Load sensing point motors

[Enterprisingwestern]

According to what I've read and seen servos frequently do not operate well on a model railway, but if they work well for you that's OK.

 

 I take it you have no practical experience of servos then?

 

Mike.

[newbryford]

It looks like Hobbyking's micro linear servo is no longer with us.

 

According to what I've read and seen servos frequently do not operate well on a model railway, but if they work well for you that's OK.

 

Apart from susceptibility to interference there's also the slightly complicated business of setting up the correct throw distance, and for several reasons that might require adjustment over time. Pressure sensing would eliminate that.

 

My servos are simply mounted in aluminium channel with a hole for the throwbar (as per Megapoints instructions).

No throw adjustment needed - they worked straight out of the box. 4 exhibitions so far and not a single issue with them.

Yes - there's a bit of "chatter" as I have the early version control boards, but it is not at all noticeable above the general background hubbub of an exhibition hall.

[Crosland]

According to what I've read and seen servos frequently do not operate well on a model railway, but if they work well for you that's OK.

There are well understood issues and well understood fixes, some related to servos, some to the wiring, some to the controller, some to the power supply. Some have even been covered up-thread.

 

Apart from susceptibility to interference there's also the slightly complicated business of setting up the correct throw distance, and for several reasons that might require adjustment over time. Pressure sensing would eliminate that.

I would never design a turnout operating unit without including some form of compliance, e.g., an omega loop. Throw distance is then not an issue.

 

If you want the best slow motion (or bounce effect for signals) you should be aiming to use the complete servo throw (or as much as possible) rather than trying to accurately set the endpoints.

[AndyID]

I take it you have no practical experience of servos then?

 

Mike.

Hi Mike,

 

Actually I have a lot of practical experience with servos and I can make them work without any problems at all. The topic isn't really about servos. It's about simple point motors driven from an inexpensive current sensing controller.

 

If people happen to like using servos, more power to them.

 

Cheers,

Andy

[AndyID]

Looks like a SN7554410 could be an option for driving the motors. One can drive two motors and they are not very expensive. I think it would be possible to make one drive four motors but I don't think it's worth the extra complications.

[NIK]

Not sure where to put this. It applies just as much to DC as it Does to DCC, but here goes.

 

I'm pretty sure most of us have cars with electric windows. Why don't model railway point motors work the same way?

 

How does an electric car window know when the window is closed? Because it senses the motor current and shuts off the power when it exceeds a predetermined value that indicates the motor has stalled, or is close to stalling.

 

Hi,

 

Its a good question.

 

Maybe the force/current to move point blades on a ballasted point is higher than that to keep them in place - I think this is how the H&M point motors worked- with a overcentre spring in the motor to keep the blades parked once the current was removed.

 

Nowadays some point motors can be used to operate semaphore signals where stalling the motor could bend or break the signal.

 

Hand made points don't usually have over centre springs so there has to be some mechanical force to keep the blades in place.

 

If the force from a point to a servo is too high the gears will move the motor and the servo potentiometer will tell its electronics to correct the movement. This might not happen if the servo is turned off once a certain current is reached.

 

Stall motors such as Tortoise presumably use the stall force to keep the point blades in place.

 

 

Regards

 

Nick

[AndyID]

Hi,

 

Its a good question.

 

Maybe the force/current to move point blades on a ballasted point is higher than that to keep them in place - I think this is how the H&M point motors worked- with a overcentre spring in the motor to keep the blades parked once the current was removed.

 

Nowadays some point motors can be used to operate semaphore signals where stalling the motor could bend or break the signal.

 

Hand made points don't usually have over centre springs so there has to be some mechanical force to keep the blades in place.

 

If the force from a point to a servo is too high the gears will move the motor and the servo potentiometer will tell its electronics to correct the movement. This might not happen if the servo is turned off once a certain current is reached.

 

Stall motors such as Tortoise presumably use the stall force to keep the point blades in place.

 

 

Regards

 

Nick

 

Hi Nick,

 

I'm only thinking about driving points, although I suppose it would work for signals etc too. The general (half-baked) idea is to use a micro-controller to drive a large number of points. The maximum force would be preset to some default value but it could also be configured to a different value for each motor. I'm assuming there will be plenty of friction in the drive to hold the blades in place, and as Crosland says, it's always a good idea to have compliance between the motor and the tiebar. I also think I would make the controller do a periodic "force test" at each motor to ensure the point blades were not drifting.

 

Of course, this is all entirely theoretical at the moment

 

Cheers,

Andy

[Crosland]

If the force from a point to a servo is too high the gears will move the motor and the servo potentiometer will tell its electronics to correct the movement. This might not happen if the servo is turned off once a certain current is reached.

