The saga of building an alternate fueled generator system.  


Affordable home scale fusion power is now a reality 


News Flash! It now runs on used cooking oil. Read below for details.

Amazing 22% overall energy efficiency.

"Free" Air Conditioning at last.

How to keep it cool.. read on

6/16/2006 added hookup diagram.

11/17/2006 Added electrics diagram

India elbows in on China's turf

3/5/2007 Added computer control schematic

5/1/2007 Ooops, leaking injector pump

5/8/2007 Added new computerized engine controller web page

5/29/2007 Added dis-assembly and inspection of the Changfa 195 injector pump

1/7/2008 Added cold weather starting

10/1/2009 Added a cost analysis of generating electricity with WVO

11/10/2009 -- is now


In the winter of 2004-2005 my home in the mountains of Northern California had several power outages that lasted for two or three hours. The first time the power went out, it was during the day and did not really cause any problems. The second time the power went out, it was on a Friday night while my family and I were watching a video. The power failed at the climax of the show. After waiting a few minutes, I decided to string an extension cord out to the RV which has a 3500 watt generator. We watched the remainder of the movie on generator power. Since I had the RV generator running, I decided I might as well hook up the refrigerator and freezer. Once the power came back on, I began thinking about getting a standby generator to keep things going in the event of an extended power outage.

The biggest problem with the RV generator is that it is 120 volts only. My home is on a well for our water, so when the power goes out, we only have the water that is in the presure tank. This is enough to flush the toilet a few times, but eventually the presure drops to zero. I also wanted a generator powered by diesel because my home is heated with "home heating oil" which is code name for "diesel". I have a 300 gallon tank which the fuel oil company keeps full, so I would never have to worry about storing fuel for the generator.


I began searching Ebay for diesel generators, and ran across a store that specialized in selling engines and generator heads for about a quarter of the price of buying the finished product. This was my first contact with the slow speed diesel engines known as "listeroids". These engines are made in India copies of Lister engines which are known for longevity. From the Powersolutions listing on Ebay, I found out about This website is run by George Breckenridge who is the guru of do-it-yourself power. I read everything that George had on his site. I ordered the "Utterpower Lister Longevity CD" and read everything there.

In addition to the listeroid engines, Joel from powersolutions was also selling Chinese single cylinder diesel engines. These engines were made by Changfa and run at 2000 rpm instead of the 650 of the listeroids and included electric start. The electric start plus the cheaper cost of the Changfa settled the issue for me. I still lust over the 12 horsepower 2 cylinder listeroids, but my checkbook keeps holding me back.

Used Cooking Oil as Fuel

Back in 2002, at a family re-union, my cousin Keith mentioned that he was in the process of converting his Dodge pickup to run on used french fryer oil. At the time I though this was the most crazy thing that anybody could do to an expensive engine. Later on, I found out that Keith was not crazy after all. It turns out that nearly any diesel engine can be made to run on vegetable oil by heating the oil to the point where it approaches the same viscosity of diesel oil. Hundreds of thousands of miles have been driven in cars powered by the oil that restaurants throw out every week. This waste vegetable oil (called WVO) can often be collected for free because the restaurant usually has to pay to have it hauled off. A great deal of information about using WVO can be found at the infopop WVO forum.

One power outage a year is hardly enough to justify spending over a thousand dollars piecing together a generator to be there just in case. Now, with free fuel and a generator large enough, I could run my entire house and lower my power bill. This was the final piece of the puzzle regarding my standby generator. 

Update 10/1/2009 

In the years since I first started working on this project, I have found that it is nearly completely impractical to try to "make money" by running a generator on WVO.  See my cost analysis at the bottom of the page.

The Purchase

In September of 2005, I bid on a 12 horsepower Changfa 195 engine with a 7.5kw generator head. Upon winning the auction, I loaded up the family and drove up to Oregon where we picked up the engine and generator. The engine and generator were in wooden crates which were strapped to a pallet. They fit nicely in the small utility trailer towed behind the van. I had to wrap the crates in a tarp, because it likes to rain in Oregon. We had rain from Portland to Eugene on the way home, but nothing got wet.

Here is my dad with a hammer and crowbar working on getting the lid off the crate. My first impression on uncrating the engine was "It is bigger than I expected". I knew it was going to be big, but somewhere in the back of my mind, I was expecting something like an oversized lawnmower engine. The Chanfa 195 engine is bigger than the engine in the Hyundai car I once owned.

Engine Inspection

For listeroid engines, you basically get a kit of pieces that you first must dis-assemble, clean and re-assemble before you even attempt to run the engine. The Changfa engines have much better quality control. In any event, I removed the crankcase cover just to be sure that everything was ok. Here you can see the crank at BDC with the safety wire in place as it should. Above and below the crank are the counterweights which do a great job of keeping this engine from jumping around like a jackrabbit. (Listeroid engines don't have these weights, and some of them are almost legendary in their ability to break their hold-downs). I added a small magnet in the oil pan to catch any bits of metal that might break loose. I also pulled the rocker arm cover and added a magnet there. The oil pickup was not quite hoizontal and needed a little adjusting. Other than that, the engine looked great.

Built in Battery Charging Alternator

when I first got the engine out of the box, I noticed what looked like a sealed voltage regulator bolted to the same bracket holding the starter switch. I assumed that it was for a generator that was not included with my model engine. However,
once I had the starter and crankcase cover off, I could see behind the flywheel, and I noticed 3 electrical type terminals. These intrigued me. Nobody had said anything about a generator on these engines. So, to get a better look, I pulled the flywheel. What I found was a 3 coil permanent magnet alternator. Since that time, George has added a webpage about "Changfa Electrics" which goes into hooking up the Chanfa Alternator. The diagram shows two wires from the alternator going to the input of the regulator and the third wire connecting to the output of the regulator. This puzzled me and turned out to be totally wrong.

The Changa charging page held the clue to solve my puzzle (Thanks Barry). I originally thought the three coils connect to the three terminals in a "Delta" configuration. It turns out that two of the coils are connected in parallel, and the third is all by itself. The two coils in parallel drive the battery charging circuit, and the third coil is used to run the headlight.

