Solenoid diesel injection pumps

[The Ignoble Troll]

I'm all on board with polymer use / o-rings. I agree that they could be used more often instead of relying on metal to metal fits. Its one of the ideas I've been kicking around.

My designs (all virtual so far) are for a unit injector with cam actuation and wedge control in place of a helix and rack. The injector and pump are actually 2 separate pieces, sealed with an o-ring between them and an o-ring where it seals to the cylinder chamber (that last one maybe optimistic..).
I plan to use a plunger type pump and use a gland o-ring seal at the top end, but I want most of the sealing for the high pressure to be done by the plunger fit.
Another item I'm struggling with is the size of the check valve, even at 1/8 scale that I'm trying to design to with a piston of ~30mm diameter (1-3/16"), the check valve ball and components end up being tiny - 1mm ball and itsy bitsy springs. I need to work on watchmaking to works with parts that small.

I probably need to concede to reality and use an external check valve to the pump, and use slightly large parts to simplify things to start.

The plan is to use a mushroom type injector, seems the easiest to make.

Here is a section view of what I've drawn up so far, with the injector and pump in a cylinder head beside a valve. For scale, the OD of the pump and injector are 3/8" (9.5mm)
View attachment 164002

I like it! Do you have a cad of the plunger driver? Curious how you plan on transmitting the force to the plunger.

 

[The Ignoble Troll]

Excellent discussion, everyone!
I am still in the unit pump/injector wannabe group and have at least a half dozen versions of the "so close you can taste it," category in the boxes of failures. Designing to a minimal number of critical sealing surfaces seems to be the most workable goal for the home shop, but I have take apart a yanmar pump-line-injector setup, and a detroit diesel unit injector, both of which use the lapped plunger with rotating helix spill design for volume control. Both of these in stock form produce a perfect cloud of vapor-fine mist when operated. The fit of the plunger into the pump body is an order of magnitude better than anything that I can produce. The fit is almost magically good, LOL. But they are made to work for 1,000's of hours. I would be happy with minutes.
Our nemesis of scaling factor is one problem. If you try to make a pump of 1/3 scale, the pump volume will be 1/27 BUT, the area of the sealing surfaces will be 1/9. So you essentially have 3 times as much leak potential as a full sized unit. I was told by a retired diesel expert, that any type of elastomeric material in the high pressure side will lead to inaccuracies in fuel delivery. I am still holding hope that a good governor will be able to overcome some of that issue. But the real injectors all seem to have an abundance of both flat and cylindrical lapped surface, all of which are critical.
Seal material choice is critical. Look up compatibility info with diesel fuel and as already noted, Viton and teflon might be the best possibilities, except for a few very expensive specialty compounds. Some material might work perfectly today, but after sitting for 2 days have swelled and locked up.
I am at the point that most of my failures have actually somewhat worked, but none have worked "well", at least for any length of time. Attention to detail, tenacity, project burn-out, are all critical factors. I am again, at the almost-there on another variation and I remain optimistic. Others have succeeded, so there is probably no reason that any of us should not be successful, too.
Lloyd

I follow your build with great interest, especially the blower build. That was some prime trouble shooting and adapting.


"If you try to make a pump of 1/3 scale, the pump volume will be 1/27 BUT, the area of the sealing surfaces will be 1/9. So you essentially have 3 times as much leak potential as a full sized unit."

Thank you for that tidbit. I hadn't thought about it that way...

 

[HMEL]

I just finished moving and am months away from getting my shop to be functional. That being said I am picking away at the issues I will need to solve to build a small two stroke diesel.

Currently my thoughts are a 1" bore with a 4" stroke.

My initial plan is to build a single cylinder proof of concept unit and supply it with shop air at 15psi.

From there I want to build a 5 cylinder version that will use a bought roots blower as the air supply.


I want to start with the fuel system and perfect that before building the engine.

To that end, I plan on designing a small injector that will use minimal lapped parts and a separate fuel pump to generate the flow to run the injector. They will be connected with some sort of schedual 80 small bore stainless pipe.

For the fuel pump I am thinking that the easiest solution is to do away with mechanically driven pumps and to use a series of solenoids with a common plunger. One coil would extend a 1mm diameter plunger into a true plunger pump, the other coil would take the place of a return spring and withdraw the plunger.

