ESC with feedback control??

[ge_rik]

Just been test running my new battery powered loco with a Mac Five controller. It's the first time I've experienced full radio control other than for test-running. My railway has 1:40 gradients and, having been used to DCC, the first thing I noticed is that the loco with its train in tow runs away downhill and struggles uphill unless I constantly keep adjusting the joystick. I'm assuming that it must be possible for an ESC to use feedback to monitor the power needs of the motor as DCC decoders do - does anyone know of an ESC which incorporates feedback control? I'm assuming it must be possible. I'd like to be able to set a train running and then switch off the transmitter.

Rik

 

[ge_rik]

Does the lack of response mean there are no ESCs with motor load feedback out there? I've tried doing a search and found a bit of discussion on some forums (usually r/c car forums) but not come across anything tangible.

Is it something you could incorporate into the next generation of your ESCs, Tony, (Walsham)?

Rik

 

[Tony Walsham]

Hi RIK.
The short answer is no, back EMF detection cannot be incorporated in my board design.
I am still using the same LMD18200T solid state motor driver I have been using for years. It does not have back EMF detection as it has never been needed. I know of hundreds of users that switch on and let it run unattended without noticeable speed differential up or down grade.
At least some of them must have had the 1 in 40 grades you mention.
The exception is of course for locos that have free wheeling gearboxes such as the Bachmann K27 etc. That loco has a 14:1 gearbox with a single start worm. Normal worm gear drives that have a 30:1 final ratio should not run away downhill.

 

[ge_rik]

Hi RIK.
The short answer is no, back EMF detection cannot be incorporated in my board design.
I am still using the same LMD18200T solid state motor driver I have been using for years. It does not have back EMF detection as it has never been needed. I know of hundreds of users that switch on and let it run unattended without noticeable speed differential up or down grade.
At least some of them must have had the 1 in 40 grades you mention.
The exception is of course for locos that have free wheeling gearboxes such as the Bachmann K27 etc. That loco has a 14:1 gearbox with a single start worm. Normal worm gear drives that have a 30:1 final ratio should not run away downhill.

Hi Tony
Thanks for getting back to me. Maybe it's something which I've noticed now because I've newly come to battery power from DCC. No doubt I'll develop a strategy for minimising it but at the moment I seem to have to have the speed set quite high to enable the train to take the gradients without stalling if I want to use cruise control - otherwise it's a case of 'driving' the train.

I must add though that I am not using one of your controllers - I might give yours a try as Mike Cullen is only a few miles away from where I live.

Rik

 

[CoggesRailway]

I just watched your video and wondered what you were using. It's a beautfiul build but one could see the driving was a challenge in places to get slow realistic responses. I have used three different solutions, using RC tx for a car/boat and the only really elegant approach I have found is Cliff Barker, it really is designed, in total, for a garden train. Sadly its £140!

I must say it is odd that a U class chassis would run away with worm drives? I hope you odn't mind me observing this from you beautiful video, i just have been there myself, and could see though well edited, the speedcontrol was a battle.

 

[whizzo]

Hi I am now worried regarding mac 5, Brian jones is unable to build one for the next 5-6 weeks - and at about £60 is this the way to go???? - a company ecm made a controller which had full inertia- but panel mounted + 25 yrs ago ( sad or what ) please can some one say if cliff barker controller does the same thing -as I also have a gradient at 1/40 and I am now confused, as I thought it would be built in regards Dave

 

[CoggesRailway]

I would call and ask Cliff. I have no material gradients so cannot comment. I know that you can configure the emergency stop feature to dynamically brake as as opposed to coast which might imply it is fairly "intelligent" What i would say is the degree of control is far better than you will get on a standard joystick rc tx. I will drop another user with a 1.20 and Cliff barker kit a PM to come and comment.
I suspect this will be really down to your gearbox.
I ordered from Cliff on a Thursday and received my full order the following Thursday.
I have used Brian for sound and he is a good guy, but doesn't build stock and you have to wait around. It was worth it for the sound!

