Atkinson Differential Engine - Making it work?

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I'm back. Sorry it took so long but I had to prove / disprove my observation of a vacuum being in the cylinder when the engine does not fire. I decided to make a very elaborate intake valve arrangement consisting of the standard poppet valve followed by a sliding piston valve (not sure of the correct term).
It turned out to be one of those things that you draw and figure it's fine and then realize it won't work so you make a change. Eventually when it looks good you end up with something possibly way more complicated than it needs to be. So please don't comment "you should have done it this way". It's what I ended up with after a lot of changes.
Unfortunately I had to throw some electronics at it.
The theory was I'd needed to keep the poppet valve because it will seal properly when there is pressure in the cylinder under normal operation. The problem to solve was how to relieve the vacuum in the cylinder (that will open the poppet valve) but don't let it pull air through the carb. Then, when the proper intake cycle begins, enable that proper vacuum to the carb. That's where the piston valve comes in. It's solenoid operated and it's set up like a directional valve. It vents the cylinder vacuum when the poppet opens but through the use of a couple of hall sensors that detect the proper intake timing it closes the vent and opens up the passage to the carb.

The picture below is the solenoid I pulled from by parts bin. You can see it is fastened to the top of the water jacket and pulls the piston that is inside the valve block. At rest the piston is in vent mode. Behind that you can see the standard poppet valve that leads to the cylinder.

1590763936023.jpeg


Below is a CAD drawing looking from the top of the valve block with the piston in vent mode. The vent hole is the small one in the piston valve and body facing you. The vacuum will come up through the poppet valve and will pull air in through the vent instead of the carb.


1590764110234.jpeg


The next image is from the carb input side also when in vent mode. The passage to the carb is blocked by the piston valve. Vacuum comes in through the poppet valve (green) and air is pulled in though the vent on top (not the carb).


1590764249589.jpeg



Looking from the top again, the next picture shows the piston valve when the solenoid pulls in. The vent is blocked off and there is a direct path to the carburetor (up in this picture).


1590764353701.jpeg


From the carb side you can see there is a direct path through the block past hole in the the piston valve where you can see the valve stem of the poppet valve.


1590764432626.jpeg



Obviously the trick to operating the solenoid at the right time is to time it with the motion of the pistons. This turned out to be fairly easy by adding a rotating magnet plate to the crank shaft and a couple of hall sensors on a stationary plate. The first hall sensor triggers a simple 555 timer (cheap) that activates a mosfet to turn on the solenoid. The second hall sensor resets the timer and de-activates the solenoid. The timer function is not really required. The chip is used as a cheap flip fop. But I took advantage of the timer function to turn off the solenoid if it gets triggered but not reset (depending on where the engine stops) after about a half second.

The picture below is the surface mount board I made for the timer / driver fastened to the frame. Building that was not without problems. WOrked fine on the bench but when installed the buzz coil ignition was triggering it so I had to tack on a few extra components to filter the noise. That's why the extra stuff scabbed onto the board.
The nice thing about all this crap I have added is it can be removed with no signs it was there.

1590765072545.jpeg



SO.... does the engine run now - NOPE.
But this addition does prove my observation of the vacuum in the cylinder when it does not fire.
Was it the source of the problem? Apparently it wasn't the only problem but IMHO one that could have been a source of mixture issues. So it was worth eliminating it before moving on to other issues.
This modification does function as expected. I can feel a vacuum on the vent port on what should be the exhaust stroke and then the vacuum on the carb seems about the same as before when the solenoid switches to intake mode. The exhaust valve also works because there is a (normal) brief period of compression in the cylinder after the vacuum is vented and the intake port is covered while the piston finishes a few degrees of motion to finish the exhaust stroke.

So what's the next step?
As some have suggested I might add a chamfer on the side of the pistons to make a sort of combustion chamber.
I also seem to have lost a bit of compression (wear perhaps due to minimal lubrication). So I may make new pistons with the chamfer and with O-rings if I can find some of a suitable size This way I can put the current pistons back in if required.

BUT. This won't happen any time soon. It's summer and there are a lot of things that need attention outside the shop.

LATER - (much later)
 
Last edited:
Although not the Gingery Design - apparently it can be made to work. Dave Perreault researched the original patents and managed to get one to work. Apparently very finicky. So maybe there's hope for me / us yet.

 
Although not the Gingery Design - apparently it can be made to work. Dave Perreault researched the original patents and managed to get one to work. Apparently very finicky. So maybe there's hope for me / us yet.


An interesting point: I just looked at the original patent drawings again which are available on the web:

https://pdfpiw.uspto.gov/.piw?PageNum=0&idkey=NONE&SectionNum=3&HomeUrl=&docid=0336505
The original design seems to have a mechanical actuator on the intake valve. I am not sure whether it actually controlled the intake or if it was intended to be a governor. I will have to take another look at that when I get a chance. I keep saying that I am not going to get sucked back into this thing but things keep coming up which renews my interest.

Gordon
 
Yeah. Me too. Interesting.
I've been doing some drawings for new pistons with the chamfers to get more combustion space and O-rings. I've noted recently that I've lost a lot of the original compression I had. But strangely I can build a good vacuum. Not sure what's going on with the that. I'm not sure if the O-rings will last though because the rings pass over the ports in the cylinder and they will probably wear them out.
I'm grasping at straws really.
It's not real clear in the video what's going on with the intake valve on his supposed copy of the patent engine.
I'll ask him.
I'll also have a look myself at the patent drawings. Most times they are sorely short of information and possibly mis-leading info (purposely).

