My little paradise has a stream that provides enough water flow and head to
run a small turbine, to provide electricity to my home. While writing this, the
microhydro plant is being implemented, and here are some photos of the

Since I usually like to start at the end, the first thing I built is the

It is an implementation of Jan Portegijs’ “Humming Bird”, with
some changes and adaptations.

The largest cost of the plant, by far, is in the piping for the
rather long penstock. When the purchase was made and the truck arrived, we
unloaded the pipes at different places, to get them as close to the installation
area as possible. Here are the main stacks:

A smaller number of pipes were stored closer to the turbine site.
All these are white PVC pipes, class 2.5. The lowest of them will be used pretty
much at their safe limit, very close to 2 bar of pressure.

Only for the last part of the run, where the pressure exceeds 2
bar, I will use blue class 4 PVC pipe.

To change from the low slope run of the white pipe to the much
steeper run of the blue one, a change of direction is required. I tried to
purchase a 135 degrees bent pipe, but it was not available despite appearing in
the catalog. The best I could buy is a 135 degrees corner piece, which is far
from optimal, but perhaps better than heat-bending a pipe myself, in this
relatively high pressure area.

My neighbor Gabriel and his son took the hard work of digging the
channel. Here you can see the beginning of the open channel that will bring the
water from the stream to the forebay. A very small dam, made from wooden boards
and some rocks, is foreseen to raise the water level a little bit to fill the

The open channel has a rounded cross section
and runs with a very small slope. I don’t know yet how high it will fill and
thus how fast the water will flow. I also haven’t decided yet whether to line
the channel with concrete (good but expensive) or with polyethylene film
(cheaper but not nearly as solid). In any case, some lining is required, or the
water will wash away the soil. Suggestions about this are welcome.

The flat area is where the forebay will be
built. The intention is to square the channel there, deepen it, provide a silt
basin with drain valve, then a trash filter, to end in a tank that has a water
level at least 50cm above the tubing. Again, suggestions for this part are

From the forebay, the 250mm PVC pipe
will run underground through this trench.

The pipe will have to cross a short stretch of
muddy terrain. As expected, when digging the trench through it, water seeped up.
I hope to get away with simply burying the pipe in the mud, which is stabilized
by grass roots. Here the pressure in the pipe is still very low (about 0.3 bar)
and the run is rather straight, so it seems possible. Suggestions are welcome,
in any case… 

After the muddy area, the trench runs through
the forest in stable soil, almost horizontally.

Then it reaches a place where it needs a bent
downwards, to start a long downslope where we will finally start gaining some
head. At this point a vent valve is planned, but I’m still a bit worried by the
need to bend the pipe so much. Gabriel and his son dug very deep at this place,
to soften the bend, but still it is considerable. We brought one section of pipe
up, and bending it looks pretty hard. This class 2.5 piping has such thin walls
that I fear melting it, burning through, or compressing and breaking the pipe,
instead of bending it. Any ideas?

This is as far as we got until now. Work
continues. The trench now has to cross the road, then go through a long stretch
of grassland with constant slope and easy terrain, then it has to cross the
stream on a wooden bridge we have to build, and then the real fun starts:
Running it through forest so dense that it’s hard to climb through!

The machinery foreseen is a Chinese brushless
synchronous generator with internal voltage regulator, mounted vertically on top
of a pyramid casing containing a Turgo runner built from 16 spoons of the
“orange” size, The
generator is waiting for me to pick it up, the spoons are already on hand, the
rest has to be built. The turbine will use four jets. For the nozzles, I intend
to use these concentric reductions: 

They will be welded to flanges in the position
shown here. The big side will get hose connections for 4″ hose, while the small
side will get homemade nozzles. The first nozzles will have 27mm outlet
diameter, and I intend to turn them from nylon and give them a simple conical
shape. Later I will probably make a range of nozzles of different sizes. The
idea is to be able to exchange nozzles simply and quickly, by removing the
whole injector unit from the turbine casing.

The four 4″ valves and assorted fittings have
been ordered but not yet arrived. I don’t know yet how to make the distribution
box that has to connect to 250mm PVC pipe on one side, and to four 4″ brass
valves on the other side, with a pressure gauge on top. Once more, suggestions
are welcome! 

News! The generators have arrived:

These are Chinese-made brushless synchronous single-phase
generators with electronic voltage regulators. The manufacturer is Mindong
Wanlong Electric.Their low price made it possible to buy two of them, so there
is a spare, just in case there is a failure.

Here you can see one of them hoisted into the air for
unpacking. Its companion sits in the back.

At the same time, I’m finishing the detail design of the
turbine. Here is a quick glimpse of a sort of schematic drawing, seen from the
side, which also illustrates one of the four water jets. It’s a traditional
vertical axis Turgo. Comments and suggestions are welcome.

