Servo stall motor for points

[melmerby]

On 10/03/2020 at 11:38, Ian Morgan said:

If you are proficient enough to consider modifying a servo, you could easily build a Merg Servo4 kit. At less than £7 and able to control up to 4 servos, the cost of the electronics can be affordable.

Or an Arduino, even less for 16 servos

£1.41 for a 16 channel servo controller + Postage £0.43

£1.38 for an Arduino Nano clone + Postage £1.04

5V PSU required

[newbryford]

1 hour ago, melmerby said:

Or an Arduino, even less for 16 servos

£1.41 for a 16 channel servo controller + Postage £0.43

£1.38 for an Arduino Nano clone + Postage £1.04

5V PSU required

 

The prices may be all well and good, but how useable is the above kit for an average modeller, bearing in mind that a reasonable number cannot cope with electrofrog points for example?

[AndyID]

All good stuff I'm sure but is it practical to use the motor and gears in an inexpensive servo as a stall motor to control points?

[TheQ]

10 hours ago, AndyID said:

 

If you mean actual servos (with all their all innards still intact) you can drive all of them from two pulse generators that run continuously via single pole switches. One pulse is long and the other pulse is short. The snag is all the servos will rotate by the same angle and you'll have to adjust the throw at the points by some mechanical means. If you extend the intervals between the pulses the servos can be made to rotate quite slowly. There is no requirement to drive them every 20 milliseconds. (I am referring to cheap servos like the SG90).

 

The pulse generators can by coded in a simple micro-controller or wired up on veroboard from a couple of 555 timers if you prefer.

Why bother with that when you can buy a pre-made servo controller for about a quid? 

https://www.amazon.co.uk/Profrssional-Consistency-Controler-Channels-Jasnyfall/dp/B07WVD5D7M/ref=sr_1_10?crid=3ABEQV4AE8XGU&keywords=servo+tester&qid=1584259414&sprefix=servo+teste%2Caps%2C192&sr=8-10&swrs=61491DA45CC368A24B690749046526DD

 

[Izzy]

I would say yes, with the proviso the actual stall current was manageable. The figures you gave previously, 0.126ma, indicate that this could be an issue at 5 times that of the tortoise/cobalt ones. The total current draw if large numbers were needed at 1amp per 8 - I think that’s correct? So if there’s a couple of dozen etc.....or more....

 

Another reason perhaps to use them at low volts for slower movement and cut power after movement stall. It’s all swings and roundabouts really....

 

Izzy

[melmerby]

9 hours ago, newbryford said:

 

The prices may be all well and good, but how useable is the above kit for an average modeller, bearing in mind that a reasonable number cannot cope with electrofrog points for example?

Probably easier than the Merg kit I was comparing with.

[Crosland]

14 hours ago, cliff park said:

OK, this is like, throwing an idea out there, may not even be new, may be nonsense. To save on the control electronics, how about if all the power supplies were common to several servos, but the control wire went through a one pole wafer switch. So you turn it to the servo you want to operate then twirl the knob or operate the switch to control the servo. Would obviously only work for basic end to end servos, no fancy signal bounce etc, but for points ? I have checked with several different servos, and they are really hard to move from the horn end, so I can't see them moving when they shouldn't. Failing the single pole idea, what about a three pole switch and switch the power leads as well ?

 

Turning off the control pulses to a servo is a well known technique to prevent high current consumption when they hit a mechanical stop, so yes this would work with a couple of provisos

 

Do not leave the control line floating. You need to tie the inactive ones to 0V.

 

Make sure the wiring is such that the control lines are not susceptible to any coupling from nearby wiring, especially traction current.

[Izzy]

4 hours ago, TheQ said:

Why bother with that when you can buy a pre-made servo controller for about a quid? 

https://www.amazon.co.uk/Profrssional-Consistency-Controler-Channels-Jasnyfall/dp/B07WVD5D7M/ref=sr_1_10?crid=3ABEQV4AE8XGU&keywords=servo+tester&qid=1584259414&sprefix=servo+teste%2Caps%2C192&sr=8-10&swrs=61491DA45CC368A24B690749046526DD

 

 

I had thought in the past that a couple of these, one each set to the opposite travel distance and ganged with centre-off switches as per a previous post might be a simple/easy control method. But the drawback would be the fixed travel distance and max speed for all servos connected, which boards such as the MERG servo4 overcome, being adjustable for both.

 

The reasons I moved to hacked servos were the combination of wanting smaller units than the Tortoise/Cobalt are, for 2mm use with such as raising/lowering un-coupling magnets as well as points, and then not being able to overcome the EMI/RFI issues I could not eliminate that many don't encounter.  I found I still had them with the single servo tester I have so didn't take it any further.

 

Izzy

 

 

[Andy Reichert]

15 hours ago, newbryford said:

 

The Cobalt Analog IP and Digital IP motors switch down the current after a few seconds after stall - to about 5mA, (enough to power the electronics)  whereas the Cobalt Omega and Tortoise continually draw current at stall - typically 25mA.

