DIY Overflow Box Questions

[cypher1024]

I originally posted this in the 'Pipe Sizing Charts and Flow Rates' sticky, but then I realised it made more sense to put it in a separate thread.

Does anyone know if I can use the figures in this thread to size my DIY overflow box? I assume the horizontal flow numbers are the most applicable, since it won't have the height to achieve the vertical flow numbers.

I did some quick calculations and it looks like the flow rate is fairly consistent - around 7.96 gallons per minute per square inch of cross-sectional area. I'm using a Quiet One 5000 at around 5 feet of head, which should output about 900GPH. I'm not sure how much extra flow I should allow for. Would 25% be enough? That would require an overflow box capable of 1125GPH or 18.75GPM.

If everything else is right, that would mean I would only need 2.36 square inches of cross-sectional area in my overflow box. That doesn't sound like much. Is there any detriment to building the oveflow box far larger than it needs to be (other than the gradual accumulation of micro-bubbles)?

This guy suggests that the outer section of the overflow box should be lower than the inner section to improve flow, but he doesn't provide any solid reasons for it. As far as I can tell, the flow would be the same in the uneven configuration (if not slightly lower). I can understand having the outer section lower to allow for a standpipe or sponge filter above the bulkhead(s), but I don't think it's going to improve flow. Am I missing something?

 

[cypher1024]

Sorry for the double-post - the edit timer ran out before I could make these images.

Here's what I'm thinking so far:







The overflow is around 370mm (14.5") wide, 160mm (6.3") tall and 210mm (8.3") deep.

The bottom W section will be bent out of one piece of clear 3mm (1/8") acrylic, as will the upside-down U (except it's black acrylic). The gap between the inverted U and the W (over the rim of the tank) is 12.5mm (1/2").

The intake strainer comb and the sides of the overflow run all the way up to the tank's lid so that fish can't jump in.

The pump should be producing about 3400l/hour (900GPH). From what I understand, 1" bulkheads should be good for 600GPH each, so I plan on using gate valves to choke the outer two bulkheads to around 450GPH each so that the overflow (and its piping) is quiet. The middle bulkhead is in case one (or both) of the outer bulkheads clog. I bought three 1" bulkheads, but I'm starting to think I should've bought a 1.5" for the emergency pipe (in case both 1" pipes stop completely).

I'm not sure if any dimensions should be different because I've never built one of these before. Any comments or suggestions are welcome.

 

[cypher1024]

Everything arrived, so I thought I'd have a go at making a prototype. As expected, it isn't perfect, but I did learn a lot along the way.

This is the saw blade I used. It yielded very nice results. I got it for AU$34 on eBay. Comparable blades at the local hardware shop were over $100. This one may not last as long as them, but given the low volume of cutting that I'm doing it should be fine. The main things to look out for if you're buying one are



Here's a close-up of the triple-chip blade:



This is my right-angle bending jig. I built it out of a few pieces of scrap wood and a pair of door hinges. It worked really well for the first bend, but then had to be reconfigured for later bends. From sdplastics.com: "Use hollow ground high speed blades with no set and at least 5 teeth per inch. Carbide tipped blades with a triple chip tooth will give the smoothest cuts. Set the blade height about 1/8" above the height of the material. This will reduce edge chipping."
I used a 60-tooth wood blade on some scrap and the results were not nearly as nice as the 100-tooth blade. The corners were chipped and there were scores along the edge.





The 3mm (1/8") acrylic was heated with a propane torch until it started to sag, then I used the bender to form it into a right-angle (or close to it). Another option is a strip heater like this. They take a little longer than a torch to heat the acrylic, but from what I've heard, they yield a nicer bend. Unfortunately I haven't been able to find one for a decent price in Australia.

Right-angle bend. I would have preferred a smaller radius, but this is fine.

 

[cypher1024]

Sorry for the multiple posts - I keep getting hit with the edit timer.

I started in the middle, making the two folds that sit over the rim of the tank, then worked my way out. As I added bends, I had to reconfigure the bending jig.





Here are the completed pieces:


As I said before, they're not perfect, but I think they're OK for a first attempt. If you look at the right side of the clear piece, you can see some marks in the acrylic. I think these are from the wood in the bending jig. I didn't sand it flat, so it left impressions in the acrylic when it was hot. There are also some scratches on the left side. Neither of these things are very visible when the box is submerged.

