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

[PaulJ2303]

Can I ask a small QOL question:
Is there a way to manually set the vertical scale of temperature displays on the dashboard?

 

[hhaase]

Looks awesome .
Remind us again what all this board will offer ? I noticed double ph probe , db9 for power strip , couple of temp probe and two ato probe ( photoelectric ). No float switch and 10v pwm output right ?

Two temp probes. Two measurement probes (Can be either Ph or Orp depending on the installed resistor). DB9 power controller output. One photo-electric water sensor.

For this version I haven't done a float switch yet, but that's on my list for the next version after this. I haven't looked at the 10v pwm yet either. I could probably still add a float switch but I haven't had time to dig up the needed circuit. Would it still be just a power/ground/signal header, with a pull-up resistor on the signal? If you can suggest which GPIO pin to attach to, I can add one or more of them easily. I haven't ordered these yet.

-hans

 

[Des Westcott]

Can you file a issue in github to track this. I need to dig deeper into this. Somethings to remember
- reef-pi stores all stats in memory to reduce disk io. This is a global strategy to avoid sad card wear and tier as much as possible,
- reef-pi will save the stats in disk (sd card ) during regular shutdown and hourly roll up .
- if you are shutting down the pi abruptly , it is expected to loose the current hours data,

I have to decipher if your condition is as expected due to those. Or we have a bug somewhere

I have filed a bug on GitHub. It is definitely happening outside the 1 hour window. ie data from a water change 6 hours ago disappears when I lose power unexpectedly.

 

[Ranjib]

I have filed a bug on GitHub. It is definitely happening outside the 1 hour window. ie data from a water change 6 hours ago disappears when I lose power unexpectedly.

Thank you . I’ll investigate as soon as I can. Meanwhile if you know the time of power outage it’s worth configuring a systemD timeR (cron job) for reloading reef-pi before the outage. That will save things on disk proactively . Keep a backup sd card ready , sudden shutdown will speed up the bad card wear n tear

 

[Ranjib]

Two temp probes. Two measurement probes (Can be either Ph or Orp depending on the installed resistor). DB9 power controller output. One photo-electric water sensor.

For this version I haven't done a float switch yet, but that's on my list for the next version after this. I haven't looked at the 10v pwm yet either. I could probably still add a float switch but I haven't had time to dig up the needed circuit. Would it still be just a power/ground/signal header, with a pull-up resistor on the signal? If you can suggest which GPIO pin to attach to, I can add one or more of them easily. I haven't ordered these yet.

-hans

Yes, the float switch circuit is exactly same as attaching any latching /spst switch . Use any of the free gpio excluding i2c, pwm and one wire.

Gpio 18 & 19 are normal 3.3v pwm output , Gate them with a npn transistor to convert them to 10v pwm. This will allow kessil light, deltec pump and variety of other equipment control

 

[Ranjib]

Can I ask a small QOL question:
Is there a way to manually set the vertical scale of temperature displays on the dashboard?

Not right now. We removed it in 3.0 due to some inconsistent behavior. We plan to reintroduce it back after we have some idea how to cleanly support this

 

[hhaase]

Ahhhh, now I see it. The PWM control falls under the 'light controller' design, with the PCA9685, correct? So there will be additional voltage regulation and such involved.

I think I'm going to handle that as a daughter-board actually. But I can very easily add on a header for that until I have time to do that extra board.

-Hans

 

[Ranjib]

Ahhhh, now I see it. The PWM control falls under the 'light controller' design, with the PCA9685, correct? So there will be additional voltage regulation and such involved.

I think I'm going to handle that as a daughter-board actually. But I can very easily add on a header for that until I have time to do that extra board.

-Hans

PCA9685 provides additional 16 channels of pwm, but Pi itself has two hardware backed pwm (GPIO18/19), I was referring to the later, as it wont increase the price/complexity significantly. Yes, generally a voltage regulator (lm2596 is what I use, but you can go with lm7810 or something similar) is use to supply the target pwm voltage (10V), frequency control is done on pi (via reef-pi)

 

[hhaase]

Makes sense. If it's just a generic NPN transistor and 10k resistor it'll be easy enough to add, I'll have to see what kind of compact regulator I can find in 10v to keep wiring simple. I still think I'll add an output header location though for additional SCA/SCL driven hardware, with 5v/10v/12v available too.

No promises it'll all make it on this revision though.

-Hans

 

[Michael Lane]

Here's a picture. Just deleted & re-added the probe.

After I type the value in:

After I go up by one:

After I go back down by one:

I haven't seen this behavior. Which browser are you using? I'll try to recreate and solve the bug.

 

[Schreiber]

I haven't seen this behavior. Which browser are you using? I'll try to recreate and solve the bug.

I'm on Firefox with a Raspberry Pi Zero W, running 3.0. I've seen the issue every time I've tried to set up the pH probe, going back to 2.5

 

[hhaase]

I opted, for now, to put all the PWM stuff off onto a separate daughter board. The cost of all the components for the PH probes makes me not want to go through too many iterations. That and doing the isolation and guarding traces on the main board took a while, and I just don't want to tear all that up right now.

So the separate daughter board is going to have the GPIO18/GPIO19 PWM signals, plus the PCA9685PW decoder. It also lets me off-board the 10v regulator for the time being. Still a few things I want to clean up, I'm concerned with connector amperage over the long term in some locations. Should have it ready to order in a couple days.

I'm also going to lock it the feature set for this version too. If I keep adding stuff I'll never get anything built.