 

Unlikely. It takes a lot of force to move a servo, even when powered off. You are more likely to strip the gears in the cheap servos recommended up-thread.

[Suzie]

The SG92R has carbon fibre gears that appear to be indestructible. The output shaft can be turned when unpowered at rest but it does take quite a bit of force to do so. I would not try that with an SG90 - it will probably break.

[AndyID]

The SG92R has carbon fibre gears that appear to be indestructible. The output shaft can be turned when unpowered at rest but it does take quite a bit of force to do so. I would not try that with an SG90 - it will probably break.

You can back drive the SG90 quite easily by applying enough pressure. It keeps going once you get it moving. I haven't managed to strip any gears yet

[AndyID]

How time flies!

 

In the interest of getting things going I thought I'd try using an Arduino Mega2560 (clone) to drive a number of servo point-motors. The idea is to detect when the servo is trying to overcome the mechanical resistance created by the "end stops" and cut off power at that point.

 

As the controller "sweeps" the servo in the desired direction, if the error is not large, the duration of the motor drive pulses is quite short but when the servo starts to encounter a lot of mechanical resistance the error becomes significantly greater which results in a much longer drive pulse. The duration of the servo's drive pulse can easily be determined by sensing the servo supply (or return) current.

[Sol]

On 19/11/2018 at 11:03, Junctionmad said:

The cobalt analog IP point motor from DCCconcepts works like you describe , at the end of the throw , the power is cut to the drive motor ( unpon sensing stall )

 

Dave

 

OK, so power is cut by the Cobalt so how is it put back on when turnout direction is to be changed?

[NIK]

8 minutes ago, Sol said:

 

OK, so power is cut by the Cobalt so how is it put back on when turnout direction is to be changed?

Hi,

 

To change the direction I think you operate a Double Pole Double Throw switch wired up to reverse the polarity fed to the Cobalt analogue IP.

 

I guess the Cobalt analogue IP has diodes across the stop switches so that when the input voltage is reversed the stop switches are bypassed and the motor can move in the opposite direction (similar to the Fulgurex slow action point motor).

 

Regards

 

Nick

[newbryford]

4 hours ago, NIK said:

Hi,

 

To change the direction I think you operate a Double Pole Double Throw switch wired up to reverse the polarity fed to the Cobalt analogue IP.

 

I guess the Cobalt analogue IP has diodes across the stop switches so that when the input voltage is reversed the stop switches are bypassed and the motor can move in the opposite direction (similar to the Fulgurex slow action point motor).

 

Regards

 

Nick

 

Cobalt point motors do not have switches to limit the travel.  (Tortoise are the same)

 

The Cobalt IP range have current sensors that cut the power off to the motor.

In the case of the Analog IP, it waits for the input voltage to reverse, for the Digital IP, it needs the command to move the other way - either by DCC, or the manual inputs via the PBS terminals.

 

(Analog stall type motors can be operated via SPDT switches to save on wiring, but that's another topic)

 

Cheers,

MIck

 

 

Edited May 24, 2019 by newbryford

[Junctionmad]

5 hours ago, Sol said:

 

OK, so power is cut by the Cobalt so how is it put back on when turnout direction is to be changed?

all looked after by the wondrous built in microprocessor !

[AndyID]

4 hours ago, newbryford said:

 

Cobalt point motors do not have switches to limit the travel.  (Tortoise are the same)

 

The Cobalt IP range have current sensors that cut the power off to the motor.

In the case of the Analog IP, it waits for the input voltage to reverse, for the Digital IP, it needs the command to move the other way - either by DCC, or the manual inputs via the PBS terminals.

 

(Analog stall type motors can be operated via SPDT switches to save on wiring, but that's another topic)

 

Cheers,

MIck

 

 

 

Hi Mick,

 

I'm attempting to create a slightly less expensive solution.

 

A MEGA2560 clone costs around $11 and should be able to control 40 servos. Servos run around $1.50 a pop. Say 40 points for around $75 (plus switches, PSU etc.)

 

A Cobalt iP costs about $15   40 points will run around $600 (plus switches, PSU etc.)

 

Cheers!

Andy

[newbryford]

1 minute ago, AndyID said:

 

Hi Mick,

 

I'm attempting to create a slightly less expensive solution.

 

A MEGA2560 clone costs around $11 and should be able to control 40 servos. Servos run around $1.50 a pop. Say 40 points for around $75 (plus switches, PSU etc.)

 

A Cobalt iP costs about $15   40 points will run around $600 (plus switches, PSU etc.)

 

Cheers!

Andy

 

Quite so Andy.

I never brought any costs into it - I was just making an observation on Cobalt motors.

 

But now you've mentioned it.

Don't forget to include servo mounts and wire to fix to the tiebar.

And micro switches - if frog switching is required. Probably a few other bits and pieces required as well.