Meanwhile, work on the penstock
installation continues. Gabriel provisionally put the tubes in the trench, to see how
they fit. Here is the first tube, beginning right where the
forebay will be. I asked him to make the trench a bit deeper
here, because as shown there wouldn’t be enough water over
the tube.

From here, the penstock runs through
forest, swamp, and more forest,


and then through grassland.

From here the pipe crosses the creek and continues on the other side, through
the forest, for its last 120 meters. 

Digging a trench through the forest is hard work! It was all done
by my neighbor Gabriel and his son.

The final 24 meters of penstock
drop sharply, developing 30% of the total system pressure. For
this last run only, blue class 4 pipe was used. All the rest is
the much cheaper white class 2.5 piping. 

And here the
penstock is seen from the turbine’s perspective!

Now the final installation of the pipe can start!
It has to be connected using rubber rings, and it has to be bent in many
places. To bend this kind of pipe, many people use blowtorches. This is a quick
method, but a dangerous one! It’s easy to burn a hole into the pipe, or at least
severely deform it. It’s also possible to cause a fire. So I invented a PVC
pipe bender. It consists of two electric quartz heating elements of 500 watts
each, installed in a wire cage that slides into the tube. An aluminium foil reflector keeps 1/4 of the pipe circumference cool,
to act as a stiff hinge, while the rest of the pipe will soften,
in a length of about one meter.

In use, the
pipe bender is fed from a portable generator. Here you can see the work
at the sharpest bend my penstock requires.

This was a typical bending setup. Gabriel
acting as center point, his son as end load, and I
as the controlling factor on the front of the pipe section.This piece of pipe had to be
bent in three places, to give enough curving without too much weakening. We worked
into the night, as you can see!

The red glow from the tube bender can be seen here. The
heating is almost exclusively by infrared radiation.

And this is the result.

To assemble the pipes, we cleaned the ends, installed the rubber seal, marked
the insertion length, chamfered the pipe end, brushed
on some soap-based lubricant, plugged the the tube ends together, and then used a lever hoist and
some ropes to pull it fully in.

The assembled pipe
is already reaching out of the forest!

And soon we were at the point where the pipe had to cross the road.
My valiant helpers dug a trench through the road, we installed the pipe there, and
then they closed the trench, all within two hours, so
that we didn’t cause an excessive interruption to the truck traffic going to the house
construction site! Everything is running concurrently here!

At several places we installed air relief
valves. Here is such a place. A hole is cut into the
tube, a collar is installed, and an extension tube is screwed in which will
later carry the valve at its top.

Before installing the top pieces of tube, I
wanted to find out for sure if I would get enough water level over
the tube, to avoid vortex formation. For this purpose, we fed some water into
the channel leading to the future forebay.

The water flowed nicely along the channel,
but it was clear that the slope was not constant, so I asked Gabriel to deepen
both the second half of the channel, and the trench
for the first two sections of tube. As you can see here,
the water flows quickly down to the rock in the middle of the picture,
and from there on is almost stagnant.

Here you can see the water pooling up where
the forebay will be. Statically, there was enough water level, but when the
water is flowing, the level will be lower. So we need some more depth for the
tube. The soil is so dry after more than a month without rain, that it soaked up
all the water we could get in! The creek was running very low, only about 10
liters per second. We fed about 2/3 of the
creek into the canal, and it took three hours
to fill to this level! After stopping the water supply, it dried up in
20 minutes. Clearly some lining is necessary!

And here ever-efficient Gabriel

has deepened the trench for the tube!

When I bought the plastic tubes, I bought a
few extra, just to be sure I had enough. I
had measured the distance by GPS only, because at the time I bought the tubes, it
was not yet possible to get through the jungle by the exact path the penstock had to take.
As we advanced with the work, at times it looked like even
this excess would not be enough! Then again it looked like only one length of tube
(from a total of 84) would be left over. Finally, we finished just right,
with all tubes being used, and only one and a half meters of the
last piece left over! Here Gabriel is cutting off the only little piece that is left! Murphy
lost the game, one time at least!

After the tubes had been installed, I asked Gabriel
to quickly close the trench, to avoid pipe degradation by ultraviolet
radiation from the sun, and also to let nature start fixing the damage we did.
If any joint leaks enough to become a problem, I guess we will notice even so, and can dig it
up again. Anyway we can’t very well pressure-test the tube without covering it, because
it would buckle up and become disconnected!

The soil is full of roots
and seeds, so as soon as it rains again, this scar should get green
and merge into the rest of the landscape.

In the forest we laid the tube into a trail, which now just
looks a bit more open than before.

