Help with 'Speedy' valve gear and other issues...

[Roger]

It's taken a few more attempts to get the rear valve to seat properly. I reduced the end of the lapping ball shaft so it would fit the Dremel, and that made life easier.

Anyway, they aren't 100% perfect, there's a slight leak at 100psi on the air test, but I'll give it a try just to see where I am with it all.

20210106_123534 by Roger Froud, on Flickr

So today I've made up the new Steam feed pipe and assembled it all with the last One Piece Cone that I made to check that I'm nipping the end of it and that seems good.

20210106_160901 by Roger Froud, on Flickr



20210106_160912 by Roger Froud, on Flickr

Out of interest, I've just tried the old cone and suprisingly that seems to work way better than it did in the scale 8X injector body.

Initially I had no idea what the regulation gap was, so I just fiddled with it to find where it worked best. It seems to pick up anywhere from 50-90PSI even with water at 30C so this is much more promising. There's too much water coming out of the overflow at lower pressures, but pinching the pipe reduces that a lot.

So this was a pretty freestle uncontrolled test just to see if it leaked or was just hopeless. I think I need to take another look at the overflow valves to see if I can get those really air tight, and I can see the Steam valve has a small leak which I ought to fix.

Anyway, it seems a different animal even on this initial test, which prompts the question why? The Steam Cone is new and has that long tapered entry, so maybe that's important after all.

So I'll let it all cool down and measure the regulation gap so I've got a starting figure.

This is the style of One Piece Cone I need to make to take advantage of the separate overflows

Test injector with modified one piece cone by Roger Froud, on Flickr

[Roger]

I've managed to get the second overflow valve to be almost a perfect seal, so I'll leave that for the moment.

However, in trying to measure the regulation gap, I can see that there's a burr on the inside of the injector body where I've added a tiny chamfer to help the O-ring enter the bore. That's going to upset the measurement so it's got to go.

So here I've turned a Steel lapping mandrel and put some fine grinding paste on the face.

20210106_203808 by Roger Froud, on Flickr

I ended up using the sprung tapping centre which misses the outlet pipe and also keeps a gentle pressure on the mandrel.

After a few goes with this I can't feel an edge there any more.

20210106_203955 by Roger Froud, on Flickr

That all needs cleaning up in the Ultrasonic tank and trying again.

Ok, having tried this again, I think the seal on the valve is more important than I imagined for maintaining a stable flow. It might be that it would benefit from a little more spring pressure too, but that's another question entirely. I can hear it rattling around and not quite seating, but that might be a chicken and egg situation.

I've taken it apart and measured the Regulation Gap I ended up with after this test, and that turns out to be 0.16mm which is smaller than I had at any point with it in the Scale body. That might account for why I wasn't getting good results.

Anyway, that's enough for one day. I'll pick it up again tomorrow.

[jma1009]

Hi Roger,

Quite inspiring!

As I posted awhile back, you ought to consider adding a water valve to the water inlet, and it ought to replicate those I think you have already made so far as the internals are concerned.

Don't overdo the lifting properties - you are never going to have to lift more than a few inches of water and there is no point giving the injector a challenge for testing purposes that it doesn't need. I would raise considerably your water bottle supplying the water intake for the test injector set up.

A self restarting automatic injector must have a lifting characteristic, but these only need be moderate in your loco.

Cheers, Julian

[Roger]

Jan 6, 2021 22:01:24 GMT jma1009 said:

Hi Roger,

Quite inspiring!

As I posted awhile back, you ought to consider adding a water valve to the water inlet, and it ought to replicate those I think you have already made so far as the internals are concerned.

Don't overdo the lifting properties - you are never going to have to lift more than a few inches of water and there is no point giving the injector a challenge for testing purposes that it doesn't need. I would raise considerably your water bottle supplying the water intake for the test injector set up.

A self restarting automatic injector must have a lifting characteristic, but these only need be moderate in your loco.

Cheers, Julian

Hi Julian,
Yes, a water valve is going to be fitted. However, the one I ordered isn't suitable, so I've ordered a different one. I could make one, but at the moment I'd rather explore other issues.

The lifting properties are just a consequence of the ratio of the Steam throat to the Condensing Cone throat. I could reduce the amount by changing that, but at the moment there doesn't seem much point. I hear what you're saying though. This is happily lifting a 8 inches or more at the moment.