 

[ge_rik]

I must say it is odd that a U class chassis would run away with worm drives? I hope you odn't mind me observing this from you beautiful video, i just have been there myself, and could see though well edited, the speedcontrol was a battle.

Probably a slight exaggeration to say it runs away - basically it decelerates up the grades and accelerates down them. To avoid stalling on the up grades I have to set the cruise speed what I consider to be unrealistically high. Whilst I do tend to 'drive' my trains during operating sessions, there are times when I just want something pottering around in the background when gardening or entertaining which is fine under DCC as it uses motor load feedback to keep the train moving at a realistic speed without any interference from me.

Maybe it will take me a little while to get used to a battery set-up - it's early days yet.

Rik

 

[whatlep]

I would call and ask Cliff. I have no material gradients so cannot comment. I know that you can configure the emergency stop feature to dynamically brake as as opposed to coast which might imply it is fairly "intelligent" What i would say is the degree of control is far better than you will get on a standard joystick rc tx. I will drop another user with a 1.20 and Cliff barker kit a PM to come and comment.
I suspect this will be really down to your gearbox.

I'm the person Ian PMed....

I too am surprised that you've got such a free running gearbox on a 1 in 40 (2.5%) gradient, but I can see it could happen with a heavy train on a lengthy gradient of that angle. My experience on a gradient twice that steep and about 20 feet long is that it's rare to get locos running away unless they are unusually free running. The LGB Rugens 0-8-0T would do it even on DCC with back EMF, but that was both heavy and particularly free running. An R/C Stainz with 10 hoppers behind it doesn't run away, but it is hard to get it to stop. Holding it on the grade requires a touch of backwards power, so I guess the potential for a runaway is there.

 

[CoggesRailway]

I would call and ask Cliff. I have no material gradients so cannot comment. I know that you can configure the emergency stop feature to dynamically brake as as opposed to coast which might imply it is fairly "intelligent" What i would say is the degree of control is far better than you will get on a standard joystick rc tx. I will drop another user with a 1.20 and Cliff barker kit a PM to come and comment.
I suspect this will be really down to your gearbox.

I'm the person Ian PMed....

I too am surprised that you've got such a free running gearbox on a 1 in 40 (2.5%) gradient, but I can see it could happen with a heavy train on a lengthy gradient of that angle. My experience on a gradient twice that steep and about 20 feet long is that it's rare to get locos running away unless they are unusually free running. The LGB Rugens 0-8-0T would do it even on DCC with back EMF, but that was both heavy and particularly free running. An R/C Stainz with 10 hoppers behind it doesn't run away, but it is hard to get it to stop. Holding it on the grade requires a touch of backwards power, so I guess the potential for a runaway is there.

But do you have any comment on you Cliff Barker stuff regards ease of general control on the hill? I know my "garden railway chip" and car tx equipped alcos were a handful!

 

[whatlep]

But do you have any comment on you Cliff Barker stuff regards ease of general control on the hill? I know my "garden railway chip" and car tx equipped alcos were a handful!

I shall stop looking at furniture catalogues with SWMBO and go out and do some testing. Thanks Ian!

 

[nicebutdim]

I've tried a couple of different esc's (not Cliff Barker's yet) and have to say the best behaved one on my incline is the mtroniks loco 10. The only problem with this esc is it can only handle up to 12v, but I find that a nice speed anyway. It doesn't slow too much going up and a little quicker on the decent.

 

[whatlep]

Right, testing completed. Surprisingly interesting: well to me anyway. I set up a rake of 8 LGB 2 axle hopper wagons and ran comparative tests with one Cliff Barker r/c fitted Stainz and one DCC fitted Stainz.

The r/c Stainz accelerated slightly downhill, but not enough to make me want to use the remote control to slow it down. Uphill it slowed enormously and needed manual intervention to feed more power to the motor. It was always under control and had no problems getting the 8 wagons up Pootank Pass (1 in 20/ 5% grade).

The DCC Stainz ran steadily at the same speed downhill and would have presumably done the same uphill EXCEPT that it developed wheelspin and couldn't get up the hill. As the locos are identical starter set models, I can only deduce that the pickup "bullets" on the DCC loco caused enough extra drag to stall the loco.