My penstock has to cross the creek. For this
we had to build a bridge. I used some old timber laying around, and engaged my
sister’s workforce for the task.In a month’s time she has to play the Baroque
flute in a concert of Bach’s Musical Offering, and joked that this was a
great way to rehearse for the concert!

The ready-cut pieces
have been carried to the construction site.

And here the bridge is in place! We installed it
on four concrete blocks, to avoid having the wood in contact with the ground,
which would make it rot too fast.

Later my trusty neighbor covered the tube
with earth, up to the bridge ends. I will build a roof
for the bridge, to protect the tube against the sun, even if here in
the forest it gets rather little sun.

Finally the canal was deepened, to maintain
enough flow speed and avoid sedimentation in it.
I prefer having any sediments settle down in the forebay. My sister serves as
measure of the depth of the canal.

Gabriel took a well-deserved week off, and in
the meantime I’m doing some steel work. Here the frame for the trash rack is
about to be welded together.Note the special model table! Its top is a leftover
from building the transport structure for my flying machine, two sides are
leftovers from building the big box I use
in the car, while the other two sides are scraps left from building my kayak.
And the bamboo legs are locally grown!   

The trashrack is ready! It is 1m tall by 65cm
wide, and will be installed underwater in a 45 degree angle. The spacing between
bars is 16mm. The mounting frame on top of it
has only three sides yet, because I ran out of angle stock due to a
last minute design change. This is the first thing I ever welded, so I’m glad that the resolution of the photo is
too low to show all the spattering and rough finishes! The next item I
have to weld is the turbine housing.

I went right to the work, and here it is, the
basic part of my turbine housing, still hot from the welding!

And here you can see it from the top. The
welding is not perfect, because I’m powering the 5.5kVA arc welder from a 2kVA
generator, which is not a very good thing to do… But I got the job done

This turbine case will mount on the frame at
left, which in turn will be cemented in. The bolts are welded to the

I built this wooden mold for the water outlet
duct of the power house. It will have a larger opening for the turbine, and a
smaller one that can be used as a general sink, or for a second, smaller
“summer” turbine. 

The turbine mounting frame got some spikes
welded on, to anchor it in the concrete.

Gabriel is digging the platform for the power

And then the hardest work started for him:
Hauling in about three tons of gravel, sand and cement, by wheelbarrow!

All this material had to be mixed by hand.
While he mixed and poured concrete, I finished the steel structure. In this
photo the outlet channel is already made. Note that this channel also acts as
main anchor for the pushing force exerted by the water in the penstock. In
addition, there are two smaller anchors in the corners of the concrete slab.

While Gabriel continued making and pouring
concrete, I cut the penstock to length, installed a cap,
and installed steel and mould for the tube anchors. 

The concrete work is ready!

After the concrete had set, I built the 
power house on it. Here you can see the finished structure. All wooden pieces
were cut to size at my little cabin, then hauled to the site and assembled
there. Incredibly enough, they actually fit! 🙂   

Metal sheeting was used for both the roof and
the walls. I will probably paint it later, when there is more time. Also the
wall sheeting still needs a few more nails. This is the first house I ever
built with my own hands, so watch it with due respect!

You might remember that my penstock has a 135
degree elbow. We cast concrete around it, to make a nice anchor block. We filled
in concrete to about 10cm above the plastic.

The power house being ready, work at the
other end of the system needed to be done: The forebay. Here is the mold and
steel structure for it, with the frame for the trash rack already installed.

The mold looks quite complicate. This corner
has the overflow chamber.

We lowered the whole thing, steel and all,
into the hole dug by Gabriel. We got caught under the behemoth (2.44 * 100 * 122
cm) and had some trouble getting free again! Then we poured the concrete for the

When the floor had hardened enough, we slowly
filled the walls.

My largest fear was how to haul all the
gravel, sand and cement in. Here we needed slightly more material than for the
power house, and the trail to this place is steep uphill. Not even Gabriel could
push a wheelbarrow full of stones up this trail! Ever resourceful, he
went to his land (neighboring mine) and came back with the perfect tool for this
task: His horse, and some sacks.

After allowing the concrete to harden
for a week, the molds were removed. This was really difficult! The molds inside
the overflow chamber had to be burned through with a blowtorch, not easy at all
with wood totally soaked by the wet concrete and the rain!

Unfortunately, what emerged under the wood
looked really terrible. Despite all our attempts to properly compact the
concrete, things went wrong and the concrete looks roughly like a Swiss cheese.
At many places the little sand in our mixture had failed to get past the many
stones, leaving lots of air spaces. You might not believe how hard it is to get
sand here! The material I got for the forebay was a natural mix of sand with
stones of assorted sizes, but clearly it had too many stones and not enough
sand.  To avoid further delaying matters, I decided to try using the
forebay like it is, rather than trying to line it with another layer of
concrete, or demolish and rebuild it, which would be the only really good thing
to do. At least, the steel frame that will hold the trash rack got a nice
coating of anticorrosive paint…

In the powerhouse, the mounting frame for the
turbine also got anticorrosive paint. 