I think the key issues are to do with the overflows and how they seat. However, the gap in the Combining cone and how drill ways deal with that are likely to be important too.

I'm just starting out really, but at least it's more promising than it was. Being able to change the regulation gap while it's running is a game changer.

[jma1009]

Hi Roger,

You might be misunderstanding my recent post - the sort of injector we attain to must have lifting characteristics and you have quite rightly designed these in same as I would do if I were doing the same - but there is no need to impose on your tests having the injector 8 inches above the water bottle.

You might be surprised the advantage to be gained by having a water valve with the same passages and operation of that fitted to your loco.

For one thing, you can reduce the water as pressure drops to probably much lower steam pressures than you have already noted. And you can easily test the restart characteristics and note what happens at the overflows.

[dscott]

I have been looking for this for some time and it just popped up at last.
Taken at the Bluebell several years ago.
Very small spaces in relation to the size.

David and Lily.

[Roger]

Jan 6, 2021 23:47:39 GMT jma1009 said:

Hi Roger,

You might be misunderstanding my recent post - the sort of injector we attain to must have lifting characteristics and you have quite rightly designed these in same as I would do if I were doing the same - but there is no need to impose on your tests having the injector 8 inches above the water bottle.

You might be surprised the advantage to be gained by having a water valve with the same passages and operation of that fitted to your loco.

For one thing, you can reduce the water as pressure drops to probably much lower steam pressures than you have already noted. And you can easily test the restart characteristics and note what happens at the overflows.

Hi Julian,
Ah, I see what you mean. I've actually got the bottle I'm feeding from sitting on the bench, so the only suction required is to start the syphon. Once it's done that initial lift, it doesn't require any suction to keep going. So although it's not identical to how it will be on the locomotive when it starts, its pretty much the same without needing a lift once it's running.

[Roger]

Jan 7, 2021 4:13:13 GMT dscott said:

I have been looking for this for some time and it just popped up at last.
Taken at the Bluebell several years ago.
Very small spaces in relation to the size.

David and Lily.

Hi David,
Now that's an interesting variation, I've not seen that one before. I note that there isn't any overflow between the outlet of the sliding Mixing Cone and the Delivery Cone.
I suppose a small volume is created as it starts to pick up though, because the Mixing Cone moves left and the space to the right of it increases slightly. I guess there isn't a vent hole there so that the pressure can build up and push the Mixing Cone back.

Another thing this seems to indicate is that the volume between the Mixing Cone and the Delivery Cone is going to be full of water. This is something I was concerned about with my design having a single overflow valve on the outlet. I imagined it filling up with water and that somehow interfering with the flow at the Delivery Cone. It would appear that this isn't an issue.

A couple of other things are interesting too. They too use a lead in for the Steam at the Steam Cone, and they also have a very short Condensing Cone with a huge bell mouth. They also have a very long and shallow tapered Mixing Cone where the angle changes from a steeper one near the start.

As with other full sized injectors, they don't seem to have any worries about small leaks on whatever system they use for the Condensing Cone overflow. I think it's more important for our smaller designs though.

[Roger]

Ok, I've probably not done the seating on one of these valves any favours with a little too much grinding, so I need to use another approach.

Now, I could use a PTFE insert, but I'd prefer to use something off the shelf which can be replaced easily. So here's my solution using a 1.5mm section O-ring with an ID of 2mm

I like the idea that I can pull them off their seats like on the real thing, and that I can also press on them if I want to simulate a stronger spring.

Overflow valve by Roger Froud, on Flickr

This is one of the bottom caps getting the hole for the shaft. Obviously this will dribble when it overflows, but that's the same as the one on the scale injector. It's not really feasible to seal this, so I'll live with that.

20210107_110234 by Roger Froud, on Flickr

I've making these trial valve stems from Brass, they'll be Phosphor Bronze or Stainless Steel on the real thing.

20210107_113921 by Roger Froud, on Flickr



20210107_115440 by Roger Froud, on Flickr

It's not easy to test if these leak because of the noise coming from the air leaking out of the shaft clearance. I'll have to dunk it all in water to prove the point.

20210107_122135 by Roger Froud, on Flickr

Ok, I've done the dunk test and they seem to seal 100% air tight, so that's good. I did notice that there are two tiny leaks at the joint where the overflow pipes meet the bodies. However, it's not enough to worry about, it won't affect the working of the injector. The overflow pipes just stop the water going everywhere.

