Valve-timing a Maggie Romulus

[matthewmakesmodels]

My dad and I purchased a Maggie Romulus (an 0-6-0 version) during lockdown (castings and drawings can be obtained from Blackgates – we have the drawings) and we are trying to valve-time it. We knew when we bought it that the timing was out, as it had a dead-spot in reverse which we experienced on the very short length of track we could try it on. The loco was ‘finished’ and almost new – it had some obvious issues (such as the whistle dangling on a piece of pipe in the cab), all of which we have now sorted. We ‘just’ have the timing issue to sort now…



To cut a long (and frustrating) story short, we’re struggling. The loco is very lumpy when running (either on compressed-air or steam and both ‘on-blocks’ and on a track). We seem to be getting admission too early (both before front and rear dead-centres on both sides) and his causes the motion to ‘lock up’ shortly before each dead-centre and then ‘release’, with a jerk, once the pressure in the opposite cylinder overcomes that on the side which is ‘locking’. The loco is designed (according to the drawings) to have a lead of 0.015” (=0.38mm – the ports are ¼” / 6.35mm) and I would guess we are getting admission around 20 degrees (maybe even more) before a dead-centre. From what we can tell, cut-off seems to be reasonably sensible, but we haven’t looked too closely at this aspect yet.



The loco has Walschaerts valve gear and is fitted with inside-admission piston-valves, making it hard to ‘see’ what is happening – no steam-chest cover to remove and peer inside!



We have tried putting the loco’s dimensions, from the drawings, into Allan Wallace’s Walschaerts simulator, but are not sufficiently familiar with either valve gear in general or the simulator’s requirements, to get something we’re convinced is matching reality. We also had a go with the Dockstader simulator, but weren’t sure what dimensions to put into some fields, to match the loco – it has a ton of variables…!



We have set the return crank with a machined jig, the design and setup for which was cribbed from another thread on this forum – the jig fits over the return-crank pin, at a dead-centre, and into the ‘pip’ in the driving-wheel centre, with the return crank being adjusted to ensure the pin fits accurately in the centre of the wheel. We have then set the piston-valve to the best position we can, given the ‘lockup’ problem, i.e. to have equally early admission close to both dead-centres. The motion ‘jerks’ at all four dead-centres (two on each side).



We have set the eccentric rod and the radius rod to match the drawings – the latter required us to make a new part as it was far too short, as built (~2cm!). According to the simulator, this should have helped (and perhaps it did, but not sufficiently). The former was close to being correct, with only a few mm adjustment needed. If we stick with our new radius rod, we will have to modify the piston-valve cylinder end-cap, otherwise the piston-rod end attachment will (just) hit the end-cap (the section in which the packing goes – this is much longer than really necessary and so can be reduced without too much of a problem, although we’ll probably make new ones, if needed.



We have measured (and indeed, re-made) the piston-valve piston. We did this to allow us to use PTFE piston rings, as the original cast-iron rings leaked pretty badly, although it was dimensionally correct, to the drawings). We also checked the piston-valve cylinder dimensions as we had to remove this to replace some gaskets – they were correct to drawings, so far as we could measure.



With the return crank set, we did the forward-reverse level check at the dead-centres and get very little valve movement. We’ve also tried adjusting the return crank to get zero movement. Neither option worked appreciably better than the other. Each time we adjusted anything, we always went back and adjusted the valve spindle to re-centre its travel in the valve.



Having done all of the above (and more – bubbles from draincocks … etc., you name it, we’ve tried it…), we’re now out of ideas. We don’t know if it’s a problem with the design or a problem with it as-built, or both, or just a problem with our attempts at setting it up. This is the first time we’ve attempted to set up Walschaerts and so I hope we’re ‘missing something’…!



It is probably fair to say that we have done the vast majority of our testing on compressed-air, so if this is a steam vs compressed-air issue, then we may have missed something. I know steam expands very differently to compressed air, so this may be a part of the explanation?



Any help/suggestions would be very welcome! I’m happy to post photos, drawings etc. here, as needed.

[borderer]

How are you setting the return crank? It is only set at 90 degrees to the crankpin if the valve gear is an "all-square" layout, which it may or may not be - refer to drawings?

Apologies if you have covered this already.

