I have a spring fed stream that runs all year at 1500 gpm avg. through a natural necked down point where I have diverted 300 gpm to power a Banki design crossflow turbine. I only have about 3' of usable fall without encroaching on the adjoining property. The turbine was built from (2) 26" 700 series bicycle rims with spokes and hubs removed and replaced with a round disk of 1/4" pvc sheet material. Once centered in the rim, a 1/8" hole was drilled through every other spoke hole into the PVC, and 1"X6-32 flat head stainless machine screws were installed. The rims each make a perfect 1/2" type A belt pulley. The center holes used to layout the disk circles were enlarged enough for a 5/8" stainless axle and stainless flange bearings. 32 turbine blades were fabricated from (8) 2' lengths of 6" SDR 35 pvc sewer pipe cut lengthwise into 4 blades, each 24" long and 4.245" wide. After sizing the blades to exact length and squaring up the ends, they were mounted between the rim assemblies. I used aluminum S2 wire lugs as L brackets and nylon 1/4-20 nuts and bolts to attach the ends of each blade to the disks. Nylon was chosen for it's low shear strength, in case a blade takes a hit from debris. I pre-load the fasteners to 5fp. This sets the shear point at an additional 4 or 5fp. After about a year of constant use, I have replaced one sheared bolt. The blade was not damaged and continued to function normally for several weeks before I found it. Nylon fasteners also help keep overall cost and weight down. Total weight of the finished turbine is about 60 lbs. The turbine is mounted in a swing assembly, hinged from a cantileverd beam that overhangs the creek enough to center it on the on the channel. I can adjust the turbine position for optimum rpm, or raise everything out of the water with a winch assembly consisting of a hand winch, a length of stainless wire rope, stainless pulley, and a triplex wire sling used as a safety, in case the winch latch fails. (3) 12" diameter concrete piers support a 20' main support beam, built with 2X10 treated lumber, the turbine swing assembly and winch hardware. A SuperAmp PMA http://www.windynation.com/products/wind/generators-and-pmas/superamp-permanent-magnet-alternator was installed using (2) 12" lengths of UniStrut and (2) 6" cushioned pipe clamps around the body. This holds it tightly in place without damage, and it's adjustable slightly front to back for easier pulley alignment. A sandbag diversion dam was built to raise the water level and pressure entering the turbine. Initial PMA unloaded test voltage is 26.26 (20.2 x 1.3 since it has a 3-wire output}. A 5" driven pulley gives me 14-16vdc into the GTI and provides 100va output along with excellent voltage stability. During high water conditions, the arc of water hitting the turbine is flattened somewhat, raising the point of impact, and slightly reducing turbine rpm and GTI input voltage 1 or 2 volts. As a result, no overvoltage protection is needed for the GTI. Loaded PMA rpm losses? I started out with a freewheel turbine speed of 120 rpm. After the PMA and GTI were connected, I have a grand total of 80 turbine rpm, which translates to 462 rpm at the PMA, a reduction of 33%. From what I understand, this is about average. I have 'fine tuned' the system since initial testing and have achieved a solid 100w. average GTI output. I have a few more ideas to try, but for now it's online and producing power 24/7. So 100w. x 24 hrs. a day x 30 days a month / by 1000 = 72 kwh per month, x my kwh cost of .115 is $8.28 per month off my electric bill. A constant 100w. also lowers my meter demand reading slightly, which lowers my overall kwh rate by a miniscule amount. I figure about a 10 year payback, or more with maintenance costs. The GTI feeder circuit back to the house was installed, utilizing an underground conduit for (3) #6cu THWN conductors (large enough for a second PMA later), a #8 grounding conductor, and a pull wire for future use. I'm getting the exact same voltage 150' from the site that I did right at the site, so I'm glad I bit the bullet and bought a 500' spool of larger than needed wire. I'm using Fenner PowerTwist adjustable link v-belt material. It's temperature stable, excellent grab when wet so it needs less tension, thus less bearing stress and more rpm, dampens vibration, and field assembled to length with no tools, although a small hemostat clamp helps. A bit pricey but well worth it. After a year of use, it shows virtually no stretch and remarkably little wear. Water pools for about 100' upstream of the diversion dam which slows down debris heading toward the two 6" PVC turbine feeder pipes. A trash rack was built from scrap #4 rebar welded into a grid, and a few scrap treated 2X4's. It keeps larger flash flood debris away from the inlet pipes. Leaves and small debris pass easily through the rack and turbine. The turbine ran free-wheel all winter 2012-13 in all kinds of weather and waterway conditions. Ice will build up on the support structure from splash water when the temp. falls to 20F or so, then break off as it warms up, with no adverse effects. If it gets below 10F and stays there long enough (like winter 2013-14), ice starts to build up on the outside of the turbine. I take it out of service then, to avoid any freezing pressure and possible damage. I perform any needed maintenance and/or upgrades during this time, and return it to service as weather permits.