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

[Ranjib]

Ranjib and all, I am not proposing this (see graphic), because it is quite divergent from our path at this time. But, when I was looking at aquarium controllers on the internet, I found this one and it peeked my interest. I captured the URL in the graphic, if you want to look at it.
The dashboard quite explanatory - the URL has an assortment of dashboard screen captures.

I have seen this before. Prior to writing reef-pi I did an extensive google search to see if anything out there would fit the bill, if so I could have rather contributed to that project instead of writing something from scratch. This one was one of the google search results. I like the dashboard UI. I think we'll eventually have something similar, but with different color themes. thanks for sharing

 

[Somnifac]

Do you see a function that handles an import/export of a settings file to be a reasonable addition to the web interface?

 

[Cary]

Copy reef-pi database file (/var/lib/reef-pi/reef-pi.db) to the new sd card , after installing reef-pi in it, and restart reef-pi, that should be it

But should I not update the Raspbian as well, prior to copying the reef-pi database file?

 

[Ranjib]

The warning section would also likely be worthwhile to have on the Dashboard.

Spot on. This is something on my mind as well..

 

[Ranjib]

Ranjib,
I agree that the minimum system health is good to monitor, but may not be "telemetry worthy". What are your thoughts on having reef-pi check the health metrics and have a telemetry section set for errors/warnings alerts. That way there is only something sent if it needs to be brought to the users attention. This could contain many safety checks for all aspects of the system (i.e. extreme temperature, alert from floatswitches that should never be triggered, etc...)
Can telemetry accept text, or only numeric values? If that is the case, maybe a list of error codes and then alerts setup within adafruit io?

Telemetry data can contain string or even pictures. Adafruit has size limit on images which is why I went with google drive integration (still under development) for the camera module.
Telemetry data in most cases is sampled, which mean only a subset of the emitted data is collected and sent to telemetry collector (reading: statsd) . But this is not the case with reef-pi, we send all data to adafuit.io, no sampling.
Regarding the error/warnings, its facinating that you mentioned this. There are dedicated services that allow capturing and analyzing errors (reading: https://sentry.io) , but I think it will be too much for reef-pi. This is something on my mind, and I will have some answer soon. My current thought process is to have notification system (email or sms) in place first. Then track and report error when it goes above certain threshold. Does that sound sane?

 

[Ranjib]

But should I not update the Raspbian as well, prior to copying the reef-pi database file?

You can , it always a good idea (since it brings latest security patches and all new goodies), but it is not required. reef-pi does not depend on anything other than a vanilla linux system.

 

[Ranjib]

Do you see a function that handles an import/export of a settings file to be a reasonable addition to the web interface?

Yes very very reasonable I would say a must have feature for 2.0. Cause without this users will have a hard time upgrading.

You all are using reef-pi while its in active development (or more accurately, even before a stable major release), which is equivalent to "drinking from firehose" , so its bit rough experience. But I do appreciate (and thankful) all the feedback , without your comments and criticism , this project wont get better over time. And I need someone other than me to test it as well. User experience in particular (like the one you just mentioned) is critical

 

[Ranjib]

No worries. When @Ranjib mentioned there was an API (I'm sure it was mentioned earlier in the thread, but I must have missed it) I went diving for the documentation.

I've flown through DFW (I assume, based on your location noted as Allen, TX) many, many times, but it's never been an endpoint to my journey.

API is the key for 1.0 release. Its not documented and fully stable yet, but the reasoning for 1.0 release (or any major release ..) is that the API will not have any compatibilty break for the entire 1.x series.. it will be maintain at least 1 year, at max 4 year (long term support release). It might get new additions, but all thing old will remain and work as it is.

reef-pi 1.0 is the ground work I need before making a reef-pi cluster, where multiple reef-pi controller will be controlled by a single mothership (via API). There are couple of folks in local university and aquarium who wants to use it, but its simply not practical to manage 20 different controller independently.
So, again I'll repeat following are my priorities in terms of reef-pi design & acrhitecture
- Priority 0: A robust and reliable controller that provides fundamental features for equipment control and sensors using opensource software/hardware
- Priority 1: Stable API, user experience., so that users can trust using it for long run ( A typical reef tank runs for decades). This also allows keeping the core system at minimal, while let others to extend it as their requirements.
- Priority 2: Provide minimal but good documentation & visual aesthetics (part of user experience . but lower priority than API or reliability).

