Steam Engine Prototype 1 Design

Towards the end of yesterday’s blog post, we mentioned our progress on the modern steam engine. This is part of our near-term development program (and part of Proposal 2011) towards upgrading our Power Cubes and the LifeTrac infrastructre to modern steam power. Yesterday we met with Robert Thomas, one of those rare individuals who builds steam and gasoline engines for fun. He built this steam tractor replica (23 hp) of a larger 1920s farm traction engine completely from scratch, including building the steam engine from heavy-walled pipe and cutting the gears:

Steam Tractor – 23 HP from Marcin Jakubowski on Vimeo.

We began a design session. Our conclusions from yesterday are to produce a proof-of-concept prototype of a modern steam engine involving Arduino to provide electronic steam injection. This is analogous to electronic fuel injection in cars. The concept looks like this. and can download this conceptual diagram in Dia here to collaborate on the design:

It is a single cylinder, 4” bore, single-acting uniflow steam engine. The inlet port is a solenoid valve operated by Arduino, with a sensor on the flywheel to provide timing. The materials cost is under $200.

Our proof-of-concept engine design is comparable to that of the hit-and-miss gas engines from the early 1900s. Actually, we were quite impressed with these when we saw them at the Northwest Missouri Steam and Gas Show last week. These are simple, 1-cylinder, water-box-radiator cooled gasoline engines which fire only when they need to – so in idle – they fire every few seconds. For anybody who lives off-grid – these are a perfect solution for a gas-sippind device that can run all day on about a gallon of gas. Does anybody know their more specific gas consumption rates. Here’s an example of one – in the 4 second clip, the engine fired only once if you listen and watch the air intake valve move:

Hit and Miss Engine from Marcin Jakubowski on Vimeo.

Here’s a basic diagram of the hit-and-miss:

Our concept simplifies the design tremendously. Strip the design down to a cylinder with flywheel, replacing the mechanical timing linkage with Arduino-controlled steam injection. This can’t get any simpler. People, it looks like we’ve got a workable DIY people’s engine design.

Outstanding questions – especially for the Arduino collaborators – are:

  1. What kind of position sensor is most suitable for timing purposes in this application?
  2. What is the maximum RPM that the Arduino sensor feedback loop will allow?
  3. What is the maximum RPM that the solenoid will allow?

We aim to run the steam engine in a hit-and-miss fashion – without injecting steam at every stroke. This solves the solenoid valve cycling speed issues. The hit-and-miss configuration is a low RPM device to begin with – such as 300 RPM in idle for the red engine above.

We have already showed you a similar concept in a blog post last year – from last year’s Steam Automobile Club of America meeting. This was a steam bicycle with electronic steam injection via hydraulic solenoids:

Steam Motor Bikes – Part 6 from Marcin Jakubowski on Vimeo.

The steam bicycle above ran at 800 RPM, and there were issues with the solenoid. The developer used a hydraulic solenoid adapted for steam use – but mentioned that he expected issues with solenoid lifetime due to its design for lower temperature operation.

Water corrosion and high temperature issues may be resolved readily by using an external solenoid linked to a proper, steam-handling valve – such as a piston or even poppet valve. We would like suggestions on the best solenoid to use for this purpose, as we have not yet looked into solenoid availability. All that we know is that cycle times for solenoids are on the order of 50 ms, which is perfectly acceptable for a simple electronically-controlled steam engine running in hit-and-miss operation. Regarding the valve itself – this is old technology and requires little development effort.

Our goal with Prototype 1 is to demonstrate a lifetime-design, low-speed steam engine for remote power applications – and we will fuel it with pelletized biomass. It is an efficient uniflow design, and because it is single-acting, it is the least complicated design possible. Our approach here may be a breakthrough for remote power applications. We have not seen any low-cost, workable, affordable, appropriate technology, remote power system out there yet. The closest in this game is the Tinytech rebellion in India.

Prototype II and onwards will feature multiple cylinders and higher speed and power. We are pursuing our usual design-for-hypermodularity. This means that we aim to bring forth a veritable Steam Engine Construction Set, such that anyone can build their own engines. With this in place, you and your grandmother can then birth raw slave-power – while reducing the number of real slaves in the world.


  1. Kevin

    how about having n solenoids in parallel, that way they fire with a frequency of f/n which may improve their MTBF.

    if you have more then 1 solenoid then a failed solenoid may not stop the machine operating unless its stuck open.