If my injector line heater takes more power than the single coil can supply, I will rewire the alternator to a delta configuration and use a 6 diode bridge to rectify the power before feeding it to the regulator. I will then run the injector line heater from the unregulated side, and charge the battery on the regulated side.

update 4/10/2006

Further investigation reveals that there are 6 magnets mounted on the flywheel. Assuming they are arranged in alternating north/south orientation, this means that all three coils are perfectly in phase. Previously, I though there was just a two or four pole magnet which would have placed the coils out of phase with each other. Based on this, I can simply run the coils in parallel and drive my injector line heater off the A/C before it is fed to the regulator.

Update 11/17/2006

Here is how I have my generator and regulator hooked up. (Click the picture for a full sized view). The first thing to note is that both therminals 2 and 3 of the battery charging alternator are tied together. This places all three coils in parallel for maximum power. Next I am using a bridge diode to rectify the A/C before feeding it to the regulator. The only reason for this is that I installed my injector line heater with one end tied to ground. By using the diode bridge, the negative side is effecivly tied to ground.

When the "ignition" switch is in the on position, the battery is hooked up so that it is charged while the engine is running.

If I had been thinking, I would have not tied one end to ground, but brought it back to the regulator and tied it in to the other input to the regulator. Then I would not have needed the bridge diode. Here is what a preferred solution looks like:

When I first had the electrics hooked up, I wired them just like the changfa manual calls for. This shows terminal 3 tied to the output from the voltage regulator. The result is that I was overcharging my battery. I also had the injector line heater hooked up so that the "ignition" switch controlled the heater. That saved me a switch, but if I forgot to turn off the switch when I stopped the engine, the heater stayed on. As a result, I would first overcharge the battery, and then totally discharge it by leaving the heater on. I managed to kill two batteries before I caught on. Since that time, I have corrected both the overcharge problem, and the discharge problem. Now the only problem I have is that I have blown my battery budget for the next two years.

Dec 8, 2006, Marty emailed asking about how to hook up his Changfa 195 which has a 4 terminal alternator. In his case, there is a common terminal, and each coil has its own terminal. Here is how I suggested he hook his charger up. Note, since he is not running an injector line heater, he may not need to have all three coils producing power, so I indicate that hooking t3 is optional. I have found that the "Regulator" is not much more than a bridge rectifier and it does not do much in the way of regulating. It is all too easy to overcharge your battery if you are not careful.

Generator Head Inspection

On the utterpower CD, George mentions that the ST style generator head sometimes needs adjusting before it can be used. Mine was no exception.

This was my first disappointment. The brush holders and some of the metal inside the head were rusted. The brush holders were crooked, and the brushes were not properly aligned on the slip rings. The slip rings had overspray from when the case was painted. To the right you can see the crooked brush holders. On the left, you can see how the brushes themselves are not aligned with the slip rings as well as the paint overspray. The windings of the stator and rotor might have been dipped in lacquer to seal them, but it does not show. The plastic around the slip rings looks as though somebody beat on it with a hammer. If I were buying a used generator, this is what I would expect it to look like.

Here are the brushes after I have cleaned the slip rings and aligned the brushes. I have also moved the brush holders so that there is no more than a quarter of an inch showing. These brushes are only about 3/4 of an inch long. The factory shipped this with half an inch of brush showing. Next the brushes need to be shaped to fit the rings. This is done by using 300 grit sandpaper between the ring and the brush.

The last thing that George mentions is pulling the bearings and replacing the Chinese "yak fat" with a normal grease. I was in a hurry to start making electricity so I decided to forgo this step. I will probably end up doing this soon, because the generator bearings growl for the first few minutes of operation. While I'm at it, I'll replace the bearings with US grade bearings.

The Assembly Begins

I decided that I would direct drive the generator using a Lovejoy coupler. I found a good deal on a Lovejoy L150 on Ebay. The only problem was that it was sized for a .875 shaft instead of the 38mm that I needed.

I bored out the shaft on one half of the coupler to 38mm and broached a 10mm keyway. On the other half, I drilled and tapped 6 5/16 X 18 holes where the coupler bolts to the flywheel adapter. Here I have mounted the flywheel half of the coupler on a mandrel on the lathe and am truing up the face of the coupler. Next I bolted the flywheel adapter onto the coupler and I trued the whole thing to about .001 inches runout George from Utterpower sells a really slick flywheel adapter for these engines. However, I decided to build my own. I started by taking measurements of the mounting holes and built a wooden pattern. Then I cast the adapter out of aluminum in my home foundry.

Next I built the frame. The frame is built from 2X2 heavy walled box tubing. Here I am milling slots in the generator head subframe I don't have a milling machine, but if I go slow and take light cuts, I can do some milling on the lathe. The reason I don't have a milling machine is that the money I was saving towards getting the mill went to buy a diesel engine and generator.

First Electricity

After bolting the engine and generator to the frame and carefully aligning and shimming everything, it was time to start it up. For a load, I used a 4 burner cooktop which draws about 5000 watts. The engine shrugged this off as if it were hardly worth noticing. After a few minutes of running, the cooling hopper started to steam (as seen above). After about 15 minutes of load, it was boiling violently (as seen to the right). At this load, the system boils away about a half gallon of water per hour. It probably splashes at least that much out which runs over the hopper and dries leaving a mess. You don't want to run this thing unattended with hopper cooling. For the final configuration, I will be using radiator, or tank cooling via thermosiphon.

Here you can see that the engine is actually producing work. On the right is the waveform while the generator is putting out 5000 watts. It is not the prettiest sine wave, but none of my appliances seem to care.

If you have a fast connection, Here is a 2.4 Megabyte WMV file of the engine running. This will give you a bit of an idea how much noise this thing makes. I have to wear hearing protectors while it is running. Eventually I'll add an automotive muffler which should help. Also, the air intake is surprisingly noisy. You will also see the amazing amount of steam this thing puts out when under load. This was recorded when the outside temp was around 35 degrees, so a little steam goes a long way.

Work in Progress

Running on used cooking oil

After 8 months of planning and building, it finally runs on vegetable oil. This photo shows that Rube Goldberg is alive and well. The wooden clamp is holding my jury rigged governor spring holder. The top valve is for the injector return line. the bottom valve switches between UCO and diesel. On the bottom valve the line from the left comes from the VO tank. The vegetable oil is yellow in color. The line from the right comes from the diesel tank. The diesel is red in color because it is the non taxed home heating oil. The color difference makes it easy to see the fuel switch occur.

