Experiments With Hydraulic Injection Pump Ideas -and- The Eventual Design and Building of A True Diesel Engine.
31 July 2017: Since
building The Homebrew Hvid, I've decided that I really, really like
engines that burn low grade fuels and don't have carburetors, spark
plugs or need batteries or magnetos.
is to come up with a simple and workable injection pump. At first
thought, Diesel pumps seem like simple variable volume high pressure
hydraulic pumps. The complication comes when you realize that the
start of injection always needs to start at the same crankshaft
position regardless of the amount of fuel the pump delivers.
Having said that, the complications start popping-up
designed a pump from scratch that I think will work although it seems
to be a bit clunky and may take more force than a relatively small
governor can produce.
It's kinda hard to read but here's my idea.
of all, we have to remember that I don't have proper machines to make
very small parts with high accuracy. Keeping that in mind, I want
to only have to do the really twiddly stuff to the pump itself.
Everything in the drawing is to scale and a reference is the pump
plunger which will be 0.12600" in diameter. I will drill, ream
and lap the bore for an extremely close sliding fit to the plunger.
Today, I did a test piece to find out just how hard it would be
to make this.
have just guessed the approximate maximum amount of fuel that needs to
be delivered to an engine with a one inch bore and two inch stroke
(1.5708 cubic inches). This is probably a lot more than enough
but I have to account for pump leakage.
In the drawing, the top
part is a simple hydraulic pump with a spill port to ensue that
delivery occurs at the same point in the stroke. The plunger,
when it is at rest at the low part of the fuel cam below it, will
uncover the 0.040" spill port that leads back to the fuel inlet.
As the piston rises and closes the port, delivery will
immediately begin. At the full fuel position, the cam is timed so
that the spill port is covered at TDC on the compression stroke of the
engine (this could be changed to suit the characteristics of the engine
it is to be used on). Fuel delivery stops when the apex of the
cam lobe is reached, about 80 crankshaft degrees from TDC.
order to keep the start of injection at the same point regardless of
the amount of fuel pumped, I rotate the cam follower in opposition to
the cam. The total rotation allowed will be 40 degrees or less.
In other words, I advance the timing as I shorten the pump stroke
but attempt to delay start of delivery to match the amount of timing
It may be a bit confusing, but I show the pump
in the maximum fuel position. Then, I show three more fueling
positions. At 20 degrees, or half fuel position, there is
actually slightly more than half of the total delivery. This is
because I have no way of machining a complex curve on the follower shoe
and am using a simple curve I can machine using the boring head on the
mill. Since the curve is simple, the timing is actually a bit
advanced at this position so static spill timing will have to
compensate for it. If you think about it, as the engine speed
increases, timing should be advanced a bit for optimum running.......
Oh, yes - Any leakage from the pump plunger will
fall to the lower part of the housing where it will lubricate the cam,
bearings, etc. There will be a drain and overflow back to the
In the future, before I actually decide whether or
not to build this pump for testing, I may fiddle with the curve of the
follower shoe to see if I can get it closer.
Now for the test piece.
My test fit piece.
off, I rooted around in my pin collection and found one that was
0.1245" as close as I could tell. I checked sizes of other pins
and found a box of pins that nominally measured between 0.125" and
0.126". The body that the bore was made in is made of leadloy
steel. I drilled a 5/8" deep hole with a #31 drill (0.120").
Then I reamed it with a 0.125" reamer. With the hole
carefully reamed, the 0.1245 pin was a slightly tight fit but, using
some very fine alumina made into a watery paste with Diesel fuel.
I got the pin to finish the bore for a relatively good fit.
After carefully washing out the bore and pin with Diesel I
blew them clean with air.
My test fixture was just a
small digital 10lb scale. With a bore of approximately 0.125", pressing
with a force of one pound on the plunger will produce about 86.4 pounds
of pressure. In testing by pumping air and using #2 Diesel
fuel as a lubricant, I could only press with about 1.5 lbs of force
(134.5 psi) before it leaked. I think this step just removed the
reaming marks left in the bore.
Since I have no way of
measuring the bore size and can only get to within a ten thousandth of
an inch (0.0001") with my micrometer, I had to rely on actual fit.
I tried some of the larger pins but they would only almost fit
the bore. I took the undersized pin and again applied alumina and
turned it in the bore until I could get one of the selected pins to fit
with only slight binding in the bore using Diesel as a lubricant.
worked this and rinsed a few times until the fit was smooth. This
time, I had to press with about 4.5 pounds of force (403.4 psi) with
Diesel fuel before it leaked. This is about as good as it got and
is not going to be good enough for a Diesel engine. I think I
will need between 1,000 and 1,500 psi so the injector will crack at a
high enough pressure to overcome cylinder compression and atomize the
fuel for combustion.
