many great ideas and setups developed by other reefers. I have even developed (and show you how to build here) my own simple reliable float switch as well, that can be used in any ATO system. I would like to provide guidance to help select the right top off system for you. In this page I detail the trade offs of most ATO setups. In the end you can pick what suits you best, although I will make some arguments that some are better than others.




Basic Assumptions

Note: Changes in water level due to evaporation show up in either the display if you don't have a sump, or the sump return section if you do. The discussion assumes you have a sump. All water level detection mechanisms are built relative to the return section of the sump.

Here is a diagram showing the very basic sump with an overflow drain, single baffle sump, and a return pump. Your sump will likely be more complex than this and will include your protein skimmer, refugium and other equipment. This diagram shows the basic water flow of an aquarium with a sump. In this standard design all water level changes due to evaporation show up in the return section.



Design 1:



First lets talk about the Elevated reservoir feed shown below. This is the most basic ATO setup. It doesn't get any simpler or cheaper than this. I've heard many people who use and are happy with this setup.

It solves the basic problem of keeping you from having to add top off water everyday or even for a very long time. The elevated reservoir can be made very large to add additional convenience. This design also keeps your salinity rock solid and doesn't add additional risk of electrical shock to you or your tanks inhabitants. It also allows the return section of the sump to be tiny - saving sump space for other desirable features like a larger refugium.

The number of happy people with these setup are checked by numerous stories of float valve failure. Note also that most float valve manufacturers (Kent, Coralife) do NOT recommend their float valves for use in this type of setup. The main issues with this design are with the reliability of the float valve. If it fails, the entire contents of the top off reservior will be dumped into your sump. This will likely cause one of two catastrophic failure scenarios.

1) Your sump will overflow causing damage to a minimum, your flooring and likely your stand, along with potential leakage to lower floors.

2) Dilution of the salinity of your tank water, killing or stressing every one of your tank inhabitants.

One other negative point to make about this setup is that you have to create or buy water and fill the reservoir manually.

You can mitigate the flood damage risks by designing the reservior sufficiently small such that if all the water were to dump in the sump that it wouldn't overflow the sump. While this does remove one pitfall it limits the benefit of long term water supplementing because the reservior is now small. Aren't we wanting to go on vacation sometime here?

Design 2:


So the next design uses the same float valve but swaps out the reservoir with a direct feed off of an RO/DI filter. The shutoff kit turns off the input to the filter to keep from generating waste water.

This is super convenience. We get all of the benefits of the previous design plus no more filling the reservoir! Sit back and change your filter every 6 months or so - that's it.

The drawbacks for this design are all the same as the previous, but worse.

1) Overflowing sump with worse damage since the supply is no longer limited.

2) Dilution of the salinity all the way to freshwater.

But that isn't all. Thereis a third drawback.

In this configuration, the float valve works by "sipping" water from the source as it is needed. Water slowly evaporates all day causing the RO filter to turn on and off frequenty supplying very small amounts of filtered water each time.

When you generate RO water through an RO filter the initial water is significantly higher in TDS. This is true since the membrane under backpressure with no waste water generation collects a few extra salts that are sitting in the waste partition. These dissolved particles are bled off into the RO output. From the information I have gathered, the first approximate pint is "high" in TDS. This is not the optimum water to top off with. But the setup shown above continually feeds only high TDS water into the sump.

The water is still significan'ty better than tap water, but it certainly isn't the water your filter is capable of generating. In order to make the most use out of it you will need to generate more than 2 gallons or so. Even 2 gallons dilutes this pint of high TDS water across 8 times that amount of very low TDS water. The next solution attempts to solve just that problem.

Design 3:


This time the float valve is swapped out with an electronic water level detection and control system. There are two float switches to control when water is fed into the sump. When the water level falls below the lowest switch, this signals the RO filter to start water generation. The RO generation continues until the water level reaches the upper float switch. The latching relay handles the control of the water source. The space between these switches can be designed to create a significant amount of volume to dilute the initial high TDS water coming out of the RO filter.



