This is my twin brother and I's very ambitious BMW e36 rotary valve project. The idea behind the project is to implement a rotary valve engine head into a modern production car. The type of valve train we have gone with is known as an axial flow rotary valve, also known as a Cross style rotary valve.
The car is a 1999 BMW 328is that is for all intensive purposes, not in very good shape. The car is however rust free because it came from a not cold wintery climate and more importantly, it is 100% stock. The stock bit is a very rare thing for these cars.
From 20 feet the car looks great!
This beautiful, near perfect condition engine didn't know what it had coming. Before tearing it apart we took it to a dyno for a baseline power figure. Impressively, with 163,000 miles on it, the engine put out stock brand new power figures and was only down a few lb-ft f torque.
To sum up the project, my brother an I designed the entire head, primarily by lifting pictures off of the internet of torn down engines and then scaling them in Autodesk Inventor. Once we purchased the car and tore down the head, we took some additional dimensions with calipers to confirm the model. No tools fancier than a set of Mitoyo 6 inch calipers were used, yet all of the holes in the head line up except one.
We built approximately 1/2 of the components at home on a little 1932 Craftsman Metal Lathe and the same model mini CNC mill that the BOM guys use in their shop. The larger and higher tolerance components we sent out to a tool shop that we both used to work at before we became full time engineers.
Without further ado, here is the engine head installed in the car.
It does indeed run, though only for rather short amounts of time. The smoke is due to a forced oil system that keeps the valves lubricated such that we don't accidentally ruin a bearing surface. It runs fine without the oil, but it is more of a precautionary measure at this point.
It also drives around sometimes, leaving a smoke trail...
The very first time it drove, it made some Polar Bears cry
As the project stands currently, we are solving a few problems. It needs a new timing belt, this will be the 4th one so far. The first two died from oil soak weakening the rubber, the most recent one died from a screw driver being stuck into it and damaging it. We are having issues with fuel control, it would appear that we are getting too much fuel and drowning out the spark plugs. We are considering buying 6 nice Mikuni carburetors and switching to to carburation due to the ease of tuning compared to a stock ECU of which we have no way to talk to it. We also seem to always be chasing vacuum leaks.
Let me know what you think!
Mini_Madness said
Feb 9, 2017
OK, so it needs work. what is your valve design? CAD or drawn images will help. but the principle is sound
pizzaman09 said
Feb 12, 2017
Some very very good progress has been made recently on the oil consumption. It was discovered that the crank case ventilation was picking up oil and pumping it directly into the intake manifold. A simple swap of pickup point to another location has solved 95% of the smoke problem. 5% is probably just pooled oil sitting in the exhaust headers.
The valve design is is known as an axial flow rotary valve. The intake is pulled through the right side of the valve as pictured below. The exhaust goes out the left side. As the valve spins, the exhaust and intake ports in the valve align with the single port at the top of the combustion chamber.
Today's video shows good progress on making it idle smoothly. It isn't quite there but getting close. Might try to take it for a spin up the driveway tomorrow.
-- Edited by pizzaman09 on Sunday 12th of February 2017 03:08:16 PM
Mini_Madness said
Feb 12, 2017
hmm, the rolling tube design(axial flow). not bad if i do say so myself. how did you sort out the intake and exhaust phases in the cylinder head? must be a pain to do with a design that doesn't go longitudinally to the engine block and the sealing of the oil/ combustion chamber. whats you opinion on the Coates style valve? and why not two banks of axial flow "camshafts" for exhaust/ intake? just a thought.
-- Edited by Mini_Madness on Sunday 12th of February 2017 04:58:42 PM
pizzaman09 said
Mar 11, 2017
Thanks.
The intake and exhaust phasing in each cylinder is quite simple. There is a ~130 dwell angle between the peak intake and exhaust openings. The phasing between each cylinder was quite tricky, not just because we had 6 valves but because every other valve spins opposite directions of each other. Add on top of the fact that the firing order is 135624, that means that all of the clockwise spinning valves fire first then all of the counter clockwise valves fire. We just worked out the angles out on paper, then we milled some timing marks into each of the valves. We have adjustability in 10 degree increments due to our 36 tooth sprockets on each valve, which really isn't optimal. We could really use some finer tuning for valve timing.