You might imagine that anyone visiting me
here has to work! When the Capt’n of the Capricornio came for three days
(look in the Nauticus section of this web site for stories about my trips
with him aboard his sailboat!), it was time to finish the feed channel. Here you
can see him doing grueling shovel work, in the midst of autumn fog. 

I lined the channel with plastic cloth, and
tried to connect it to the forebay with some concrete. But the night after I
tried this, it rained unexpectedly. Water got under the cloth, lifted it up,
washed away much of the uncured concrete, and destroyed my work.

To prevent this from happening again, I then
connected the cloth to the forebay using a steel frame installed with screws.
It’s not 100% watertight, but with some luck it might be good enough. At this
time, winter is coming, the creek is rising, and I want to get my system running
as soon as possible!

Now I have to install the technical
equipment. This is a tube collar with valve and hose barb, of which I will have
four in the power house.

Unfortunately the collars and nipples are of
horrible quality. The collars are cast aluminium, terribly porous, almost as bad
as the concrete of my forebay… The threads seem to be made in casting. And the
nipples are zinc dipped cast iron. This material is brittle, and many chunks of
the threads are simply missing. On top of that, the threads of the nipples are
far smaller than the specs require, so they screw fully into the collars before
their tapers even start to adjust. The valves have better threads, but very
much shorter than the minimum required by the specs of the BPST thread. They fit
the nipples so loosely that they almost fall off! I tried to wind on an almost
grotesque amount of teflon tape, and assembled the stuff, but I fully expect
these assemblies to leak everywhere. If the leaks are not too bad, I just will
build dams and canals in my powerhouse to channelize the leak water away… but
if they spray all over the place, I will have to find another solution.
Suggestions are welcome.  

Here you can see a detail of how the nipple
protrudes deep into the collar, where later the plastic penstock will be. I
solved this problem by grinding off the protruding parts, after taking this

The collars come with their rubber seals. Oh
well. Two were usable. Another looked like this, with a big break. And the forth
simply was missing. I had to make replacements for those two, but I can’t make
them with the two ridges, which might be necessary to seal against the very
rough cast aluminium. It’s really a disgrace to have to deal with such low
quality products, but unfortunately these are the only I could get.

While fair weather days were used to do the
outside work, on rainy days I did indoor work like machining the nozzles. Once
again, my hobby lathe is providing good service! I can even make the tapered
threads on it!

Here is one finished nozzle, mounted on a
reducer. It’s conical inside, while the outside is a simple cylinder to allow
easy gripping in the lathe.  It’s made from a polyamide 6. The exact brand
name of the material is Ertalon 6SA. 

Looking through the reducer and nozzle! The
perspective is a bit misleading in this closeup photo. In truth, the hole is
about one quarter the diameter of the thread.

And here you can admire all four nozzles, to
be used initially. Three of them have the nominal 27mm bore, for 12 liters per
second. The fourth is a “half nozzle”, with a 19mm bore that will give a flow of
6 liters per second.

The turbine case was completed next. The top
plate got some cutouts to allow easier mounting of the injector assemblies, and
was welded on top of the side walls in a somewhat twisted position. While
unorthodox, this position allows the easiest access to all screws. Nobody has
ever cast in stone that all things have to be perpendicular to each other!

Test fitting the turbine case to the
generator. The case can be lifted by a single person, but to bring it close to
the generator without bumping it, and then installing the bolts, I preferred to
use my hoist. Anyway I needed the hoist to move the generator to the edge of the
container door, because thatone cannot be lifted by one

The task on hand now was welding the
concentric reducers to their mounting flanges. The flanges allow fine tuning of
the position of the complete assembly, but the coarse alignment has to be done
by rolling the reducers in the flanges, before welding them. For this purpose, I
installed a few turbine spoons, with the rear piece of the runner, so I could
aim the reducers and tack weld them.

Very coarse aiming can be done visually, but
this is of course not precise enough! For a photo it’s nice, though!

I made a simple injector alignment tool.
It’s an unfinished nozzle that has a straight bore of 12mm diameter, into which
a 12mm brass rod fits snugly. The rod can be pistoned in the pseudonozzle.

Using this tool, it’s easy to see exactly
where the water jet will land. The only problem was to decide where exactly
on the spoons the jetshouldland! Good information is hard to
come by. I finally made the center line of the jet hit the spoons about 8mm
below the top surface, at the point where the spoon center line is
perpendicular to the jet. Also, I placed the jets on a slightly smaller radius
than the center of the spoons are, because this will keep the outer parts of my
rather thick jets from missing the spoons. In any case, the position of the jets
can be fine tuned later about 8mm to any side.   

