Maximising Adhesion using Electronics

andrew jukes
Posts: 212
Joined: Thu Mar 05, 2009 4:15 pm

Maximising Adhesion using Electronics

Postby andrew jukes » Tue Apr 18, 2017 2:17 pm

A possibly new thought is the use of clever control electronics to maximise the use of available wheel/rail adhesion. I'm not aware of any attempts to do this (perhaps I should check on MERG).

Most clever electronics has focussed on running motors as slowly as possible without stalling. A maximum traction design would be doing almost the opposite, trying to increase the motor's output but only until there is a sudden acceleration in rotational speed. The aim is to keep the drivers on the point of slipping, increasing the torque until that point had just been exceeded, backing off a little and then increasing torque again and repeating the cycle. With a control system like this you should be able to pull the heaviest trains that available adhesion would allow and (rather boringly) open the throttle wide with the train at a standstill and see it accelerate away without drama.

It's possible that such a device would best be loco mounted, rather like a DCC chip. And being ignorant of what current DCC systems include, I may be simply describing something that is already standard DCC practice.

Regards Andrew

Crepello
Posts: 98
Joined: Thu Jan 31, 2013 11:32 am

Re: Maximising Adhesion using Electronics

Postby Crepello » Tue Apr 18, 2017 4:20 pm

With multiple motors in a loco, possibly one per axle, it might be feasible to compare the voltage and current draw for each, and rein-in whichever is accelerating due to loss of adhesion.

junctionmad
Posts: 247
Joined: Tue Dec 08, 2015 3:38 pm

Re: Maximising Adhesion using Electronics

Postby junctionmad » Tue Apr 18, 2017 10:19 pm

andrew jukes wrote:A possibly new thought is the use of clever control electronics to maximise the use of available wheel/rail adhesion. I'm not aware of any attempts to do this (perhaps I should check on MERG).

Most clever electronics has focussed on running motors as slowly as possible without stalling. A maximum traction design would be doing almost the opposite, trying to increase the motor's output but only until there is a sudden acceleration in rotational speed. The aim is to keep the drivers on the point of slipping, increasing the torque until that point had just been exceeded, backing off a little and then increasing torque again and repeating the cycle. With a control system like this you should be able to pull the heaviest trains that available adhesion would allow and (rather boringly) open the throttle wide with the train at a standstill and see it accelerate away without drama.

It's possible that such a device would best be loco mounted, rather like a DCC chip. And being ignorant of what current DCC systems include, I may be simply describing something that is already standard DCC practice.

Regards Andrew



I suspect the gain in traction would be minimal compared to the complexity of having to derive a reference signal from an undriven wheel

andrew jukes
Posts: 212
Joined: Thu Mar 05, 2009 4:15 pm

Re: Maximising Adhesion using Electronics

Postby andrew jukes » Tue Apr 18, 2017 10:43 pm

No, I was thinking of a typical 6-coupled chassis with a single motor. In concept, what would be needed is a rotational speed signal from the motor shaft, perhaps a toothed wheel with a device counting the passing frequency of the teeth. As long as the loco was accelerating without slipping, the passing frequency would increase smoothly. When slip starts, there would be a discontinuity, the electronics would dump some volts (by experiment, just enough to quench the slip) and then almost immediately start adding volts again. More experiment would be needed to establish how quickly to add volts and probably some clever algorithm would be needed so that a reduced rate of voltage increase would apply if slip was detected at too close intervals.

None of that sounds beyond current technology and I'm wondering if anyone has tried something similar.

Regards Andrew

User avatar
Le Corbusier
Posts: 1443
Joined: Wed Feb 11, 2015 3:39 pm

Re: Maximising Adhesion using Electronics

Postby Le Corbusier » Wed Apr 19, 2017 6:06 am

If I understand correctly, you mean treating the wheels individually as far as detecting slippage is concerned but collectively as far as reaction is concerned ... as soon as any slippage is detected the power is cut across all wheels and then reapplied. I am no mathematician, but wonder if this might not actually reduce traction as it would set the operation of the chassis to the wheel with the lightest tractive effort ?