The next step was bonding these pieces to the sides. I didn't take any pictures of this process because I was alone and I didn't want to get the solvent on my brother's camera. It was Weld-On #16. I ran a bead of it along the edge of the clear section, then stuck it to the side. This guy suggested doing it the other way around (running the bead of solvent on the flat side piece, then placing the bended piece on top. Now I know why. I was expecting the #16 to be around the consistency of silicone, but it's a lot closer to maple syrup.

In the end, I managed to get everything in place. I sandwiched the whole thing between two pieces of plywood, then put about 4kg (8lb) of weight on top and left it to cure overnight.

The next day, I routed it. Unfortunately I don't have any pictures of this process either. Routers are scary and I value my fingers.
I do have a shot of the router bit I used though. It's a 'flush trim' bit. It's twin-fluted and 9.5mm (3/8") in diameter with a ball-bearing guide of the same size.



It wasn't until I peeled off the protective sheet from the sides that I saw the bubbles in my joints. Sorry about the water - this shot was taken after leak testing. The main thing to pay attention to are the joins.



I think there are two reasons for this. The first is that that there were gaps between the side pieces and the bent pieces of about 1.5mm (0.05") in some places. The second is my solvent application method. After putting the bent piece in place on the flat piece, I pushed down hard for about 30 seconds, I think this simply spread out the solvent under the big gaps, then when I released the pressure, bubbles formed. By the time I applied the 4kg weight, it was too late.

Some sections worked really well though. With crystal-clear, water-tight joints. It shows me that good results are possible with the right preparation. I'm sure that I can fix up all of the five or six leaks in this first prototype, but I'm too paranoid to run this box permanently. The nice thing is that it only cost me 1/6th of my clear sheet (~400mm x 1200mm or ~16" x 48").

This is my idea for a 'perfect edge' jig:



It's basically just a pair of flat feet that clamp to your bent piece so you can run the acrylic over a router. As long as the router table is flat, this should yield a perfect edge... Well, at least in theory...

 

[Pharaoh]

Nice start. You're making it look easy. Makes me want to try and make one even though I don't need it.

 

[cypher1024]

Thanks man! It's definitely worth trying if you've got most of the tools and some free time.
I'm already thinking about other things I could make out of acrylic.

 

[cypher1024]

That's right ladies and gentlemen, it's time for another episode of:



Sorry. I thought about that while I was working today and couldn't get it out of my head.

Today I ran a bead of Weld-On #16 along all of the joints that I could get to. Amazingly, it fixed all of my leaks! I punched a few 40mm (~1.5") holes in the back section and installed the 1" bulkheads. Using a hole-saw on acrylic is pretty easy. Make sure it's sharp, and preferably have a piece of wood behind the acrylic while you're drilling.

Here is the overflow installed on the tank. I'm aiming for the 'crispy' water that you guys are always talking about. In the sump is my "Quiet" One pump, which at this head height should yield a turnover rate of about 70 times per hour.





Side shot:



The water in the tank rose higher than I was anticipating. I didn't really have any hard numbers to go on, so I assumed it would be around 10mm(3/8"). In reality it was closer to 25mm (1"). Interestingly, the choke point seems to either be the inverted U over the top of the tank, or the baffle in the rear section. I blocked over half of the intake with a piece of acrylic and the water level hardly moved (it's hard to see subtle changes when you're pumping ~700 gal/hour into a 10 gal tank).

As expected, it isn't great at self-purging air bubbles, but this isn't a huge problem because I already have an aqua-lifter which will be connected to the final model.

Another action shot:



The water tumbling over the rear partition made noise, as did air being sucked into the bulkheads and water going down the pipes. I was expecting these, and they will all be eliminated in the final model with a trick I learned in Chompers' overflow thread. I will be putting ball or gate valves on the end of the two outside pipes so they will be filled with water and not suck air. This should make them both silent. If crap bulds up in the valves and restricts flow further, the excess will go down the middle pipe (which should be noisy).

To test this theory, I manually restricted the flow on the two normal outlets.





It worked very well. As far as I could tell, the overflow was silent. Unfortunately it raised the water level in the tank, so this will have to be considered in the next model.