My final test to just see if it would work
any better, I switched to 30 weight motor oil. This time, I
coudn't make it leak air past the fuel lubricant no matter how hard I
pushed on the plunger. Since the scale only goes to 10 pounds,
all I can tell is that the fixture makes more than 900 psi.
doing this test, I think I can be confident that a serviceable
injection pump can be made in my shop. I know I can do the rest
of it. I will order a one eighth inch cylindrical lap, a couple of extra
replacement barrels and one or two 0.126" precision steel perforator
pins if I decide to go ahead with this design
are other pumps I can make. The simplest is one that has a manual
timing adjustment. Since, without changing the timing adjustment,
only mediocre performance can be achieved at varying loads.
also studied the fuel pump used in a large, old Diesel engine that
always starts delivery at the same time with a fixed stroke but uses
lever, eccentric and pushrod arrangement tied to the pump rod to raise
the inlet valve to the pump and stop fuel delivery. My guess is
that it will take a substantial amount of force from the governor to
counteract the force of the inlet valve lifting pushrod when under
1 & 2 August 2017: Looking
in one of my old technical books, I found what I think is a very simple
and buildable pump that has constant start of delivery and varies end
of delivery to control volume. This one came out of a 1925 book
and is simply titled "Typical Arrangements of Mechanical Injection
"Typical Arangements of Mechanical Injection Pump" from 1925 book.
This side shows my take on the design .
like the method of timing the end of delivery although, at over 1,000
psi, it may take a bit of force to crack the inlet valve. If the
load eccentric is designed right, the load on the governor will only be
significant when the pump is making pressure.
think the design I'm working on (above right) will work. I've
still got to dimension the drawing and add some refinements. I
will try for the main fuel tank to be mounted directly to the injection
pump housing and feed fuel by gravity. If there is too much
leakage in the pump, I may have to go with a small overflow day tank
for the injeciton pump with a lift pump to raise fuel from the main tank.
3 August 2017: I
woke up this morning and had a "DOH!" moment. As is my usual
practice, I made too much of a simple concept. After watching
Find Hanson's YouTube channel and thinking about it for a while, his
simple and effective method of controlling injection really makes sense
to me. I've seen the system before in books about larger engines
and just didn't think it would work with something small.
Find does is to insert a governor controlled "floating" wedge between
the cam follower and the pump plunger. Because the wedge floats,
as it is moved in and out between the follower and the plunger, varying
the stroke of the plunger. In ooperation, it tiltsto take-up
any space (there is never any clearance between the cam follower and
the plunger, so that the pump always starts delivery at the same time
but stops delivery when the cam finishes rising. Of course, the
pump is simply a precision built pump with check valves.
may be a pause in this narrative while I await another book on Diesel
principles and think this whole thing over. In the meantime, I
may modify ENG12-2.DWG, making it ENG12-3.DWG which uses the floating
wedge for control.
Since this whole precision pump thing is new
to me, I will make the pump as a separate, self contained mechanism so
it can be modified in the future without having to partially
disassemble the engine.
7 August 2017: While
awaiting delivery of another Diesel fuel injection book to study, I
made a major revision to ENG12-2.DWG. This design should be
easier to build and I have fleshed it out with some fittings, a day
This should be buildable.
substituted a flat bottomed wedge for the swivel and eccentric
arrangement. I've added a fuel lift pump and a day tank.
There will be an inline filter between the outlet of the lift
pump and the day tank inlet. Note that the overflow from the day
tank flows to the cam/tappet/wedge housing and from there it overflows
back to the fuel tank. Any leakage past the injector plunger
naturally goes back to the tank. There will be a trough inside
the housing to direct the overflow from the day tank to the middle of
the assembly to make sure fuel flows over all of the moving parts.
is assumed that #2 Diesel will have enough lubricating properties to
take care of the moving parts of the fuel injection pump assembly.
The cam bearings and cam followers are unsealed ball types.
entire day tank and the facing cover of the bottom mechanicals are
to be made of 1/4" plexiglass so I can see what'e going on.
note that the main bearing mounts are independent from the front and
back covers. Pipe fittings are approximately scaled to 1/8NPT.
Virtually all of the screws shown are 6-32, except for the day
tank which are 2-56.
awaiting "The Book", I went ahead and made the parts for the day tank.
As it is, I have to go to Fastenal to get the 2-56 X 1/2 screws
before I can put it together. Here are the plexiglas parts.
There was one change made becaause when I started looking at the
scale of things, I substituted some 1/8" steel tube for the 1/8 NPT
fittings which looks a lot nicer. This tubing can be partially
threaded with a 10-32 die and the female parts threaded with a matching
plug tap and they fit really nice.