[matthewmakesmodels]

Thanks for the reply. I'm not sure that we understand when a valve gear is "all square". Does this mean that the cylinder needs to be level (not inclined) or that the centre line of the main piston and the wheels is the same, or both? Our Maggie has neither of these! It has slightly inclined cylinders and these are above the wheel centres.

The jig we used to set the return crank position is described in: modeleng.proboards.com/thread/13604/locomotive-return-crank-setting-walschaerts by John Baguley.

[borderer]

The "all square" layout is when the pivot pin at the base of the expansion link (to which the eccentric rod connects) lies on the centre line of the motion - that is, an extension of the cylinder centre line through the wheel centre. This is when the link is in its mid position - obviously the pin moves on a radius.

In many designs the expansion link is mounted higher than this - hence the eccentric rod will be inclined relative to the motion centre line, and the return crank setting has to take account of this. Obviously the jig can be made to suit this but you need some dimensions first.

Cylinder inclination in itself is not a factor.

Martin Evans gives a brief explanation and one method of setting in Manual Of Model Steam Loco Construction, page 101 - 102 in my edition.

[jma1009]

As Mark has provided us with much important detail he deserves a considered response.

I am not familiar with this design.

I am also reminded that the amount of pre-admission of steam before the dead centre point relies on the angle of advance for pre admission for lead as well as the more common stated amount of lead for Walschaerts valve gear. (The late Don Ashton was very keen on this).

The lead component in Walschaerts is provided by the combination lever proportions and unlike Stephensons valve gear is pretty much a constant.

Don Young advocated large leads for Walschaerts without considering the angle of pre-admission. Don Ashton considered that very little lead was required for Walschaerts. Don Ashton would consider a 4 degree pre-admission passable but no more. I was quite happy with 6 degrees!

A matter of importance is the quality of the steam and whether it is superheated.

One ought to be able to tabulate all the valve events via suction from the drain cocks. Setting a valve gear on air via a compressor will not provide the level of accuracy required.

A simple check of the return crank position and eccentric rod length will deal with whether 'all square' or not and whether the return cranks correctly set and the eccentric rods the correct length.

A 20 degree angle of pre-admission is way way too much and indicates something else is seriously wrong.

Cheers,
Julian

[stevep]

I do not claim to be an expert in valve gear design, but I do understand (to an extent) the mechanics and mathematics of it. So at the risk of being shot down in flames, here is my method of setting up Walshaerts valve gear.

I am assuming that the design is sound, and the parts are all made exactly to dimensions.

The movement of a steam engine valve can be broken down into two separate movements - one in phase (or out-of-phase, depending on whether it's a piston or slide valve) with the main piston, and another which is exactly at 90 degrees to the piston movement. With Walshaerts valve gear, these movements are derived from two completely independent sources, and must be set separately.

The first part is to set the reverser EXACTLY to mid -gear, so that the radius rod (the one that connects to the expansion link at the rear, and the combination lever at the front) doesn't move at all when you rotate the wheels. You then set the valve on its rod so that it opens equal amounts at each end. If everything has been made accurately to size, that's all you can do - there's no other adjustment.

If there are no setting holes for setting a piston valve, I believe that a small amount of air pressure into the cylinders, with rubber pipes from the drain cocks to jar of water can help.

The second part requires you to set the piston to EXACTLY front or rear dead-centre. I have an article, which may have been referenced on this forum at some time, which describes how to do this. Send me a PM with your e-mail address, and I'll send it on to you.

When the piston is set at that position, you should be able to wind the reverser from full forward to full rear gear, and the radius rod (and the valve) should not move. If it does, then either the return crank is set at the wrong position, or the eccentric rod (that connects the return crank to the tail of the expansion link) is the wrong length.

In practice, it is not unusual to make a dummy and adjustable eccentric rod to get this length right before making the final rod - moving the return crank, and adjusting the length until the desired effect is achieved.

Hope that helps.

[matthewmakesmodels]

Hi Both and thank you for your responses. We do have Martin Evan's book and so can try that method. I think we tried this method some time ago, since the book had a bookmark at the right page, however since then we have made a lot of changes (not least making a radius rod of the correct length) and so it is probably worth re-visiting.