I've just tested these, and there was too much back pressure and that prevented the injector from lifting at all. In other words, Steam came out of the overflows, but the water wouldn't draw. I think there may have been a little suction, but not enough to start the syphon.

Anyway, that's not satisfactory, the springs are obviously too strong, even though they're feeble. So with a little experimentation, unscrewing the covers, I could feel that it was compressing the springs by about 0.5mm, so I machined 0.4mm off the stems to reduce the compression.

That solved the problem, and it lifts nicely now.

This is another quick test, but this time using the One Piece Cone that I made to try the Radial O-ring, but with the O-ring removed of course. It still has the sharp edge on the entry to the Condensing Cone.

So the main difference between these two cones is that one has a long bell mouth on the Condensing Cone and the other doesn't. It doens't seem to make an appreciable difference on the face of it. 

What's interesting is this shot of the Injector delivering 95PSI static pressure into a closed valve from just under 85PSI boiler pressure. So it's surprising that the pipework from the Injector to the boiler is considered to be important. Obviously the flow rate will be much reduced if there is too much restriction, but it looks like it ought to work to some degree.

20210107_165026 by Roger Froud, on Flickr

[Roger]

Ok, enough playing around with the Test setup, I think that now does what I need it to do.

So first up is a new One Piece Cone to explore the size of the overflow holes and clearances. I'm ignoring the figures used in D.A.G Brown's book because I can guarantee absolute alignment across the gap, and that might make a difference to what you can get away with.

I'm pretty happy with how strong this is even in Brass with the smallest diameters being 3.3mm

The gap in the Combining Cone from Brown's data is 0.889mm (0.035") and Bob Bramson says anything between 0.762mm-1.02mm (0.03"-0.04")

Just push the limits to see what's possible, I'm starting with 0.6mm (0.023") and 10 of them.

Presumably if this is too small, there will be too much back pressure and it won't pick up. I imagine this is going to be worse at the highest pressure. Now, you may wonder why I'd start this low when we already know what does work? My thinking it that the issue with any gap is that the flow has to go across it, so the shorter the gap the better. Of course, I might be wrong about that, knowing that a jet of water has a vena contracta when it leaves an orifice. Quite what effect a scalloped edge has is anyone's guess.

However, if I don't try something small, I can't put any limits on what works. To my way of thinking, the less the continuous taper is interrupted, the better.

One piece cone with 0.6 x 10 combining holes by Roger Froud, on Flickr

These are dead easy to make. This version has three O-rings, so those are going in first.

20210107_180840 by Roger Froud, on Flickr

Then the same tool was used to add the clearances where the cross holes are going. Just thinking aloud, I could make a Form tool to do all of these in one hit if I wanter to make a batch.

The longest operation is to drill the centre hole. I spotted the end with a 1mm PCB drill, then drilled 0.5mm at a time to 16mm depth with a #74 drill.

20210107_204216 by Roger Froud, on Flickr

Then it was onto the 4th axis for ten 0.6mm and six 0.8mm holes for the overflows. I've actually reduced the Delivery overflow from 0.95mm to 0.8mm too.

So this is my starting point. After this, only one variable gets changed at a time!

20210107_210617 by Roger Froud, on Flickr

20210107_214303 by Roger Froud, on Flickr

The setting collars are already at the right positions on the reamers, so they just need the 25mm Slip gauge to get them the right distance from the face of the cone and then wind them in the required amount.

20210107_215232 by Roger Froud, on Flickr

The cone was parted off long and then trimmed to the right length so I can use the correct figure from the end for the second reamed.

20210107_220914 by Roger Froud, on Flickr

Job done. Even doing this in a knife and fork operation like this has taken less than an hour. With a form tool and a bit of practice, I reckon I could easily make it in half an hour.

I'd be interested to know how long it takes to make a set of three cones, even when you're up to speed.

20210107_221054 by Roger Froud, on Flickr

Anyway, it's too late to start playing with this now, I'll give it a go tomorrow. I won't be in the slightest bit surprised if it doesn't work at all.

[jma1009]

Hi Roger,

What does it tell you that you feed at a delivery pressure 10psi above the water pressure?

Why do you need a spring on the test injector body check valve for the gap (or your holes) between the two parts of the combining cone? Jim Ewins used to test his injectors upside down (traditional design with no spring). I thought the spring on the combining cone to delivery cone gap (holes in your case) was to achieve a different effect?