 

[Ryan115]

Telemetry data can contain string or even pictures. Adafruit has size limit on images which is why I went with google drive integration (still under development) for the camera module.
Telemetry data in most cases is sampled, which mean only a subset of the emitted data is collected and sent to telemetry collector (reading: statsd) . But this is not the case with reef-pi, we send all data to adafuit.io, no sampling.
Regarding the error/warnings, its facinating that you mentioned this. There are dedicated services that allow capturing and analyzing errors (reading: https://sentry.io) , but I think it will be too much for reef-pi. This is something on my mind, and I will have some answer soon. My current thought process is to have notification system (email or sms) in place first. Then track and report error when it goes above certain threshold. Does that sound sane?

Sounds reasonable, my thought was that reef-pi itself would not be handling the alerts, but setup through io.adafruit.
I have been playing a little with the triggers on there. So far just have temperature alerts.

 

[Ranjib]

Sounds reasonable, my thought was that reef-pi itself would not be handling the alerts, but setup through io.adafruit.
I have been playing a little with the triggers on there. So far just have temperature alerts.

Yeah, I also expect the same. This is only for sake of completeness, that if someone is not using adafruit or telemetry, reef-pi, the controller itself should provide minimal support, like the charts we have in reef-pi, they are not as feature rich as adafruit io, but they can do the job for the most basic use cases , where telemetry integration is absent.

This is also the reason I have not done it yet, I plan to visit this at the end , once all the major blockers for 1.0 is done.

 

[jarekt]

API is the key for 1.0 release. Its not documented and fully stable yet, but the reasoning for 1.0 release (or any major release ..) is that the API will not have any compatibilty break for the entire 1.x series.. it will be maintain at least 1 year, at max 4 year (long term support release). It might get new additions, but all thing old will remain and work as it is.

reef-pi 1.0 is the ground work I need before making a reef-pi cluster, where multiple reef-pi controller will be controlled by a single mothership (via API). There are couple of folks in local university and aquarium who wants to use it, but its simply not practical to manage 20 different controller independently.
So, again I'll repeat following are my priorities in terms of reef-pi design & acrhitecture
- Priority 0: A robust and reliable controller that provides fundamental features for equipment control and sensors using opensource software/hardware
- Priority 1: Stable API, user experience., so that users can trust using it for long run ( A typical reef tank runs for decades). This also allows keeping the core system at minimal, while let others to extend it as their requirements.
- Priority 2: Provide minimal but good documentation & visual aesthetics (part of user experience . but lower priority than API or reliability).

Ranjib, you have my support with the priorities. At the end the most important is to have a functional controller handling the menial stuff to be done around a tank. All the rest can wait as it is secondary to the main goal and to work reliably, dependent on the main functionality.

 

[stefanm]

Hi, just a couple of questions, unfortunately there isn't a search this thread on here.
What does think controller do for lighting, especially for those who have diy LED setups with meanwell ldd drivers, sunset/sunrise? cloud cover? lightning storms? Also will this be able to control a heater and a chiller?
Sorry for the questions, I did read through at least 15-20 pages.

One more, wouldn't this work on Allo Sparky SBC boards, they're quite cheap (at least in my part of the world and they're 7" touch screen is really cheap)?

 

[Somnifac]

@stefanm

You can control the lights a couple of ways.

#1 is a basic on/off timer, if you don't have lights that have a dimming control that you can hook up.
#2, for lights that can be dimmed via PWM or 10V analog control, you can ramp the lights up to the levels you wish throughout the day, and then fade them down the same way. Multiple channels can be controlled if you have multiple lights, and/or a light with multiple channels. You can use that to simulate sunrise/sunset cycles, and dimming/brightening throughout the day to simulate cloud cover I suppose. More complicated programs, like storms, etc. aren't part of the functionality. I'll let @Ranjib speak to future possibilities for lighting programs.

There should be no reason that once you have your temp probe wired up that it wouldn't properly provide control for a heater in conjunction with a chiller.

For the search, the search box in the upper right corner of the page allows you to search individual threads. There's a checkbox you can check to confine it.

 

[stefanm]

@stefanm

You can control the lights a couple of ways.

#1 is a basic on/off timer, if you have lights that have a dimming control that you can hook up for control.
#2, for lights that can be dimmed via PWM or 10V analog control, you can ramp the lights up to the levels you wish throughout the day, and then fade them down the same way. Multiple channels can be controlled if you have multiple lights, and/or a light with multiple channels. You can use that to simulate sunrise/sunset cycles, and dimming/brightening throughout the day to simulate cloud cover I suppose. More complicated programs, like storms, etc. aren't part of the functionality. I'll let @Ranjib speak to future possibilities for lighting programs.