Based upon the definition provided, the Maggie design is not 'all square', so our jig may be wrong? We probably should try other methods (e.g. Martin Evans' method) before returning to this (if at all).

Below is a table we put together in an attempt to model the valve gear using Allan Wallace's simulator. The letter designations refer to the 'annotated' diagram which that simulator provides. Some of the parameters (notably those around the lifting arm angle and position) are 'set to work' since the simulator doesn't model the dual-lever arrangement used on a Maggie. This aspect, in particular, may well be wrong, but I think this would only impact the total valve travel and not the point of admission.


Allan Wallace Walschaerts Timing Simulator Parameters_excl_measured-on-loco by Peter Kenington, on Flickr

And here is a copy of the GA drawing. Note that this was pieced together from A4 scans and so dimensions which cross page boundaries can't be relied upon.


Maggie_GA_Drawing by Peter Kenington, on Flickr

[jma1009]

I had a bit of a look at the GA provided today.

I haven't examined yet the Wallace simulator details, and I could not understand what the OP refers to as an extra arm.

My suspicion is that the piston valve bobbins for inside admission have the heads too close together so apparently providing the excessive pre admission quoted by the OP as 20 degrees before dead centre.

I may be entirely wrong on this.

I would posit that if the combination lever (which provides the lap and lead component) do not match the valve and the piston valve liner openings then this would be my initial area of investigation.

'All square' and dare I say 'backset' are minor issues compared to the above.

But I am guessing. The fact that the OP found a pre admission of 20 degrees before dead centre is the thing that strikes me as the issue of concern. Don Ashton would have said 2 degrees and 4 degrees at a stretch for Don Ashton for miniature locos.

[John Baguley]

I've had a quick look at your figures and put the dimensions into the Wallace simulator. There are some dimensions missing so I've had to guess those i.e. length of the lifting arm LM, die block to hanger pin BM and the droplink x dimension DGx (the drop link is forward of the little end of the crankpin). Some of the vertical and horizontal dimensions will be slightly out if you have taken them from the drawing and not taken into account the inclination of the motion.

The results are therefore not very accurate but good enough for a look at the valve events. The eccentric rod length gives an error so that is not quite correct.

The first thing to note is that the lead is considerably more than 0.015" - it's actually 0.0275"(0.7mm). This calculated from the dimensions of the combination lever, the stroke and the lap of the valve. So that will give you earlier admission to begin with.

What position of the reverser are you measuring the point of admission? At full gear?

What you have to remember with Walschaerts gear is that the admission point gets earlier and earlier as you notch the gear up towards mid gear. In other words, the steam is admitted earlier and earlier before top dead centre even though the lead is constant. At 75% cut off the steam may be admitted at say about 5° before dead centre but at mid gear cut off it is admitted much earlier and quite possibly 20° or more before dead centre.

It does sound to me though that the valve gear has excessive lead which is causing the lumpiness. I would suggest checking the dimensions of the combination lever and the piston stroke if you haven't already done so. You have checked the piston valve and the ports so they should be ok.

John

[suctionhose]

It's been a while since I've done Walschearts' but I would question John B's belief that the admission point becomes earlier with notching up. How? The link is central on TDC and lead is fixed by the comb lever?
The point that Don Ashton laboured, if I understood him correctly, was that lead, being constant, became a greater proportion of the valve travel as the travel was shortened by notching up.
As for the lead on this engine under discussion, 25 to 30 thou is too much. Don Young advocated this nonsense and I followed it once. With the benefit of experience I can tell you it results poor starting at the very least.
While not being the pure solution Don A would have liked, lengthening the lap on the valve by 20 thou would be the simplest way to reduce the lead. I.e. make the centre section of the valve bobbin 40 thou shorter (20 thou each side - assuming lead is 27 thou now)

[John Baguley]

No need to take my word for it Ross. Just look at any example on, say, the Wallace simulator and you can see it clearly if you step through the crank angle and see the angle before dead centres at which the port opens for different cut offs. It can be anything up to 30° earlier at mid gear. The amount the port is open at dead centre doesn't change but the point at which the port opens does.

You can see it on the steam diagram as well where the lines for the steam admision point versus percentage of the stroke shows that the opening point advances as the cut off increases. If the port opened at the same time for every position of the reverser then the lines would be vertical at the same percentage of the stroke no matter what the position of the reverser.

