<This design is currently in draft status - There are some minor modifications planned for the final implementation - Check back later>
Pretext: I’ve had my ATO fail while on holidays. This caused the sump return chamber to completely evaporate leading to a subsequent failure of the return pump. After fixing the ATO and replacing the return pump, the ATO did what it was supposed to do and pumped 10 gallons of saturated Kalk water (pH 12.4) back into the system. The result was a pH spike reaching upwards of 9.5. While I was able to recover from this without a tank crash and better yet, without losing any wildlife, it drove home the importance of a reliable ATO. In my opinion, it is one of the more critical components and can potentially fail without warning.
Requirements: In keeping with the guiding principles, this aspect of the system has to be completely redundant all the way through from the breaker panel to the tank. It needs to be constantly monitored and should be able to identify and alert a failure event. In the event of any single failure, it should continue to operate without impact the tank.
Addition considerations include:
• RO units are not very efficient when run for a short duration. Therefore, the ideal situation is to fill the reservoirs in one cycle of the RO unit.
• The Tunze osmolator water pumps seem to lose their prime when they run dry. I’ve often had to suck water into the pump to get it started. Perhaps it’s just my pump, but the system needs to prevent these pumps from ever running dry.
• Kalkwasser is to be dosed through the ATO system, so there should be some safety measures in place.
• The return chamber of the sump does not hold sufficient water volume to survive a failure of the ATO system for an extended period of time. Making this chamber bigger would have meant sacrificing space elsewhere.
Diagram Conventions: The following table depicts the conventions used in this document.
System Schematic: The design I came up with is probably overkill if the ATO was the only consideration. However, many of the redundant systems utilized by the ATO are also being used by other systems.
The Solenoid Valves: The solenoid valves are closed by default. They are opened by the Apex controllers and solve the problem of an inefficient RO unit during the first few minutes of operation. It’s not a question of water high in TDS making it into the tank as the DI resin takes care of this. It’s really a cost savings measure designed to protect the RO membrane and the DI cartridges.
The easiest way to control the filling of the reservoir would be to have the sensors that are installed in the reservoir open the solenoid. It’s easy enough with the Apex controller to open the solenoid for a specified period of time (say 1 hour) whenever it detects a sensor event. However, I don’t really trust these sensors enough to completely rely on them. I am comfortable using them as a backup but I want a more reliable means of refilling the reservoir on a routine basis.
The solution is really simple. Each day at a pre-set time and for a pre-set duration the solenoid opens and the reservoir is allowed to fill. The amount of time required to refill the reservoir (assuming it is nearly empty) is estimated and a safety factor added. While the reservoir is filling, the Tunze ATO is shut off so that once full, no water is removed from the reservoir until the cycle is completed. The RO units are equipped with shutoff valves and will automatically stop when once the resevoir float valve creates sufficient back presure.
Another benefit of filling the reservoir based on time is that if the float valve in the reservoir were to fail open, the water damage would be limited. The solution provides a double shut off. The first being the float valve, and the second being elapsed time. I could install a high water sensor in the reservoir as a third safety, but I’m not sure that it is necessary.
The Primary and Secondary ATO systems are programmed to fill the respective reservoir at different times during the day. In the event that the rate of evaporation is higher than anticipated on any given day, the sensors in the reservoir will open the RO solenoid and fill the reservoir as would otherwise be expected.
The final benefit of a timed system is that the reservoirs would never randomly start filling at the same time.
Controlling pH: The use of Kalkwasser in the ATO necessitates a brief discussion of pH. The tank will have a natural pH swing based in part on the light cycle and the consumption of CO2 through organic processes. Using the Apex controller and pH sensor, we can manipulate the natural swing a bit.
NSW (Natural Sea Water) has a pH of between 8.0 and 8.3, but is generally estimated at 8.2. For a variety of reasons, I will target 8.2 as my “ideal” pH. I will consider of range of 8.1 to 8.3 to be within the acceptable ideal. I will tolerate a swing of 8.05 to 8.35 before trying to further adjust the system.
It is suggested that minimizing the pH swing is more important than hitting a precise value. For example a pH swing from 7.9 to 8.1 is probably better than a pH swing of 8.0 to 8.4 even though the average is closer to ideal.