I don't think there's any issue with the Cliff Barker kit. It's very easy to use. On the wider topic of r/c and DCC you pays your money and takes your choice. R/c means you have to drive your trains on gradients, which is either a pain or more realistic, depending on your mood. DCC means you may have to add more locos on hills, or reduce train length, which is either a pain or more realistic, depending on your mood.

 

[CoggesRailway]

Excellent post Peter. Things look set fair for follow up testing Friday...

So the answer is the CB is not speed compensating in anyway.

My RC cars actually need brakes and they use dynamic braking. With 60a burst one can get in a pickle...
http://www.youtube.com/watch?v=wYB2TfSVLr0
joking aside this shows what modern batteries will do- for half an hour at a time. One LGB motor is nothing...

 

[ge_rik]

OK - thanks for all the testing folks - very reassuring.

An update on my progress. I seem to have cracked slow running. The 'stalling' which I encountered before was actually wheelspin. What caught me out was that the wheelspin was not on the stretches of track I expected. Closer inspection revealed that the lower edge of the bracket supporting the Walschaerts valve gear was catching on the track -

I've pared it down as much as I dare and it has alleviated the problem. If anyone has a U-class loco to hand I'd appreciate someone letting me know what the clearance should be - I can't see why it's now catching when I'm sure it wasn't before I dismantled and reassembled the chassis. It's almost as though the wheels have somehow shrunk in diameter.

Anyway - her performance is now much improved. There is a discernible difference between her speed when ascending and descending gradients but it's now not so alarmingly different. In fact I was very impressed by how slowly I could set the speed and she still climbed the gradient without problem. It looks as if the ESC does somehow respond to the increase in load. She does accelerate down the grades but not as dramatically as before I shaved a bit off the bottom of the valve gear bracket.

I still think an ESC which includes back emf load control would be a real boon as I'm sure I can't be the only one who would appreciate the additional level of precision which this would bring.

Rik

 

[Tony Walsham]

Rik.
Was there still a DCC decoder in the loco and you isolated the track pick ups and simply powered the whole loco from the ESC output?
If so that may explain why it was not working correctly. You must remove all DCC decoders when battery powering. They do not like PWM being fed into them.

 

[gregh]

Anyway - her performance is now much improved. There is a discernible difference between her speed when ascending and descending gradients but it's now not so alarmingly different. In fact I was very impressed by how slowly I could set the speed and she still climbed the gradient without problem. It looks as if the ESC does somehow respond to the increase in load. She does accelerate down the grades but not as dramatically as before I shaved a bit off the bottom of the valve gear bracket.
I still think an ESC which includes back emf load control would be a real boon as I'm sure I can't be the only one who would appreciate the additional level of precision which this would bring.
Rik

Ah, the problem of being here on t'other side - by the time I read the post and I can think over the problem, someone else has solved it.
I was going to suggest the loco must have had a problem.
But back to the esc - is it easy to get at the esc output/motor wires Rik? If so can you measure the volts on the motor when going up and down hill easily?
I will predict they are rock steady and don't change more than a few tenths of a volt with load - ie there is no 'load compensation' or feedback. And some of that change may be the battery volts anyhow. (try measuring that too if you can.)
The hobbyking escs that I use just put out a constant voltage dependent on the throttle setting. So any speed variation is just due to the motor armature resistance (or the battery internal resistance)

My experience with feedback /speed compensation is this. For my last steam loco I built a picaxe controller. I measured the voltage across a 2.2 ohm resistor in series with motor, to calculate the motor current and thus calculate the 'true' armature voltage=speed, so I could better control the chuff rate (I just use voltage control - no magnets/switches etc for chuffs). Having done that I though 'why not control the motor speed so it is constant?' So I made the program to hold the speed rock steady. BUT I DIDN'T LIKE THE EFFECT. I like to see some variation in speed as a train wanders around, with subsequent change in the chuff rate. But with the extra control, it just sounded the same all the time and got boring.
So I took the feedback feature off and actually added an extra 1 ohm series resistor to increase the effect of speed changing with grade.

see....
http://www.gscalecentral.net/ESCs-and-backemf-m239142 < Link To http://www.forum.gscalece...39142 post #10

 

[ge_rik]

Rik.
Was there still a DCC decoder in the loco and you isolated the track pick ups and simply powered the whole loco from the ESC output?
If so that may explain why it was not working correctly. You must remove all DCC decoders when battery powering. They do not like PWM being fed into them.