Click 'play' to see a video of switching to UCO. Watch for the color of the fuel to change from red to yellow in the line feeding the injector pump. Switchover time is about 30 seconds when driving a 5kw load. At low load, switchover is about 2 minutes.

When I first switched to UCO, I was expecting the engine to die because there was a bit of air caught in the line. Instead of dying, it kept running without any change. My first indication that the engine was indeed running on UCO was the change in exhaust smell. Everybody talks about how nice the exhaust smells, but I have to tell you that it does not smell nice. It certainly does not smell like french fries or anything else that I would want to eat. To my mind, it smells most like what I smell when I spray 'Pam' on my hot barbecue before I put the meat on.

Cooling System and UCO Pre-Heat

Before adding the radiator, I cooled the engine by hooking up a garden hose to the drain valve on the engine. I adjusted the flow of water to keep the temperature at 180 to 190 degrees F. Unfortunately, I had to keep adjusting the flow of water to match the load on the engine. Eventually I added a thermostat and then a radiator.

To the right is a diagram of how the cooling and UCO preheat connections are done. (Click to see a full size view).

One goal of the design was to not require a water pump to move the coolant. As a result I had to route the hoses carefully to prevent any high spots which might trap air and stop the thermosiphon. I was initially concerned that the UCO preheat loop would not siphon because of the small size of tubing. Once I got it hooked up, it siphoned just fine even with a few inches of upward loop at the bottom.

This diagram shows how the Diesel/UCO valve is configured.

Overall energy efficiency

In researching different generator installations, I ran across Jack Belk's Listeroid installation where he has a power meter hooked up to keep track of the amount of power generated. This seemed like a good idea, so I scrounged around on ebay and came up with a used power meter and added it to the mix

My first run after hooking up the meter involved running a week's worth of laundry. This week we needed three loads to get everything clean. I was surprised to find that this needed only 18kwh of electricity, or about $1.44 savings on my power bill. I'm going to need to do a lot of laundry before my engine can pay for itself.

My second test run, I decided to measure the efficiency. I used a dipstick to measure the amount of fuel in the tank before the run and after the run. Since the tank is essentially a cylinder on end, it just takes a bit of math to determine the amount of fuel used. It turns out that I used up 2 13/16 inches of fuel which works out to about 2.7 gallons. These 2.7 gallons of fuel produced 11kwh of power. This works out to .127 gallons of fuel per kwh generated. This is almost exactly the same that George reports on his website for a Listeroid 6/1.

Assuming 120000 BTU of energy available in a gallon of vegetable oil, and 1 BTU equivalent to .293 watt-hours, I am using the energy equivalence of 4.466 kwh to produce a single killo-watt-hour of usefull electricity. This represents 22 percent overall efficiency. To increase the overall efficiency, the easiest step will be to utilize the otherwise waste heat at the radiator.

Thermosiphon cooling adapter plate

In order to run this engine unattended for long periods of time, I decided to convert from hopper cooling to thermosiphon cooling. This is done by taking off the cooling hopper and installing an adapter plate. Again, George from utterpower sells these premade, but for me the making is part of the fun. Here is my first attempt at casting the adapter plate. It turns out that the thermostat mount takes up more room than I planned on, so I had to recast this without the lug in the center. Details about casting the hopper converter are on the lost and foundry page.

The plate is designed so the thermostat is mounted on the end of the cylinder closest to the head. This is where most of the heat originates. With the cooling hopper on, you can see the water boiling up from that end of the cylinder. The lugs on the cooling plate are where the return line from the radiator will connect. The reason for the raised lugs is there will be a 3 inch tube on the underside leading as low as possible to promote the thermosiphon.

UCO fuel tank

The UCO tank is made from an old 20lb propane tank. I have far too many of these since I had to replace them with the new kind with the overfill protection valves. The tank has a half inch copper tube running through from top to bottom. This carries coolant from the engine to pre-heat the oil. Outside of the coolant line, there are fittings for fuel supply and return from the injector. Below and left of the tank, you can see the final UCO filter which is a 10 micron "Goldenrod" filter wrapped with the hot water lines.

A word of advice for anybody wishing to modify old propane tanks... This tank was at least 15 years old, and had not had any propane in it for at least two years. I had left the valve open for at least a month before I started to work on it so I knew there was no propane in it. I took the valve off, turned it upside down on my drill press and proceeded to drill the hole in the bottom. Everything went fine until I took it off the drill press. A thick oil had run out of the tank onto my drill press table and it smelled just like propane. Apparently the oderant which is added to propane does not vaporize entirely with the propane as it is used. Over time it builds up as oil in the tank. My drill press and shop smelled like propane for a week.

Thermostat Housing

The thermostat housing serves as a thing to hold the thermostat and a place to hook up the radiator hose. Here I am machining the relief on the bottom where the thermostat sits. Initially I installed the thermostat and ran low presure water into the drain valve at the bottom of the cylinder. This worked well until I got the radiator hooked up.

Cooling tank/radiator

Somewhere, I saw a guy with a genuine Lister 6/1 with a 20lb propane tank for cooling. My initial idea was to use two old propane tanks stacked one on top of the other for cooling, but as I think about the amount of heat I need to shed, I expect that I will find this to be inadequate. I suspect that the guy with the Lister is not loading his engine down very much.

A friend gave me a radiator from an old Chevy car. Hopefully it does not have any leaks, but it should be more than adequate. I may not even need a fan.

Update June 5, 2006 I have finished making a frame to hold the radiator and have cobbled together some hoses to hook it all together. Napa auto had exactly the right hose to go from the bottom of the radiator, but the longest 1.5 inch hose was about 6 inches too short. I made do by using a 3/4 inch heater hose. I took the radiator cap off my 1961 chevy pickup, and pulled the guts out of it so that it was merely a cap instead of a presure valve. I hooked up a hose from the overflow into a bucket nailed to the wall. After tightening a couple of loose and leaky connections, I fired up the engine and waited to see what happened. The engine warmed up to 190ºF and kept climbing. Water boils at my elevation around 204ºF, so pretty soon I started seeing bubbles come out of the overflow tube into the bucket. The temperature appeared to stabilize around 210ºF, but I was only seeing steam come out of the overflow tube. Obviously, the engine was being cooled by boiling water instead of the radiator.