What is this different entry to first part of the combining cone? You haven't detailed this previously unless I am mistaken.

Cheers,
Julian

[jma1009]

My apologies, for "water pressure" in my above post please instead read "steam pressure".

(My iPhone won't allow the 'edit' function!)

[Roger]

Jan 7, 2021 22:40:25 GMT jma1009 said:

Hi Roger,

What does it tell you that you feed at a delivery pressure 10psi above the water pressure?

Why do you need a spring on the test injector body check valve for the gap (or your holes) between the two parts of the combining cone? Jim Ewins used to test his injectors upside down (traditional design with no spring). I thought the spring on the combining cone to delivery cone gap (holes in your case) was to achieve a different effect?

What is this different entry to first part of the combining cone? You haven't detailed this previously unless I am mistaken.

Cheers,
Julian

Hi Julian,
There was some debate about whether an obstructed flow in the delivery pipe would cause the Injector to stall. That's why I turned the valve to the off position so the output from the Injector was completely blocked. All it proves is that there's potentially a 10psi difference available to overcome any restrictions in the piping. Obviously these need to be kept to a minimum, but It's good to know what sort of overhead is available.

The springs are just to make sure that the valve returns to its seat. It may be that it's not necessary if there's enough suction to draw the valve shut. However, since I'm trying to keep air out of the Injector, I thought it made more sense to make sure that closed is default position. The valve only need to be open enough to allow the excess water and steam to escape. I might have confused you by showing a section view with both types of valve on the same view. It was just a compare and contrast view, rather than the intention to make different valves for the two places. I'm made them both with the O-ring type of valve now.

The entry to the Condensing Cone on this version is a sharp corner with just the edge broken. The previous two I made for the scale body had a sharp edge on one and a big radius on the other. On that initial test, it didn't seem to make much difference, so I thought I'd start this with the simplest arrangement ie just reaming the holes and very lightly taking the sharp edge off. It may well be that further down the line this gets a radius, but I want to establish a base line with this layout first.

[don9f]

Hi Roger, this is great stuff, can’t wait to see what happens....even making the form tool would be fascinating.

Cheers Don

[Roger]

I couldn't resist designing a form tool for this since I'm going to be making a fair few of the cones.

I've opened and saved a copy of my standard 20mm x 4.5mm thick Form tool as the starting point and copied and pasted the profile edges I need to recreate.

The intention is to turn the end face and the OD to 6mm so that this tool only has to create the O-ring and clearance diameters. I can't realistically add the edge chamfers, those are only 0.1mm and will be added with a needle file. That will only take a few moments, they're just there to take the sharp edges off.

I've drawn a line that's 0.3mm away from what would be the 6mm diameter and then extented all of the lines beyond the 6mm diameter to meet it. Then I've added 0.25mm radii on all of the internal corners to show me what the profile will look like when I use a 0.5mm cutter. I'd like to use a bigger one, but the cuts are smaller than that.

Anyway, the sketch is closed with the 5mm side and the long side clear of the stock. That's then extruded as a cut to chop away the end of the bar...

Form tool desired profile by Roger Froud, on Flickr

... like this.

Form tool projected profile by Roger Froud, on Flickr

That's then projected onto the X/Y plane from the angled one, to give this open sketch. The / and arrow symbol indicates that it's tied to the other sketch so that any changes to that are carried over.

This becomes the path the we need to machine.


Form tool CAM profile by Roger Froud, on Flickr

Here you can see it highlighted in Orange and the Triad on the right is the zero point of the coordinate system.

This is showing a 5.5mm cutter removing the bulk of the material before starting the finer cuts.

Form tool bulk rough by Roger Froud, on Flickr

This is only going 3.5mm deep because that's all I need. I'm leaving 0.08mm of stock on the part to make sure it clears the stock in the widest gap. It's taking 0.1mm deep cuts and very slowly since I'm using a 0.5mm cutter, and a long one at that.

Form tool fine rough by Roger Froud, on Flickr

The finishing pass is at 0.2mm deep.

The total machining time is about 3-4 hours depending on if I need to slow the feedrate below the one I've chosen. We'll see.

Form tool finish profile by Roger Froud, on Flickr

Anyway, it's simple enough to do, and that can be doing its thing while I'm playing with the one I've just made.