No, Tony. I've removed all vestiges of track power - pick-ups, wiring and decoder. Just two wires from motor to ESC.

Rik

 

[ge_rik]

Is it easy to get at the esc output/motor wires Rik? If so can you measure the volts on the motor when going up and down hill easily?
I will predict they are rock steady and don't change more than a few tenths of a volt with load - ie there is no 'load compensation' or feedback. And some of that change may be the battery volts anyhow. (try measuring that too if you can.)

Yes - the ESC is easy to get at but not sure how I'd measure the volts on the move. Presumably run a couple of wires to the meter mounted in a truck? I think I'd need a digital meter to get a clear reading - unfortunately mine is an analogue meter so it might be tricky to get a decent reading. Can digital multimeters be set to record a min and max reading (now that would be useful)?

The hobbyking escs that I use just put out a constant voltage dependent on the throttle setting. So any speed variation is just due to the motor armature resistance (or the battery internal resistance)

My experience with feedback /speed compensation is this. For my last steam loco I built a picaxe controller. I measured the voltage across a 2.2 ohm resistor in series with motor, to calculate the motor current and thus calculate the 'true' armature voltage=speed, so I could better control the chuff rate (I just use voltage control - no magnets/switches etc for chuffs). Having done that I though 'why not control the motor speed so it is constant?' So I made the program to hold the speed rock steady. BUT I DIDN'T LIKE THE EFFECT. I like to see some variation in speed as a train wanders around, with subsequent change in the chuff rate. But with the extra control, it just sounded the same all the time and got boring.
So I took the feedback feature off and actually added an extra 1 ohm series resistor to increase the effect of speed changing with grade.
see....
http://www.gscalecentral.net/ESCs-and-backemf-m239142 < Link To http://www.forum.gscalece...39142 post #10

Actually, as I watched her going round once I'd sorted out the binding on the valve gear, I thought it looked a bit more realistic than the DCC powered train which I ran with the same train for comparison. I don't have sound so don't have that extra bit of sensual feedback (if you get my drift). I think what was concerning in my earlier posts was the dramatic difference between uphill and downhill. As I was setting the speed to that which would just overcome the wheelspin you can imagine the difference between that apparent crawl and what happened when she had free rein.

The wheelspin occurred at the crest of the gradient - presumably there was just enough upward bend in the rail to foul the bracket. The great law of S.O.D. dictated that the wheelspin occurred when the loco was hidden by plant growth and so I was assuming it was a stall. There's nothing like getting down on hands and knees and peering between the plant fronds (a la Rowan & Martin's Laugh-in) to gather evidence to solve a technical problem. All part of life's rich tapestry when one chooses to run model trains in the garden methinks(!)

Rik
PS Couldn't get the link to work
PPS Just tracked down the thread and to my embarrassment see that it was me who started it on exactly the same topic - it must be an age thing!

 

[gregh]

Yes - the ESC is easy to get at but not sure how I'd measure the volts on the move. Presumably run a couple of wires to the meter mounted in a truck? I think I'd need a digital meter to get a clear reading - unfortunately mine is an analogue meter so it might be tricky to get a decent reading. Can digital multimeters be set to record a min and max reading (now that would be useful)?



..... (a la Rowan & Martin's Laugh-in) ......

Yes, I'd just put the meter in a wagon. Analogue should be OK. You can get min/max reading meters but big bucks.

Funny you should mention Laugh-In. When you used the word sycophantic in another post I thought I'd better "look that up in my Funk and Wagnells".