With off the shelf solenoids, a 1mm plunger pump should easily generate 10,000psi.

The stroke length and timing would be arduino controlled. The dual coil solenoid pump would only need a single double pole relay to control it.

I have a few quadrat sensors somewhere that could be used to adjust timing, injection duration and injection volume.


The plunger pump will have a positive pressure feed pump, which will also allow for a fuel filter to be used.

Can anyone see any foreseeable problems with this set up?

Very interesting post. I have been researching this subject myself. I believe its possible to mechanically drive the pump but have had a very difficult time with the issue of fuel metering. The smaller the air flow the more difficult it seems to be. I like the idea of solenoid but have never given the thought you could control the pump stroke that way. I have also considered the injector design and I am of the opinion you just need to get the right amount of fuel into the cylinder at the right time. At a models size I have doubts that the spray pattern is going to be all that important. That is just an opinion and does not mean its true. The injector will have to be balanced so the pressure area works against the closing spring. I would not be surprised if you put a solenoid on that. I could not set up a electrical system like you propose as to me it would be an additional learning curve. I am also thinking that you might be able to use a spring to return the plunger but removing it does lower the force the solenoid needs to pressurize the pump. I cant see any reason why your design concept would not work given the electronics are fast enough for all the variables. I hope you are able to continue your work on this as many are very interested in this subject. Good Luck

 

[ninefinger]

I don't have a good drawing or model at the moment as its been a few months since I was looking at this with some dedication, and Fusion and / or I have managed to break my assemblies with too many redesigns and edits.
In words, as its a 2 stroke, the cam actuates rockers for the exhaust valves and the plunger with different lobes for each, exactly as in full size practice for the engine type I wish to model. The difference is that between the lobe for the rocker(*) there will be a moveable wedge. I think Roger B has shown this style quite well, just having a hard time finding the post / forum he posted it in, but in his case the cam lobe acted directly on the wedge and plunger. I'll be drawing it again sometime, as I plan to build a test bed for just the fuel system before I go and try and build an engine. Again, following how Lloyd-ss made a test bed for his injector (if I'm remembering correctly).

(*) - I may try designing the moveable wedge to be at the other end of the rocker - between it and the plunger pump. Depends what will work (1st priority) and what may look more scale (secondary priority for now).

 

[Lloyd-ss]

................................. edit

(*) - I may try designing the moveable wedge to be at the other end of the rocker - between it and the plunger pump. Depends what will work (1st priority) and what may look more scale (secondary priority for now).

FindHansen has at least one good YouTube video (among his many excellent videos) demonstrating the wedge adjustment for fuel volume and for timing adjustment. I think it was in a video of his dark green vertical diesel where he has all of the parts laid out and explains them. I am not sure, but I think the only time in the operating cycle that the governor will be strong enough to move the wedge is when there is the momentary lash clearance in the valve train during each revolution.
Lloyd

I follow your build with great interest, especially the blower build. That was some prime trouble shooting and adapting.


"If you try to make a pump of 1/3 scale, the pump volume will be 1/27 BUT, the area of the sealing surfaces will be 1/9. So you essentially have 3 times as much leak potential as a full sized unit."

Thank you for that tidbit. I hadn't thought about it that way...

Thank you for the kind words, but more like half trouble shooting, and half fumbling. The build of the roots blower was where the scaling factor became painfully obvious to me, and I had to make the blower bigger than expected to compensate for leakage.

(Edit- I had to rethink the leakage factors. Still not sure how correct it is..)
For example, with a 1/4 scale blower, the blower volume is 1/64, but the lineal length of all the friction seals along the lobes of the blower is still 1/4 of the original. So even if you can cut the seal clearances in half in the model, the losses along the seals might be 1/8, but the volume pumped is still only 1/64. So the actual leakage potential of the model might be 8 times as great as the full sized unit. Luckily, you can overdrive the blower quite a bit to make up for some of the losses and won't be an obvious uh-oh in the model. The blower will just make more noise and the flow increase is (I think) the square of the blower speed increase. Having a belt drive to the blower makes changing the overdrive ratio a bit simpler.
Lloyd

 

[Lloyd-ss]

@The Ignoble Troll It is a bit serendipitous that you started this thread last week. I had set the diesel aside and was working on the Arduino Hall/Optical tach. Your post pulled me back to the diesel and I had to finish up a few things on the injector, coincidentally, incorporating that it looks like all of us have been contemplating.