You can see that holding things together for
tack welding is not so very easy! A helper would have been welcome, but I had

And one more photo, where you can see
one nozzles and the alignment tool in place. Tack welding the underside
nozzle was particularly funny, lying on my back in a forest ants nest in
front of the container, and welding overhead at short distance!

After tacking was complete, I fully welded
the injector assemblies, and then these assemblies and the casing finally got a
much deserved coat of anticorrosive paint, which was then hidden under mandarin
red paint, a tribute to the Chinese generator this turbine will move!

The finished turbine case, complete with
rubber seals, ready for installation!

The machined pieces for the turbine runner
are the best made parts of the whole system. They were made from stainless steel
in Oscar Quijon’s workshop in Santiago, to my design. Here you can see the
main piece, mounted on the generator axle.

The complete runner was test-assembled, to
check the fit of the spoons, which have to be machined a little to be used on a
16-spoon runner.

This shows how the spoons sit on the runner
back piece. 

When I was satisfied with the fit, I
assembled the runner, complete with water thrower disc, and with Loctite thread
sealant in all threads and mating surfaces.The ugly scratch on top of the water
thrower plate was my fault, when my Dremel tool slipped while making the key
cutout in the disc, which was not foreseen originally and thus not made by the
workshop. The scratch is very shallow, in any case.

And here the same thing from below. I torqued
the bolts to the optimal force calculated by my friend Michael, who is an expert
in the matter.

The only problem was that the Chinese-made
Allen wrench wasn’t quite up to this torque!

Allow me some artistic expression in this
series… such asHydromechanics and nature!

Finally, the runner was stored on the tip of
the generator axle, waiting for action!

Looking from “below”. Once assembled, a
bolt with large washer will hold the runner to the axle. This same hole is
threaded with a larger thread (which was funnily forgotten by the machinist in
Santiago, and had to be made in Temuco later!). Using this thread, a long bolt
can be inserted and used as extractor when the runner has to be taken off the
generator for bearing maintenance. 

My storage container is starting to look

Final installation has started! Here you can
see all the plumbing, in my car, on its trip to the powerhouse.

Hey, after cutting all these holes, I could
now start a flute factory!

The four valves and the manometer have been
installed. The next step is feeding water into the penstock, and watching where
it will burst or leak! There is no question that itwillleak – the
only questions are how much, and at how many places!

I also installed the air release valves along
the buried penstock.

The high voltage line from the powerhouse to
my little cabin was also built. First, I made two dozen of these assemblies,
which would be used for joining the conduit sections. 

The line consists of two 16mm conduits, each
carrying a single 1.5mm2 wire, of the common household type that has a PVC
insulation rated for 600V. I will run this line at 2kV between wires, with a
center connection at ground potential, so that each wire sees only 1kV over its
insulation plus the conduit. That should be safe even if water gets into the
conduits! The line was provisionally laid over the forest floor, but will be
buried for safety by Gabriel. Anyway I will use it at 220V only, with low power,
until the transformers are ready.


Here is a splice point. The conduits and
wires come in 100 meter lengths. I soldered the wire ends together, insulated
them with Scotch Super 33+ tape (the only insulating tape I canreally
recommend!), and then slid my joining pieces over the ends of the conduit, and
torqued them shut. These joints should be pretty well sealed.

The next step was testing the penstock and
plumbing. For this purpose, I needed some water in my forebay. The work started
by cleaning away some of the dead brush that was completely covering the creek
at the place where the weir needs to be built. 

A simple wooden weir lined with plastic cloth
soon had a trickle of water flowing into my forebay, not enough to run the
turbine, but enough to slowly fill the penstock.

To keep the dirt out as much as possible, I
provisionally installed a spaghetti sieve in the penstock inlet!

A piece of chicken wire over the main filter also helps, and the
penstock is filling!

In the power house, meanwhile several new springs were showing up. The rubber rings that came with
these collars were hopelessly useless. No amount of tightening the collar bolts would
make them seal, because the pipe deformed under them.

And the teflon tape in the threaded joints didn’t do a better job.
These threads are unfortunately of such low quality, that the teflon tape simply
cannot seal them. 

Water was dripping everywhere! And this was with the penstock just
starting to fill, with a pressure that was only 1/4 of the final working

So, I disassembled everything again, cleaned out the teflon
residues, and reassembled the threads using a tixotropic anaerobic thread
compound. According to its label, it would harden in 12 hours. I let it harden
for two days. But it didn’t harden at all, not even the slightest bit! When I
did the next water test, the situation was much worse than with teflon tape!
Now, absolutely every thread leaked!