Tim
Tim Lee

User avatar
Jol Wilkinson
Posts: 804
Joined: Mon Jul 21, 2008 7:39 pm

Re: Maximising Adhesion using Electronics

Postby Jol Wilkinson » Wed Apr 19, 2017 7:01 am

As - presumably - the wheels/axles are all mechanically linked, then measuring the motor voltage and current draw would surely be the simplest way. If adhesion is lost, then slip will occur and, if my recollection of my engineering training is correct, the current drawn will fall as the power drawn by the motor falls.

I expect that could be sensed by an on board electronic device, but it would take someone here with more electronics expertise then me to say whether it is then possible to use that data to control the supply voltage to reduce the motor speed,while at the same time leaving overall control still at the main controller. What about the interaction between the motor and controller? Will the controller react in response to the current draw/motor BEMF?

It sounds like the sort of thing DCC might be able to provide. However, it it was, why hasn't anyone done it yet? For the modeller on the Clapham omnibus, DCC is often seen as the answer to loco running problems. AS RTR loco performance is usually measured by how many coaches/wagons it can pull, then this would seem an obvious fix if it were practical.

I rather think this is a potentially complex or expensive solution looking for a problem. Ultimately the gains might be minimal. and could be achieved with other, simpler means.

User avatar
Tim V
Posts: 2343
Joined: Tue Jul 29, 2008 4:40 pm

Re: Maximising Adhesion using Electronics

Postby Tim V » Wed Apr 19, 2017 9:28 am

Maximising adhesion is more about weight distribution - usually via compensation/springing which, perhaps, is where we usually look.

DCC can bring smooth acceleration, in a way that DC cannot hope to match.
Tim V
Scalefour News Editor

andrew jukes
Posts: 212
Joined: Thu Mar 05, 2009 4:15 pm

Re: Maximising Adhesion using Electronics

Postby andrew jukes » Wed Apr 19, 2017 11:37 am

What we’re mostly talking about is taking advantage of the fact(?) that static friction is greater than sliding friction. Equalising axleloads (after allowing for weight transfer effects, which are often ignored) and making sure the suspension design keeps axleloads equal is a logical starting point in maximising drawbar pull. But when you’ve done everything you can in that direction, you still have the problem of keeping motor torque at the maximum level possible without the drivers collectively ‘letting go’.

The sort of system I described is aimed at keeping the drivers as close as possible to that slip/slide boundary for as much of the time as possible when trying to accelerate a heavy train or climb a serious gradient. So not a substitute for all the good things you can think of that aid adhesion but a (possibly?) interesting add-on.

I suggested an approach using motor speed-sensing directly as I’m pretty sure any system based on detecting current or back emf changes would be far to slow in response. I picture a 30-tooth wheel on the motor shaft with monitoring of the tooth passing frequency. If the start of slip could be detected by 3 teeth passing when there should only have been one, on a loco with 30:1 gearing, the drivers would have only rotated 1/300th of a turn before the clever electronics would start acting to quench the slip.

I'm certainly willing to believe that some of the massive investment that’s gone into DCC could have included traction control at the slip/slide boundary - but I don’t know much about DCC and perhaps someone can point us to where we might find such a feature? As to generalisations about DCC v DC, it would help to know what’s being compared with what. Looked at from a DC motor’s point of view, a smoothly rising DC voltage would seem the ideal if you want a DC motor to accelerate smoothly. The things that may make DCC ‘better’ for a model loco are likely to be remote from the motor’s terminals.

Regards Andrew

User avatar
Guy Rixon
Posts: 637
Joined: Sun Mar 27, 2011 6:40 pm

Re: Maximising Adhesion using Electronics

Postby Guy Rixon » Wed Apr 19, 2017 4:15 pm

andrew jukes wrote:I picture a 30-tooth wheel on the motor shaft with monitoring of the tooth passing frequency. If the start of slip could be detected by 3 teeth passing when there should only have been one, on a loco with 30:1 gearing, the drivers would have only rotated 1/300th of a turn before the clever electronics would start acting to quench the slip.


How would you detect the passing of teeth? You need some kind of encoder, and I'm not seeing any really small ones when googling. Most makers would consider a 25mm-diameter encoder to be super-compact.