    1. Marcin

      Good strategy – f/n is another way around issues with solenoid response time. Any suggestions on the solenoid of choice?

  2. Kevin

    depending on the flow cycle time of the solenoid/valve you may want to consider 2 valves in series to allow you shape the duration of the steam input pulse.

    as an example, if the solenoid takes 250mS to open and close then the highest frequency you can drive is it < 4Hz. if you add 2 in series, then by offsetting the activation of solA N mS after solB then the effective steam pulse is 250-N mS which when coupled with the parallel configuration provides a larger operating frequency using slower components all still controllable by the micro controller. This also allows a failed "stuck open" solenoid to be removed from the cycle by disabling its series partner.

  3. Marcin

    Here’s an interesting popular article for those of you who do not know about the potential of modern steam power:

  4. Benjamin Gatti

    Very cool;
    2 or more cylinders are very important in reducing the speed or flywheel size without a drop in efficiency. Steam needs time to expand, and a way to cool down without being exothermic. A turbine is very efficient in part because each section experiences only a small thermal shift, but the steam passing through moves from hot to less hot in many stages, yielding its energy at every stage.

  5. Benjamin Gatti

    Arduino can handle these speeds in its sleep (literally, it has a sleep mode), These same chips are use for brushless motor controls in which the switching speeds are in kilohertz. The only choice for sensor is magnet and hall-effect sensor – perhaps even the latching model. (optical sensors could be used, but grit would be pox.) The hard questions are the cylinder and valve choices. The magnets should be placed at the start (and perhaps the end) of the steam injection cycle.

  6. Søren Poulsen

    I think it would be possible to use a optical switch ( The maximum RPM the system would be able to handle depends on the resolution needed. From the H21A1 datasheet it seems it takes about 60 microseconds to switch in on and off witch gives a maximum of about 16k switches pr sec. Going at 3000 RPM this gives a maximum resolution of about 300 switches pr revolution. At the this switching speed the Arduino would have just under 1000 clock cycles pr switching witch should be plenty to do any kind of regulating.

  7. Evan

    I like the single acting uniflow idea, thats pretty slick.

    I looked around as some industrial suppliers and did a back of the envelope calculation for the speed of the valve. It looks like your going to need a valve about an order of magnitude faster than what is commonly available.

    I think a good speed for the engine would be 360rpm, that could be geared 1:10 to drive a single pole generator head, or 1:5 for a two pole. At that speed one cycle is 166ms, guessing that the exhaust ports will be uncovered for 10deg(WAG) before and after BDC that gives 170deg and 78ms from TDC to get steam into the cylinder.

    Omega-flow makes a valve designed for steam but its open time is 100ms. You’re going to need a much faster valve.

    Anyone have ideas?

  8. jim moore

    And here is the link to the company mentioned in pop sci article:

    I really like cyclone power’s steam engine.
    The design takes the incoming air and does a two stage preheating of the air to increase the efficiency. First the air is heated by cooling down the steam, and then it is further heated by using a heat exchange with the exhaust from the combustion chamber.

    But I am not sure you will be able to super critical water in your design. Water at 3200 psi and 1200 F sounds kind of dangerous with the wrong materials.

  9. Evan

    I looked around at some commercially available steam solenoid valves, is representative of what I found, its opening time is ~100ms. I did an back-of-the-envelope calculation, at this speed you’ll only be able to get a few dozen rpm.

    I think a good speed would be 360rpm, this could be geared up 1:5 to run a two pole generator head at 60Hz, or 1:10 to run a single pole head. At that speed one cycle is 166ms, assuming the exhaust ports are open 10 deg before and after BDC (this may be too little or too much, I don’t know) that gives 78ms for the expansion stroke.

    At these speeds it seems you’ll need a valve about an order of magnitude faster than what’s commercially available.

  10. Mimarob

    So simple it might actually work, what precision is needed in the cylinder?

    There are probably tons of sensors that would do the work. I thought of drilling a hole and using an infrared LED and a detector like old floppy drives of the past. One of the simplest sensors might be those magnetic types used in cheap bicycle speedometers. Software needs to handle contact bounce in some way.

  11. Bustednuckles

    I am duly impressed!
    The very first obvious question I did no t see addressed was what are you using for a voltage source to power up the control module?