I was naively hoping that I would not need a fan on the radiator. I scrounged around and found an old room fan and hung it where it wouild blow through the radiator. This helped, but the engine was still boiling water. Now I suspect that the 3/4" heater hose is too restrictive to allow a thermosiphon to function properly. I'll find a way to use 1.5" hose and report back.

More update. I found a piece of 1.25 rigid conduit which bent to a nice sweep and was just right to hook up my 1.25" radiator hose. I fired up the engine and I still had problems. I could feel the bottom radiator hose getting hot before the top radiator hose. Then I remembered the trick about drilling a hole through the thermostat to allow any air to get past the thermostat and provide a slight bit of circulation to keep the system from trying to siphon backwards. This time it worked. As the temperature came up to 190 I could feel the top hose getting warm, and then it got hot all of a sudden and the temperature dropped instantly from 195 to about 160 as the engine got a slug of cold water out of the radiator. It also turns out that the fan will be absolutely necessary. Without a fan on the radiator, the system boils water. With a small fan the engine can run with a 2000 watt load without boiling water. With a 16 inch room fan blowing through the radiator, I was able to run the engine with a 5000 watt load for half an hour with the temperature pegged at 198ºF.

Fuel Switching Valves

I picked up two 3 way valves to switch from diesel fuel to UCO. The first valve goes just ahead of the injector pump and connects to the diesel filter, and the UCO filter. The second valve goes on the Injector return line. After hooking up the valves, I found that the amount of fuel returned from the injector is very small. After running for 2 hours, the quarter inch clear hose collected about 2 CCs of fuel. I won't need to bother switching this valve, I'll just leave it routed back to the UCO tank. I was initially concerned about the low amount of fuel bypassing the injector. When I switch between UCO and diesel, I will need to burn all of the fuel in 12 inches of injector line. Based on the smell of the exhaust, the engine with a full load will be purged of UCO within 30 seconds. A lightly loaded engine will be purged in about 2 minutes. If there is any UCO still in the injector when the engine is shut down, it will be very difficult to start the engine once it has cooled down.

note 5/4/2006 It is very important to purge the UCO even when planning on an immediate restart. If the engine is shut down on UCO, followed by an immediate restart, it takes several seconds of cranking before the engine fires, and then it is a only half hearted 'pop'. It is necessary to keep cranking with these half hearted 'power strokes' until it eventually catches on and starts up with a great cloud of smoke.

More notes 6/15/2006 After I added a temperature sensor to my injector line, I found that I was not getting the UCO nearly hot enough. I have added some insulation and rewiried the heater. Now I am able to shutdown on UCO and restart within a minute or two without any difficulty.

In all of the discussions on the Infopop forum, everybody talks about looped versus return to the tank. I decided that I wanted to return to the tank to simplify purging the lines of air. Once I got it hooked up, I found that the return line returned next to nothing. My confusion about the volume of fuel in the return line was caused by my experience with the Perkins engine on a Massey Furguson 265 tractor. While repairing a leak in the fuel tank, the engine was started with the return line off and I was able to see fuel flowing out. What I did not realize was that most of the fuel was coming from the injector pump and NOT from the injectors. It seems that some injector pumps bypass a great deal of fuel into the return line.

The injector pump on the Changfa does not have any bypass, so the only fuel in the return line is that which comes from the injector. Only a tiny amount of fuel comes out, but also some air comes out along with the fuel. I just leave the return line valve set in the 'return to UCO' position. In 78 hours of running, the return line has about 12 inches of fuel in it.

Injector Line Heater

Over the last few months, there has been a lot of discussion on the infopop WVO forum about injector line heaters. Since I will be heating my UCO with coolant via thermosiphon, I decided that it was unlikely to be able to get the oil temperature up to 160 degrees F prior to the injector pump. Since by nature the highest temperature in the thermosiphon is at the top, this means that the injector pump will be at the very bottom of the coolant loop with the coolest temperatures. Currently fattywagons is selling a 30 inch heater for $22. I contacted John at fattywagons and asked him about how to deal with a single cylinder engine with only a few inches of injector line. He made a special version of his heater for me. This heater is 14 inches long, and draws about 6 amps of power.

In this picture, the injector line is wrapped in the orange high temp silicone tape that comes with the fattywagons heater. You can also see that the engine is running on UCO based on the yellow fuel in the line. The line on the upper left with the red diesel is the injector return line.

After running for about 10 minutes, the injector line at the injector is too hot to touch. Currently the UCO is heated with engine coolant lines tied to the UCO lines. This only gets the UCO warm enough flow through the filter and injector pump easily. The injector line heater brings the temperature up to the point where it injects like diesel.

Update 5/15/2006 I have installed an LM35 temperature sensor on the injector line just ahead of the injector. What I found was that the temperature was not nearly as hot as I was expecting. I was only using 2 of the coils in the battery charging alternator to run the injector line heater. As a result, I was only getting about 140º to 150ºF max. I rewired the electrical so that I'm using all 3 coils, now the line gets up to about 170ºF. When I'm charging the battery, the heaters only get to 150ºF. I'm now scrounging around to find the right sort of insulation to wrap around the injector line. I am really wanting temperatures approaching 100ºC, but at best I get 78ºC (172ºF). If that does not do the trick, I may try shortening the injector heater to cause it to draw more power, but my little charging alternator may not be up to the task.

Update June 5, 2006 I found some aluminum backed adhesive foam rated for "hot or cold pipes" which looked promising. The foam is about 1/8" thick. With a single layer of insulation on the injector line, I now am seeing 121ºC (250ºF). This is much higher than the 110ºC rating of my temperature sensor. I suppose at worst I'll melt the sensor, or at best just get an out of spec temperature reading. Even though the insulation lists "works on hot or cold pipes", I suspect they weren't thinking of pipes this hot. Now I will be able to charge the battery and keep the injector lines hot.