It might be possible to improvise a re-counting device with a permanent magnet on a disc driving a fixed coil, such that you count either whole revs or perhaps quarter revs (4 magnets). With 30:1 gearing that would give slip detection after ~ 1/22 of a wheel revolution. Getting it to work reliably might be hard :-/

One might also build a home-made, really-coarse, optical encoder. That might be easier than the thing with the magnets.

User avatar
Will L
Posts: 1702
Joined: Sun Jul 20, 2008 3:54 pm

Re: Maximising Adhesion using Electronics

Postby Will L » Wed Apr 19, 2017 5:11 pm

But isn't a back EMF based feedback controller (DC or some DCC) effectively controlling how fast the motor is turning?

andrew jukes
Posts: 212
Joined: Thu Mar 05, 2009 4:15 pm

Re: Maximising Adhesion using Electronics

Postby andrew jukes » Wed Apr 19, 2017 6:26 pm

For what I'm suggesting, it's not motor speed that needs to be measured nor the rate of change of motor speed - it's abrupt changes in either motor speed or the rate of change of motor speed that's required.

I now picture a thin slotted disc on the motor shaft as likely to be best for detection with components of the size of those inside an optoisolator mounted on each side to count the passing slots. I don't know whether response times would be fast enough but optoisolator performance should be a good guide.

Regards Andrew

nigelcliffe
Posts: 526
Joined: Mon Jul 21, 2008 8:31 am

Re: Maximising Adhesion using Electronics

Postby nigelcliffe » Thu Apr 20, 2017 7:43 am

Will L wrote:But isn't a back EMF based feedback controller (DC or some DCC) effectively controlling how fast the motor is turning?


Yes, it is. It will attempt to keep the motor speed constant regardless of load. But, if the rate of change of the motor speed is too rapid, the loco will appear to "kangaroo" or "buck" as it changes speed too rapidly - this can be seen with high BEMF settings and consisted (double headed) locos on undulating ground. So, Andrew's hypothetical sensing system needs to overcome this matter.

I'd fix the problem with careful weight distribution in the loco.


- Nigel

JFS
Posts: 709
Joined: Wed Dec 16, 2009 3:47 pm

Re: Maximising Adhesion using Electronics

Postby JFS » Thu Apr 20, 2017 11:30 am

andrew jukes wrote:For what I'm suggesting, it's not motor speed that needs to be measured nor the rate of change of motor speed - it's abrupt changes in either motor speed or the rate of change of motor speed that's required.




Well, I would never be one to dismiss an idea out of hand, *but* ... to me, this is not how model locos slip at all. If we think about a real loco - say a GWR King with known "good" adhesion characteristics - when the driver opens the throttle in somewhere close to full gear, he is unleashing potentially a couple of thousand horsepower, but - initially - the wheels do not turn because of the friction with the rail and the resistance of the train. If that friction is lost, the power is released to revolve the wheels at uncontrolled speed until the driver is able to shut the throttle and stop them.

For a demonstration, watch this - https://www.youtube.com/watch?v=P15HEIVCUVA at about 1:05, the driver takes his hand off the sander lever to notch up...

The model situation is completely different. Because of the low inertia but high friction, to move our train from rest - even a heavy one - requires a tiny amount of power compared with that needed to hurl it round at 90mph, and if we replicate the prototype situation of opening full throttle it will - often - just race away at full tilt. thus, even when trying to move off a heavy train, we apply just enough power to turn the wheels and no more. (this is even more true of "programmed" acceleration as in DCC [and - Tim - many DC] systems). To pose that contrast in a different way - who has ever driven a model train by opening up to full power and allowing the train to gather speed in its own good time? Yet that is exactly how the prototype is driven.

So it seems to me that the situation of "uncontrolled run-away" which characterises the real thing never arises on models - unless you can demonstrate otherwise Andrew?

Hence - as others have said, as model loco builders, we must use our skill and ingenuity to design locos which are up to the job.

