Back EMF and Gradients

[Matt]

I've been a DCC user for a decade or so and have generally installed decoders with back EMF without any issue.  As i am now starting to get the upper level of my layout operational i am starting to run trains down a helix between levels which has an approx gradient of 1 in 40.  In early testing i have noticed that trains running say 6 coaches behind a locomotive on the descent seem to "hunt" continually coming down the gradient - looks as if the weight of the coaches starts to push the loco which then cuts power and brakes the train before then speeding up again.  Apart from seeming inelegant, after a few runs down the helix one of my Bachmann class 37s started to sound like a bag of nails.  Have i missed something with back EMF and gradients - i.e. should i be isolating the back EMF functionality and allowing the train to gain a bit of speed on the descent?  I've never heard any discussion on this before but would appreciate any thoughts on this.

 

Thanks,

Matt

[Alan Kettlewell]

Hi Matt.  It's worth checking the documentation for the chip as nowadays back EMF can be adjusted via CV settings.  Sounds like it may not be calibrated quite right and may need a tinker.

 

Cheers... Alan

 

 

 

[Mike Buckner]

What decoders are you using?

[Grovenor]

IMHO this has more to do with the drive train than the decoders, it happens on DC and on DCC without feedback, probably even more so..

It is generally a feature of worm drive. When the train is trying to run faster than the motor setting then the gears push the wormshaft up against the end thrust bearings, these are commonly plain bearings and the thrust creates extra friction that has a braking effect. Good back emf should be able to compensate for this increased frictionbut will need very careful setup, usually they are not optimised for downhill running.

As the braking effect slows down the train the pressure on the thrust bearing is reduced and the motor will speed up again giving the jerky motion. Nany motors have back bearings that deal with the thrust much better than the front bearings so the effect will be worse one way than the other. The cure is to improve the bearings, sometimes an additional front bearing can be added ahead of the worm. On many modern locos with gear towers washers can be added either end of the worm to reduce friction and reduce clearance for the worm to move fore and aft.

The old Hornby Dublo vertical motors were well designed in this regard with ball bearings at both ends of the motor shaft and a screw fitting for the top bearing to give precise adjustment.

[WIMorrison]

The issue is with the design of  worm gear as you cannot normally drive the worm gear from the pinion, only from its shaft (there are special occasions when you can). When the loco is going down hill and wants to go faster than the motor is driving it then the friction of the worm gear being driven by the pinion will momentarily brake the loco until the motor shaft has turned sufficiently to free the lock up, which will then allow the loco to move until the next lock up and the cycle repeats.

 

A well designed worm gear will minimise this and in rare cases where the angle of the worm is sufficiently steep you can actually drive the worm from the pinion. This was the case in the old Hornby motors as the worm was very small with a steep worm onto a large pinion. Modern worms are large with shallow angles and you cannot turn the worm from the pinion - try turning the wheels and see how they lock up, this is what s happening as you loco goes down the gradient being pushed by the carriages. A multi-start worm generally has a similarly steep worm angle allowing the worm to be driven by the pinion.

 

The solution is to reduce the gradient, the weight of the carriages or put brakes on the carriages.

[RAF96]

See also ‘Cogging’

https://www.precisionmicrodrives.com/content/cogging-torque-in-permanent-magnet-motors/

A bit like backlash in a gearbox but electro-magnetic.

[34theletterbetweenB&D]

For all the mechanical reasons already mentioned, this actually doesn't need any BEMF action to occur. On a I in 40 there is every prospect of the vehicles going faster than the motor can turn to keep pace, even at whatever is full power.

 

There's a potential simple solution worth trying that is very 'traditional railway'. Creep down gradient at dead slow. A little experiment will determine the maximum permissible speed before the hunting effect kicks in. Acceleration to line speed can be phased in very near the bottom.

[Matt]

All - thanks for the feedback.  Sounds like i was barking up the wrong tree!  I cannot really reduce the gradient of the helix but will try some of the other ideas such as controlling speed on gradient to a creep and see if helps.  M

[Matt]

On 15/08/2020 at 20:48, Mike Buckner said:

What decoders are you using?

 

A wide variety TBH - i would need to check for loco i was testing with.  I use Gaugemaster, Hattons, Bachmann, Lais and probably others at various times.  This case was probably Lais as have tried them recently.

M

[Alan Kettlewell]

Hi Matt,

 

I have several helices on my H0 layout, some at less than 1 in 40 gradient and I haven't found the same issues you refer to.  For interest here's a clip of the layout in its early days before any scenery was added  when  I was testing a few trains on the spirals.  There are several other videos on my YouTube page made public if you wish to view more.   Most of these clips will feature long trans, some with 8 coaches, running up and down spirals without a problem.  These are continental trains from top manufacturers such as Roco, Fleischmann, Lilliput etc and are generally fitted with  good spec chips eg Lenz, Zimo, Roco, ESU.

 

I still maintain it may be worth looking at the CV settings for back-EMF adjustment to see if a tweak or two can resolve the problem.  I'm not saying it will work but just worth a try.  Good luck with it.

 

Cheers ... Alan      

 

 

[Alan Kettlewell]

Another ..

 

 

 

[Alan Kettlewell]

and another ...

 

 

[Crosland]

2 hours ago, Alan Kettlewell said:

I still maintain it may be worth looking at the CV settings for back-EMF adjustment to see if a tweak or two can resolve the problem.  I'm not saying it will work but just worth a try.  Good luck with it.

 

Any motor control system using feedback requires tuning for the best results, especially with the wide variety of decoder, motors and drive trains. Unfortunately it's not a simple process. I think the Zimo manuals give the best description of BEMF tuning  but it cannot neccessarily be applied directly to other makes.

 

To prove the worm drive theory, disable BEMF completely in the decoder and run the train again.