Reef-Pi Controllers: New all-in-one with power bar, and the original Pico

Stephen2

Community Member
View Badges
Joined
Jul 31, 2014
Messages
48
Reaction score
28
Rating - 0%
0   0   0
Will one of the modules be for relay control?
Any idea on pricing?
 

Stephen2

Community Member
View Badges
Joined
Jul 31, 2014
Messages
48
Reaction score
28
Rating - 0%
0   0   0
Will there be a module for the relay boards that has an option for 12v and 5v control signals?
Any idea on pricing?
 

Stephen2

Community Member
View Badges
Joined
Jul 31, 2014
Messages
48
Reaction score
28
Rating - 0%
0   0   0
What is the best way to test if i fried the pcb?
I had a short on a motor driver that went through a pca9865
After that no i2c devices get detected and my relays wont work through the db9 connector
 

t951

Community Member
View Badges
Joined
Feb 5, 2016
Messages
55
Reaction score
44
Location
Orlando, FL
Rating - 0%
0   0   0
Relevant SpaceBalls The Movie quote:

Lord Dark Helmet: what the heck am I looking at? When does THIS happen in the move?
Col. Sandurz: NOW. You're looking at now, sir. Everything that happens now is happening now.
Lord Dark Helmet: Go back to then!
Col. Sandurz: What?
Lord Dark Helmet: THEN!
Col. Sandurz: I can't!
Lord Dark Helmet: Why not?
Col. Sandurz: We passed it!
Lord Dark Helmet: When?
Col. Sandurz: Just now!
Lord Dark Helmet: When will then be now?
Col. Sandurz: SOON!
 
OP
OP
theatrus

theatrus

Valuable Member
View Badges
Joined
Mar 26, 2016
Messages
1,941
Reaction score
3,329
Location
Sacramento, CA area
Rating - 0%
0   0   0
So, I've been experimenting a lot with the ReefMod hardware, and there are some definite advantages, but a lot of compromises especially with how the whole thing assembles. In short, I'm not ecstatic with it, and since this is a part-time hobby project as well its been hard to make progress with the setup.

The main driver is the casing and assembly in an enclosure is just a mess. I'm working on bringing in some pretty big mechanical capability in house for a number of reasons (cut a nice check to Haas, primarily for the lighting side), but the number of little face plates, adapter plates, and mounting on frankly pretty tiny PCBs is a giant pain, and a pain to make and install all the spacers for end users, and in the end to have a very compromised connector footprint and spacing.

I'm sorry to say I may have to scuttle the current Mod design and board design, and move on to one of my earlier ideas which attached all of the modules, sensors, etc over a different interface: USB, WiFi, and CAN. I had a number of back burner designs based around that form factor, using ESP32 modules. For example, here is a full 8 channel 1A LED driver based on the form factor I was thinking about:

1597013713834.png

1597013954246.png


The goal is to setup the same case form factor with the common features: Input/outputs for digital sensors, pH sensor, PWM outputs, analog input, relay drivers for power modules (though smart plugs may be more desirable). The Reef-Pi is external and either connects over USB or WiFi to the Core module. This requires driver work for ReefPi, but nothing substantial, and leaves the entire Pi available for other connections.

So, instead of plugin in IO modules, its... plugin IO modules, just externally.

I've revived some older work in this department and will have some prototypes to develop with soon, as the bulk of this is basically a copy and paste of circuits from the existing modules :)
 

inktomi

ReefMan
View Badges
Joined
Feb 17, 2018
Messages
427
Reaction score
409
Location
Los Angeles, CA
Rating - 0%
0   0   0
I think external modules would be fine! As users, we can build little 6" cables if we need to as long as the connectors are easily available - or you could even include them with the kits optionally.

It's always discouraging to have to toss work that you've invested time into - but as engineers (I do software, not hardware) - it's important to know when to cut losses and just start fresh with a different avenue.

Do you have a github repo that we could sponsor, or anything like that?
 
OP
OP
theatrus

theatrus

Valuable Member
View Badges
Joined
Mar 26, 2016
Messages
1,941
Reaction score
3,329
Location
Sacramento, CA area
Rating - 0%
0   0   0
I think external modules would be fine! As users, we can build little 6" cables if we need to as long as the connectors are easily available - or you could even include them with the kits optionally.

It's always discouraging to have to toss work that you've invested time into - but as engineers (I do software, not hardware) - it's important to know when to cut losses and just start fresh with a different avenue.

Do you have a github repo that we could sponsor, or anything like that?

The original Pico was here:


I didn't create one for the MOD but I may as well for history's sake. The others are coming soon :)
 

hotdrop

Well-Known Member
View Badges
Joined
Jan 29, 2019
Messages
608
Reaction score
567
Rating - 0%
0   0   0
external is fine but just be cognizant that it adds a lot of new complexity and failure points. Also cost of each enclosure can be a problem unless you can spec into something super available. I’ve had a number of diy projects where the enclosure and faceplate cost more than the electronics and that’s something you really want to avoid.

