..Injectors 2


I tried pressure testing the old Flying Scotsman boiler on Monday and it leaked like the proverbial sieve! So much for using it as a test boiler. I think it would have taken a lot of work to cure all the leaks so I abandoned that idea and decided to use the loco instead.

I rigged up some temporary pipework so that the injectors would be easy to get at and gave them a try. Needless to say not one of them worked! Plenty of steam and water out of the overflow pipe but none into the boiler. To get a better idea of what was (or was not) happening I rigged up a pressure gauge on the outlet of the injector. This indicated that none of the injectors were able to overcome the boiler pressure in order to feed water. I spent another day playing about with various combinations of cones with little success so I decided to check all the dimensions (which I should have done in the first place of course!). I also checked the alignment of the cones and found several problems with the combining cones. One had a slight burr on the end of the taper partially blocking the hole and one was out of line. Both these problems were easily solved by removing the cones and refitting them. Checking the steam and delivery cones showed that the gaps between them and the combining cones were also slightly out which would not help matters. On these very small injectors the dimensions are obviously very critical, any errors and they just won't work.

Test setup

By mixing and matching the cones I eventually made up an 8oz injector that was pretty close to the required dimensions and gave that another try. Still no joy! I then disconnected the outlet pipe and tried it then. It shot a very healthy jet of high pressure water out of the end so it must have been working after a fashion. I did notice that the jet was leaving the injector at a slight angle so the alignment of the cones is still slightly out. I then re-connected the outlet pipe whilst the injector was running and lo and behold the pressure gauge leapt up to boiler pressure and the water gauge showed that it was slowly feeding water into the boiler. Turning the steam supply on and off showed that the injector would pick up most of the time and feed water. Perhaps the fact that the outlet pipe was full of air to begin with prevented the injector working?

Although the injector was now feeding after a fashion there was a lot of water coming from the overflow so it's still not right. Derek Brown suggests this is because the annular gap between the end of the steam jet and the entrance to the combining cone is too large so I tried reducing this with not a lot of success, in fact the injector stopped working altogether. I played around adjusting the gap but the injector refused to work again so I decided to abandon the project until I've got time to make a proper test setup to do the job properly. It's a pain trying to test them using the loco as there's too much to do at once. As well as mucking around with the injector you've got to keep the boiler in steam all the time which is not easy. Over the last couple of days I reckon I've had the boiler in steam for about 4 hours at a time over several sessions and got through gallons of water and loads of coal! Still, at least it was good practice keeping the fire going!

I've got another rally this Sunday at Sutton Coldfield MES so it's back to working on Helen. I've also got to put the driving trolley back together so I won't be doing any more with injectors till after then. I'm going to make a small gas fired vertical boiler from a piece of tube I've got which will be much easier to use and also make the injector tester described by Derek in his article. This is based on the one designed earlier by Bill Carter and consists of a cylinder containg a piston. The injector outlet connects to one side of the piston and the other side is connected to the boiler and so is at boiler pressure. The piston is normally held at one end of the cylinder by the boiler pressure but when the injector is working the output moves the piston to the other end of the cylinder opening a port that allows the water from the injector to drain away. The output can thus be easily measured by noting the amount of water fed over a certain period of time. The cylinder also has a pressure gauge fitted to the injector side so that the outlet pressure from the injector can be monitored.

Incidently for those interested the relevant articles by Derek appeared in ME Volume 184 issues 4111, 4113, 4115, and 4117.



Over the last few weeks I've been working on the test boiler and fittings inbetween other jobs. I had a piece of 3¼" diameter by 1/8" thick copper tube spare so I used that for the shell. I knocked up a couple of tube plates from 3/32" copper to take a central flue of 5/8" diameter and seven outer flues of 3/8" diameter, mainly because these sizes were to hand as offcuts from Helen's and the A1 boilers. Various bushes were turned up to take the fittings and the whole lot soldered up with easyflo. The boiler was then hydraulically tested to 220psi but would probably take 300psi with no trouble.

Two views of the finished test boiler ready for the fittings

The boiler will be gas fired for convenience and will be mounted on a base board to take the complete test rig. The board will have a large hole cut in it over which the boiler will be mounted. The base will then sit on top of the domestic gas cooker with the boiler over the smallest burner. This will avoid the need to make a seperate burner assembly and using domestic gas is much cheaper than the bottled variety!

The boiler fittings will be kept to the minimum - a gauge glass, water feed clack, pressure gauge, safety valve, and a steam turret with two bought in globe valves (I couldn't be bothered to make these!)


The test rig is now finished and operational and although a bit Heath Robinson appears to work!

Boiler mounted on base with pump and fittings

The pump is a bog standard design of 3/8" bore and the safety valve is set to a working pressure of 90psi, the same as Helen's. The two globe valves are mounted on a brass manifold screwed into the boiler and these proved to be quite unreliable, leaking badly under pressure. I managed to get them reasonably steam tight in the end by lapping the cone on the end of the spindle onto the valve seat but they still leak slightly. One of them is left open all the time so a leak does not really matter. This supplies steam at boiler presssure to the artificial load. The other one however controls the steam to the injector and a slight leak here heats up the injector if the water is not turned on and stops it starting properly.