At the risk of hijacking your thread, which I will not do, here is a brief video I took this AM of what I finished up last night. Here isa thumbnail description. I will post a detailed video and description on my thread about the "2 stroke diesel, will it ever work."
https://www.homemodelenginemachinis...e-will-it-ever-work.31110/page-16#post-415187

Unit injector with no lapped parts. Plunger is1/8" dia polished 17-4ph. Stroke approx 0.13". Will have low pressure supply pump. Mock-up has a weighted column to give approx 120psi feed pressure. Final high pressure is generated in the injector. Plunger rod passes thru a single machined teflon bushing which is also the fuel inlet for the pump. The HP fuel pops a pressure relief valve to exhaust the fuel thru the injector tip. The injector tip is simply a stub rod with a countersunk 4-40 tapped hole. A matching countersunk SHCS threads into th e injector tip hole as if to seal it up. But it has tiny scratches to form nozzle vents for the fuel. It seems to work well but needs clean-up. Still need to make the governor controlled spill vent to control the pulse duration.
Getting to this point seemed almost impossible at times.

A quick video.

 

[The Ignoble Troll]

I don't have a good drawing or model at the moment as its been a few months since I was looking at this with some dedication, and Fusion and / or I have managed to break my assemblies with too many redesigns and edits.
In words, as its a 2 stroke, the cam actuates rockers for the exhaust valves and the plunger with different lobes for each, exactly as in full size practice for the engine type I wish to model. The difference is that between the lobe for the rocker(*) there will be a moveable wedge. I think Roger B has shown this style quite well, just having a hard time finding the post / forum he posted it in, but in his case the cam lobe acted directly on the wedge and plunger. I'll be drawing it again sometime, as I plan to build a test bed for just the fuel system before I go and try and build an engine. Again, following how Lloyd-ss made a test bed for his injector (if I'm remembering correctly).

(*) - I may try designing the moveable wedge to be at the other end of the rocker - between it and the plunger pump. Depends what will work (1st priority) and what may look more scale (secondary priority for now).

I ended up on a return solenoid because a return spring would have to be beefy to overcome the plunger pumps packing. I did not want the load of the spring competing with the solenoid.


Again, I'm months away from chucking metal in my lathe, so updates will be non existent on real world progress for a while.

When I do get to work on it, if it works, I will be sharing everything freely, including what ever code that drives the unit. That way anyone could replicate the unit without having to redesign it.

 

[The Ignoble Troll]

@The Ignoble Troll It is a bit serendipitous that you started this thread last week. I had set the diesel aside and was working on the Arduino Hall/Optical tach. Your post pulled me back to the diesel and I had to finish up a few things on the injector, coincidentally, incorporating that it looks like all of us have been contemplating.

At the risk of hijacking your thread, which I will not do, here is a brief video I took this AM of what I finished up last night. Here isa thumbnail description. I will post a detailed video and description on my thread about the "2 stroke diesel, will it ever work."
https://www.homemodelenginemachinis...e-will-it-ever-work.31110/page-16#post-415187

Unit injector with no lapped parts. Plunger is1/8" dia polished 17-4ph. Stroke approx 0.13". Will have low pressure supply pump. Mock-up has a weighted column to give approx 120psi feed pressure. Final high pressure is generated in the injector. Plunger rod passes thru a single machined teflon bushing which is also the fuel inlet for the pump. The HP fuel pops a pressure relief valve to exhaust the fuel thru the injector tip. The injector tip is simply a stub rod with a countersunk 4-40 tapped hole. A matching countersunk SHCS threads into th e injector tip hole as if to seal it up. But it has tiny scratches to form nozzle vents for the fuel. It seems to work well but needs clean-up. Still need to make the governor controlled spill vent to control the pulse duration.
Getting to this point seemed almost impossible at times.

A quick video.
View attachment 164041

Hijack away, it'll come back eventually.

