A real Ferrari fuel injector hanging over the mouth of a variable height trumpet.
One of mine
The black tubing is the conductors for the coil and the fuel is fed directly into the injector via the fuel rail
This section of fuel rail had to be made to check the o-ring seal and alignment before cutting o-ring grooves on all ten injectors
The fuel is fed through a channel in the arm of the fuel rail and the channel is 0.8 mm in diameter
Could this be the worlds smallest fuel injector?
You can see the 0.5mm cross section X 3 mm OD HNBR o-ring
The o-ring is only 0.5 mm in cross sectional thickness so the fuel channel which crosses the injector mounting hole at an angle must be above the o-ring on the feed side and below on machining side.
As far as I can tell, the same as on the real engine.
To give an idea of the scale
The body of the injector and fuel rail is 6082T6 aluminium. The coil cover is from 303 stainless with a wall thickness of 100 microns. Its micro bored using a Horn Cutting Tools Supermini® Type 105 boring tool, as are a lot of the parts. These tools are amazing, astronomically expensive but worth every penny. The initial investment is quite large to get a decent stock of inserts but I can now precision bore diameters as small as 0.2 mm (200 microns!). Surface finish has to be seen to be believed. The little “chimney” that takes the conductors away from the coil is CNC machined from Delrin. The nozzle / orifice holder is also from 303 stainless.
The problem with very small.... When I first embarked on this project I didn’t set out to make every nut bolt and o-ring. The problem is that everything on a full size F1 engine is as small and light as is conceivably possible. I soon found that it was one thing designing and drawing parts, it was another one finding someone to supply the various bits and pieces you would ordinarily take for granted , o-rings for instance. After weeks of searching and numerous phone calls, the only company I found that supplies o-rings with a CS of less than about 1mm was in the USA and wanted an obscene amount on money. So I decided to make my own. The first problem was finding a supplier that would provide small quantities of the uncured product.
One of the problems with this project is getting people to take you seriously. The first questions is always what are you making / cutting.... what’s the application ... the then after they’ve stopped laughing they can’t wait to get you off the phone. Persistence is the answer. There are still some companies and people around that seem to care. After a lot of phone calls this company came up trumps - A. Littlejohn Limited They could not have possibly been more helpful and even went to the trouble to providing me with a “beginners guide to moulding rubber” It restores your faith in mankind when you’re lucky enough to find such nice people (thanks terry)
The moulds are made as two halves; the faces are machined in the Cowells CW lathe with a 0.5mm ball end mill to a depth of 0.25 mm and reamed through 1.6mm.
The two halves are clamped together with the uncured HNBR blank in between on a 1.6mm shaft that has threaded ends to pull everything together.
Terry recommended using a diluted solution of original fairy liquid as a release agent so I sent the long suffering other half to Tesco’s to buy “release agent”. It didn’t work so well so I tried a solution of grannies soap flakes which seemed better although it’s not an exact science and it takes ages to clean the mould for re-use. I’ve since invested in some different types of industrial release agents and mould cleaning agents from Ambersil. I haven’t tried them yet as I’ve made a stock of o-rings. There are a number of steps here that need to be followed to get this process to work properly so I’ll do a proper section when I get more time. The whole thing is baked in the oven for 4 minutes at 165 centigrade to cure the rubber.
Other problems with making things this small is finding cutting tools and tool holding sufficiently accurate. Most operations involved in making the injector are done under a microscope with a video camera displayed on a monitor. I can capture clips like these but the capture quality is pretty poor. This is a 100 micron hole in the end of the injector orifice an the white thing is one of my hairs still attached to its follicle!
You soon realise that you’re onto a hiding to nothing with your 30 quid Jacobs chuck and start learning lots of new words like “Albrecht” and “Schaublin”. The only thing I’ve found that has sufficiently little runout and repeatability are the Schaublin B8 collets supplied by Colin at Cowells Small Machine Tools for my CW Lathe. Even the Albrecht chuck which is balanced to 50,000 RPM wont hack it with cutting tools as small as this. Huge testimony here the Cowells CW. I won’t have a word said against this fantastic machine. More on Cowells later in the workshop section
I haven’t had a chance to properly evaluate measuring the flow using the Kern scales as yet. No doubt they’re extremely accurate but when you’re dealing in 1/1000s of grams, just the positioning of the cylinder on the measuring plate can make a big difference. Even a small downdraught from moving an arm can have a sample putting on a few milligrams. I need to build a proper testing system and pass fuel rather than water. I haven’t gone to town with this at the moment as this really needs to be done when the new workshop is properly operational. Never the less, the initial results are extremely encouraging with a pretty linear delivery. The tail off at very low pulse widths isn’t a concern. Most commercial fuel injectors don’t even open until the pulse width gets up to the region of 1 millisecond so anything in the range from 100 microseconds to 1000 microseconds is a bonus. I’m more concerned with fuel delivery at the top end and it’s pretty linear to 20,000 RPM.
Fuel injector video from a slightly different angle showing the spray pattern at different frequencies (RPM) and pulse width ratio (throttle position) There is some lag between the data overlaid and the video. At low RPMs due to the frame rate, some firing events occur between frames making the injector look as though its miss firing. !
I really must sort out a better way of filming. This one was done using a web camera so the quality isn’t very good. I couldn’t overlay any data as I don’t have everything set up in the new workshop properly. I decided to stop using water in the injector as it plays havoc with the ferrous components. Methylated spirits seem like a good choice for now. The injector is mounted upside down so the fuel is sprayed upwards. The mist is then ignited using a Weller Pyropen. When I get the chance, It would be nice to set this up to see how it all copes with spark ignition. This as all beginning sound like an engine management system.
Drilling a 100 micron hole This is the beginning of the injector orifice. I've already started the hole before the video starts The stock is rotating although that's not clear in the video. The view is actually looking through a single droplet of WD40 on the end of the work used to help flush out the swarf.