A year or two back a friend and I visited the home layout of a well known P4 modeller not too far from Manchester. It is a large main line continuous run with with long trains of very heavily weighted stock. The first train was a "Jubilee" hauling twelve or thirteen, and I was struck by a kind of continuous and quite loud grinding noise. I feared a disaster in the making until I realised that the noise was caused by a continual and very slight slipping. But notice that the loco had no trouble moving the train from rest - only when the frictional load increased with speed. Having felt the weight of the rolling stock, I was convinced that I was talking to someone who REALLY knew how to make a loco grip the rails - and was prepared to work it to the very limit!

Finally, I agree with Will - Back EMF controllers do indeed cut the power when the revs rise and they do so very well - and they prove that such an approach does not prevent slipping in the model context.

Best wishes,

User avatar
Tim V
Posts: 2343
Joined: Tue Jul 29, 2008 4:40 pm

Re: Maximising Adhesion using Electronics

Postby Tim V » Thu Apr 20, 2017 1:07 pm

JFS wrote:To pose that contrast in a different way - who has ever driven a model train by opening up to full power and allowing the train to gather speed in its own good time? Yet that is exactly how the prototype is driven.


That is exactly how trains can be driven on Clutton :D

Using DCC, by setting the acceleration rate, by setting the top speed....
Tim V
Scalefour News Editor

andrew jukes
Posts: 212
Joined: Thu Mar 05, 2009 4:15 pm

Re: Maximising Adhesion using Electronics

Postby andrew jukes » Thu Apr 20, 2017 2:42 pm

I think we all tend to look at things from the perspective of our own modelling interests.

Mine are how to operate trains weighing up to 3Kg on a main line with 1 in 100 gradients. A particularly demanding requirement is how to accelerate a train from standstill in the storage sidings to a scale 80mph as it emerges from Welwyn South tunnel with approx. 5metres of track (including a rising 180º curve) to do it in. For that particular feat, setting the throttle to the right setting for 80mph and letting slip control cope as best it can seems like a good approach. Elsewhere, just having a loco that is more sure-footed seems quite desirable.

I’m afraid I don’t recognise Howard’s description of what’s going on when a train gets underway. When the throttle is opened, the wheels do turn, hopefully starting to move the train and not spinning. So long as the friction at the rail is sufficient to cope with the resistance of the train, the train will accelerate. If this ceases to apply - for example because the gradient increases or rail conditions deteriorate - the coupled wheels will 'let go’ and start slipping. The point at which a slip starts should be identifiable, both on the prototype and on a model.

As for Kings having ‘good’ adhesion characteristics, that’s rather like me filling my locos with tungsten and being pleased at how well they pull the train!

I’m currently pursuing my idea on the MERG forum and learning of some interesting possibilities.

Regards Andrew

User avatar
Will L
Posts: 1702
Joined: Sun Jul 20, 2008 3:54 pm

Re: Maximising Adhesion using Electronics

Postby Will L » Thu Apr 20, 2017 3:01 pm

JFS wrote:...A year or two back a friend and I visited the home layout of a well known P4 modeller not too far from Manchester. It is a large main line continuous run with with long trains of very heavily weighted stock. The first train was a "Jubilee" hauling twelve or thirteen, and I was struck by a kind of continuous and quite loud grinding noise. I feared a disaster in the making until I realised that the noise was caused by a continual and very slight slipping. But notice that the loco had no trouble moving the train from rest - only when the frictional load increased with speed.


And If I've identified this layout aright, and I'm pretty sure i have, the 1:300 gradients to be found on this line will also have had a part to play..
Having felt the weight of the rolling stock, I was convinced that I was talking to someone who REALLY knew how to make a loco grip the rails - and was prepared to work it to the very limit!

While I'm still tying to persuade him of the virtues of CSBs I would be the first to agree that his chassis are well engineered well though through, significantly heavy and not considered satisfactory until they will do at least as well as you observed.

Finally, I agree with Will - Back EMF controllers do indeed cut the power when the revs rise and they do so very well - and they prove that such an approach does not prevent slipping in the model context.


What I have discovered when trundling my own locos plus big loads round this layout and Tony M's Ambergate plus, both of which present significant haulage challenges, is that the difference between running up hill slip free (and able to accelerate to the controller setting), reliably making the grade but with some constant slipping apparent and reduced speed, and just slipping to a stand, can be as little as a single additional vehicle, certainly so when it those 200gram coaches the Jubilee was in charge of.