IMO the problem with most available Controller hardware is that it’s not designed for failure. Any single failure can result in almost catastrophic runaway and any time you add additional failure modes it’s becomes even more complicated to design around. On one board you can have secondary monitoring once you get away from that architecture everything acts independently and you need a whole new set of failsafe logic.
 
OP
OP
theatrus

theatrus

Valuable Member
View Badges
Joined
Mar 26, 2016
Messages
1,941
Reaction score
3,329
Location
Sacramento, CA area
Rating - 0%
0   0   0
external is fine but just be cognizant that it adds a lot of new complexity and failure points. Also cost of each enclosure can be a problem unless you can spec into something super available. I’ve had a number of diy projects where the enclosure and faceplate cost more than the electronics and that’s something you really want to avoid.

IMO the problem with most available Controller hardware is that it’s not designed for failure. Any single failure can result in almost catastrophic runaway and any time you add additional failure modes it’s becomes even more complicated to design around. On one board you can have secondary monitoring once you get away from that architecture everything acts independently and you need a whole new set of failsafe logic.

Yes, absolutely. Enclosures are brutal. Plastic isn't out of this realm, but a simple aluminum extrusion, fasteners and PCB end plates (the cheap approach as opposed to machined+printed plastic or metal), easily adds $20 to the component cost, which increases the product price by at least 2x that amount.

The goal for resilience is to avoid anything complicated, and unneeded coordination and communication. If you can run a device with instructions totally stand-alone then you improve the failure modes, as only one function may break.

That's the idea behind controllers like https://www.reef2reef.com/threads/theatrus-perfect-temperature-controller.707407/, which are totally self contained, but can share status with other controllers. Its how I end up running my current tanks, made up of a bunch of independent control units (lights, top-off, heat, calcium reactor, chiller).

Reef-Pi of course isn't quite built for offload yet and mixes all the functionality into one monolithic box, but it doesn't have to be like that and retain the same UI functionality long term.
 
OP
OP
theatrus

theatrus

Valuable Member
View Badges
Joined
Mar 26, 2016
Messages
1,941
Reaction score
3,329
Location
Sacramento, CA area
Rating - 0%
0   0   0
Long weekend design ruminations - 8 up power bar with relays or SSD modules, with vertical space left for some low voltage control (sensor inputs, float switches, 1wire, etc). Built in power supply. USB link to a Raspberry Pi (isolated, of course) for control.

Its nice to hit reset on a project every now and then to get somewhere better?

1599467377518.png
 
OP
OP
theatrus

theatrus

Valuable Member
View Badges
Joined
Mar 26, 2016
Messages
1,941
Reaction score
3,329
Location
Sacramento, CA area
Rating - 0%
0   0   0
Some basic fit checking for I/O. The goal here is not to build a all in one for everything, but a power module for the bare essentials you'd want

- Power control (8 outlets)
- Temperature sensor (1 wire or analog)
- Level sensors / float switches

As you can see I'm trying to "fit in" pH. Jury is still out if its even needed here - tempted to scrub it for the sake of cost and simplicity.

The goal is to power up a Raspberry Pi and do the communication over a USB link, so on-board 12V and 5V are available, with enough power for the Pi4. ReefPi can do all of the reading and control with a sidecar program which writes to files used for the File driver, and takes care of the USB endpoint communication. Since some sensors are on board, autonomous limits can be added so for example there can be a heater failsafe control to turn the outlet off if its exceeding a limit. Same for floats.

Later revisions may add an ESP32 for Wifi control. Trying to keep the firmware builds simple right now.

The RJ10 port is the CAN bus link which can link sensors to other control points in an autonomous way (just broadcasting messages for sensor values), in lieu of using the USB port for active control. I've been putting it on most designs where I can, even though its a pretty big connector, even though right now my personal usage is limited to the remote-pump auto-water changer I'm building up.

1599543469201.png
 
OP
OP
theatrus

theatrus

Valuable Member
View Badges
Joined
Mar 26, 2016
Messages
1,941
Reaction score
3,329
Location
Sacramento, CA area
Rating - 0%
0   0   0
Soon?
Too soon?

Targeting EOY for initial units, assuming I don't run into a supply hiccup on parts :)

The software at this stage will likely be primitive, but will make sure any firmware is field upgrdeable to not repeat the pH circuit fiasco of 2019 ;)
 
OP
OP
theatrus

theatrus

Valuable Member
View Badges
Joined
Mar 26, 2016
Messages
1,941
Reaction score
3,329
Location
Sacramento, CA area
Rating - 0%
0   0   0
Some updates:

I have basic power boards already design up, and a few sample cases in house. The form-factor is nice, though slightly tall compared to your average power strip. Probably not a major issue, as its a fair bit less wide. I'll report back when I have some first samples assembled. The power board has a number of provisions, including protection for the relay contacts with inductive loads, without using a plain MOV or other circuit which leaks power. This adds a few dollars of cost per outlet, but ensures the relay will continue working for its entire rated life, no matter what is plugged in, and that the relay fully switches off with light loads.

I have a companion sense board designed to give current per strip using a CT (fully isolated). This will require some calibration when building to ensure an accurate current figure.