It confused me at first when I was timing the piston valves on my Helen Long. I noticed that the valve was opening earlier as the cut off increased. I though it was the lead that was increasing (which it doesn't with Walshaerts) but it was actually the pre-admission that was increasing. I can't remember now whether I discussed this with Don or not.

Just noticed this on Don's website where he discusses lead and pre-admission:

Stephenson’s link motion with open eccentric rods generates increasing lead towards mid gear whilst that of Walschaerts’ gear remains virtually constant, though the preadmission increases towards mid gear in all cases.

That's probably where I learnt about it.

John

[matthewmakesmodels]

Many thanks for all of the replies - very informative. One of the reasons for pursuing this in the way we have is that we want to learn and the discussion has certainly been useful in this regard!

We had assumed, up to now, that the design was 'correct', i.e. so long as it was built to design (which it wasn't, notably in relation to the radius rod), then it should just be a matter of setting the return crank and the valve spindle and all would be fine. We had also assumed that it was built to drawings, initially, and only found that it wasn't when we couldn't get it to work. Sadly correcting the radius rod (which was a long way out) and the eccentric rod (which was only a couple of mm out) didn't help, hence our plea for help.

What I'm picking up from the discussion is that the design is not correct - this is a conclusion we were coming to, but were reluctant to believe. The loco chassis was supposedly built by its designer, with the cab/boiler etc. being added by its previous owner. We have recently seen a video of another example of this design, which was definitely built by the designer, and it clearly sounds 'off-beat'. Again, this added to our emerging realisation that there was/is something not quite right with the design.

We can certainly make new bobbins for the piston valves (after all, we've done it once already, to accommodate PTFE piston rings), although this will take a little while, as we don't have the material to hand. The main question is how to pick the correct dimensions, so we don't end up making half a dozen... Phosphor bronze takes a dent out of the pocket money! For example, do we need to measure accurately the point of pre-admission (probably easiest to do as a proportion of full-stroke) to do this? Our 20 deg was a 'by eye' guess, not a scientific measurement - it was basically a way of saying 'lots'. We had assumed that the design was right and that we were missing something in the many methods we used to try and set it up. This now looks like a wrong assumption. Would an alternative option be to re-design the combination lever (easier and cheaper to do, certainly as an experiment before making a nice-looking version)?

Regarding the lifting arm arrangement, this does not tie up with Wallace's assumed format in his simulator (so far as I could see), hence we were unsure how to include these dimensions. If you look at the GA, the lifting arm (at least as I understand the term) lifts a further arm (not sure of its name) which, in turn, lifts the die block. It is this 'two-lever' arrangement which we referred to in our initial post. Given this differing arrangement, what parameters would be useful to measure from the drawings? I'm happy to do this, if it will help with diagnosis. The parameters we assumed for these parts of the design ,when we tried to use the simulator, were based purely on making the simulated arrangement 'look right' when going from full-forward to full-reverse. We figured that if we had inadvertently gone slightly beyond 'full forward' or, the opposite, were notched-back a little at 'full forward', then neither of these would significantly impact pre-admission, but would just cause slightly too much (or too little) valve travel. It looks like we were wrong about that, though!

In regard to our testing, we have done this mostly in full forward or full reverse, but notching back (which we tried a few times) didn't seem to make much difference to our pre-admission, although we were only interested in whether or not this produced an improvement and didn't pay close attention as to whether it made things worse!

Thanks for the interesting discussion.

Matthew and Peter

[uuu]

One of the difficulties I find with these simulators is the interpretation of the results. So you can look at a graph for your own loco, then tweak one of the components and get a different graph. But is it better or worse? I've no idea what I'm looking at. What does an ideal loco look like?

Wilf

[jma1009]

The last edition of Don Ashton's book is available from Camden and although not an easy read does provide a fundamental guide to Walschaerts and Stephensons valve gear.

The analysis of the combination lever with the maths includes a sort of criticism of Martin Evans' book that incorrectly assumed that full port opening was required for steam inlet, which it patently does not require.

How to interpret matters.

I can see where Wilf is coming from, but I don't think it is that difficult.