You've been one of my silent mentors on deciding to build my own unit, I'm completely fascinated with your build.

The two stroke egg is tough to crack but I have few doubts that once cracked, it will open a floodgate of mini two stroke true diesels.


I love the crisp sound of your injector in the video!

Does that single teflon bushing keep the fluid from leaking all on its own? I am worried about teflons creep. I have both HDPE and Teflon sitting by to make the plunger bushing. Not sure which is the best option for longevity.

Edited for clarity

 

[The Ignoble Troll]

Very interesting post. I have been researching this subject myself. I believe its possible to mechanically drive the pump but have had a very difficult time with the issue of fuel metering. The smaller the air flow the more difficult it seems to be. I like the idea of solenoid but have never given the thought you could control the pump stroke that way. I have also considered the injector design and I am of the opinion you just need to get the right amount of fuel into the cylinder at the right time. At a models size I have doubts that the spray pattern is going to be all that important. That is just an opinion and does not mean its true. The injector will have to be balanced so the pressure area works against the closing spring. I would not be surprised if you put a solenoid on that. I could not set up a electrical system like you propose as to me it would be an additional learning curve. I am also thinking that you might be able to use a spring to return the plunger but removing it does lower the force the solenoid needs to pressurize the pump. I cant see any reason why your design concept would not work given the electronics are fast enough for all the variables. I hope you are able to continue your work on this as many are very interested in this subject. Good Luck

As modelers I think we can also get around spray pattern drama by injecting onto a baffle that's designed to break up the spray.

That and indirect injection chambers.

 

[Lloyd-ss]

Hijack away, it'll come back eventually.

You've been one of my silent mentors on deciding to build my own unit, I'm completely fascinated with your build.

The two stroke egg is tough to crack but I have few doubts that once cracked, it will open a floodgate of mini two stroke true diesels.


I love the crisp sound of your injector in the video!

Does that single teflon bushing keep the fluid from leaking all on its own? I am worried about teflons creep. I have both HDPE and Teflon sitting by to make the plunger bushing. Not sure which is the best option for longevity.

Edited for clarity

I am glad you are online! I haven't been this excited about the project in quite a while and I think this is a major milestone. All of the "golden arm" adjustments and fits are gone.There are still a number of potential leakage points but they are mainly housekeeping issues, not functionality problems.

Just an FYI, the pump as demoed in the video is putting out 0.006 grams per stroke which is too much, I think. Injection volumes controllable in the 1 mm^3 are needed.

Here is the basic circuit description of the setup in the video.
Fuel under 120psi enters the pump body via 1/8" copper tube. No check valve on the inlet. The plunger cylinder is stainless, but the inlet hole section is a Teflon bushing with a SS bushing above it to make for adjustable packing compression. The Teflon is fully captured so creep might not be an issue. UHMW might not have the temperature capability. PEEK or a filled Teflon might be suitable. The plunger has a slightly tapered point that passes thru the teflon bushing. All cheese-grater effect holes have been eliminated. In the center of the bushing, there is a side inlet hole of about .050 dia. When the plunger is all the way up, the tapered point of the plunger is still in the bushing but has just a few thou of clearance to allow the 120psi fuel too get by to fill the plunger cylinder. Optimum stroke length is a TBD. The main injector return spring is 130 lbs/inch from previous builds but the Teflon has much less resistance so the spring might be big overkill. The end of the plunger cylinder dead ends at the pressure control check valve. Everything from this point on must be compact with no extra volume. Pressure check valve is a .055 hole and tiny heavy spring and machined seal-tip. Getting this spring, tip, pressure set-up correct is difficult because travel is minimal and the spring preload is critical, and difficult to bench-test. I had to wind a spring from .041 SS spring wire. The fuel passes thru the spring pocket into the 4-40 threaded hole of the nozzle. It is built as if the 4-40 screw is supposed to be plugging up the fuel injector outlet. But some minimal scratches on the screw give a precise bypass and spray arrangement. The 4-40 screw can be removed for cleaning the tip. No impossible-to-clean .006" injection holes to cuss at. In the video there appears to be some residual fuel at the injector tip. Stiffening up the injector spring should fix that. The injector spring operates almost a coil-bind.
The spill port will be between the tip of the pump plunger and the injector check valve. Controlling that with a cam and wedge or a solenoid (ding-ding) ought to work.
Getting the spill incorporated into the injector body will be a challenge because it is now at the point where it either works, or the incorporation of the spill arrangement ruins a bunch of good work.
Lloyd

 

[The Ignoble Troll]

I am glad you are online! I haven't been this excited about the project in quite a while and I think this is a major milestone. All of the "golden arm" adjustments and fits are gone.There are still a number of potential leakage points but they are mainly housekeeping issues, not functionality problems.