User avatar
Will L
Posts: 1702
Joined: Sun Jul 20, 2008 3:54 pm

Re: Maximising Adhesion using Electronics

Postby Will L » Thu Apr 20, 2017 3:16 pm

nigelcliffe wrote:Yes, it is. It will attempt to keep the motor speed constant regardless of load. But, if the rate of change of the motor speed is too rapid, the loco will appear to "kangaroo" or "buck" as it changes speed too rapidly - this can be seen with high BEMF settings and consisted (double headed) locos on undulating ground. So, Andrew's hypothetical sensing system needs to overcome this matter.


After 15 yaers of exhibition running made much smoother by the use of Compspeed controllers, I can't say I ever observed single loco's doing this, and given the sampling rate is quite high, I wouldn't have though the difference in motor speed between samples would have been enough to cause such behaviour. I do however agree that trying to run double headers smoothly required well matched loco's.

John Palmer
Posts: 604
Joined: Fri Jul 15, 2011 11:09 pm

Re: Maximising Adhesion using Electronics

Postby John Palmer » Thu Apr 20, 2017 7:05 pm

JFS wrote:To pose that contrast in a different way - who has ever driven a model train by opening up to full power and allowing the train to gather speed in its own good time? Yet that is exactly how the prototype is driven.

andrew jukes wrote:I’m afraid I don’t recognise Howard’s description of what’s going on when a train gets underway.
That was also my reaction, and I have a strong suspicion that Howard is not in the habit of getting his plougher under way by pushing the throttle hard against the stop.

On the Midland/LMS compounds, immediate full opening of the regulator when starting initiated full compound working, with the attendant risk of stalling your train. To avoid this, you needed to get away on first valve. Similarly, in the video to which Howard posted a link, it's noteworthy that the King's regulator is never seen more than half way across the quadrant. You might well want to run with a well opened throttle and vary power output on the cut-off, but 'getting your shoulder under it' from the outset is liable to be counter productive!

I can see the merit in Andrew's proposal to meet the particular operational requirement he describes, and I guess that there may be plenty of layouts having a similar requirement upon exit from the storage roads. However, the method by which motion is imparted to a 4mm model of a steam locomotive (gear trains of varying complexity) is so different from that of the full sized article that I doubt I will ever see a 4mm model move with the same fluidity as the prototype it seeks to represent. For the same reason I am pessimistic that the characteristics of the prototype's controls will ever be reproduced realistically in model form, whether by electronics or otherwise.

JFS
Posts: 709
Joined: Wed Dec 16, 2009 3:47 pm

Re: Maximising Adhesion using Electronics

Postby JFS » Fri Apr 21, 2017 8:15 am

Will L wrote:
nigelcliffe wrote:Yes, it is. It will attempt to keep the motor speed constant regardless of load. But, if the rate of change of the motor speed is too rapid, the loco will appear to "kangaroo" or "buck" as it changes speed too rapidly - this can be seen with high BEMF settings and consisted (double headed) locos on undulating ground. So, Andrew's hypothetical sensing system needs to overcome this matter.


After 15 years of exhibition running made much smoother by the use of Compspeed controllers, I can't say I ever observed single loco's doing this, and given the sampling rate is quite high, I wouldn't have though the difference in motor speed between samples would have been enough to cause such behaviour. I do however agree that trying to run double headers smoothly required well matched loco's.


It does happen - I run multiple units in multiple - ie you have two units coupled together each with its own power unit. In this case, there is an observable phenomenon that the motor under load slows down and the one off-load speeds up - because the high BMEF from the one is masking the low BEMF from the other, thus, The back unit speeds up until the buffers touch, the front one gets a nudge and it then speeds up until the couplings tighten, which gives the back one a tug, which then speeds up.... It only happens at very low speeds but is a pain, and unavoidable.

I also have a problem with a couple of my steam locos, whereby when working them off my hand held Pentrollers, they can get into a surging mode when run full tilt. Fortunately, that never happens in regular operation.

Before Tim or others start spouting about how DCC would overcome all of this, I would say that I have four different MU type which are all inter-couplable so to speak, that is 16 different "consists" to programme in and every one would be needed during a timetable sequence - let alone coping with the variation between the individual units in each type.