I had also replaced the original hard rubber rings that came with
the collars by much softer ones, hoping they would better seal the collars to
the penstock. Some actually worked, but others did not. The pipe was already too
deformed from the pressure of the hard rubber rings.

The situation was really between funny and ridiculous. Note the
manometer. I have 0.1 bar in the penstock, and still all joints are

The solution to this problem was silicone caulk. I again
disassembled everything, cleaned the black useless sealant away, discarded the
rubber rings, and reassembled everything using silicone caulk. Now everything is
watertight, at the full working pressure! There was only one tefloned joint
left, at the center hose barb. I left it just to see if this joint of rather
well finished plastic with brass would hold water. When I connected the hose and
tried it, I found that it leaked too, so I had to seal it with silicone too!
Lesson learned: Teflon tape is useless for big threads, and gasfitter thread goo
is even more useless. On the small threads of the manometer connection instead
the teflon tape works well.

The metal water canal in the photo is there to take care of some
water dripping through the valves. The factory new metal to metal seals need to
set in first, before sealing hermetically. But that’s not a problem.

And then the rain came, putting a sudden and violent end to the long lasting
drought. It rained 400mm in a few days, and while I write this, it’s still
raining, with only one day of of break in between. My little almost dry creek
quickly swelled into a big torrent. Its bed clogged with dead plants, almost the
full creek went into my canal and forebay, resulting in damage to the canal.

I tried opening the weir, but couldn’t. Everything I tried just
worsened the situation, so I’m now pretty helpless and watching how nature has
its way. Specially bad is that a lot of water is going under the lining, causing
much damage. 

The cloth is moving constantly between the two layers of moving
water, and this motion makes it chafe against rocks, roots, twigs, etc, and all
such places are becoming damaged. Here you can see a tiny twig that worked its
way through the canal lining. So, the lining, a 200 dollar investment, has
essentially been lost, even before the turbine has been installed!

The canal overflowed and eroded the soil at three places. But at the
same time I’m getting about 180 liters per second of water into my forebay, more
than three times as much as my turbine can swallow! Happily, the spill gate of
the forebay works well, and there has been no damage so far.

Having enough water, it was time to start up the system! After a
solid week of continuous rain, there was one single fair weather day. My friends Manuel
and Ariela came over to help me with the hard job of hauling the generator to
the machine house. We used a two-wheeled cart for the task.

At the steep places we tied ropes to trees and used mountaineer’s
methods to slowly let the generator with its cart down the slope.

And here it is, in the machine room, dangling from the hoist, with the
turbine already attached!

A last look at the turbine, before installing the machine on the
outlet channel.

Then the machine was tilted into the vertical position and lowered
onto its pad.

A last look through one of the nozzle mounting holes, before
installing the nozzles for good.

With the nozzles installed, and two of them connected, I opened
the water valves. But I got no voltage, and the generator ran very slowly. A
moment later, it started smelling of hot plastic, and another moment later,
smoke came out of the generator’s AVR box!

It turns out that at the factory the box was miswired, so that the
AVR was disabled, and  a dead short placed across the output. This shorting
wire was a thin one, and this was the one that smoked, when the generator put
who knows how many amperes through it! Here you can see the miswired box. Note
the two power leads of the AVR, coming from THE SAME post of the voltmeter! And
then follow the thin wires between the left output post and the circuit breaker,
and you will see the short circuit.

I don’t know if this miswiring is done on purpose, for safety
reasons. But if so, they should mention it in the manual!

After flipping the two wires, everything started
running! I adjusted the voltage to 230V. The frequency is held to 50Hz +/-1Hz by the
ELC. I had preadjusted it in the lab, before moving, and no further adjustment
to the ELC was needed.

I did some load step tests, and everything seems
fine. Each full size nozzle seems to be producing a tad over 2kW, while
the small size nozzle produces about 1kW. The generator needs several hundred watts for its base
loss (excitation, friction, ventilation). With one large and the small nozzle active, I
had about 2.3kW of output power. This promises close to 8kW when
using four large nozzles, and this would be on the upper side of
my expected range! That’s good news.  Pressure drop in the penstock was very small with
these two nozzles in use, so it seems that the math was correct in this
regard too.

So I closed the machine room and went home to enjoy my hydroelectricity. But
it didn’t last long! The lights went out even before I arrived. An
impressive amount of organic debris had clogged the intake screen. Here is
a small sample of it: Some leaves, lots of twigs, and many larger pieces of
rotten wood. I have a fine wire mesh screen in place that holds all this back,
but that can’t work without clogging. Leaving just the coarse screen of parallel
bars, much of this debris would get through, so it would clog slower, but then I
fear that my nozzles will clog faster than the screen! And they are harder to

Fortunately the problem for the most part solved itself.
This huge amount of debris came in because of a combination of the flooding rain
after long drought, and running the system for the first time, which flushed all
loose debris in the channel. After cleaning everything, the self-cleaning of the
filter is now operating perfectly. There is enough flow into the loading chamber
to cause some turbulence, which keeps even the fine meshed filter clean. Often a
twig or leaf settles momentarily on the filter, only to be dislodged and flushed
over the spill gate seconds later. As I write this, the turbine has been running
for exactly three weeks without any manual cleaning required.