I've ordered some Weld-On #3 in anticipation of better-fitting joints next time as well. As always, comments & suggestions are welcome!

 

[cypher1024]

The Weld-On #3 still hasn't arrived (they're still waiting for stock from the U.S.), but that's OK because I've had plenty of other stuff to play with.

I've been testing different methods for bending the acrylic. As you can see in the bottom-right corner of the last picture in my last post, the bends weren't great. A big part of the problem was that it's difficult to concentrate the heat from a propane torch. I ended up heating a wide area, and that meant sloppy bends, even with my jig.

The first thing I tried was a halogen lamp system, based on a design I found online (but can't seem to find now). Basically, it was a heating element out of a photocopier (which is really just a long halogen lamp), with some watercooled bars so that the heat is focused on a thin line across the acrylic, and a PWM circuit to control the heat. It seemed to work quite well, but I couldn't replicate it becuase I'm not very good with electronics, and I couldn't find a lamp wide enough (~500mm or 19.6"). The other issue is that halogen lamps will burn out quickly if they aren't run near their preferred operating temperature.

Instead I tried a few small, cheap halogen lamps ($2 at the hardware store) set under a similar watercooling system. My idea was that I could vary the temperature by changing the distance between the lamps and the acrylic.

Building the watercooling bars:



Watercooling bars attached to the platform:



Insanely dangerous, "do not try this at home" lamp rig:




Unfortunately I don't have a shot of it all assembled. I put it together, and was too focused on trying not to die that I forgot to take pictures.
The bend turned out OK, but it was clear that the sections directly over the lamps were heated to a higher temperature than the sections over the gaps. Given how dangerous this rig was to operate, I decided to try other methods.


The next thing I tried was a hot wire, since I remember cutting polystyrene with one when I was a kid. I found some information online and ended up with this:



It's essentially just a hot-wire cutter. It has a strand of 0.33mm (0.013") stainless steel wire stretched between two hooks, with a spring to keep it tight. I tried a few different power supplies, each highly recommended on different forums:



Left to right: modified computer power supply, downlight transformer, battery charger.

All three of these got the wire hot enough to cut foam, but not hot enough to bend the acrylic. I suspected that a 24V battery charger would probably work, but I didn't want to spend a bunch more money testing a hunch, so I tried the next best thing: two 12V batteries in series. It worked surprisingly well, but the problem was that the only way I could control the heat output was to move the alligator clips along the wire. When I moved the clips, the wire moved, once getting too close to a piece of acrylic and cutting it.

The best result with that wire was achieved in the watercooling platform, with the hot wire carefully threaded through the maze of metal:



Unfortunately, the stainless steel didn't seem to like running at the temperatures I needed for bending acrylic, so it would occasionally snap. I hit the internet, looking for a solution.

 

[cypher1024]

I found this guy's site: http://www.bishoptechnology.com/proj_acrylic_v2.html

I used his basic design, but expanded on it and made it a little safer.



It's got two strands of nicrhome wire twisted together, running between a machine screw and one end, through an aluminium channel, and through a spring at the other end. The aluminium acts as a heat reflector. I've built box sections on each side so it's impossible to touch any of the wiring or the live machine screws. I've also grounded the aluminium U channel and the L-shaped guide on top.

The chuck sitting on the acrylic is to ensure that it stays flat against the table. The middle section of the table is thin, so I can make bends close together.

His design has a light dimmer in series with the heating element. I've done the same thing, but added some power sockets so that I can plug in an auxillary load. This gives me much finer control over the temperature of the wire. The switch next to the dimmer is the main power switch, and the one between that and the sockets is a bypass so that I can run it without an external load.

The other nice thing about running an external load is that it limits the total current in the circuit, so even if there was a dead short between the posts of my contraption, the total current will never exceed what the external load will draw by itself.

I've made two bends with this thing so far and it's been almost perfect. It takes about three minutes to heat up, then the acrylic is bent by hand roughly into shape. It's then clamped in a right-angle jig and allowed to cool.

The only issue I've had so far is the ends of the acrylic occasionally cracking. I think the problem is that the aluminium guide acts as a heatsink, so the acrylic doesn't get quite hot enough on that edge. I've been experimenting with spacers and it's looking promising.

 

[merob5407]

cool