This system adds a significant advantage over the float valve control. Water generation is now in controlled batches. Generating 2 or so gallons of water creates the optimum water that your filter is capable of generating.

The downside is very similar to the previous RO feed system. A single failure in: The top float switch, the latching relay, or the solenoid valve will spell the catastrophes already mentioned. Now there are three different places where it can occur and all for different reasons. This system is arguably less reliable than Design 3. Three extra downsides are added this time as well:

1) Salinity swing. That 2 gallons of fresh water that your system adds when needed will make salinity go up and down all day long. This might be acceptable if you have a large tank and 2 gallons (or whatever you setup to produce) is a small percentage of the total water volume.

2) Electronics in the system. There is always some danger of having electronics near or in salt water. This risk is mitigated by the use of low voltage. The drawback to the mitigation is that the 12 volt adapters leak current and will add to the e-bill just by being plugged in (Not a lot but, feel one sometime - it will be warm). I've had my own death defying experiences with electricity and my tank so I am very cautious about what electronics (and how) I put near the tank. Every once in a while I read of someone in a forum using 120V on their float switch. Not only will this reduce the life of the switch is pressents an uneccessary risk to the aquarist (and livestock) Always ask if something can be accomplished without having wires in the water. Reduce these risks how ever you can.

The other side of electronics in saltwater is that the corrosive and wet environment can increase the likelyhood of equipment failure. Leaks in wiring and "sealed" components all can contribute to system failure.

3) Waterfall in the sump. Since the waterlevel now is designed to vary significantly to get the optimum TDS filtered RO water, the trade off is that the sump now will have a waterfall in the return section. Waterfalls create noise, salt spray, and microbubbles - none of which we want in our system.

Design 4:


Design 4 I bring up just for completeness.



It really has no pros compared to Design 1.

This setup adds no advantages and only adds the salinity swing problem, electicity waste and risk, along with extra complexity to a system that doesn't give you anything in return. If you are going to do this, use the float switch instead (Design 1). It is the better alternative of the two.

But if you can't tell so far there is sort of a progression going on here. We will get to bigger and better things as we go.

Design 5:


Design 5 Improves a little on design 3. The design adds a redundant float switch.



The redundant float switch guards against float switch failures that will cause a flood or dilution scenario that I previously mentioned. This is an improvement over Design 3

This design is still has a single line of defense against solenoid, or relay failure. The extra switch does not protect against these types of failures. Also since the switches are identical whatever conditions that caused the first switch to fail are also present to cause the second switch to fail. To mitigate this somewhat a second brand/style of switch might be used.

This design is also still affected by salinity swing, electicity waste and risk, and having a waterfall along with a slight increase in system complexity.

Design 6


Design 6 Improves a little on design 5. The design adds a float valve instead of a redundant float switch. Because of this we have to implement a shutoff kit



The float valve is better than the extra float switch since it's failure mechanism is very different than the float switch lessening the likelyhood of both the top switch and the valve failing together. More importantly, the valve backs up not only the top switch, but also against solenoid or relay failure.

We have to add the shutoff kit which is another chance of failure however it is backed up by the solenoid.

This design is also still affected by salinity swing, electicity waste and risk, and having a waterfall along with a slight increase in system complexity.

Design 7


Design 7 Is a different approach which doesn't depend on gravity to feed the RO supply. A single float switch controlls a submersible pump through a relay. As soon as the water falls a tiny bit, the switch activates the pump to keep the water level constant.



This design is a little simpler than those previous. It doesn't depend on valves. It's a small advantage that the pump failure scenario is to jam or stop pumping in some way which keeps from feeding water. Compare that to the failure scenario for a normally closed solenoid valve - flood and dilution. Also, salinity stays rock solid.