• The Kalkwasser doser will always be operational except where the pH rises above 8.3. There is no consideration for night and day. Only the pH of the system is considered. If the pH rises above 8.3 the controller will stop operation of the secondary ATO system and only the primary ATO will function. However, If any of the low water level alerts are triggered, the pH condition will be ignored and the secondary ATO will be turned back on. I figure it’s better to allow Kalkwasser to be dosed without regard to pH, than to have the entire system fail.
• If the pH rises above 8.2, the Calcium reactor will be activated. This system is independent of the ATO system, but will serve to lower the pH and keep it closer to ideal.
• The Kalkwasser mixer and Calcium reactor operating in tandem provide a number of benefits which we can discuss in another section.
The obvious question at this point should be “Do you trust you pH measurement instruments enough to rely on them”?
There is a 2 part answer to this question.
1) It doesn’t really matter if they are completely accurate as it is more about the swing then the precise pH value. I consider it an acceptable error if the pH is actually swinging from 8.0 to 8.2 rather than the target of 8.1 to 8.3.
2) I will have some insight into the calibration and operation of the pH probes because I will have two pH probes connected to two separate Apex controllers. Once calibrated, these probes should have identical readings. If either probe produces readings outside of an expected baseline, I will know further investigation is warranted.
Water Lines: The difference between the high water mark and the low water mark does not have to be very big. It should be greater than the rate of evaporation between the two water levels such that the primary pump is not activated during a kalk mixing cycle (1h 10m) or a reservoir fill cycle (up to 2h).
Primary ATO: The primary ATO holds the water level at the ‘Low Water Line’ and is always on. The only exception is while the reservoir is being filled, which is programmed to occur an hour before the lights come on. This time was selected because of the predictably low pH at this time.
This is pretty standard stuff here. The only extras are the float sensor attached to the Apex one of which is in the sump and the other of which is in the reservoir.
Secondary ATO: The secondary ATO holds the water level at the ‘high’ mark and is turned on and off according to the following logic:
• Kalk Mixer On (10m) – Tunze Osmolator Off (1h 10m)
• Kalk Mixer is On approximately 10m every 4 hours.
• The Tunze Osmolator is off during a reservoir fill cycle.
• The start time for a reservoir fill cycle is programmed to coincide with a kalk mix cycle, but is estimated at 2H.
What I’m trying to avoid here is pushing water through the Kalkwasser mixer and into the tank while the mixer is in operation. The kalk needs about an hour to settle after it has been mixed.
The reservoir is filled just before the lights go off. This time was selected because of the predictably high pH at this time.
Dual ATO: The two independently operating ATO systems provide a backup to each other. The system can withstand any single failure without crashing the tank.
Limitations: The systems not perfect. There are some failure scenarios that warrant a note. Although unlikely to occur, there is always a possibility.
• If either of the Kent Marine Float Valves fail, there is a potential that the reservoir will overflow during a scheduled fill cycle. To mitigate this, the overflowing water would pool in the sump area of the stand. This area is sealed for leakage and should hold a significant amount of water before spilling onto the carpet. A water on the floor alarm could also be triggered.
• The RO/DI units could spring a leak before the check valve. They are installed in the furnace room which has a floor drain. I am confident that water would be contained in this area.
• The John Guest fitting could fail at the reservoir. This could potentially be bad, but is reasonably unlikely to occur. The damage would be limited to my basement carpet and the tank should be just fine.
• If the primary ATO fails for any reason, then the secondary ATO is forced into an on state. This effectively disables the pH controls. However, the rate of kalkwasser addition to the tank is still limited by the rate of evaporation and the Calcium reactor would still be on. I don’t anticipate a significant pH spike.
• Larger reservoirs might make things a little easier and reduce the margin for error. However, they would also take up more space which is really at a premium under my tank.
• It would be nice if the two Apex controllers were aware of each other. Unfortunately, they are not.
Feed Cycles: I’d like to be able to stop the return pumps for feeding. I use a check valve on the return lines to limit the amount of water that back siphons into the sump. Don’t worry, the check valves are not safety features, rather just a convenience factor (the sump can hold the water if need be). However, the water level inevitably changes with the small volume of extra water and the high level alarm goes off. The other problem is when the pumps start up the water level in the sump drops while the overflow stabilizes and the ATOs start trying to compensate.
The solution is to shut off both of the Tunze Osmolators during the feed cycle and turn them back on 10 minutes after the feed cycle completes. The Apex controller has this feature built in.