There should be no reason that once you have your temp probe wired up that it wouldn't properly provide control for a heater in conjunction with a chiller.

For the search, the search box in the upper right corner of the page allows you to search individual threads. There's a checkbox you can check to confine it.

Thanks, I'll go through the searches.

Though I'd like to know about the feasibility of using other SBC's as I had mentioned.

 

[Ranjib]

Hi, just a couple of questions, unfortunately there isn't a search this thread on here.
What does think controller do for lighting, especially for those who have diy LED setups with meanwell ldd drivers, sunset/sunrise? cloud cover? lightning storms? Also will this be able to control a heater and a chiller?
Sorry for the questions, I did read through at least 15-20 pages.

One more, wouldn't this work on Allo Sparky SBC boards, they're quite cheap (at least in my part of the world and they're 7" touch screen is really cheap)?

reef-pi the software itself allow sunrise/sunset and on-demand dimming features. The standard electronics generate 5v pwm , you can convert that to any type of required signal using hanful of transistor/capacitors, depending upon your lighting equipment. The build guide covers how to do it for Kessil lights.
https://reef-pi.github.io/build-guides/lighting/
There are folks who have used it with blackboxes, .
Storm and other features might be available in next version, and not a priority for the upcoming 1.0 release (around X-mas)

reef-pi the software itself does not have anything raspberry pi specific and should run anywhere if linux is available. The standard installation files are available on both ARM6 (pi zero) and ARM7 (Pi 3). If your board uses one of these processor, then you should be able to use the packages as it is. Otherwise, you have to compile it on your own.
BTW, i just looked at the price of this board, it looks more expensive than pi3 (wifi), definitely more than pi zero (10$)

 

[Ranjib]

@stefanm

You can control the lights a couple of ways.

#1 is a basic on/off timer, if you don't have lights that have a dimming control that you can hook up.
#2, for lights that can be dimmed via PWM or 10V analog control, you can ramp the lights up to the levels you wish throughout the day, and then fade them down the same way. Multiple channels can be controlled if you have multiple lights, and/or a light with multiple channels. You can use that to simulate sunrise/sunset cycles, and dimming/brightening throughout the day to simulate cloud cover I suppose. More complicated programs, like storms, etc. aren't part of the functionality. I'll let @Ranjib speak to future possibilities for lighting programs.

There should be no reason that once you have your temp probe wired up that it wouldn't properly provide control for a heater in conjunction with a chiller.

For the search, the search box in the upper right corner of the page allows you to search individual threads. There's a checkbox you can check to confine it.

Thank you @Somnifac .

A work in progress version of the temperature controller can be found here: https://reef-pi.github.io/build-guides/temperature/

 

[Cary]

Scary Quiet. I guess we are all busy.

 

[AshwinRavi]

I got called out for a last minute project across the country. missing my tank

 

[Somnifac]

Scary Quiet. I guess we are all busy.

Then I'll ask a question:

Has anyone explored pumps that could be used w/ reef-pi in a wavemaker/controllable capacity (once the feature is added (double parenthetical: I'm making an assumption that DC pump/wavemaker controls will make it in eventually; maybe not 1.0, but eventually))? I've seen that pumps like the Jebao powerheads are controllable, and have an accessory that allows them to connect to an Apex controller. I'm sure there are others out there, but I'm still exploring.

 

[Cary]

Sorry to hear that, AshwinRavi.

Somnifac, I have both Sicee and Koralia nano wavemakers connected to a Koralia controller and have had no problem with them. I have them switching current (across the length of the tank each minute (1 min).

They have been in this configuration for at least 7 months and I have had no problems.

I have a question about electronics purchasing. Does everyone purchase on ebay or amazon, or do you have local electronics stores? All the Radio Shacks in my area are closed and I find that everytime I want to start building something, I have to order from one of those two places.

Also, Has anyone noticed the ATO kit that robo- tank sells. It has a mount, pump, optical sensor (like Ranjib specs) and a pump for $45.95 plus shipping. I haven't checked the shipping from Canada, but it couldn't be too bad.

I don't know anything about the kit or company, I just thought that I would throw it out there. I guess if you run two sensors you wire them together.

later