Lazy Governor

In testing the generator with various loads, I noticed that the unit was not putting out 60 hertz power when it was loaded down. George talks about lazy governors on listeroids, but claims that the Changfa is much better. With no load, I set the speed to 61 hertz. By the time I add 5000 watts of load, the engine speed is down to 56 or 57 hertz. The voltage has sagged down to 218 (down from 245). I know the engine has enough power to carry the load because I can manually re-adjust the speed back up to 60 hertz, but once the load goes away, then I end up making 65 hertz power at 250 volts. In asking at the Lister Engine Forum, it was suggested that I replace the governor spring with a longer spring which does not have as much change in strength per unit of stretch. The stock spring is about 1.5 inches long. I tried a spring of similar strength, but 2 inches long without any noticable effect.

Next, I cobbled together a mounting plate and a threaded rod which can hold springs up to 8 inches long depending on where I clamp the thing. Using various springs, mostly those from junked desk lamps, I tried 4 6 and 8 inch springs.With a 4 inch spring, the engine sagged down to 57.5 hz. With a 6 inch spring it sags to 58 hz. By adding the 2 inch spring onto the 6 inch spring I came up with an 8 inch spring. With the 8 inch spring, the engine sags down to 58.5 hz. However, with the 8 inch spring, the engine seems to hunt slightly plus or minus about .25hz.

To date, my best spring is one about 2.5 inches long and is not as strong as the stock spring. This spring behaves like the 8 inch spring without the speed hunting up and down as badly. My jury rigged spring mount will need to be replaced by a lever of some sort. Right now, to kill the engine, I have to reach under the diesel tank and hold the governor lever shut.

Broken Muffler

After ten hours of use, the muffler started to crack around where the pipe enters the muffler. I brazed the thing shut, but I don't expect it to last very long. For the final installation, I will need to have the muffler isolated from the engine. The brazing job broke again at 25 hours runtime. I still don't have a good muffler, so I brazed the thing shut again. This time I added lots and lots of brass. As of 50 hours runtime, my patch job is holding.

update 5/11/06 At 54 hours runtime, the muffler started rattling with yet another break. I still don't have the generator shed ready, and without a good place to mount a muffler, I am putting up with the stock unit. Eventually, the bottom of the muffler will be a series of concentric breaks all brazed together. I'm now just about of oxygen for the torch, so hopefully this weld lasts long enough for me to get a new set of tanks.

Leaking Diesel Tank

After I got the engine running on UCO, I decided to go on a "shake down cruise" with the generator powering the house. I decided to treat it just like my wife would. So I turned on the hose and let it run, turned on the dishwasher, started a load of laundry, and loaded the dryer. To top it all off, I turned on the tv and watched a movie. At the peak, the generator was putting out 6700 watts (The highest I've seen so far). Running on UCO at this load, the engine smokes, but the smoke is more greyish in color and not soot black like when running on diesel.

After running for 5 hours this way, I began to notice oil spots showing up on the concrete under the engine. I assumed that oil was leaking out from the crankcase cover so I shut down, and tightened the bolts. I found a loose one, so I thought the problem was cured. I ran up to town on an errand, when I got back, there was a pool of diesel soaking into the concrete under the engine. What I thought was a crankcase leak turned out to be a leaky diesel tank. After 5 hours of heavy load, one of the welds on the hold down bracket broke and caused the leak.

For permanent installations, George suggests that the diesel tank be removed and be replaced by something not mounted on the engine. I was putting this off until I got ready to install the engine in its own shed. I have now built a diesel tank similar to the UCO tank, except that it does not have the heating pipe through the center.

Fuel Line Blowout

At 68 hours runtime, I started the generator, switched the house to generator power and went out to the back field to move some sprinklers. When I walked back to the shed, I saw water spraying out of the engine. What I thought was water turned out to be vegetable oil spraying out of the line leading to the injector pump. Even though this line is only under gravity pressure from the UCO tank, the pulses of the injector pump was spraying oil 10 to 15 feet. I got very lucky on this one. The split could have happened where it sprayed over the muffler, or into the shed. As it was, I just have a bit of oily gravel and driveway.

My lesson is that clear vinyl tubing is wonderful stuff when trying to set up a fuel system for a diesel engine because it makes it trivial to locate air bubbles. However, long term vinyl is a disaster waiting to happen. I will need to replace all of my fuel line with something which won't crack.

Load Testing

The real test of the generator happened when I let my family do the testing. I was merely the observer while I watched my family start a load of laundry, turn on and leave on most of the lights, run the microwave, and fire up the George Foreman Grill. The grill was the 'straw that broke the camel's back'. I was reading a book when the lights dropped from normal bright light to a pulsating glow. I ran to unplug the grill then lectured my son about running the grill while the dryer was running. His response was "What am I supposed to do when I'm hungry?". I suggested a peanut butter sandwitch instead of a grilled cheese sandwich. He was not convinced, so he ate the cheese cold.

On another day when things had warmed up a bit, I tried running the Air Conditioning. I don't know the rating of our unit, but it has a 30amp breaker. The A/C had been sitting all winter when I first started it on generator power. It groaned and came to life. It seemed to be making more rattling noise than usual, but otherwise seemed ok. Since the day was not really warm enough to justify running the A/C I shut down until it got warmer. The next warm day, I tried again. This time, however, the generator was not happy at all. I watched the Kill-A-Watt meter sag from 61hz to about 50Hz. and keep dropping. I switched to reading volts. The voltage was down to 100 volts and dropping. The Kill-A-Watt started fading out at75 volts while I ran to the breaker panel to shut off the A/C. Outside at the panel as I switched off the A/C I could hear the Changfa really plugging away with HUGE clouds of really black smoke. Once the A/C was off, the Changfa picked up and was happy.

I switched back to line power and tried starting the A/C again. This time, I had my clamp on ammeter on the line so I could see the current. As the A/C started, the current peaked for a fraction of a second at 56amps, then settled down to 10 or 11 amps. I don't know if it was just a fluke that the Changfa could start the A/C the first time, or if there was other load at the time I tried starting it the second time. In any event, it appears that I may not be able to run the A/C on UCO power. this is somewhat of a disappointment because I was looking forward to a nice cool house without cringing at the sight of the power meter spinning.