Back at it sooner than I wanted to be I guess.
 
I got a pretty quick email answer from Dave P.

The 10" flywheel Atkinson Differential engine has atmospheric valves on both intake and exhaust.
The mini Atkinson Differential engine has mechanical intake and exhaust.
Being so small, I found the need to come up with a different valving approach.... So that is when I decided to design my own valve arrangement.
Using atmospheric valves on this engine would make it very hard to build.... This slide valve simplified that whole section.


He didn't specifically say the patent replica engine but I guess that's the only 10" flywheel engine.
It's all very confusing with so many versions out there.
 
I came across these engines on Youtube (again) and thought I hadn't seen them before. Turns out I had and I even contacted him via the comments on the first video (which you can read). Seems they are from the Gingery plans (with some embellishments). He says he figured it out first by building a wood model. Not sure if he stumbled on the solution to it's problems of if he's a more careful builder than I am. Anyway the results speak for themselves I guess.
I guess I've probably posted all this before. Sorry.




and another from the same Youtube page

 
Last edited:
Dave,
I am considering and of the opinion that O Rings for the Pump Piston will likely increase the compression, however multiple posts here in HMEM, indicate/suggest O Rings are not suitable when there are cross holes involved, seems logical to me. Will be trying that on my Mini Atkinson, I am of the belief that it will benifit yours for Compression as well as Vacuum.
 
I made piston rings for mine from cast iron, old an trued Trimble method, they were made at 0.045 Inch thickness instead of the specified 0.062 inch, so far engine has very decent compression and suction from the carburetor.

I had some issue with the internal diameter of the rings, it was specified bigger than what my calculations gave.

Did not took much to make the rings sit well in the cylinder and make compression, ( no power assisted turning to set them was needed).
I may have been lucky, who knows we will see if it runs or just makes me place it on the shelf of failed projects.
 
Ken:
I have been THINKING about this for a while but I've done nothing. I have abandoned the idea of O-rings for the reasons you mention i.e. I think they will be torn up by the ports in the cylinder.
My recent thoughts will be to add a chamfer to the piston like you have to form a bit of a combustion chamber. I'm concerned that it will reduce the already low compression so I'll do some CAD measurements to optimize the size before doing anything.
Since I'm making new pistons I'll also make new rings, hone the cylinders and start again. I think I lost the original compression because I pulled the pistons out several times while messing around. This time I'll avoid removing them.
I could change my thoughts AGAIN before I find the motivation to make chips.
 
Last edited:
Ken:
I have been THINKING about this for a while but I've done nothing. I have abandoned the idea of O-rings for the reasons you mention i.e. I think they will be torn up by the ports in the cylinder.
My recent thoughts will be to add a chamfer to the piston like you have to form a bit of a combustion chamber. I'm concerned that it will reduce the already low compression so I'll do some CAD measurements to optimize the size before doing anything.
Since I'm making new pistons I'll also make new rings, hone the cylinders and start again. I think I lost the original compression because I pulled the pistons out several times while messing around. This time I'll avoid removing them.
I could change my thoughts AGAIN before I find the motivation to make chips.
Dave, I have decided to remake my pistons so I can try slightly wider/thicker rings, the original spec's were for .050" so I'll try increment of .010. Hey, this is a hobby after all😉.
 
I finally found the enthusiasm to have another look at the Atkinson. I decided to connect a fuel tank pressure sensor to the intake port in place of the carb. I connected my scope and also triggered the scope from the ignition to get a TDC reverence. Attached you can see the waveform captured. The horizontal boxes on the screen divide the 360 degree full cycle of the Atkinson into 90 degree increments. You can ignore the vertical boxes they are for a four cylinder engine. You can also ignore the color of the boxes. Again they are for a 720 degree normal cycle. But they can rougly be labelled left to right Power, Exhaust, Intake, Compression.
The yellow waveform is the ignition trigger. The multi-sparks start on the hall sensor going low on my engine.
The green waveform is the vacuum at the intake port after the intake valve (where the carb would be).
As you can see - as expected - there is a huge vacuum pulse when the engine does not fire and the pistons travel down the cylinder to the intake end to the intake port where the valve opens. Then you can see the weaker official intake vacuum but it's not very strong and it appears to oscillate I think because it has trouble overcoming the valve spring pressure so the valve bounces and fades away as the stroke ends. The spring is about as weak as I can make it.
As I expected I can imagine the huge pulse would be sucking on the carburetor but immediately following that there would be a short period of pressure as the piston passes the intake port on the way to the exhaust port and the pressure would eject that fresh charge out the exhaust port. I think the real intake vacuum is not strong enough to bring in a suitable charge and that's why it won't fire.
If I could get it to fire then there would be pressure in the cylinder instead of that large vacuum. If the real intake vacuum could be improved it might make a good charge and allow it to fire. Sort of a catch 22 affair.
I don't have any solution to the problem. I'm just reporting more findings - for what they're worth.

1611429808366.jpeg
 
Last edited:
BTW. The above green vacuum waveform is flat on top (clipped) because I think the sensor was max'd out. It would have continued much higher further emphasizing the difference between it and the weak intake vacuum pulse.
 

Latest posts

Back
Top