Yes - still cheaper than the cost of a Cobalt, but all "hidden extras" on top of your $75 and then you've got to program it. Which in itself will put many folk off.

Great for those that are into that sort of thing, but there's always a market for "Plug n play"

 

Cheers,

Mick

 

 

[AndyID]

I have no commercial interest. I'm just exploring alternatives available to anyone. IF and when I can get the code running I will make it available to anyone who wants it (for their personal use of course).

 

Normally a low cost servo like the SG90 is driven by sending it a pulse every 20 milliseconds. Varying the pulse width between 1 and 2 milliseconds rotates the servo horns approximately 90 degrees. But it's not necessary to use a constant pulse rate like 20mS. It's possible to initiate another command pulse from the controller as soon as the internal logic in the servo removes power from its motor. The controller can determine when that happens by monitoring the servo current.

 

By issuing a fast stream of small position increments to the servo the controller can "tip-toe" up to the point where the servo encounters serious mechanical resistance. When that happens there is a significant increase in the duration of the servo current consumption. That "tells" the controller when to stop sending pulses.

 

At least that's the theory. I'll try to put it into practice. This may take some time.

 

 

 

 

[Paul4256]

Hi, I have also been using an Arduino mega to work point motors/mimic lights linked to home made signals & two other Arduino Uno's that constantly scan IRDOT detectors. The Arduinos are all powered by 9v PP3's to keep them isolated from the Gaugemaster mains transformers powering the track & lighting.

All works fine, except for the servo power supply. I power them from a dedicated 5volt transformer but when one of my older locos crosses points there is a momentary electrical blip which sends the servos into a frenzy.  Anyone else had similar or has found a solution?

[Nigelcliffe]

13 hours ago, Paul4256 said:

Hi, I have also been using an Arduino mega to work point motors/mimic lights linked to home made signals & two other Arduino Uno's that constantly scan IRDOT detectors. The Arduinos are all powered by 9v PP3's to keep them isolated from the Gaugemaster mains transformers powering the track & lighting.

All works fine, except for the servo power supply. I power them from a dedicated 5volt transformer but when one of my older locos crosses points there is a momentary electrical blip which sends the servos into a frenzy.  Anyone else had similar or has found a solution?

 

the "blip" is electrical noise from the loco - probably the brushes arcing somewhat, causing electrical noise.   The rails and the track wiring then act as an aerial to broadcast this noise.  Its picked up by the servo leads.

 

Possible solutions:

1 - swap the servo for a different brand.  They seem different in how they respond.

2 - alter the servo cable routing, particularly make it perpendicular to rails if you can

3 - consider pull-up or pull-down resistors on the servo drive circuitry

4 - arrange the servo control so it removes the power supply from the servos after movement of the servo is complete, without power the servo cannot twitch.

 

I'd suggest (1) and (4) may be the simplest. 

[AndyID]

On 04/06/2019 at 09:00, Paul4256 said:

Hi, I have also been using an Arduino mega to work point motors/mimic lights linked to home made signals & two other Arduino Uno's that constantly scan IRDOT detectors. The Arduinos are all powered by 9v PP3's to keep them isolated from the Gaugemaster mains transformers powering the track & lighting.

All works fine, except for the servo power supply. I power them from a dedicated 5volt transformer but when one of my older locos crosses points there is a momentary electrical blip which sends the servos into a frenzy.  Anyone else had similar or has found a solution?

 

Hi Paul,

 

As Nigel says it's most likely the servos are reacting to RF noise produced by arcing, either at the motor itself or between the power source and the motor (rail contact, pick-up contact etc).

 

The noise immunity of the inexpensive SG90 servo is (to be polite) not great, particularly when the servo input is low (logic zero). One method that seems to work for some is to stop sending pulses and hold the servo input high until it's necessary to move the servo position again. (BTW, that does not "power-off" the servo. It can still react to RF noise.)

 

If that doesn't work well enough for you it's also possible to "front-end" the servo with an optoisolator. I've found that to be pretty much bomb-proof. There's a thread here somewhere about it. Everyone seems to think I'm bonkers (I cannot deny that) but it does work and it's not very expensive

 

Andy

[Paul4256]

Hi Folks,  Many thanks for your suggestions. I'll have a go.  Cheers.

 

Update... Tried a few things - re-routed wiring that crossed under the section causing the issue,  moving servos circuit to battery power, fitted 100uF capacitors to servos, but in the end Nigel's first thought worked. I swapped the servos for other identical looking ones and the issue seems to have gone away. Looking closer, the 'faulty' units all had the brand name 'Lofty ambition' and a slightly different diameter shaft, otherwise identical to the Tower pros that work fine. We live and learn!

Edited June 13, 2019 by Paul4256
update