Just an FYI, the pump as demoed in the video is putting out 0.006 grams per stroke which is too much, I think. Injection volumes controllable in the 1 mm^3 are needed.

Here is the basic circuit description of the setup in the video.
Fuel under 120psi enters the pump body via 1/8" copper tube. No check valve on the inlet. The plunger cylinder is stainless, but the inlet hole section is a Teflon bushing with a SS bushing above it to make for adjustable packing compression. The Teflon is fully captured so creep might not be an issue. UHMW might not have the temperature capability. PEEK or a filled Teflon might be suitable. The plunger has a slightly tapered point that passes thru the teflon bushing. All cheese-grater effect holes have been eliminated. In the center of the bushing, there is a side inlet hole of about .050 dia. When the plunger is all the way up, the tapered point of the plunger is still in the bushing but has just a few thou of clearance to allow the 120psi fuel too get by to fill the plunger cylinder. Optimum stroke length is a TBD. The main injector return spring is 130 lbs/inch from previous builds but the Teflon has much less resistance so the spring might be big overkill. The end of the plunger cylinder dead ends at the pressure control check valve. Everything from this point on must be compact with no extra volume. Pressure check valve is a .055 hole and tiny heavy spring and machined seal-tip. Getting this spring, tip, pressure set-up correct is difficult because travel is minimal and the spring preload is critical, and difficult to bench-test. I had to wind a spring from .041 SS spring wire. The fuel passes thru the spring pocket into the 4-40 threaded hole of the nozzle. It is built as if the 4-40 screw is supposed to be plugging up the fuel injector outlet. But some minimal scratches on the screw give a precise bypass and spray arrangement. The 4-40 screw can be removed for cleaning the tip. No impossible-to-clean .006" injection holes to cuss at. In the video there appears to be some residual fuel at the injector tip. Stiffening up the injector spring should fix that. The injector spring operates almost a coil-bind.
The spill port will be between the tip of the pump plunger and the injector check valve. Controlling that with a cam and wedge or a solenoid (ding-ding) ought to work.
Getting the spill incorporated into the injector body will be a challenge because it is now at the point where it either works, or the incorporation of the spill arrangement ruins a bunch of good work.
Lloyd

What about controlling the wedge with a servo? If you used say a 70ish lbs 180° servo but used levers to change 180° of rotation into a few mm of wedge travel, the mechanical advantage and resolution should be enormous? It would also mean that you could precisely derive the wedges location without dropping a few hundred on an encoder to directly encode it's position.


I was originally thinking of doing this, the servo/wedge, before I decided to tackle injection volume at it's source.


You have fuel entering your system at 120psi but no check valve, is this because the plunger covers the inlet hole and provides a stop to back flow?


On injection volume, I read somewhere that the ultra large ship engines only inject around 130cc's of fuel at a time. I'm not sure if that's in total or from each injector and I misplaced the source... either way the amount we are trying to inject, as you pointed out, is insanely small.

I wonder if on a single cylinder test engine if it might be easier to have a fixed volume injection but run the injection hit or miss to maintain a given rpm.

Using a pwm signal to extend the solenoid plunger gives me hope that it will produce a variable injection volume, but that's pure fantasy right now.


My other thought on using solenoids was to use a servo to actuate the distance between the pump and the solenoids. That would allow the solenoids to be run at full travel aka no fussing with drivers, but would vary the pump output by changing the displacement relative to the solenoid.

Do you happen to have a link to a diagram of your injector... it sounds rather unique?