Update 6/1/2006 At the suggestion from the listerengine forum, I ordered a "hard start" capacitor kit off of ebay. The thing is called a "Supco Starter POW-R-PAK'. It is a regular mallory motor starter capacitor with an electronic relay glued to the top. The instructions call for the capacitor to be hooked in parallel with the run capacitor on the motor. When I took the cover off the air conditioner, I was pleased to find a full schematic pasted on the inside. The schematic includes location for a "hard start capacitor". The capacitor is hooked in parallel with the start capacitor. I hooked the "POW-R-PAK:" to the location indicated in the schematic, and bolted it to the frame in the location indicated. I first tried starting the A/C on line power, and it started just fine. Next, I fired up the ChangFa and had my son turn on the A/C while I watched the ammeter. To my complete joy, the unit came on without a grunt. The current peaked at 28 amps while starting.

Now that I can run the air conditioning, all I need is a really hot day. We had two weeks of record breaking temperatures in May, but those disappeared to be replaced by two weeks of unsettled weather and 3 days of frost.

Generator Shed

Right now, the engine and generator live in the woodshed. This is convenient for working on the engine, because there is a heavy workbench and a vise within easy reach. However, based on the amount of noise this thing makes, I don't want it so near the house.

Update 11/17/2006 My two sons and I spent a weekend rebuilding an old shed on my property. Now I need to get backhoe to dig a trench for the power lines. In the meantime, since I don't have power hooked up to the shed, I still have the engine in the woodshed (which is quite full this time of the year).

Update 6/2007 The shed is finished, but has since been filled up with bycycles, go-carts, fertilizer spreaders and other miscellaneous junk that is too good to leave outside, but not good enough to warrant a location closer to the house.  Now I need to build another shed just to clear out space to put the generator where it belongs.

WVO/UCO filtration

I have been collecting rice bran oil from a local restaurant since last October. This has been sitting in the shed in a mostly frozen state. I purchased a set of plans from Dana Linscott. These plans are a great starting point for anybody interested in UCO as a fuel source. Based on a recent story about a guy burning down his house while trying to heat UCO on his stove, I have decided to not use electricity to heat the oil. I am using engine coolant routed through about 30 feet of copper coils in the prefilter tank.

Yes I know this looks like a mess. I managed to take a day off, and I spent the time trying to cobble things together to get the engine running on vegetable oil. As time started running short, I started getting creative in hooking things up. If you want to know how all this works, you need to get a copy of Dana's plans. My only change is to use coolant heat instead of the method that Dana describes in his plans.

Bypass oil filter/Emergency shutdown

Since I'm planning on running this engine for a long time, I would like to add an oil filter. While I'm at it, I will add an oil presure sensor and rig some way to shutdown in the event of overheat or no oil presure. George sells an adapter that allows a pressure switch to be plumbed into the oil line feeding the rocker arms. In my case, I won't be able to install the pressure switch behind the flywheel, because the built in alternator sits in that spot and covers that end of the oil line. For shutdown, the easiest solution might be a solenoid which pulls the compression release lever. It stops the engine, but it makes a bunch of smoke coming out of the intake, so I would not want to do it very often. As an experiment, I tried stuffing rags into the air inlet to see if I could stop the engine that way. It kept on running without noticing my attempts. I guess that duct tape might do the job, but if the engine was running away, it is likely that I would be running the other way.

Computer control

Eventually I would like to have the whole system automated. With auto-start and auto switchover to UCO and an appropriate purge time before shutdown. This is way in the future, but certainly feasable. At this point, I'm considering an Atmel ATMega8 cpu which includes built in analog to digital conversion and lots of I/O ports. For thernal sensing, I'm considering LM35 temp sensors. The only problem is that the temperature of the injector line after the heater may exceed the 110 degree C limit of the LM35CAZ sensor.

Update 11/17/2006 I have been playing around with various Atmel cpus. I first picked up a 'MR-8' cpu board from Jameco for $35.00. This has an Atmel ATMega8 and all of the associated stuff on a tiny circuit board with 28 pins. This was an ideal platform to learn how to program the cpu. Unfortunately, the folks who designed the board decided to tie the Analog voltage reference to +5 volts instead of leaving it float ,or tie it to a capacitor. This means that I cannot use the internal 2.56volt reference. Next, they had lousy quality control, because two of the pins on the SMT cpu where shorted together with a solder bridge. I managed to use a soldering iron and remove the solder bridge, but when I tried to lift the AREF pin, I ruined something. Now the cpu works, but it gets extremely hot, so I must have fried something.

I still think the Mega8 is the right cpu for the job, but in the meantime, I'm playing with an ATtiny26 cpu. This is a 20pin dip with 2k of flash memory and 128 bytes of ram. I can program it in C, but with 128 bytes of ram, I can't do much. I've been dusting off my assembly language skills and learning to write AVR assembly code. It has been at least 20 years since I last programmed in assembly, but I still end up typing old 8080 mnemonics, but I'm learning. At this point, I have built a software UART and got the A/D conversion working properly. Next, I will rig the unit so that it reports the temperature over the serial line.

Update 3/5/2007 Here is my first cut at a schematic for my computer control:

I found a slick stepper motor and lead screw assembly on ebay which I will use to actuate the speed control on the engine. I will pick up RPM by using a reflective optical sensor. The stepper motor is driven using IRLZ24N mosfets. These have a logic level gate which simplifies things a bit. I am using a ULN2003 driver chip to drive the various relays on the system.


  • Run/Stop (stop = low). start or stop the engine.
  • kill -- initiate emergency shutdown.
  • tachometer, 1 pulse per rev.
  • stepper limit lo: goes low when stepper motor reaches the limit at the low end of travel.
  • stepper limit hi: goes low when the stepper reaches the high end of travel.
  • water temp: analog voltage 0-2.56 volts (VDO sensor).
  • oil pressure: analog voltage 0-2.56 volts (VDO sensor).
  • Injector Line Temperature: analog voltage 0-2.56 volts (LM35 sensor).
  • serial data in: from LCD display unit .


