reef-pi :: An opensource reef tank controller based on Raspberry Pi.

[Mandelstam]

Funny story, my primary heater failed 2 days ago. Back up heater kept the tank at the 76 its set to do. Luckily amazon had free one day shipping on cobalt neos so had up and running the next day.

I think you could probably accomplish alternating with the timer function, if ability to add a second heater pin.
Put each on say 4 hour timers,
Downside is if both are used equally frequently they will fail around the same time. I get this all the time with batteries in my wireless mouse and keyboard. 1 day mouse dies, next keyboard, when changed at the same time.

Yes, ability to add more heaters would be a great start. And I like the idea with alternating as you probably double the lifespan. And when one fails, yes the second one is probably going to fail soon too, but it gives you some time to go get a new pair.

 

[Ranjib]

Another thing to the wishlist (again I'm sorry if this has already been discussed):

In the ULM thread I came across this post about redundancy and backup systems surrounding heaters and I found it very interesting. And it would be very cool if Reef-Pi could have ways to set up different ways of redundancy and backup systems surrounding this very vulnerable part of our system. In the post below sbidny talks about two different ways of dealing with it, either in a load balancing way or an alternative way. Some may prefer one before the other. If Reef-Pi were able to control multiple heaters it would be cool with an option to choose between load balancing or alternating solution. A way to detect faulty heaters would of course be super valuable also.

https://www.reef2reef.com/threads/t...rstv-investigates.341149/page-20#post-4333504
"Ideally, a backup should be running just as often as your primary, either in a load balancing (both heaters running) or an alternating solution. This way, you know both heaters in a functional state.

Then you just need a way to alert when one of the heaters is no longer working.

There is very little chance both heaters are going to fail at the exact same time (maybe around the same time but not at the exact same time). There is a much greater chance the backup won't work when you need it (having been submerged in saltwater for multiple months/years or improperly configured at some point).

To detect a non-functional heater, you could employ one of a few methods using a controller.

1. Detect a change in the current via the outlet

2. Detect a change in how long it takes to heat the tank

3. If alternating the heaters, detect a fall past a low temperature point while trying to heat with each of the heaters

As for load balancing versus alternating, this comes down to whether heaters fail more often from increased cycling or increased usage. Load balancing decreases the overall usage but not the cycling. Alternating cuts the cycling on each heater by half and also reduces their usage.

I personally choose alternating, but this does usually require more advanced controller programming knowledge."

I love reading through this type of posts :-) . Thank you for sharing your thoughts.
Heater crash are probably the most common reason for tank crash, and reef-pi should definitely let users mitigate this. My thoughts:
- We need to support multiple sensors, multiple heaters . This is a predicate based on which whole redundancy based approaches to reliability works.
- Changes in current draw is not necessarily means heater gone bad. A generic anomaly detection is more important. For example, I have found fascinating things about the heaters in three of my tanks, after monitoring them for almost 3 months. Heaters are really really good at keeping temperature within 78.3 to 78.1. In fact from the temperature graph I can actually spot the time when heater is kicking off, and when its shutting down. I can also approximate the shut off time. This is very stable data, but it does change, depending upon the heater age/power, tank size and ambient temperature. Given tank size & power (wattage)is fixed we can actually machine learn the trend and detect when heater is acting up. I have a feeling, thats gonna give us early warning.
- The UI for a more sophisticated temperature controller will be some work. I think we can instead build and expose a programming capability, which reefers can use to combine multiple sensor data, and make corrective actions (multiple equipment control). Once we have found a pattern that works for us, we can standardize it using the UI.
- For multiple heater system, I'll always vote for all active all the time. i.e. run both heater all the time. Probably get the second heater after 3 months of the first heater.

I am really fascinated by the trends reef-pi is showing for temperature water. I have some similar thoughts about ATO also. I think we can get a lot more stable (less variance for longer period) and reliable (detects anomaly and takes corrective action) by just analyzing past data, and then employing a timer or PID controller based feedback system.

 

[Ranjib]

What is the latest fritzing diagram for current build? I would like to make KiCad version, I would like to play a bit and to make some shield v.1 prototype maybe. What do you think?

https://github.com/reef-pi/DesignFiles
I am not sure if they'll be of use for your purpose. I primarily use them for the breadboard images in reef-pi official docs. I was not able to use them for PCB. I still have to learn it :-(

 

[Ranjib]

What is the latest fritzing diagram for current build? I would like to make KiCad version, I would like to play a bit and to make some shield v.1 prototype maybe. What do you think?

It will be so much awesome if you do that. The combined AIO circuit does not have a fritzing file yet. The best I have is an actual photo of my perfboard, shared in the reef-pi 1.0 release thread. I can cross post it here, if you want.