 

[ninefinger]

I love the idea of using the plunger as the check valve, it achieves the fundamental requirement without additional parts. Definitely going to borrow that idea in my next iteration of my design. I had already planned on a pressurized fuel supply of around 50 to 80psi to help encourage flow and not be dependent on suction to replenish the pump. I figured on 80psi max as a reasonable achievable pressure for a home built gear pump with diesel fuel as the fluid. Eventually the gear pump would find itself driven by the engine.

One other detail I didn't share previously was that in my design, I had intentionally made it so that if the plunger was over driven manually, it would open the injector and allow flow through bleeding of the system.
With the plunger as the check valve I'll need to reconsider how bleeding would be achieved.
Everything has trade offs...just like your trouble with incorporating the spill port into the injector body...

Mike

 

[Lloyd-ss]

I love the idea of using the plunger as the check valve, it achieves the fundamental requirement without additional parts. Definitely going to borrow that idea in my next iteration of my design. I had already planned on a pressurized fuel supply of around 50 to 80psi to help encourage flow and not be dependent on suction to replenish the pump. I figured on 80psi max as a reasonable achievable pressure for a home built gear pump with diesel fuel as the fluid. Eventually the gear pump would find itself driven by the engine.

One other detail I didn't share previously was that in my design, I had intentionally made it so that if the plunger was over driven manually, it would open the injector and allow flow through bleeding of the system.
With the plunger as the check valve I'll need to reconsider how bleeding would be achieved.
Everything has trade offs...just like your trouble with incorporating the spill port into the injector body...

Mike

Mike, (I am a 9 finger, too, LOL) The bleed method sounds good. Mine currently has no good method to bleed and it is a major pain.
Re: the gear pump. Something I have already tried and it works for the low pressure fuel, is to use a single piston pump driven off the crank and synched/timed with the injector to put max pressure into the inj inlet just when it needs it. An adjustable spring and ball pressure relief in the pump would allow you to tailor the pressure going into the injector. No accumulator is needed, either.
Lloyd

 

[Lloyd-ss]

What about controlling the wedge with a servo? If you used say a 70ish lbs 180° servo but used levers to change 180° of rotation into a few mm of wedge travel, the mechanical advantage and resolution should be enormous? It would also mean that you could precisely derive the wedges location without dropping a few hundred on an encoder to directly encode it's position.

I have locked myself in to staying true in principle to the Detroit Diesel as much as possible. The wedge is less untrue than any electronics, LOL. Plus, I think governors are pretty cool.

I was originally thinking of doing this, the servo/wedge, before I decided to tackle injection volume at it's source.

Yup

You have fuel entering your system at 120psi but no check valve, is this because the plunger covers the inlet hole and provides a stop to back flow?

Exactly!

On injection volume, I read somewhere that the ultra large ship engines only inject around 130cc's of fuel at a time. I'm not sure if that's in total or from each injector and I misplaced the source... either way the amount we are trying to inject, as you pointed out, is insanely small.

Just looked thru my old notes (I try and keep a project notebook) and had calculated that stochieometric ratios for a 50cc cylinder volume, a 25 to 1 air fuel (power load) 2.8 cubic mm of fuel per stroke. (.0024 grams).
At a 40 to one (cruising) air fuel 1.7 cubic mm per stroke. (.0015 grams of fuel).
So my current volume of .006 grams per stroke is too much but that shouldn't be too hard to bring under control.

I wonder if on a single cylinder test engine if it might be easier to have a fixed volume injection but run the injection hit or miss to maintain a given rpm.

Must stay true to the DD, but give it a try.

Using a pwm signal to extend the solenoid plunger gives me hope that it will produce a variable injection volume, but that's pure fantasy right now.

Playing with solenoid impact force by only changing the voltage has shown me that the solenoid forces are definitely adjustable by changing the electrical input parameters.

My other thought on using solenoids was to use a servo to actuate the distance between the pump and the solenoids. That would allow the solenoids to be run at full travel aka no fussing with drivers, but would vary the pump output by changing the displacement relative to the solenoid.

I like that. If the solenoids don't hit hard enough, you could have them hit a mechanical advantage lever arm.

Do you happen to have a link to a diagram of your injector... it sounds rather unique?

I need to clean the drawing up and bring it up to date, but yes, I will share that, most likely on my build thread. The crappy winter weather makes more inside time available!!
Lloyd