    I would imagine you would need some kind of battery.

  12. Offlogic

    A ring magnet on/in the flywheel could be used to provide position feedback to the controller. Allegro Micro makes a nice logic-level output quadrature encoder (sine/cosine) sensor that works at good clearance from the magnet. This would give the controller high precision position information so that the solenoid timing could be adaptively controlled to maintain RPM over varying loads.

  13. Dennis

    Couple thoughts from a retired steamfiller. What steam pressure are you planning to operate? Old Time steam operated at high pressure up to 100 psi. This can be dangerous in a rupture or breach but I think you need the pressure to do useful work. Turbines need dry steam beyond the capability of simple DIY I think. Old time small gas engines used belts to drive household appliances. Why not drive and air compressor and use AIR to drive your appliances? Can you drive a standard agricultural Power Take Off (PTO) that opens up whole bunch of equipment sharing common drive shaft rpm requirements All in all looks a fun practical and very useful build

  14. Chris

    The hard part of steam engine design would not seem to be timing the release of steam into the cylinder, but controlling the boiler to have enough steam available to meet power requirements but not so much as to waste fuel or blow up the boiler.

    Any ideas on how to handle that?

    1. Marcin

      Agreed. We plan on a coil-shaped, flash steam generator. One possible design rationale for steam injection would be to inject steam when a pre-set pressure threshold is reached. The design and size of the generator will determine the allowable steam injection frequency limits. The concept of hit-and-miss is used to provide flexibility in steam boiler output. The design consideration would be to aim for steam injection under the conditions of optimal steam expansion ratios. Setting different steam injection frequencies would allow steam generators of various sizes to be used efficiently. The steam generator could be controlled in part by the firing rate.

      All of the above is just speculation. Real experience will tell. We don’t know of anybody who has built this particular concept, though we aim to pick up much insight from the aging gurus from the Steam Automobile Club of America meeting on Sep. 15-19 in Berrien Springs, Michigan. Our goal is to gain insight on flash steam generator design and controls, as well as insight on the desirable steam injection cutoff and timing – for attaining the highest possible efficiency. In principle – it appears that hit-and-miss operation can allow very high thermodynamic efficiency – simply by allowing the steam to build up to very high temperature and pressure regimes by proper selection of timing. This kind of flexibility is introduced by the electronic control scheme. The devil’s in the details.

  15. Dennis

    Capacity of your steam generator may well be the limiting factor, If you saw traction engine at Steam Thresher Shows, most of the mass was a big “locomotive” type boiler. A Single pass fire tube boiler, not overly efficient but with lots of tubes and surface to allow steam generation. Perhaps by sizing your pistons and estimating strokes per minute you can guess at amount of steam you’ll need per hour. Then work backward to your steam generator but its going to be trial etc..

    Usually steam is contolled by pressure control, one for cycling heat source, another safety to shut down on overpressure plus at least one safety relef valve and some form of low water feeder and low water safety shutoff

    Good Luck seems doable just first one is tougher

  16. Mimarob

    As I understand it, the hazards of steam are greatly reduced if the boiler is avoided in favour of a flash steam generator i.e. a coil of pipe heated by a fire or other heat source.

    So the worst case scenario for a steam-related accident would be equivalent to a pipe burst as opposed to a boiler burst on a conventional steam engine.

    I lack experience on this, but I guess that scorning is the most probable injury? How much steam do you really need to produce a serious burn?

    There simply wont be enough mechanical energy stored to produce a large blast.

    Hopefully this setup will minimize the amount of steam existing in any given moment, but the average production rate would still be a function of the heat input and efficiency of the steam generator.

    Probably you want to put an anti-drainback valve between the steam generator to the solenoid and a safety valve to get rid of excess steam.

    Perhaps individual anti-drainback valves on the output of each f/n solenoid would prevent backward leakage in case the previous solenoid is not closed when the next one in turn fires?

    Also the idea of having a closing solenoid behind the one nearest to the cylinder seems sane. You also need steam storage “enough for one puff” between the solenoids in this case.

    I think one could do a lot of reasonably safe testing and evaluation using pressurized air, but eventually one has to move on to steam to render this device useful.

