PDA

View Full Version : DIY Biocube29 compact drop in UAS - testing



mulcmu
12-26-2014, 09:14 PM
I've been planning to build a drop in UAS for the back of a BioCube29. This final design is still in progress; testing some ideas has started on a separate feeder tank. The goal is to have a pair of screens crammed in the rear compartment to allow alternating cleaning cycles. To maximize the width of the water channels, the lights need to be as thin as possible as the width of the rear chamber is about 3 inches wide. Current concept is to use 0805 sized SMD LEDs on a custom PCB.

5793

Testing some ideas for the final design has started in a 20 gallon feeder tank with a DIY hang on glass UAS. Tank is about half filled with brackish water (16 ppt), 72°F. No tank lighting. About 1 inch of aggregate. About 1g of pellets per day. No algae in the tank. Nitrates run quite high (dark orange to red with API test kit). Screen is about 3x5, roughed up on both sides. Flex of screen keeps it in place on side U-channels.

I was planning to use rigid airline tubing for the air supply. Using the rigid tube for this test setup took a few tries to get right. Very sensitive to angle, hole size, and air flow. After scrapping a few failed attempts that had one end of the tube plugged, I found running the same airline to both ends seems to resolve some of the sensitivity. The Calfo-style return manifold kind of inspired this setup for the airline. Seems like some water gets into the tubing under the holes with a thin layer of air maintained along the top of the tubing. A slug of water still gets pushed around in the center.

A small prototype PCB was fabricated with some generic white leds from the parts bin. There are 30 leds crammed into this area less than a square inch. Rated power is 2.4W (including the current limiting resistors). Lights can be seen in the reflection in the above picture. During initial testing the light was taped right to glass and running at 75% rated power and later about an inch from the glass. At this power output the center of the screen had no growth but there was some coloration visible around the outer edges. Setup ran for about 2 weeks in this configuration without any growth in center. Output was then dropped to 40% of rated power. Photo shows first week of growth at 40%. Lights have been running using a 18 hours on, 4 hours off cycle. (I realized later it would typically be 6 hours off, but used 4 for some reason when I wrote the software for the timer / led driver. Haven't had a need to change the code in the controller yet.)

5792

The PCB plus LED thickness is 0.1 inch. Running the current lights at 75% was pushing the thermal limits. 60°C was measured with a thermocouple on the LEDs. 0805 660nm LEDs on order. Will try 64 LEDs over 4 square inch area next. Hopefully more space between chips will run cooler.

SantaMonica
12-26-2014, 09:34 PM
Welcome.

Yes I was going to say, the white LED's are probably the bottle neck here. They usually don't grow at all. After replacing with reds, and running for 10 days, let's see how the first growth starts.

Flow/bubbles look good. Only concern will probably be clogging of the air outlets, but I'm sure it's easily replaced.

mulcmu
01-12-2015, 07:37 PM
Got a batch of 660nm red leds in. First test was to confirm that they are 660 with spectroscope. Hard to take a picture to show but the top line is a standard red led and the bottom led is the 660nm led. The 660 was more concentrated and agreed with the scale on the meter. Scope picture of the white leds with the scale from the scope visible included for reference.

5808

5809

I modified the in tank section to use 4 magnets. They are glued in place on this section. A mating external housing was fabricated for the light side. This side has 4 matching magnets but they are removable. Steel washers are glued to the back of the housing to keep the housing tight to the magnets. This allows the light housing to be removed to inspect the growth without removing the in tank section. Works ok, some magnets are more attracted to the washers than the corresponding magnet in the tank. Holds the lights in place much more consistent than the original tape/wire method.

5815

5811

The new PCB has 60 of the 0805 leds soldered to an area about 3.5 square inches. Current limit resistors are on the back side. Spaced them out way more than the first test with the white leds to distribute the light better.

5812

5813

The red leds have been running for 3 days now. 18 hours On, 4 hours Off. Started running at about half power. I want to add a heat sink to the regulator on the controller before testing at full power. The glass was cleaned and half the screen when they were installed. Bits of green are starting to mix in with brown already.

5814

5810

SantaMonica
01-12-2015, 09:38 PM
I think this will work much better.

mulcmu
01-18-2015, 06:30 PM
Daily growth pictures over first week with the red LEDs.
Cleaned the browner regions from the left hand side of the screen.

5823

Looks promising, starting to design a real PCB with the 16 chips per square inch for under water testing.

Still need to make a proper enclosure to hold the power supply so I can increase the current.

SantaMonica
01-18-2015, 07:53 PM
Let it fill in now, maybe 14 days.

mulcmu
02-02-2015, 08:04 AM
Starting to fill in. No other changes to setup.

5841 5842

SantaMonica
02-02-2015, 01:50 PM
Time to scrape.

mulcmu
02-02-2015, 02:25 PM
Cleaned and rinsed with fresh water

Before cleaning:
5847

After cleaning:
5848

Do the white patches look like pod growth or issue with light?

SantaMonica
02-02-2015, 04:40 PM
Yea I think they were pods. Try to get the rest of the black stuff off too.

mulcmu
02-11-2015, 08:21 PM
Growth before second cleaning with red LEDs
5865

Still have some white patches
5866

SantaMonica
02-11-2015, 08:52 PM
More light.

mulcmu
04-04-2015, 02:06 PM
Had the 0805 chip leds maxed out, no improvements in the growth. Tried reducing the air flow to match lighting but still no improvement.

Giving up on the 0805 low profile for now. 3W bead leds and aluminum PCBs have arrived.

Crude lux testing with DMM show the 3W are 4 times brighter than the 0805 board. This was with the 3W running at only 10% (about 75 mA) without a proper heatsink or cooling. These run much cooler too, only 15°F above ambient. They are only twice as thick, 1/4 inch vs 1/8 inch.

5974

5975

SantaMonica
04-04-2015, 04:33 PM
3w usually do work much better.