 

[Ranjib]

Another thing to the wishlist (again I'm sorry if this has already been discussed):

In the ULM thread I came across this post about redundancy and backup systems surrounding heaters and I found it very interesting. And it would be very cool if Reef-Pi could have ways to set up different ways of redundancy and backup systems surrounding this very vulnerable part of our system. In the post below sbidny talks about two different ways of dealing with it, either in a load balancing way or an alternative way. Some may prefer one before the other. If Reef-Pi were able to control multiple heaters it would be cool with an option to choose between load balancing or alternating solution. A way to detect faulty heaters would of course be super valuable also.

https://www.reef2reef.com/threads/t...rstv-investigates.341149/page-20#post-4333504
"Ideally, a backup should be running just as often as your primary, either in a load balancing (both heaters running) or an alternating solution. This way, you know both heaters in a functional state.

Then you just need a way to alert when one of the heaters is no longer working.

There is very little chance both heaters are going to fail at the exact same time (maybe around the same time but not at the exact same time). There is a much greater chance the backup won't work when you need it (having been submerged in saltwater for multiple months/years or improperly configured at some point).

To detect a non-functional heater, you could employ one of a few methods using a controller.

1. Detect a change in the current via the outlet

2. Detect a change in how long it takes to heat the tank

3. If alternating the heaters, detect a fall past a low temperature point while trying to heat with each of the heaters

As for load balancing versus alternating, this comes down to whether heaters fail more often from increased cycling or increased usage. Load balancing decreases the overall usage but not the cycling. Alternating cuts the cycling on each heater by half and also reduces their usage.

I personally choose alternating, but this does usually require more advanced controller programming knowledge."

On this note,
I'll be adding support for sht31-D temperature/humidity probe in reef-pi 2.0. I have a feeling my ATO intake anomaly has to do with these two

 

[Ranjib]

Redundancy is somewhat already in place for this. If a heater were to fail, the tank temp would drop and you would be alerted through the telemetry. You can also view the amount of time the heater has been on for each hour of the last 7 days. You would need a room temperature sensor and a lookup chart for reef-pi to know that it was taking too long for it to heat at a particular ambient temperature. Mine runs much more often if the room temp is 65 than if it is 75.

I observed the same. My room temp is at 72, but I need to confirm

 

[Ranjib]

In my opinion an alarm is not the same as redundancy or a backup system. Many people have some form of backup heater installed, often set to a lower temp so it will only kick in if the main one fails. A disadvantage with that, as the linked post pointed out, is that you rely on the backup to be functioning when in reality you don't really know, as it's rarely used. Therefore, an alternating system or a load sharing system is much safer.

Heaters are a weak point in a reef system so having some extra safety features in place seems reasonable to me. Especially when having the power of a controller to back it up.

In my opinion an alarm is not the same as redundancy or a backup system. Many people have some form of backup heater installed, often set to a lower temp so it will only kick in if the main one fails. A disadvantage with that, as the linked post pointed out, is that you rely on the backup to be functioning when in reality you don't really know, as it's rarely used. Therefore, an alternating system or a load sharing system is much safer.

Heaters are a weak point in a reef system so having some extra safety features in place seems reasonable to me. Especially when having the power of a controller to back it up.

Agree. Just alerting won’t suffice most people need corrective measures, and that measure needs to be executed periodically to ensure it works

 

[Ryan115]

For the simultaneous dual heater redundancy in the mean time you could split the controlled power outlet, or bridge the control pin to two relays and therefore two outlets.

What is the most common heater failure others have experienced in the past? I would say my most common have been complete failure to heat or stuck thermostat.

 

[Ranjib]

For the simultaneous dual heater redundancy in the mean time you could split the controlled power outlet, or bridge the control pin to two relays and therefore two outlets.

What is the most common heater failure others have experienced in the past? I would say my most common have been complete failure to heat or stuck thermostat.

complete failure of heature, and cooking up the tank.

 

[Noahs Home Zoo]

When alternating your heaters do have a perfect overlap for the control limits? Or are you setting them offset? i.e.
Heater 1 Runs For An Hour Then Shuts Off, Then heater 2 turns on for an hour?
VS
Heater 1 lower limit of 78 upper limit 78.75
Heater 2 lower limit of 78.50 upper limit 79.25
in the process of coding mine up and planning the best approach. :)

 

[nele]

It will be so much awesome if you do that. The combined AIO circuit does not have a fritzing file yet. The best I have is an actual photo of my perfboard, shared in the reef-pi 1.0 release thread. I can cross post it here, if you want.

Yes please, also I don't know if you are aware but there are some relay boards (still with just 1 or 2 relays) which are wifi controlled. They have ESP-01 wifi board on relay board. For powerbar this would meen that pi zero is not needed any more, just couple of those boards. If you like I can post here the link. I just ordered one for testing.