  • start: +12 volts supplied via relay to starter solenoid on engine
  • decompress/kill: 12 volts supplied via relay to decompression solenoid
  • UCO/Diesel: 120VAC via relay. on = UCO, off = diesel
  • stepper motor: 4 coils driven via MOSFET
  • serial data out -- to LCD display

Sofware description:
To start the generator, the "remote start" switch is closed. This supplies power to the CPU as well as drives the run/stop input line. The CPU starts and sees the "run" signal. If no tach input, then the system advances the speed control to the start position, and engages the start relay for a number of seconds or until the RPM reaches a specific speed indicating that the engine has started.

After starting, the engine speed is set to about 800 rpm to allow the engine to warm up. Oil presure must reach a specific level within a fixed amount of time. Once the engine has reached a minimum temperature, the speed is advanced to 1800 rpm.

The engine speed is controled using a "PID" (proportional / integral / derivative) controller implemented in software. The normal Changfa governor is not perfect, and allows the RPM to drop when load is applied. The PID controller will keep the RPM at 1800 rpm as long as the engine is not overloaded.

If at any time, the "kill" signal is detected, or the water temperature exceeds a fixed limit, the oil pressure drops below a fixed limit, or the RPM exceeds a fixed limit the controller will execute an emergency shutdown. An emergency shutdown involves closing the "rack" on the governor, and opening the decompression lever until all rotation stops.

In normal operation, once the engine temperature reaches a set point, and the injector line reaches a set temperature, the system will switch from diesel to UCO fuel.

To shut down normally, the "remote start" switch is opened. DC power is still supplied by the 12volt unregulated power, but the "run/stop" signal is gone. When stop is indicated, the system switches to diesel fuel for a fixed amount of time to purge the UCO out of the injector line. Then the "rack" is closed and the system waits for all rotation to stop. Once rotation has stopped, the system resets the rack to the normal location for 1800 rpm operation. This provides a fallback method for operating the system without CPU control. If the battery is totally dead, the engine can still be cranked by hand and it will run just fine, except it won't have the PID controller providing "rpm sag" support.

Manual operation mode: If the cpu wakes up, and it detects engine rotation and oil pressure, and the "stop" signal. In this case, the cpu assumes that the engine was started manually. In this mode the cpu provides monitoring of temperature and oil pressure, but does not attempt to control anything else. To switch out of manual mode to active mode, close the 'remote-start" switch. The system will see the "run" signal, but will notice that the engine is already running, so it won't try to start an already running engine.

5/4/2007 State Diagram Added I am nearly finished with the hardware debug phase. In order to debug the hardware, I had to write software to perform all of the functions that the final controller will need to do. The state diagram below shows all of the phases of operation of the controller.

5/8/2007 New web page just for the computer control. Since the computer control project is more complex than the entire generator project, I have decided to give it its own web page. You can read about my ongoing program on my Automatic Generator Controller Page.

This is what the engine looked like as of 6/2007.  I'm in the process of working the kinks out the the engine controller.  You can see the rats nest of wires hooked to the engine and stepper motor.  The white mark on the flywheel is the reflective tape which is picked up by the tachometer sensor.  I'm a little embarrased by the mess of wires and the oil drip catchers under the engine.

Leaking Injector Pump

Around 150 hours of runtime, I noticed that the oil level appeared to be rising. My first thought was that I had overfilled the oil on the most recent oil change. I kept an eye on it and it really was rising. I then decided that I had been running the engine at fairly low load levels which was causing unburned fuel to leak past the rings and end up in the oil. Then winter came, and it was harder to filter my oil, so the engine sat in the shed for most of the winter.

This spring as things warmed up, I started getting the engine ready to run. When I checked the oil, I nearly fell over backwards, the crankcase was full of fuel. The dipstick showed a black layer at the bottom, with an inch or more of diesel fuel floating on the top of the oil. The only explaination is that the injector pump is slowly leaking fuel into the engine. I assume that it is leaking slowly because if it was leaking rapidly while the engine was running, I would expect the oil to be fully black, and not have a "clear" layer on the top.

Now that I'm just about ready to hook my CPU controller to the engine, I'm going to have to dive into the guts of the engine to track down a leak. Hopefully it is something simple like a loose bolt or a failed "O" ring. Worst case, I may have to replace the injector pump.

5/29/2007 update Over memorial day weekend, I tore into the injector pump to find what was wrong. I found nothing... You can read about the dis-assembly and inspection of the Changfa 195 injector pump. For now, I'm turning off the fuel when the engine is not running.  I suspect the leak was caused by a bit of debris under the seat between the pump barrel and body.  If the delivery valve nut was not completely tight, it may have allowed fuel to leak around the pump barrel and end up in the crankcase.

Cold Weather Starting

The power rarely fails here unless the weather is lousy.  Usually power failures are when it is snowing and the engine has sat in the unheated shed for several months without starting.  The battery will be barely able to turn the engine over, so it will need to be cranked by hand.

I have had to start the engine three or four times under these conditions.  The first couple of times, I just kept cranking away until I started to get feeble pops from the engine.  These pops were enough to dis-engage the crank (so I couldn't keep cranking), but not enough to carry the engine past another compression stroke.  I consider myself to be reasonably fit, but at the end of 10 minutes of this kind of workout, I found myself looking for a better method.

The Changfa 195 has a little fitting which screws into the compression chamber which can serve as a "non electric" glow plug.  The idea is to take the glow plug out, and heat it with a torch until it is very hot.  Then install the fitting as quickly as possible and try starting.  I have used this approach once when the temperature was around freezing and it worked well.  The problem is that you need to somehow get the plug into the head without burning yourself.

In January 2008, a series of cold storms swept through Northern California and knocked out power for many people.  I never did have the power go out for very long, but I decided to fire up the generator just to get the exercise.  This time, I removed the air filter, and stuck a lit propane torch down the air intake for a minute or so.  By the end of the minute, the intake was too hot to touch on the outside.  I buttoned up the air filter, and tried cranking.  This time, it fired and kept running on the first compression stroke.  This was far easier than heating the glow plug (no tools needed and less chance of burning my fingers).  I suppose on a really cold day, I might need to do both, but the torch in the intake is my preferred method at the moment.

Electricity production cost analysis

One of the early justifications for building this generator was that I could lower my electricity bills.  That has turned out to be impractical.  I routinely get emails from people asking for advice on how to "run their meter backwards" using the "free electricity" from their generator.