-Hans

 

[Ranjib]

I opted, for now, to put all the PWM stuff off onto a separate daughter board. The cost of all the components for the PH probes makes me not want to go through too many iterations. That and doing the isolation and guarding traces on the main board took a while, and I just don't want to tear all that up right now.

So the separate daughter board is going to have the GPIO18/GPIO19 PWM signals, plus the PCA9685PW decoder. It also lets me off-board the 10v regulator for the time being. Still a few things I want to clean up, I'm concerned with connector amperage over the long term in some locations. Should have it ready to order in a couple days.

I'm also going to lock it the feature set for this version too. If I keep adding stuff I'll never get anything built.

-Hans

I feel its wiser to keep the pwm outputs in the board and ph as separate. Its a relatively expensive. standlone module (ph). In its current form that can be used, sourced, assembled or added/removed from an existing build easily (just hook up the i2c wires). The same is not true with PWM, they have much bigger use cases, they are cheaper and you will have lesser complexity on the board. Doing the pwm bits by hand with through hole components is rather tricky i feel.
Your call at the end, its DIY and opensource , so nothing stops you really from tinkering with the design. I do think theres might be some niche use cases with dual ph probe.

 

[hhaase]

All good thoughts. I'm still figuring out how I want to sub-divide everything in the long term to be honest. Particularly with me not having a reef tank again yet (it'll be a month or two till I can set it up).

Right now I don't have anything I can PWM control, so it's a lower priority for me,. But I have a freshwater tank and my RO storage system I can monitor. So I've been focused mostly on the monitoring portion.


-Hans

 

[Des Westcott]

I've always though that modular stackable hats are the way. Start with a Pi. You only want equipment control? Stack an equipment control hat onto the Pi. You want to control lighting? Add a Light control Hat to the top of the stack. You want a few dosers 6 months down the line? Stack a doser hat on top.

Or am I missing something as usual? maybe limit it to 3 hats that when used together will give you full functionality, but you can choose to have partial functionality if you prefer to start and then increase later really easily.

 

[Michael Lane]

I've always though that modular stackable hats are the way. Start with a Pi. You only want equipment control? Stack an equipment control hat onto the Pi. You want to control lighting? Add a Light control Hat to the top of the stack. You want a few dosers 6 months down the line? Stack a doser hat on top.

Or am I missing something as usual? maybe limit it to 3 hats that when used together will give you full functionality, but you can choose to have partial functionality if you prefer to start and then increase later really easily.

The difficulty with stacking hats is ensuring gpio compatibility between the modules. It either requires standards, or a disciplined design and everything from a single designer/vendor.
It's an interesting idea though. I've considered making a pi tower based on that concept.

 

[hhaase]

My other big issue with stacked hats, or multiple boards of other flavors, is because of the unique aspects of reef electronics. It's fairly common, due to convenience, to see various reef electronics mounted inside a cabinet that also has an open sump. This is a perfect atmosphere for oxidation, corrosion, and dendritic growth (tin whiskers). Every piece of electronics on my old tank suffered from this to one degree or another.

There are a lot of ways to environmentally seal electronics on a circuit board, either via conformal coatings, or urethane potting, if the boards are designed to make this possible. However, it's really hard to affordably make connectors protected from a salt environment. So the less connectors you have, the better.

 

[Des Westcott]

Dewd's Big Cube Reef-Pi build

RTC installed Nice is the RTC working like its supposed to

www.reef2reef.com

Yeah. Electronics and the sumo environment don’t mix. Have a look at my build thread above. I’ve separated the electronics from the sump environment entirely to avoid that exact issue.

 

[Ranjib]

The difficulty with stacking hats is ensuring gpio compatibility between the modules. It either requires standards, or a disciplined design and everything from a single designer/vendor.
It's an interesting idea though. I've considered making a pi tower based on that concept.

Yes.. I too do not like HAT as a general-purpose solution. Its not impossible to really make an all-inclusive design, but I think its pretty hard, the form factor itself imposes a challenge.
First of all, lets call out a key thing, an official Pi HAT is a very specific thing, it has to conform to certain guidelines, including a specific EEPROM spec for auto-detection. Only then it can be called as HAT. Otherwise, we should just call it Addon board, with HAT's form factor (i.e. it sits on top).

The problem I see with stacking HATs are
1) The form factor becomes weird. You get a brick-shaped thing where pi connectors (HDMI, sd card slot etc ) are in one place.. now for exposing all PWM, outlet, inlet, ph probe you'll need connectors in another place. So if you conceptualize the end product, you have a brick-like thing with connectors popping up from multiple directions:-/ . I dont like that. I prefer the traditional rectangular shape with connectors spanning in one side only, this eases mounting and we know how to do brackets etc.. Its not impossible to get other designs, but they required significant design thinking/work.
2) The power delivery can get complicated, this typical perf board or pcb boards are less than 3A rated. The power requirements are kinda blocker for arbitrary stacking.

 

[Des Westcott]

The difficulty with stacking hats is ensuring gpio compatibility between the modules. It either requires standards, or a disciplined design and everything from a single designer/vendor.
It's an interesting idea though. I've considered making a pi tower based on that concept.

That's kind of how I envisaged it. One supplier / designer made 3 or 4 modules that worked together, but weren't necessarily compatible with others. Although it would be fantastic IF that could be the case. I would definitely support that one guy. Maybe not all at once, but once I bought into his system, I would definitely go back to him to add to the system.

Kind of like a reward for taking the time to design and produce a full package deal.