14 August 2017:
Day tank parts
got "The Book" yesterday and am studying it to see how my idea for a
pump compares with what the "professionals" do. So far, I have
seen some really good designs that are proven by time to work well on
modern engines but they are too complicated for my shop. Later
today and before I do any more, I will finish the book to make sure I
don't make too many stupid mistakes on this project.
12 August 2017: "The
Book" is read, the decision made. I will use my design, above.
It is the simplest and has the greatest possibility of being
built successfully. Having said that, here is the day tank.
The back cover is loose because it has to be off to be able to
tighten the drilled bolt to the injector pump.
13 August 2017: Okay, I've finished the injection pump itself (not the rest of the system).
I am at the start of the day.
Ready to be assembled.
Looks nice but
I've made the pump
body out a piece of ductile iron left over from another project.
On the left in the left hand image above, you can see the
cylindrical lap. Then, there's the extractor pin that I will use
for the plunger. They are ground and polished and very accurate.
This one has a diameter of 0.126". I drilled the piston
bore with a 0.120" drill then reamed it with a 0.125" reamer.
The bore actually was about 0.1255" due to the cheap reamer set I
The lap is a rod with a split brass end piece. There
is a tapered plug in the end that can be screwed in to increase the
size of the lap. Since this is the first really close lapping job
I've done, I used some very fine alumina powder made into a thin paste
using Jet-A (kerosene, parrafin). I chucked the lap in the lathe and took my time and slowly increased the bore by the
approximately 0.0005" until the pin just fit with a slight
interference. I hope that's good enough for this engine because I
don't know how I can get the fit any closer.
Then, after lapping
the ball valve seats, I put it all together and filled the day tank
with Number 2 Diesel. The first thing I noticed when I worked the
plunger was bubbles coming out of the intake port. Not good!
At that point, I quit for the day. Tomorrow, I will take it
apart and see if I can improve the valves.
14 August 2017: Spent
the day fiddling with the injection pump valves. Both leaked
under pressure. I ended-up making a lap for the outlet valve seat.
The 45 degree lap.
Lapping the inlet seat.
lap was made from a piece of drill rod, turned to fit into the 1/4-28
threads of the outlet valve cavity. The outlet valve ball seated
well enough to not leak gasoline with just gravity holding it to the
I also lapped the inlet valve seat. No matter how
much I tried, I could not get it to hold any kind of pressure.
There is an issue with there not being enough space for the
spring and it continually cocked to the side, making the ball unseat.
I got frustrated with it and ended up making a poppet valve out
of a flat head 4-40 machine screw. Because the little poppet valve
stuck out from the inlet bolt seat much less than the ball did and had
a larger head, there was room for a larger diameter spring. I
used the lap to widen the seat in the bolt then lapped the 4-40 valve
to th14 August 2017:e seat. It also holds well enough for the pump to prime and
move Diesel fuel..
The pump undergoing initial tests.
pump slowly leaks down to the fuel cup but, when I put my finger over
the outlet hole and press the plunger, it still squirts no matter how
hard I press the outlet. Tomorrow, I will again thrash with the
pump valves. 15 August 2017: I
got an email from David Jones in Texas and he sent a couple of images
of a McCormick Deering WD-40 injection pump. Looking at them,
they are very similar to the ones in the 1&2 August post.
McCormick Deering WD-40 injection pump, courtesy of David Jones.
David very politely set me straight on my faulty design. The way
I have the latest edition, as the fuel delivery is decreases, the start
of injection is later and later. NOT GOOD!
The bad news is
that I've got to go back to CAD and re-design the whole outfit so the
governor controls the inlet valve of the pump. That way, the pump
stroke is constant and injection always starts when the plunger starts
moving. Injection is terminated when the eccentric causes the
inlet valve to open earlier and earlier in the pump stroke.
That's why I try to get all the input I can on these brainstorms.
My best work has always been when I'm getting input, criticism,
kibbitzing and argument from other technical types. This working
in isolation causes me to do a lot of back tracking. Thanks,
I also got an email from Chris in Illinois with
suggestions on how to make ball valve seats seal by "coining" the seat
with a ball that is disposed of after banging it into the seat with a
punch. I didn't think it would work in cast iron and steel.
I have done it in aluminum and brass with success. If I
make the next pump with ball valves, I'll give it a shot.
Aside from all of that, I did get the gauge and made up some plumbing to hook it to the pump.
Pump test fixture.
you can see, it will hold about 150psi for a quite a while after
pumping. That's the good news. The other good news is that
the pump will make at least 1000psi when pushing the plunger with my
thumb (that's about as hard as I can push it without it getting
painful. Because of the leaks, the pressure drops down within a
second or two. The bad news is that I'm gonna have to re-think
the whole thing and start over. At least all i've gotten done is
the pump itself and that was good practice for the real thing.
Well, I guess if it was easy, everybody would be doing it.