Matthew is getting tied up in knots over the lifting arrangements for the radius rod - which can make a difference - but it is an inconsequential possible error that is minor compared to the apparent huge pre admission before dead centres.

[John Baguley]

To be honest, I'm not sure what the 'ideal' graph should look like either! I just fiddle around with the variables until I get the valve events at each end of the cylinder as identical as possible i.e. the lines showing the opening and closing of the valve for inlet and exhaust are as close together as possible.

For example, here's the Events Diagram for the 2½" Gauge Canadian Switcher in forward gear after I had redesigned the valve gear to give better valve events:

You can see that the pairs of lines representing the inlet and exhaust for each end of the cylinder are almost identical. Whether the lines are the right shape I don't really know. Reverse gear is not quite so good but how often do you run a loco in reverse?

I actually used Don's spreadsheet to give the basic valve gear dimensions and then tweaked the figures to get them spot on.

John

[ettingtonliam]

When I was redesigning 'Locomotion's valve gear which is weird, and probably not suitable for any of the analysis programmes, I reverted to drawing it out twice full size, and plotting the valve movement as the wheel revolved, which was quite informative.

[springcrocus]

Jul 14, 2022 22:58:00 GMT jma1009 said:

How to interpret matters.

I can see where Wilf is coming from, but I don't think it is that difficult.

Come off it, Julian, that sort of throwaway remark helps no-one! That's almost the same as saying "I know how to read the charts but can't be bothered to explain it to you"

John Baguley's post that followed yours is much more informative. Why don't you try explaining things occasionally, instead of throwing Don Ashtons book in the ring at every opportunity. You, yourself, admit that it's a difficult read and not everyone is a member of MENSA.

Regards, Steve

[John Baguley]

Jul 15, 2022 5:59:13 GMT ettingtonliam said:

When I was redesigning 'Locomotion's valve gear which is weird, and probably not suitable for any of the analysis programmes, I reverted to drawing it out twice full size, and plotting the valve movement as the wheel revolved, which was quite informative.

That's one of the advantages of using 3D CAD. You can 'build' the valve gear on the computer model, revolve the wheels and see exactly what the valve is doing at any time. I appreciate that CAD is not everyone's cup of tea though.

John

[suctionhose]

As a response to John B's earlier remarks on pre-admission, I read DA's Walschaerts' chapters today and concede the pre-admission angle seems to advance with notching up. It must be a result of the summation of the two harmonic motions - one constant, one variable - that is behind it. Interesting!

Actually, you jogged my memory back to 1983ish with my first and only Walschaerts design from scratch per Henry Greenly and observing the two motions compounding the behaviour of the valve during the 'open' part of the cycle. I never realised the effect on pre-admission.

These last 20 years I've built two traction engines and an Allan gear loco none of which conform to the preset configurations of the Wallace Simulator.Suffice to say, I've not used Don's spread sheets or Wallace for any real purpose. Graphical solutions have new purpose thanks to the precision of CAD drawing!

Anyway, your revamped Switcher must be quite something! Those lines are so close!

Regards,
Ross

[jma1009]

Jul 15, 2022 6:45:30 GMT springcrocus said:

Jul 14, 2022 22:58:00 GMT jma1009 said:

How to interpret matters.

I can see where Wilf is coming from, but I don't think it is that difficult.

Come off it, Julian, that sort of throwaway remark helps no-one! That's almost the same as saying "I know how to read the charts but can't be bothered to explain it to you"

John Baguley's post that followed yours is much more informative. Why don't you try explaining things occasionally, instead of throwing Don Ashtons book in the ring at every opportunity. You, yourself, admit that it's a difficult read and not everyone is a member of MENSA.

Regards, Steve

As Steve dislikes his own threads being hi-jacked, and as Matthew has not requested what Steve suggests I attempt (which will be a very lengthy post that will hi-jack this thread), perhaps we ought to start a new thread specifically on the points raised by Steve and Wilf.

I would however remind forum members interested in valve gears that Don Ashton's invaluable website has been maintained since his death in 2020, and does provide the 'what to aim for'. Don also wrote a series of articles in EIM.

For a 'working' range of cut offs, you need to try and get equality of events for the front and rear of the cylinder events, be able to make the parts accurately, and then set the valve gear accurately.