Building things that switch power, and are at high power loads from the wall, means testing extensively when developing and shipping. Part of that is ensuring every outlet can be fully loaded to specification, and the strip is running at maximum power, in addition to providing multiple points of calibration load. About 10 years ago I helped design a number of factory power test fixtures based on high power, non-inductive heat sink mounted resistors, nicknamed "George Foreman" after the line of home grilling appliances. I'm now basically repeating that work with GeorgeForeman, Mk2, which can give up to 15A of load at 120VAC (1800W), at least for some period of time until heatsinks are too toasty for the fans to keep up. The basic premise is a large enclosure, multiple 300W non-inductive resistors, lots of heatsinks, lots of fans, and some relays to switch loads on and off:

1600479971176.jpeg


Final count is x12 100ohm 300W resistors parallel switched across 4 heatsinks running the full 12" length of that case, each with a switch board:

1600480102494.png


Plus 2 much smaller series switched resistors of about 500-1k to allow for low load. Its the world's most expensive heater, but is much cheaper than giving Chroma a ton of money for a fancier AC load at 5x the price. Power supplied will be on a Instek AP-7100, and a 20A wall outlet for full power runs (since the AC source, despite weight in at 100lbs and connected up to a 240V outlet, has a 1000VA / 8A limit).

Inductive loading needs to be tested as well, and I'm looking at options (aka, what large transformers are laying about I can use).

As mentioned in the ReefPi thread, I've been doing sensor qualification. Currently the biggest challenge is.. finding labels that will stick to themselves. Super annoying. Some quick teasers:

image0(3).jpeg


In a batch of 20 sensors, I had 5 sensors outside the 0.5C range, and 1 sensor a whole 1.5C out - and this the "high quality" import option. A ton of other sensors all used fake DS18B20 chips and were even much further out, so as part of figuring out the screening and calibration system went straight to the ewaste pile. Each sensor was in the calibration dry-well for 30min, so its a pretty time consuming process. Keep an eye out on the website for sensors soon, once I get labels working...
 
OP
OP
theatrus

theatrus

Valuable Member
View Badges
Joined
Mar 26, 2016
Messages
1,941
Reaction score
3,329
Location
Sacramento, CA area
Rating - 0%
0   0   0
1600751409347.png


Just a sneak peak of the assembly. There are a few components I'm still sizing mechanically, such as strain relief for the outlets to the top case, and ground wire bonding. I still need to work on the logic board side of this, but so far we have:

- USB out to a Raspberry Pi including powering. The strip has a 20W supply which is plenty for some sensors, the relays, and a RPI4. Something like a doser module will require a second supply - I'd like to include a bit more but space is at a premium.
- Relays are TV8 rated, and designed for large inductive load switching with suitable suppression to prevent relays from both leaking at low load and arcing which can weld them closed over time.
- There is a current transformer (can't see it in the picture) which provides current monitoring of the whole strip.
- Planning on USB, a serial CAN bus (for more strips and modules), a temperature probe 1-wire port, as well as a few water level connectors. Nothing extra fancy, but from a "basic aquarium function" standpoint it gives the bulk of whats needed.

As I mentioned, prototype boards are ordered. Just don't have them in hand yet.

The mechanical design for the testing load is also coming along nicely (bonus pictures). No you won't be able to have one (unless you really do want an 2000W rack mount space heater..., its not cheap since the resistors alone run $600). Heatsinks are 4x 12" long segments of the HeatsinkUSA 4.6" extrusion, combined with 120mm fans pushing 300-400cfm each.

1600752648253.png
 

Ranjib

7500 Club Member
View Badges
Joined
Apr 16, 2016
Messages
9,825
Reaction score
17,041
Location
Pleasant Hill, Concord
Rating - 0%
0   0   0
you are not worried about house panel circuit's tripping ? :) these test setups will draw almost near capacity (20A is the normal limit ?) , i think the dryer outlets are 40a/30a
 
OP
OP
theatrus

theatrus

Valuable Member
View Badges
Joined
Mar 26, 2016
Messages
1,941
Reaction score
3,329
Location
Sacramento, CA area
Rating - 0%
0   0   0
you are not worried about house panel circuit's tripping ? :) these test setups will draw almost near capacity (20A is the normal limit ?) , i think the dryer outlets are 40a/30a

Its all about destructive testing - the strip is fused at the inlet but its important to know when things fall apart. I have a 30A 240V and 50A outlet for the TIG welder run in the garage, but its 240V. Going to have to run a neutral leg for on one of those or see if I have any breaker space left for the destructive testing at 120V :). Thankfully the lab space is on the same wall as the electrical panel so running a circuit is straight forward, assuming I just mail order small EMT conduit sections to avoid having to go to a store.

Its important to know what happens when everything goes very wrong. I might have to build a second load if I need to push towards 50A to actually explode something serious - I expect wiring to start failing at these levels so not sure its even worth it.

Part of the test rig is running with a GFCI+AFCI combo outlet in series specifically to make sure that it doesn't trip, since both GFCI is a great idea for tanks, and AFCI is on new electrical code builds (and can be finicky)
 
Last edited:
Back
Top