The single float switch creates a new problem that we haven't talked about. Short Cycling. This occurs when you turn a device on and off rapidly. The float switch will hover at the point of contact. As soon as it makes contact, the pump engages and starts filling the return section. Very quickly the return section water level starts rising, opening the switch contacts. This is likely to be a very small amount of water. It will therefore take little time for that amount of water to evaporate and cause the pump to turn on again. This causes the pump to be turned on and off potentially a hundred times a day. This causes wear and tear on the pump and is less than efficient in terms of power consumption.

This effect will be heightened by surface water turbulence if it exists in the sump.

This design still suffers from Electrics/Water risks, Wasted Electricity, Creature interference, and manual filling.

Design 8


This design adds an extra switch to handle the short cycling issue.


The extra switch removes the short cycling issue from Design 8. The water level now varies between the switch positions, creating a fixed pump on off cycle.

This design suffers from all of the other shortcomings of Design 8: Electrics/Water risks, Wasted Electricity, Creature interference, and manual filling.

This effect will be heightened by surface water turbulence if it exists in the sump.


This section is still under contruction.

Design 9


I designed this switch setup. I uses a water tight snap action switch which forces the switch to move a fixed distance before turning on to off and off to on. The switch is connected to a float arm on a pivot which allows the water level to move the arm up and down closing and opening the switch. This system addresses the concerns with most float switch implementations.


The design removes the short cycling issue from Design 8. The water level must move a fixed amount. In my implementation it must move 3/16" which translates to 1 pint of water. I evaporate 1-2 gallons a day, which means 8-16 cycles a day. This strikes a nice balance between providing consistent salinity and having a high pump cycle rate. The float is adjustable up or down for a total of 5". Because the float is the only thing that touches the water, critters really can't get to it and even if they could there is nothing for them to jam or weigh down. Corrosion doesn't effect the float either making it maintenece free. There are no electrical wires in the water (the pump's wire is for now unavoidable). The pump power supply is controlled directly with 120V so no relay or ac/dc adapter is needed, meaning less components to fail and no wasted power. The switch is sealed as well so the corrosive environment is not a concern. The one I chose is reated at 2,000,000 cycles which turns out to 342 years even at 16 cycles a day. There are no parts to fail due to magnetism.

This system is controlled by a single switch, which means single point of failure. However, I do believe that the reliability of the switch setup has significantly improved over other implementations because there are less components, That being said, I am currently trusting system to this ATO setup. Another con is this float takes up more space than a typical magnetic float switch.

This part is under construction as well


Some more diagrams for consideration. (And for me to add descriptions to :))





A Discussion about switches

Standard Switches

Your everyday switch (house wall switch, reed switch, appliance switches, etc.) suffer a few drawbacks from their over simplicity in design. There are just two metal contacts in there. When the switch is depressed, the motion of the switch level (the thing you touch) directly makes the metal plates come in contact with each other. Sure the lever will eventually "snap" to one position or the other but the way the contacts internally work, the contact is made and released from you moving the lever. The key here is that you can back out of the on or off transition at any time. And that is where the main drawback comes from. As the switch hovers near the point of contact, and as the contacts slowly move toward each other something happens before the two conduncting contacts actually connect. Arcing. Yes even the small voltage will cause this to occur. Try it at home on a switched plug with an incandescent bulb. Slowly move the switch from off to on (or the other way - doesn't matter). Resist the switches spring force to flop to the other position. Somewhere in the middle of the transition from off to on you will hear a ZZZZZZZZ sound out of the switch and the lamp will start to flicker. At this point the contacts are not connected and arcing is occuring through the air between them. This is hard on your relay and your pump as well, not to mention somewhat unsafe at high voltages.

Snap Action Switches

They are better and don't suffer from this point of contact. The motion of the lever is designed in such a way that as you reach the middle area of the switch motion the internals are latched over making a solid and secure contact. If you deside to back off of pressure of the switch lever it's too late the internal have already made contact fully. Now you'll have to move the lever back some distance before the internals will snap back to the other position. The point here is it's impossible to get the switch to perform the arcing example that we posed earlier. That protects your electronics by having the motors get a good clean start up voltage, and makes it such that a float switch will actually work in this case.