[uuu]

If you wanted to speed the machining, and get uniformity, could you use your mill as a lathe
(as you have previously demonstrated) ?

Wilf

[Roger]

Jan 8, 2021 7:17:50 GMT uuu said:

If you wanted to speed the machining, and get uniformity, could you use your mill as a lathe
(as you have previously demonstrated) ?

Wilf

Hi Wilf,
That's certainly a possiblilty, but it does take a fair bit of time to set it up. Once it's set up and running, you want to knock out all of the ones you're going to need because it's a pain to set up again. Twenty parts is the absolute minimum I'd set it up for, preferably fifty. It takes maybe half a dozen parts to get it settled down and the program adjusted to get the right depths of cut and strategy.

It would make it possible to peck drill the deep hole, and that would save a lot of time. I had considered it for the drilling, but thought better of it. For mass production, it would almost certainly be the quickest way for me to make them. The advantage of being able to do all of the first Lathe operations in one go, automatically, is certainly attractive.

If it didn't take so long to sort out the programs and set it up, I'd do it now. Creating the programs with the Fusion360 Lathe CAM output would be the obvious thing to do if I was using that. I have a pretty clumsy way of preparing the programs because I have to use the Milling output from Alibre. At the moment, I don't want to get involved with Fusion360, I've got too much else to do.



I certainly won't rule it out if I'm spending too much time making them this way. A form tool will just leave the centre hole to do, so it's still a huge saving on time.

[jon38r80]

Your new set up looks far more workmanlike and less cobbled together. The original was getting a bit messy from adaptions.
Good luck with the next round of testing.

[Roger]

Here I'm reducing the diameter of the support for the central O-ring, so that when it's removed, there's a significant path for any flow between the two overflow zones.

Basically, I'd like to find something that works really well with the O-ring in place, and then find it works just as well with the O-ring removed. That may be a pipedream, but that's the ideal. I might be able to live with a compromise in performance as long as it's not too great. We'll see, there's still an enormous amount of work to do.

20210108_164103 by Roger Froud, on Flickr

This is how I'm measuring the Regulation Gap. I've got a shortend arm on the clock so it's more sensitive, but it does mean that I can't rely on the figures on the dial.
Instead, I set zero on the DTI and the DRO and then use the DRO to tell me what the distance is when the DTI is reading zero on the Steam Cone.

20210108_124556 by Roger Froud, on Flickr

So here's the setup for machining the Form Tool for the outside of the Injector body. I have a standard 8 degree wedge that matches the 3D model .

I use the LH edge as the machine zero and set the front by eye. At long as it cuts all of the way along, that's fine.

20210108_113235 by Roger Froud, on Flickr

I'm using the High Speed Spindle at about 20KRPM with the 0.5mm PCB cutters. They're so fragile, but just about usable if you don't mind a few breakages. The flute length is miles too long for this sort of thing really, but that's how they come. The point is, they're rediculously cheap, so it's worth being patient and letting them do their work. Tiny cutters tend to be very expensive.

20210108_124237 by Roger Froud, on Flickr

This is how far I got with the first cutter, so not too bad. I'm currently on the fourth, and it's not much further on to be honest. So I've upped the RPM and tried a different cutter style. Strangely enough, burrs seem to work better than fluted cutters.

20210108_155533 by Roger Froud, on Flickr

Anyway, I did conduct a few tests today, and concluded that the 10 x 0.6mm holes for the Combining Cone overflow isn't quite sufficient for creating enough lift. Once it was cajoled into working, it worked fine with water a 10C and again at 30C, although it was less keen to start at the higher temperature.
Clearly this was a step too far, so I've changed the holes to 6 x 0.75mm for the next test.

I've also modelled a two hole row version that I might give a try too. The idea being to stagger the smaller holes so that the flow only has to jump a hole rather than a complete gap.

Watch this space. I'm going to persist with the form tool first, it will make life so much easier.

[terrier060]

Jan 7, 2021 4:13:13 GMT dscott said:

R
I have been looking for this for some time and it just popped up at last.
Taken at the Bluebell several years ago.
Very small spaces in relation to the size.

David and Lily.

Really sorry about breaking into your thread Roger - brilliant work by the way, But Dave and Lilly have got me really exited.  Could that section be throug a Terrier injector.  It looks very much like the one on my thread.  If so I could give you both a kiss, as I have been asking for some time if anyone knew whT the inside of one looked like.