With the turbine running smoothly, it was time to make
the transformers, to be able to transport the full power to the house. The 600
meter long line was designed to have low loss when operating at 2000 volts.
At 230V, the losses are extremely high at anything over a few hundred watts. So,
here I go…

Here is the material. You can see the nice stacks of
silicon steel E-I laminations, two coils of enameled copper wire, two rolls of
insulating material, a can of impregnating varnish, and some spaghettis. Not
shown in the picture are some bolts, steel angle stock, the terminal
strips, and cotton straps.

The E-I laminations measure 30x25cm, and the full stack
(for the two transformers) is a bit more than 30cm high. Don’t let these low
numbers fool you: The steel weighs no less than 140kg! And there are 55kg
of copper wire, but not all of the wire was used. Only about 80%.


I do have a transformer winding machine, a
gift of my friend Enrique from many years ago, but for this large transformer
(10kVA), the machine was too small. So I crafted this setup to wind the
transformers: A simple wooden form with a crank on it, mounted on a steel tube
clamped to my desk. Below it is a support for the wire coil. The wooden form is
designed in such a way that the pieces can be individually removed once the
winding is complete. Each wooden piece was individually wrapped in several
layers of kitchen plastic wrap, to act as a demoldant and keep the
impregnation varnish from sticking to the wood. 

The bobbin for the transformer was made
simply from two U-shaped pieces of 1.5mm thick Pressspan. For those of you who
don’t know it, this is simply a very hard and stiff cardboard. It’s antique, and
only rated for class B (that tells how much heat it survives), but I couldn’t
get anything better for the purpose. Since it will be serving mostly a
structural purpose, it’s OK despite the fact that the rest of the transformer is
built with class F materials.

The two U pieces are epoxied together, and
compressed to keep them from buckling.

Two layers of class F insulating material were wound on the bobbin. I used
0.19mm thick Nomex-Mylar-Nomex laminate. Then the high voltage winding was made,
in eight nice layers, with one sheet of NMN laminate between each layer and the
next. To keep the wire from buckling out during winding, it has to be pre-formed
in the opposite sense while winding, as shown here. I used wool gloves, to avoid
wearing through my poor thumb!

The first layer of the high voltage winding is ready. Cotton rope is used on
each side to keep the wire from getting near the edges. This rope is later
removed. The flat cotton straps instead, which bind the wire turns together and
keep the whole thing from falling apart once the form is removed, remain in

The work is simple, but a bit exhausting when it comes to the low voltage
winding, which uses quite thick wire that’s not easy to bend into the proper
shape. Anyway, here you can see the completed winding, with the last layer of
the low voltage winding showing, prior to installation of the protective cover,
made from several layers of NMN laminate.

Note the many wires coming out of the form. There are eight wires from the
low voltage side, and four on the high voltage side. This is because each
winding is made in two halves to provide a center tap, and because there
are actually two low voltage windings, which allow configuring the transformer
for different modes of operation. 

The moment of truth: When opening the form, will the winding hold together,
or spring apart?

It holds together, fortunately. In this picture it’s shown ready for
impregnation. Only the top cover of the form has been removed, and the bottom
side has been sealed with some more kitchen wrap. I then applied the
impregnation varnish by simply pouring about two liters of it into the
winding. It will soak into the Nomex, wet and cover every piece of wire,
gluing everything firmly together, moisture-proofing it to some degree,
improving the insulation, and aiding thermal transfer.

The varnish I used is intended for oven-drying at 140 to 150
degrees Celsius. I don’t have an oven large enough to fit this coil, so I heated
it by applying roughly 20A of direct current to the high voltage winding. I
monitored the temperature by measuring the resistance change of the wire.

After drying the varnish and removing the wooden form, the coil
assembly feels as solid as a rock. There are some remains of roasted kitchen
wrap, which look ugly black, but otherwise the only dark color comes from the

As a last step, the voids left by the cotton rope are filled with
neutral-curing silicone sealant, which is an excellent insulator and will
survive the high temperature this transformer might attain. I didn’t do a
particularly beautiful caulking work here, but technically it’s good enough.

The core is now assembled.

Inserting the very last E!

Then the steel angle pieces and bolts were installed, taking care
to properly insulate everything, to avoid unnecessary losses and heating. I even
insulated the holes through which the bolts go!

