cypho's 36" DIY reef (1 Viewer)

[cypho]

Slight regression back to the mechanical side of life. I may have missed how you auto rinse and remove liquids from the cups.

To clean the cup, I use the same type of peristaltic pumps that I am using for everything else.

For rinsing I just pump RO water into the cup.

For the drain I have a tube that runs down to the bottom of the cup. There is a pipet tip on the end of the tube that touches the bottom of the cup. Thanks to water's tendency to stick together it pretty much gets 100% of the water out. I drain, rinse, drain, and repeat 3 times so if there is any water not removed it should be almost pure RO water. And pure RO water in any quantity has no impact on the results.

You can see the drain line in the photos - it is ziptied to the pH probe to hold it in place.

 

[cypho]

I think I have a solution for the pH probe.

By adding 2 T's and 4 one-way valves I think can plumb the rinse pump drive the drain line when run in reverse. That will free up a pump that I can use to fill the probe.

__________________________

I've also continued to ponder self calibration of the pumps. And I think I have a solution for the sample pump.

Verify that HCl, NaOH, and EDTA pumps are properly calibrated as previously described. Then,

Test #1 - Repeat the HCl vs NaOH titration again, but this time use the drain pump to remove 50% of the sample before titrating. The result should be exactly half of the first HCl vs NaOH titration. If it is not, adjust the drain pump's calibration.

Test #2 - If I add a sample of aquarium water, test it for something (ca/mg/alk) then
Test #3 - If I add a sample of aquarium water and then drain 5ml, and test it for the same thing.

The the result of #3 is not exactly half of the result of #2, adjust the calibration of the sample pump.

 

[cypho]

The alk titration repeatability really went downhill recently.

The titrator is programmed to repeat the test if the alk has changed more than 0.05 meq/l, and then continue re-testing until two consecutive tests vary less than 0.05. Before when the test needed to be repeated it almost always confirmed the first test was accurate. But recently it has been having to repeat the test 5+ times before two consecutive tests would agree. At first I assumed it must be a software issue as I rarely go a day without changing something in the software. When I reverted all recent software changes, and saw no improvement I decided it must be a hardware issue.

After a lot of troubleshooting I decided that the problem may be that the peristaltic pump tubing was worn out. I ordered some new tubing.

The pumps came with 1mm ID tubing. I could not find 1mm tubing on amazon and I am not in the mood to wait for the slow boat from china. But Amazon had 0.5mm ID tubing for really cheap, so I ordered that. The new tubing is about 3x slower (1 rotation of the motor pumps 1/3 of the volume) so it should be even more precise than the original tubing.

When I removed the old tubing it was somewhat flattened, I'm hoping that is a sign that I found the problem. The new tube is calibrated, crossing my fingers that I will get good results tonight.

 

[cypho]

So, I have been insanely busy with non-fish related stuff for most of the last year. On several occasions I went multiple months without even verifying that there was still water in the aquarium. Finally things are calming down and the slow process of rebuilding begins. It was not a total loss, but let's just say it will be awhile before I post any full tank pictures.

Anyway, the rebuilding is going to begin with an overhaul of the controller. The current/old controller is working fine, but that it be piece I am most excited about and it keeps me from dwelling on the livestock I lost.

So, controller version 2 helped a lot with the huge rats nest of wires. But there are still a lot of wires and managing them has continued to be an annoyance. The other big problem with V2 is that the raspberry pi central hub is a single point of failure. If a new bit of software crashes, or if I short a wire while working on something, or get something wet that was not waterproof, it could down the whole system.

To address the those issues, version 3 is going wireless. Everything will still need a power cord because charging batteries on sensors that run 24/7 would suck, but instead of 100 sensors each a wire running back to the central hub, each sensor will be able to function as a stand-alone unit which can talk to everyone else via wifi.

 

[Riley]

Bad looking tank or not, it's awesome to see you on here again. I have no doubts that V3 won't be anything other than impressive and I'm looking forward to your updates.

 

[cypho]

Well time for an update.

My first design was for an 8 port 0-10v analog output (for controllable pumps and lights). And it did not work. I did not read the datasheet carefully enough and one of the components does not do what I thought it would. I've redesigned using the correct components(hopefully) and now I'm waiting on the new parts to arrive. But I am not particularly confident that version 0.02 is going to work either as 10V control is the one thing that I never got working on my current controller. Which is why I started with it on this one of course....

Even though it create a 0-10v analog signal, the part worked well enough to confirm that my master plan is feasible and allowed me to start working on the software.

 

[cypho]

And then on to failure number two.

The next part I designed was a multi-purpose sensor input module.

The plan for this one was to use a bunch of tiny digitally controlled switches enable/disable a bunch of parts, so it can be reconfigured on the fly for a bunch of different uses. I was quite excited about this one, but in hindsight it was a bad idea.

The problem I had with this one was that the word tiny just does not adequitely describe how small the switches are. I think the design is solid, but I lack the hand-eye coordination to actually assemble it. If any of the switches are mis-alligned by more than 0.2mm, then nothing works. 0.2 mm is really really really small. And to make it even more fun, the connections are on the bottom of the part so you can't even get out a microscope and visually inspect the connections. The only way to figure out which part is misaligned is to start randomly removing thigns. And each time I add or remove a part, there is a good chance I will accidentally move mess up something else. I've speent the last week trying to assemble one, and so far I have not managed to get more than 50% of the switches mounted correctly. So I'm througing in the towel on this one too.

I could redesign it using larger (easier to assemble) parts, but the larger parts are 2x the price and it was already getting more expensive than I wanted. So I am going to take a big step back and simplify.

Instead of including a bunch of parts on the board, most of wich will not be needed for any particular sensor, I am just going to embed the whatever extra parts are needed into the sensor's plug.

In addition to making it 1,000x easier to assemble, financially it also makes way more sense. It will increase the cost of each cord by $1.00, but it will decrease the cost by $20 by leaving out all of those unneeded parts. And it will also be more flexable, since no matter what I put in there, it will never cover every scenario.

 

[cypho]

And here is a picture of what one of my (non-functional) parts when fully assembled. Even though it is not working yet, I am still really excited about the idea of a network of thumb sized aquarium controllers.

And those little black square parts. Those are the ones that I am unable to install with the required 0.2mm precision.

Sent from my Pixel using Tapatalk

 

[RSnodgrass]

Those are tiny for sure. Part of your business model is the replacement of lost components isn't it

 

[cypho]

This one works (mostly)!

The software still needs a lot of work (like the ability to set/store calibration values), but I am getting valid salinity readings! Even without calibration, just based on theoretical values from the amplification circuit parts, the readings are in the right ballpark.

I've also been having fun designing 3d printed cases.

https://cad.onshape.com/documents/c...9321013fbc9aa9a183/e/43637be52996e9d3978ea9ca

Sent from my Pixel using Tapatalk

 

[RSnodgrass]

It's so little, could put that just about anywhere.