The two banks of axial flow rotary valves is possible. We went the route we did because we were familiar with it. We had built 3 Briggs and Stratton engines prior to this build using the same style valve and we were quite successful on our third attempt. The flow possible in our design is much higher than that of the dual valves that run parallel to the crank shaft. It would have been much easier to implement the timing chain on that style but then we would have had to have made two sliding sealing surfaces in the combustion chamber instead of just one.
The Coates style valve isn't too far off of what we are doing. The spherical shaped valves take advantage of a simpler and possibly more robust but much higher stressed sealing setup. Once can seal the combustion chamber to the rotating valve with a circular ring as opposed to the half pipe we are using in our engine. The Coats style valve necessitates a more complex head to channel intake flow but it does allow for individual valves to be parallel to the crank shaft which makes the running gear simpler.
There were only really two pains in making the rotary valves perpendicular to the crank shaft. First, we wanted to use the stock timing chain which was super long so we had to mount the power take off from that chain much higher than we really needed. The rotary valve head is about 6 inches shorter than the over head cam head that came out. Second, we found it was difficult to get the tensioner and belt set up so that we could actually transmit enough torque to drive all 6 valves. We have been through 4 $100 timing belts thus far. It seems now we have the drag situation sorted out and we have also started to protect the belt from oil spray so it doesn't swell and loose all of it's strength.
Mini_Madness said
Mar 11, 2017
I can see your point about sliding sealing surfaces, However the oil lost would be a product of the surface area covering the valve? would it not? An idea to shrink your tolerances would be to build a catch can with an oil heater before starting your engine meaning that the engine would use less oil when running, due to the tighter tolerances.
On the subject of complexity different strokes for different folks however i can't help but feel like you would have a difficultly finding the right timing on each cylinder and the wear that would occur with a belt stretching over time, Peugeot's timing belts stretched 1inch when i replaced it, as well as it having no teeth on the belt. (Non-interferance engine thank Christ).
Drag might not be the immediate problem with that set up rather, axial torque, making the drive unstable from the timing chain to the valves, due to the pulsing delivery from the crank.
pizzaman09 said
Mar 11, 2017
We aren't particularly concerned about oil loss honestly. I see what you are getting at though. The valve seats are leaded bronze so they can take some amount of running without oil before anything serious happens.
The timing can potentially change over time due to belt stretch. We noticed this especially when the belt would swell due to oil or fuel soak. Now that we have the seal drag dialed back the belt loading is much much lower, total system torque is likely lower than the original cam shafts.
We did have some pulsing issues at one point due to a poorly pinned sprocket allowing +/- 20 degree swings in valve timing. Since we got that under control the system seems pretty smooth.
The biggest issue we are having right now seems to be drowning the plugs out in fuel. The plugs fire when they are outside the cylinder but once installed they quickly get wet and stop firing. This is confirmed by a timing light. The ECU appears to be functioning correctly but I can't know for certain. There are many times where I wished I had stared with a carburetted car with a set of points in a distributor.
Mini_Madness said
Mar 11, 2017
Starting out with a rich mix isn't so bad when tuning, you can cheat the O2 sensor with a resistor(or length of wire) till you can buy a programmable ECU. and dial the thing in.
Consider swapping to a colder or hotter plug? Hotter would be my bet.
I think you will end up chasing distance from intake or messing with injection placement. something like this. youtu.be/nx7ugia9hN8 if you don't want holes in your intake plug them with pipe plugs. Or equivalent.
Other than that start changing valve timing, maybe, because it has better air delivery and fuel. more CFM!
pizzaman09 said
Mar 12, 2017
Those are some very good points about the O2 sensor. In all reality, the O2 sensor might just need to be replaced. At this point, the engine has been run for so long with uncontrolled amounts of oil accidentally being pumped into the intake, there is little doubt in my mind that the O2 sensor is quite fouled.
I have done a fair bit o reading on the spark plug topic. All of the old builders of rotary valve heads seem to switch to a hotter plug. Unfortunately it isn't much of an option for us. The one design mistake we made was backing ourselves into using an odd ball long reach 8mm diameter NGK plug that is original to a few Honda motorcycles. The result is we really only have one plug choice. We have done our best to match the resistance of our current ignition system to that of the OEM BMW coil on plug setup. We are running 10k Ohms of resistance compared to the BMW 7k Ohms only because we can not seem to source an appropriate non resistor spark plug boot.