At this point we could say that the transformer is ready and fully
functional. The only job that remains to be done is dressing the connection

And now it’sreallyready!   And then I had to
make the other one…

After finishing the second transformer, the big job that remained was moving
one of the transformers down to the machine house. My friends from Temuco were
planning to come over and help me, when suddenly, out of the blue, my long
missing neighbor Gabriel showed up. He had heard the turbine running (it’s right
across the fence from his property), and seen the outflow, so he knew it
was finally running, and wanted to see it in detail, to enjoy at least a little
bit of satisfaction from his hard work. I told him about the transformers,
and he proposed to immediately bring down one of them. Why not? So I went
for my pry bar, to use it to carry the transformer, which at its 92kg of
weight can be carried by two people without too much trouble. But Gabriel will
ever be Gabriel. “92 kilos? I can carry that alone! Don’t bother with your pry
bar.” And so he did! 

This trip made Gabriel remember the many others down this slope, with the
wheelbarrow full of gravel and sand. 

Provisionally the transformer was seated on some wooden pieces, to keep it
clear of possible minor flooding. The conduits with the high voltage wires were
fixed in such a position that in the event that condensation accumulates and
drips out, it won’t drip onto the transformer. Also, the wires have drip loops
and are in free air, so that there is no insulation problem. But anyway, at only
1000V each against earth, insulation isn’t really a problem. The center tap of
the high voltage winding is grounded.

The other transformer was installed on a small wooden pad on concrete blocks,
next to my little cabin. Later it will be installed indoors in the definitive
round house.

For weather protection, this transformer got a nice doghouse-style
structure over it. It’s sheeted in thin polyethylene, which should hold up long
enough ( a few months) until the transformer can be moved to the round house.
The roof of the doghouse seems oversized, but I did that because in the apexes
of the sides there are ventilation openings, and I don’t want the wind to blow
rain into them. These openings, and also the big one below the transformer,
are covered with wire mesh against rats, insects and birds. 

With the transformers in place, I could enable an additional nozzle at my
turbine, so it’s running on two large and one small nozzle now, generating a tad
over 4000 watts in this configuration, and consuming 30 liters per second.
Extrapolating, and considering the losses, with the four large nozzles (its
maximum configuration) it should reach very close to 7000 watts, at 48 liters
per second. Not bad for a microhydro installation conceived for a baseline
power of 5kW, with any extra being welcome. And there still is the possibility
of installing even larger nozzles and running higher power. The penstock was
designed for 70 l/s, and the turbine is supposed to accept nozzles of up to 34mm
diameter, while my “big” ones now are 27mm. At the higher flow, the penstock
loss would become noticeable (it’s almost unmeasurable now), so the power
increase wouldn’t be that large, but it might be possible to coax the system up
to 9kW, which would be pretty much at the limit of the 10kVA rating of the
generator and the transformers.

For now, however, I won’t do it, because the 4kW generated at this time
provide ample power to my little cabin, including very comfortable heating while
on the outside it’s below freezing. So I will stay at 4kW for some more time,
maybe until next winter, when I will be living in the big round house, and
needing more power.

The frequency is keeping very stable at 49.5Hz. When I next remember bringing
a small insulated screwdriver along, I will correct it to 50Hz. The voltage is a
bit less stable, as the AVR has some temperature dependance. In this photo it’s
particularly low, because I was testing the system disconnected from any
external load, with the 4kW being dissipated inside the machine house, so it was
very warm there. But normally the voltage stays between 224 and
228V.  Which means that after the drop in the transmission line and
transformers, I end up with 220-224V at my cabin, which is much more stable than
what the public grid can provide!

This is the outflow from my turbine, at 30 l/s. Poor tired water, depleted of
most of its energy! 🙂

The switching of the TRIACs in the ELC, coupled with the
inductance of the generator, causes some significant distortion to the

To get rid of at least the hardest flanks, I installed a 20uF
capacitor in parallel with the generator. This is the result:

It’s not perfect, but workable. Anyway, later on I will build a
more sophisticated ELC, using high frequency pulse width modulation, with full
filtering. Until then, I can live with this slightly distorted waveform.

After all this work, I celebrated with a turbine-styled apple pie!
After making this photo, it was baked in my electric oven, using microhydro
power, and eaten along with tea brewed using an electric kettle powered by
microhydro, in a house heated by microhydro and lit by microhydro too, while
some beautiful Schütz music was playing in the microhydro-powered music
equipment. Ah, that’s life!

After close to a year using the system in the
configuration shown here, I added a new ELC installed at the home, that
distributes the power optimally among several useful loads. If you are
electronically inclined, have a look atmy homemade ELC, and general ELC

Back toImplementing my

Read More


Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.