The artificial load I made for the injector is a very simple device compared to the one designed by Bill Carter but seems to work ok. It consists of a brass cylinder with a cover with a 1/4" x 40 connection at each end. The cylinder contains a freely moving stainless steel piston with an O ring seal to make it steam tight. The reduced end of the piston is just to act as a stop and reduce the total movement of the piston to about 3/16". The injector end of the cylinder has a connection for a pressure gauge (to show the injector output pressure) and a hole in the bottom for the water from the injector to escape. In use, the boiler end of the load is at boiler pressure which pushes the piston to the injector end of the cylinder and seals the hole in the bottom of the cylinder. When the injector is operating, the output pushes the piston back against boiler pressure, uncovers the hole in the cylinder, and the water flows out into a convenient container. Thus the injector output can be measured by seeing how much water is pumped in a given time.

The artificial load

Artificial load connected to injector output

At the moment the water feed to the injector is a bit of a bodge being a tin can with an aquarium air valve as a water valve. This will be improved later but I was keen to try the rig out!

The first injector tested was one of the original 8 ounce made earlier. This needed the steam cone shimming out further from the body by about 10 thou for it to start reliably at 90psi. It pumped well but has a constant dribble of water from the overflow which I could not stop by regulating the water supply so it needs some further work on it. However it pumps from 90psi right down to 20psi and packs up completely at about 15psi which is pretty good. Measuring the output gave about 7oz per minute.

The output pressure of the injector seems to be a constant 10psi more than boiler pressure. It starts well so long as the water is turned on first and is dribbling out of the overflow before the steam is turned on. If water is not coming from the overflow it will not start and just blows steam out of the overflow. I think the problem is that the natural water flow through such a small injector is very small and if the injector is at all hot (due to the leaking steam valve as mentioned above) the water flow is not sufficient to cool the injector down and it refuses to pick up. I found that when the steam valve was shut off after running the injector water would not continue dripping from the overflow as it should and so the injector would not pick up again when the steam valve was opened again. Cooling the injector with a damp cloth started the water flow again and the injector picked up immediately.

Water, water, everywhere! Testing the 8 oz injector

Still pumping (and dribbling!) at 30psi boiler pressure. Thermometer shows temperature of output water

Still going at 20psi

The next test was on a commercial injector which my brother bought to fit to his Rob Roy. This is a number 2 with a stated output of 11oz per minute. The connections on this are 5/16"x40 for 5/32" dia. pipe so I had to make up another set of pipes to test this one.

This injector refused to pick up at 90psi and just poured water from the overflow with no output at all. This is usually a symptom of the annular gap between the steam cone and the combining cone being too large so I skimmed a bit off the flange of the steam cone so that it would go further into the body, thus making the gap smaller. This made no difference but I did not want to take any more off the cone in case my brother decided to take the injector back if it didn't work!

By fiddling with the water supply and restricting the water flow the injector suddenly started working perfectly and pumped well. By carefully reducing the water supply it ran perfectly dry with nothing at all coming from the overflow (as should be the case). It worked well down to about 40psi and by carefully reducing the water as the pressure dropped it continued to run dry. However, it would not restart unless the water supply was increased again. The 'modus operandi' is obviously going to be to turn the water on full, turn on the steam, and then reduce the water supply until it picks up.

I did wonder if the steam valve was restricting the steam flow to the injector (a common fault) thus causing the pick up problem so I drilled the passages out as much as I could. This made no difference though so that probably is not the problem.

The injector also suffered from the infamous 'chirping' noises when running dry which is a sign that air is being drawn into the injector via the overflow pipe and in fact putting a finger over the overflow showed a strong vacuum here. I did try reseating the ball valve (a bad seal can cause this effect) but to no avail. However the injector does at least work and should be ok to use but the air drawn in would be a problem with a steel boiler. This vacuum can be reduced by very careful regulation of the water supply but it's a very fine balancing act and not practical to do this when running. I think there has been quite a few accidents caused by drivers fiddling with injectors when they should be watching the road ahead!!

Incidentally, the output of this 11oz injector turned out to be about 18oz ! (which agrees with the diameter of the delivery cone which I measured earlier)

Testing the commercial No.2 injector (unknown make)


I've now also tested the 4 ounce injector as well and even this one works quite well. There is a drip from the overflow when running but not as bad as the 8 ounce ones. The actual output was measured at about 5 ounces. That does not surprise me as I did not have a small enough drill for the delivery cone for this one so I had to use the smallest I had which was a number 80.

I'm going to try all the injectors on the loco before going any further to see if they will work in 'real' life. In theory if they don't then the problem should be with the pipework and valves on the loco rather than the injectors themselves.


Previous Page Next Page

Page 1 3 4