We have played considerably with the valve timing. It seems to like a bit of advance compared to what the original design calls for. When the engine is running it very much likes the mid to high range rpms, the extra flow allowed by the rotary valves really seems to make it frisky at the top end.
The one possibility that we could be fighting right now is just cold temperatures. The engine is much more likely to start on a warm day over 50 deg F. Today it was in the 20s. I am going to strongly considering purchasing some new O2 sensors for it. The ones on the car are likely original with 163k miles on them.
pizzaman09 said
Mar 7, 2019
If anyone has interest in this project, I did finally make a video explaining it all.
-- Edited by pizzaman09 on Thursday 7th of March 2019 02:25:17 AM
This is my twin brother and I's very ambitious BMW e36 rotary valve project. The idea behind the project is to implement a rotary valve engine head into a modern production car. The type of valve train we have gone with is known as an axial flow rotary valve, also known as a Cross style rotary valve.
The car is a 1999 BMW 328is that is for all intensive purposes, not in very good shape. The car is however rust free because it came from a not cold wintery climate and more importantly, it is 100% stock. The stock bit is a very rare thing for these cars.
From 20 feet the car looks great!
This beautiful, near perfect condition engine didn't know what it had coming. Before tearing it apart we took it to a dyno for a baseline power figure. Impressively, with 163,000 miles on it, the engine put out stock brand new power figures and was only down a few lb-ft f torque.
To sum up the project, my brother an I designed the entire head, primarily by lifting pictures off of the internet of torn down engines and then scaling them in Autodesk Inventor. Once we purchased the car and tore down the head, we took some additional dimensions with calipers to confirm the model. No tools fancier than a set of Mitoyo 6 inch calipers were used, yet all of the holes in the head line up except one.
We built approximately 1/2 of the components at home on a little 1932 Craftsman Metal Lathe and the same model mini CNC mill that the BOM guys use in their shop. The larger and higher tolerance components we sent out to a tool shop that we both used to work at before we became full time engineers.
Without further ado, here is the engine head installed in the car.
It does indeed run, though only for rather short amounts of time. The smoke is due to a forced oil system that keeps the valves lubricated such that we don't accidentally ruin a bearing surface. It runs fine without the oil, but it is more of a precautionary measure at this point.
It also drives around sometimes, leaving a smoke trail...
The very first time it drove, it made some Polar Bears cry
As the project stands currently, we are solving a few problems. It needs a new timing belt, this will be the 4th one so far. The first two died from oil soak weakening the rubber, the most recent one died from a screw driver being stuck into it and damaging it. We are having issues with fuel control, it would appear that we are getting too much fuel and drowning out the spark plugs. We are considering buying 6 nice Mikuni carburetors and switching to to carburation due to the ease of tuning compared to a stock ECU of which we have no way to talk to it. We also seem to always be chasing vacuum leaks.
Let me know what you think!
Some very very good progress has been made recently on the oil consumption. It was discovered that the crank case ventilation was picking up oil and pumping it directly into the intake manifold. A simple swap of pickup point to another location has solved 95% of the smoke problem. 5% is probably just pooled oil sitting in the exhaust headers.
The valve design is is known as an axial flow rotary valve. The intake is pulled through the right side of the valve as pictured below. The exhaust goes out the left side. As the valve spins, the exhaust and intake ports in the valve align with the single port at the top of the combustion chamber.
Today's video shows good progress on making it idle smoothly. It isn't quite there but getting close. Might try to take it for a spin up the driveway tomorrow.
-- Edited by pizzaman09 on Sunday 12th of February 2017 03:08:16 PM
hmm, the rolling tube design(axial flow). not bad if i do say so myself.
how did you sort out the intake and exhaust phases in the cylinder head? must be a pain to do with a design that doesn't go longitudinally to the engine block and the sealing of the oil/ combustion chamber.
whats you opinion on the Coates style valve?
and why not two banks of axial flow "camshafts" for exhaust/ intake? just a thought.
-- Edited by Mini_Madness on Sunday 12th of February 2017 04:58:42 PM
The intake and exhaust phasing in each cylinder is quite simple. There is a ~130 dwell angle between the peak intake and exhaust openings. The phasing between each cylinder was quite tricky, not just because we had 6 valves but because every other valve spins opposite directions of each other. Add on top of the fact that the firing order is 135624, that means that all of the clockwise spinning valves fire first then all of the counter clockwise valves fire. We just worked out the angles out on paper, then we milled some timing marks into each of the valves. We have adjustability in 10 degree increments due to our 36 tooth sprockets on each valve, which really isn't optimal. We could really use some finer tuning for valve timing.