 

[nele]

Here are Front layer and back layer files. There is some additional work to add silkscreen layers and to do some fill but PCB will be 38mm (1.5 inch) x 25mm (1 inch) in total :)
If I have time I will finish it tomorrow so it can be produced for testing.

 

[Ranjib]

Here are Front layer and back layer files. There is some additional work to add silkscreen layers and to do some fill but PCB will be 38mm (1.5 inch) x 25mm (1 inch) in total :)
If I have time I will finish it tomorrow so it can be produced for testing.

Nice :-) . This was done using KiCad ? I am wondering if similar things can be done in fritzing. I would love to get handful of these boards printed using oshpark or some similar service ?
I got too busy with something else related to reef-pi. I'll post some details tonight, once i get my basic setup working,

 

[nele]

Nice :) . This was done using KiCad ? I am wondering if similar things can be done in fritzing. I would love to get handful of these boards printed using oshpark or some similar service ?
I got too busy with something else related to reef-pi. I'll post some details tonight, once i get my basic setup working,

KiCad. I found kicad easiest to work with. I've used fritzing diagram posted on reef pi webpage as base for this. Also resistors are hand soldering SMD, saving some space on the board. I will change pcb layout to place audio connector and power supply on board. Is that better or to leave as it is now (just male 2 or 3 pin connectors left)?

 

[Ranjib]

KiCad. I found kicad easiest to work with. I've used fritzing diagram posted on reef pi webpage as base for this. Also resistors are hand soldering SMD, saving some space on the board. I will change pcb layout to place audio connector and power supply on board. Is that better or to leave as it is now (just male 2 or 3 pin connectors left)?

I think pcb mount connectors are better,
thank you :-). You have any idea how much these boards will cost? OSHPark states 2"x2" is $5 USD , double sided. Dont know if there is a minimum number limit.

 

[Ranjib]

KiCad. I found kicad easiest to work with. I've used fritzing diagram posted on reef pi webpage as base for this. Also resistors are hand soldering SMD, saving some space on the board. I will change pcb layout to place audio connector and power supply on board. Is that better or to leave as it is now (just male 2 or 3 pin connectors left)?

Having the i2c pins in the pcb that will be awesome :-). I am working on to support humidity/temperature sensor and LED display/buttons, most of these ICs are i2c based.

 

[Kampo]

My Fresh water light, based on Reef-Pi is done. went together first try, first time using a soldering iron in years, and first time messing with a raspberry pi, but it booted up and worked first go...i'm slightly worried i've used up way to much good luck on this project lol. salt is next...bit more soldering on that one, need to order another raspberry pi for it though since i wasn't sure if this controller would work (it works and its pretty dang slick)

Pics below. disregard the mess of a tank its over its a healthy but plant neglected tank in my fishroom, lights going over a 20in cube in my living room, there will be a matching Reef light for another 20in cube in said livingroom.

 

[Mandelstam]

When alternating your heaters do have a perfect overlap for the control limits? Or are you setting them offset? i.e.
Heater 1 Runs For An Hour Then Shuts Off, Then heater 2 turns on for an hour?
VS
Heater 1 lower limit of 78 upper limit 78.75
Heater 2 lower limit of 78.50 upper limit 79.25
in the process of coding mine up and planning the best approach. :)

I think I would go with just a simple timer so both heaters would run equally long. And if one fails change them both out. When alternating like that I also think you could run them a lot longer before changing them out completely.

 

[Ranjib]

My Fresh water light, based on Reef-Pi is done. went together first try, first time using a soldering iron in years, and first time messing with a raspberry pi, but it booted up and worked first go...i'm slightly worried i've used up way to much good luck on this project lol. salt is next...bit more soldering on that one, need to order another raspberry pi for it though since i wasn't sure if this controller would work (it works and its pretty dang slick)

Pics below. disregard the mess of a tank its over its a healthy but plant neglected tank in my fishroom, lights going over a 20in cube in my living room, there will be a matching Reef light for another 20in cube in said livingroom.

Cool beans :-)
So this means at least some one has tested reef-pi with meanwel driver. The dimming works as expected?
CPU/Memory graph will build up with time.
Do you plan to use any feature other than lighting ? I am working on sht31 series sensors, which will allow ambient temperature and humidity monitoring.

 

[Kampo]

Cool beans :)
So this means at least some one has tested reef-pi with meanwel driver. The dimming works as expected?
CPU/Memory graph will build up with time.
Do you plan to use any feature other than lighting ? I am working on sht31 series sensors, which will allow ambient temperature and humidity monitoring.

Yup works fine. Only issue is 100 percent is off but I think you have a fix for that in 1.1?

The lights are just going to be lights. Only other feature I'm planning on using is putting a camera in them to give me a too down webcam to check health of tank remotely.

I'm playing with the idea if a fishroom controller using this though. What are the chances we can get 2 pca9635s on the same controller?

The humidity and ambient temperature features would be helpful for this controller as well.