To start with, assume that I  have a system already in place to net meter  electricity.  Earlier, I found that I can produce 1kwh of power with .127 gallons of WVO.  A single gallon of WVO can produce roughly 8kwh of power.  The average price of electricity in the US in 2008 was $.15/kwh.  That single gallon of WVO can lower my electic bill by $1.20.  In my mind, the WVO was free, and I saved $1.20, so I could easily see how I could make money.

My house consumes about 700 kwh per month.  My engine can reliably produce 5kw.  I would need to run the engine at full power for 140 hours to create the power consumed by my house in one month.  I would need about 90 gallons of WVO.   The engine requires a crankcase oil change every 100 hours.   Delo 400 oil costs about $4.00 per quart, and I need 3 quarts.   The cost to create one months worth of electricity is $18.   Electricity in my area retails for about $.1 per kwh, so I am spending $18 to produce $70 worth of electricity.  Again, I can see how easy it is to make money this way.

Now lets get practical.

Strike 1:  My WVO supplier produces between 10 and 20 gallons per month, so at best, I could only produce between $10, and $20 of savings on my power bill per month.

Strike 2: Running your meter backwards is not easily done.  There are home power websites that describe how to hook up an induction motor and powering it from your engine.   This will actually work, but it is illegal in every place that I am aware of.  In order to do net metering (run your meter backwards).  You must guarantee that your system will not continue to supply power to the lines in the event of a power failure.   Your little 12 horse engine pushing a 240V induction motor and a bit of stray capacitance could easily continue supplying power even when the power has failed.  That same 240 volts could easily kill the power company guy trying to fix the broken wires.  Don't even think of doing something stupid like this.

The proper way to do net metering is to purchase a grid-tie inverter system which has been approved for that application.   Most grid-tie inverters are designed to work with a solar system and batteries.  The SMA windy boy is one of the few inverters which is designed to work without batteries.  A 6kw inverter retails in 2009 for about $3400.  It will take 170 months to pay that off if I only save $20 per month on my power bill.   (By that time, my engine will have run 23000 hours.)

Strike 3: In California, power companies are required to buy power from anybody who wants to sell.  However, they are only required to net-meter power (buy power back at retail rates) if the power was produced using solar, wind, or select "renewable" energy sources.  WVO is not on the list of renewable sources, so the power company will be selling me power at $.10 per kwh, and buy it back from me at $.02.  Now instead of a gallon of WVO producing $1.20 of electricity, it only creates 16 cents.  I would need to produce 3500 kwh of power to offset my $70 power bill.  To create that much power, my engine would need to run for 700 hours per month.  There are 744 hours in a month, so I have 44 hours in which to perform 7 oil changes at $12 each, so I have now spent $84 in motor oil to produce $70 in electricity.  This is beginning to sound like a business in need of a government bailout.  (After the bailout, I can spend $100 to produce $50 in electricity.)

The value of the "Free" WVO as of 2009 is between $1.00 and $1.40 per gallon.  Instead of burning 90 gallons of WVO to produce $70 of power, I could just sell the WVO to a rendering company and get $90 to $126 per month without the bother of running a generator. 

In the final analysis, electricity is so cheap, that it flat out is not practical to use WVO as a power source.

If you are off grid, then you already know how expensive it is to produce your own electricity.   Using WVO instead of diesel or gasoline makes a great deal of sense in that case.

The great storm of 2010

On Jan 18, 2010 it started to snow.  The temperature was just at 32 degrees so the snow was quite sticky.  The wind was steady from the south.  these conditions didn't change for 5 days.  After 2 days,  every tree, power line, bush and house had a heavy layer of snow stuck to the south side. 

On Wednesday Jan 20, I watched the power lines outside sag down to the level of the phone lines on the same pole.   At 10:00am the power line outside my house snapped.  I was ready.  The boys had dug a trail out to the generator shed the night before.   I went to the shed and got the generator started with some help from my foundry propane burner.  I threw the transfer switch and we had lights, heat and water again.

Two days later, the power came back on.  In all, the generator ran for 36 hours and consumed about 10 gallons of diesel.  My WVO was mostly frozen, so I used diesel instead.  Each night, I shut the generator down around 10:00pm, refueled and checked the oil.  I started up again around 6:00am.  We luxuriated in hot running water showers and electric lights. 

We had neighbors who had gas powered generators, but by the end of the first day, they were out of gas, and no gasoline could be purchased because the gas stations didn't have power.  In one case, a man died of Carbon Monoxide poisoning because the exhaust from his gas powered generator flowed back down the trench in the snow and got trapped in the space around his house under the eaves.

Siskiyou county was declared a disaster area by the governor.  Trees which had stood for hundreds of years, just collapsed under the weight.  About one tree in ten had the top broken out.   We have had deeper snow many times, but because the temperature stayed right at freezing, the snow stuck and didn't come off.  If it had been colder, the snow wouldn't have been sticky.  If it had warmed up, the accumulated snow would have slid off.

Final words

It is now officially spring, and we made it through the entire winter without a single power outage. I guess that just like washing your car makes it rain, building a generator prevents power outages. As the snow melts off the ground, I will be spending more time on projects other than the generator. Hopefully I'll have this thing in its own shed smelling like french fries before snow flies next winter.

3/5/2007 update We had one extended power outage this winter. The Changfa had not been run for several months, so it took a fair amount of cranking before it finally caught on. Once it was running, we turned on all of our lights just so the neighbors could sit in their dark homes and be jealous. The ambient temperature was below freezing, With just the lights in the house, the UCO preheat loop provided so much cooling that the engine would not come up to temperature. I finally turned on the oven to provide more load for the engine. I'm considering adding a load dump in the form of a water heater that can be switched in and out automatically. That will have to wait until I get the computer control finished.

If you want to reach me by email, you can use my gmail account. It is the same as my google pages account. 

For those people who have lost their Changfa owner's manual, I have a scanned copy available here (7.7Mb)

I would like to thank the following people for their inspiration and help

  • George Breckenridge at utterpower for providing the inspiration for all of this.
  • Joel Koch at Powersolutions for making it easy to buy this stuff.
  • Dana Linscott for sharing so much of his WVO wisdom on all of the forums where he posts.
  • Dave Gingery and Lindsay books for their excellent books on home foundry.

Self portrait done back before I gave up art