  17. Canute

    Hate to be a wet blanket, but the exact timing of the steam injection is going to gain you little in comparison to making the boiler efficient, double or triple expansion of the steam, insulating the cylinder(s) to prevent premature condensation, and recycling the condensate to recover the heat. The thermodynamics far outweigh the mechanics. Get those things optimized and then tweak the timing.

  18. Steady Growth | Open Source Ecology

    […] supply design, and to help the world get off-grid via local biomass – we need help on the modern steam engine. These are just a few things on the plate. Contact us at opensourceecology at gmail dot com if you […]

  19. Alan

    What about a car fuel injector as a valve? They are designed for fast-cycling electronic control.

    Example here:

  20. […] have not been able to post sooner because we have been away at the Steam Automobile Club of America Meeting. We will cover the exciting developments that emerged over the next 2 or 3 blog posts – since […]

  21. […] on Open Source Ecology’s proposal for the scalable, steam-powered, electrical generator that we discussed in a former post. This talk puts the power generator in context with other OSE product ecologies. It also discusses […]

  22. Marcin

    Here is an electronic valving application that goes to several thousand RPM –

    Very promising.

  23. Marcin

    Response from Dave Bowes of EVIC:

    Hi Marcin,

    A solenoid operated valve can be made to work for this application. The size and type of solenoid required depends on the valve design. The steam pressure is going to act on the valve face and create resistance that has to be overcome by the solenoid. If you are using a spring to close the valve the spring force has to be considered as well. Solenoids can be made to respond fast enough to run your proposed engine to several thousand rpm. You may find the control board more of a restriction. My 18cc Mk4 EVIC engine develops over 1 kW output at 8000 rpm. I will need to see your proposed valve design to provide any more guidance and to tell you if my controls could be adapted to run your engine. If you wish, send me a drawing or a sketch of the valve design. My engines use poppet valves opened by solenoids and closed by valve springs. I have used commercial solenoids but now I make my own. There are solenoid plans in my book and in my student guide.

    You are going to have to add a solenoid driver to the controls. To get the solenoid to operate quickly you will likely need at least 24 volts, perhaps higher. I am currently using 42 volts on solenoids with a nominal rating of 3 volts. The voltage is reduced to less than 3 volts to hold the valves open.

    Your proposed engine won’t run like a hit and miss engine. The H&M engine is a 4-stroke engine that holds the exhaust valve open between power strokes. As a result the piston can return to TDC without having to compress the residual exhaust gases in the cylinder. The design is not very efficient because exhaust gases/air are constantly being pushed out and sucked back in through the exhaust valve. This results in relatively high pumping losses which is why these engines are not widely used today. It is more efficient to throttle the engine thus reducing the pumping losses. It is even more efficient to use variable valve timing to further reduce the pumping losses as BMW has done with Valvetronic and Fait is doing with their new Multiair system. Your proposed system is like a 2-stroke IC engine, simple but perhaps not the most efficient. It won’t be too bad a low speed where the exhaust steam has sufficient time to escape. But when free wheeling between power strokes you are constantly compressing and expanding the trapped steam. This means that you need a heavy flywheel to keep the speed variation between the compression and expansion strokes small. You may want to consider at least 2 sesnors for speed measurement; TDC and BDC to help smooth this out.

    Thank you for your interest in my EVIC engines.

    EVIC products available from Dave Bowes:

    All prices include shipping to Canadian or US destinations.

    EVIC 111 Book Rev. 6 $48.00

    Complete plans for building the 12.5cc Mk2 Single Cylinder Engine

    Includes drawings, machining instructions, schematics and software source code

    Complete set of drawings for the new 18cc Mk4 Engine

    EVIC Student Guide CD $30.00

    Includes airmail postage to any location.

    The Student Guide Provides information on adapting the EVIC technology to larger engines including calculating required solenoid force and selecting solenoids.

    Book plus Student Guide CD $68.00
    EVIC 111 Mk7 Electronics Package from $225.00

    Electronics packages are custom built to order for the Mk2 and Mk4 engines.

    The electronics packages are fully assembled and tested but do not include the solenoids, ignition coil or batteries.

    International shipping for the book or the electronics packages add $8.00

    I accept payment in either Canadian or US dollars. At the current exchange rate you can save a little by sending me Canadian dollars.

    Sorry I do not accept Pay Pal or any credit cards. I will accept cheques drawn on Canadian or US banks. I accept international money orders. Many international customers have sent me their payment by Western Union.