The two banks of axial flow rotary valves is possible. We went the route we did because we were familiar with it. We had built 3 Briggs and Stratton engines prior to this build using the same style valve and we were quite successful on our third attempt. The flow possible in our design is much higher than that of the dual valves that run parallel to the crank shaft. It would have been much easier to implement the timing chain on that style but then we would have had to have made two sliding sealing surfaces in the combustion chamber instead of just one.
The Coates style valve isn't too far off of what we are doing. The spherical shaped valves take advantage of a simpler and possibly more robust but much higher stressed sealing setup. Once can seal the combustion chamber to the rotating valve with a circular ring as opposed to the half pipe we are using in our engine. The Coats style valve necessitates a more complex head to channel intake flow but it does allow for individual valves to be parallel to the crank shaft which makes the running gear simpler.
There were only really two pains in making the rotary valves perpendicular to the crank shaft. First, we wanted to use the stock timing chain which was super long so we had to mount the power take off from that chain much higher than we really needed. The rotary valve head is about 6 inches shorter than the over head cam head that came out. Second, we found it was difficult to get the tensioner and belt set up so that we could actually transmit enough torque to drive all 6 valves. We have been through 4 $100 timing belts thus far. It seems now we have the drag situation sorted out and we have also started to protect the belt from oil spray so it doesn't swell and loose all of it's strength.
On the subject of complexity different strokes for different folks however i can't help but feel like you would have a difficultly finding the right timing on each cylinder and the wear that would occur with a belt stretching over time, Peugeot's timing belts stretched 1inch when i replaced it, as well as it having no teeth on the belt. (Non-interferance engine thank Christ).
Drag might not be the immediate problem with that set up rather, axial torque, making the drive unstable from the timing chain to the valves, due to the pulsing delivery from the crank.
The timing can potentially change over time due to belt stretch. We noticed this especially when the belt would swell due to oil or fuel soak. Now that we have the seal drag dialed back the belt loading is much much lower, total system torque is likely lower than the original cam shafts.
We did have some pulsing issues at one point due to a poorly pinned sprocket allowing +/- 20 degree swings in valve timing. Since we got that under control the system seems pretty smooth.
The biggest issue we are having right now seems to be drowning the plugs out in fuel. The plugs fire when they are outside the cylinder but once installed they quickly get wet and stop firing. This is confirmed by a timing light. The ECU appears to be functioning correctly but I can't know for certain. There are many times where I wished I had stared with a carburetted car with a set of points in a distributor.
Consider swapping to a colder or hotter plug? Hotter would be my bet.
I think you will end up chasing distance from intake or messing with injection placement. something like this. youtu.be/nx7ugia9hN8 if you don't want holes in your intake plug them with pipe plugs. Or equivalent.
Other than that start changing valve timing, maybe, because it has better air delivery and fuel. more CFM!
I have done a fair bit o reading on the spark plug topic. All of the old builders of rotary valve heads seem to switch to a hotter plug. Unfortunately it isn't much of an option for us. The one design mistake we made was backing ourselves into using an odd ball long reach 8mm diameter NGK plug that is original to a few Honda motorcycles. The result is we really only have one plug choice. We have done our best to match the resistance of our current ignition system to that of the OEM BMW coil on plug setup. We are running 10k Ohms of resistance compared to the BMW 7k Ohms only because we can not seem to source an appropriate non resistor spark plug boot.
We have played considerably with the valve timing. It seems to like a bit of advance compared to what the original design calls for. When the engine is running it very much likes the mid to high range rpms, the extra flow allowed by the rotary valves really seems to make it frisky at the top end.
The one possibility that we could be fighting right now is just cold temperatures. The engine is much more likely to start on a warm day over 50 deg F. Today it was in the 20s. I am going to strongly considering purchasing some new O2 sensors for it. The ones on the car are likely original with 163k miles on them.
If anyone has interest in this project, I did finally make a video explaining it all.
-- Edited by pizzaman09 on Thursday 7th of March 2019 02:25:17 AM