But going back to Andrew's original post, I am not sure what stage your layout is now up to Andrew, but for what you want, I really do feel that DCC with feedback chips WOULD be exactly what you need - even given that you already have non-DCC locos. Have you carried out any trials yet? I would also say that DCC has developed some pretty sophisticated solutions and this development is likely to continue at an ever increasing pace. By contrast, the chances of anyone developing a NEW solution such as you describe AND maintaining it into the future (how many "great" solutions have come and gone in the control field?) is not great I feel.

For my part, I only rejected DCC because of the above challenge - but it would solve another issue - MAKING a loco slip!! As a Southern modeller, I will indulge myself with a Bulleid Pacific. SO what I need is a DCC "consist" controlling a tender drive doing all the work, with the drivers driven loaded only just enough to keep them on the road. Thus, I can control the train speed with the tender drive, whilst slippiong the drivers as much as the real thing!

Best Wishes,

billbedford
Posts: 701
Joined: Tue Oct 02, 2012 2:40 pm

Re: Maximising Adhesion using Electronics

Postby billbedford » Fri Apr 21, 2017 8:23 am

andrew jukes wrote:Mine are how to operate trains weighing up to 3Kg on a main line with 1 in 100 gradients. A particularly demanding requirement is how to accelerate a train from standstill in the storage sidings to a scale 80mph as it emerges from Welwyn South tunnel with approx. 5metres of track (including a rising 180º curve) to do it in. For that particular feat, setting the throttle to the right setting for 80mph and letting slip control cope as best it can seems like a good approach. Elsewhere, just having a loco that is more sure-footed seems quite desirable.


Powered bogies on the rolling stock?
Bill Bedford
Mousa Models
http://www.mousa.biz

JFS
Posts: 709
Joined: Wed Dec 16, 2009 3:47 pm

Re: Maximising Adhesion using Electronics

Postby JFS » Fri Apr 21, 2017 9:37 am

Tim V wrote:
That is exactly how trains can be driven on Clutton :D

Using DCC, by setting the acceleration rate, by setting the top speed....


Indeed so, and of course that is the clever electronics doing the controlling of the power to replicate (and improve on) what we used to do with our knobs. (if you catch my drift). We also work that way using the panel-mount Pentrollers on Minories - especially in the situation where three of us are trying to drive four trains - but here the electronics only comprise a capacitor / resistance network :D :D

That said, setting all the theory aside and speaking as one who looked at but rejected DCC *for my particular situation*, I really do feel that it is exactly what Andrew is looking for in his case. Having played with a couple of very large P4 layouts which are DCC, there is no way I would consider DC for a that kind of situation - despite the challenges which still remain - such as the complexities of the different controller concepts, instructions written in Anglo-Sino-German, ease of making operator cock-ups under pressure etc.

Best wishes,

nigelcliffe
Posts: 526
Joined: Mon Jul 21, 2008 8:31 am

Re: Maximising Adhesion using Electronics

Postby nigelcliffe » Fri Apr 21, 2017 11:36 am

JFS wrote:...........
It does happen - I run multiple units in multiple - ie you have two units coupled together each with its own power unit. In this case, there is an observable phenomenon that the motor under load slows down and the one off-load speeds up - because the high BMEF from the one is masking the low BEMF from the other, thus, The back unit speeds up until the buffers touch, the front one gets a nudge and it then speeds up until the couplings tighten, which gives the back one a tug, which then speeds up.... It only happens at very low speeds but is a pain, and unavoidable.
...........

Before Tim or others start spouting about how DCC would overcome all of this, I would say that I have four different MU type which are all inter-couplable so to speak, that is 16 different "consists" to programme in and every one would be needed during a timetable sequence - let alone coping with the variation between the individual units in each type.



It should be soluble in DCC, with a few one-off settings in any half-decent DCC chip. You're looking for the CV to reduce the BEMF influence. Will need a little trial and error to come up with the value for each unit, but once set that will be it, and they will all consist together nicely.


- Nigel


Return to “Control Concepts”

Who is online

Users browsing this forum: No registered users and 1 guest