    I encourage customers to build their engine before purchasing electronics. That way you get the latest technology and I am not spending my time building boards that end up just sitting on a shelf!

    My electronics can be adapted to larger engines up to about 100cc. This will likely require custom software. E-mail me with details of what you are thinking of doing. I do custom software at very low cost for hobby applications.

    I offer assistance to university students wishing to learn more about this technology by answering questions, making suggestions and reviewing drafts of student papers.

    I have used EVIC 111 and EVIC 211 electronics packages for sale. E-mail for prices.

    Please note that I do not sell complete engines, engine parts or kits.

    Dave Bowes, 1095 Afton Road, Peterborough, Ontario, Canada, K9J 8L1

    Dave Bowes

  24. Marcin

    Here is the latest discussion on issues with solenoid lifetime:

    Hi Dave,

    Thanks for your thoughtful response – especially about this issues regarding hit-and-miss operation.

    What is the expected lifetime of the types of solenoids that you make, or that we could make, if run at about 2000 RPM, in the 1 kW range application? Is your system pretty robust, or are there in general poor lifetime issues?

    Do you show the complete designs for high-performance solenoids? Are the solenoids relatively easy to fabricate – or is it an artform? Do you suggest buying them off-shelf if we are interested in long life?



    Hi Marcin,

    The solenoid shafts are the shortest life part. Mine last perhaps 25 hours of operation. For longer life you can add a bronze bushing and use a very hard steel shaft. That might get you to 100 hours. Most commercial solenoid are rated for a certain number of operations, often around 1,000,000. At 10,000 rpm on a 4 stroke engine that is 5,000 solenoid operations in a minute. 1,000,000 / 5000 = 200 minutes or a little over 3 hours. I use harder steel shafts in my solenoids but they still wear. Solenoid coil life hasn’t been a problem. I have had a few solenoid armatures break as a result of making cuts in them to reduce eddy current losses.

    If you have a lathe and can do simple metal turning the solenoids are not difficult to make. Commercial solenoids are moderate performance and certainly are not optimized for valve operation or for high speed operation. Life is going to depend very much on the size (mass) of the valve, the solenoid stroke, and the force required. Your design should keep the moving mass as low as possible and the solenoid stroke short.

    Dave Bowes

    Check out the latest EVIC engine developments at:

  25. Ken Boak


    Just picking up on this thread.

    I have been involved in converting a 600rpm 6 hp single cylinder diesel engine to spark ignition using an Arduino to control the spark ignition timing and firing the spark plug.

    The converted diesel now runs on wood gas – produced from wood chips from recycled pallets.

    The 6hp engine is of a type first built in the 1930’s, and still manufactured in India, where its used for running pumps, generators and small ag. machines.

    It’s slow, durable and easy to fix. Ideal for driving stationary agricultural tools or a 3kW generator set. I use it for home heat and power, and am moving towards an entirely wood/biomass fuelled household.

    3 lbs of woodchips produces 1kWh of electricity and about 4kWh of useful recoverable heat.

    The conversion was done over a weekend at All Power Labs in Berkeley CA – where I was attending a wood gasification course.

    More info here:


    With over a century of perfecting the valving of steam engines, we as a race pretty well perfected it. I own a book over an inch thick dedicated to just the valving of the sliding valve system, and another on the adjustment and optimization of the corliss valve. With proper maintenance, these systems last well over 100 years in constant use. examples of this are all over the world. In my daily life, building exhibits, I have rarely seen arduinos in constant use last over 3 years, and without constant debugging. The durability of electronics has a proven track record of complete fallibility. In the infallible words of the K.I.S.S. statement, why is someone trying to valve a reliable and durable steam engine with electronics? This seems like trying to compute the last number of Pi with a lever and 2 gears. Let computers compute, and let mechanics do their work.

  27. Kevin J Yancey

    I think the valving issue has already been solved. Koenigsegg is building a camless engine, with each valve actuated by a electromagnet, that can run at insane rpm’s only capable of being matched/exceeded by steam turbines. For a steam locomotive, running on 80″ drivers, to reach 80 mph, it has to run at 350rpm. So, the fastest steam locomotive in the world never achieved higher than 525 or so rpm. No modern car even idles that low! So, in principle, you can run a steam engine with electronic controls with no electrical limitations.