IME, switching off power completely to any of the above does not cause any issue of the tiebar relaxing and allowing the blade to move away from the stock rail with the potential for derailment.

It is quite difficult to backdrive a Tortoise and as good as impossible to do it to a Cobalt without damaging the gears.

 

I don't see the major advantage of switching down the current unless it's unreasonable in the first place. 30/25mA stall current means that a single 1Amp, 12v, Wall Wart will run 30/40 point motors. Bigger layouts than that will likely have a lot more running trains and so have extra power needs regardless.

 

If making point motors as a DIY exercise, IMHO, the most important aspect is making a very sound and reliable mechanical connection to the throw bar. Nothing stops pleasurable layout operation more than having to continually fix broking or misaligned point throws.  Finding motors and electronics is the easy part.

 

Probably the other important factor is having a system where the points are (a) always set to a particular preferred (or just known) direction when first powering up the layout and (b) all their actual set positions are clearly indicated by either lights or toggle switch position.  I suspect using a rotary switch as mentioned, on start-up, would mean having to set every turnout from an unknown to a wanted position before any train could be safely run.

 

Andy

[AndyID]

21 hours ago, TheQ said:

Why bother with that when you can buy a pre-made servo controller for about a quid? 

https://www.amazon.co.uk/Profrssional-Consistency-Controler-Channels-Jasnyfall/dp/B07WVD5D7M/ref=sr_1_10?crid=3ABEQV4AE8XGU&keywords=servo+tester&qid=1584259414&sprefix=servo+teste%2Caps%2C192&sr=8-10&swrs=61491DA45CC368A24B690749046526DD

 

 

Because a single pulse generator can drive an unlimited number of servos. All you need is a single pole changeover switch to control each servo.

[AndyID]

Answering my own question it seems that not all cheapo servos are alike. The one I happened to select for the initial test was very "free running" and it always reverses on demand. Others seem to be newer versions (the splined output boss has a smaller diameter) and they are less consistent.

 

Some of the "newer" servos can be back-driven quite easily but others lock-up. I'm not sure if that's because the motors are different or because the gear-train tends to wedge.

 

I'm going to muck about with pulsing the current to see if that makes a difference (if I can resuscitate the pulse generator code I wrote for an Attiny44A a few years ago   )

[Junctionmad]

On 15/03/2020 at 13:47, Andy Reichert said:

 

I don't see the major advantage of switching down the current unless it's unreasonable in the first place. 30/25mA stall current means that a single 1Amp, 12v, Wall Wart will run 30/40 point motors. Bigger layouts than that will likely have a lot more running trains and so have extra power needs regardless.

 

If making point motors as a DIY exercise, IMHO, the most important aspect is making a very sound and reliable mechanical connection to the throw bar. Nothing stops pleasurable layout operation more than having to continually fix broking or misaligned point throws.  Finding motors and electronics is the easy part.

 

Probably the other important factor is having a system where the points are (a) always set to a particular preferred (or just known) direction when first powering up the layout and (b) all their actual set positions are clearly indicated by either lights or toggle switch position.  I suspect using a rotary switch as mentioned, on start-up, would mean having to set every turnout from an unknown to a wanted position before any train could be safely run.

 

Andy

 

teh issue is for the Classic cobalt is the stall current  can be very high , 90mA on 12V for example, ( a dirty little secret of Cobalts, as the internal voltage is clamped to 6.5V and the rest is lost in resistors )  ) , so if you have 50 of them ( as I have ) thats a big draw. the Analog IP drops this to 5mA . 

 

If you are using the " magic sleeper" approach , then the simple sprung steel rod is the best 

 

agree re startup , 

Edited March 18, 2020 by Junctionmad

[Andy Reichert]

2 hours ago, Junctionmad said:

 

teh issue is for the Classic cobalt is the stall current  can be very high , 90mA on 12V for example, ( a dirty little secret of Cobalts, as the internal voltage is clamped to 6.5V and the rest is lost in resistors )  ) , so if you have 50 of them ( as I have ) thats a big draw. the Analog IP drops this to 5mA . 

 

 

 

An interesting choice for someone West of the Pond.  I wouldn't have thought it economic?

 

Andy

[AndyID]

And....I don't think it's going to work. There seems to be too much variability between actual servos and inconsistency in the servo mechanisms themselves. I think it's mainly a function of tolerances in the gear train. Hardly a surprise considering how inexpensive they are. It's a pity because the motors do seem to be quite consistent.

 

It might work with servos that have metal gear trains and ball bearings but they are a lot more expensive which sort of defeats the purpose of the whole exercise.

 

It looks like the the motor has to receive around 1.5 volts DC then some mechanism blocks the current at "full travel". That's what the hysteresis microswitch version (see above) does. I have an idea for a way to simplify the travel adjustment.

 

 

[newbryford]

15 hours ago, Junctionmad said:

 

teh issue is for the Classic cobalt is the stall current  can be very high , 90mA on 12V for example, ( a dirty little secret of Cobalts, as the internal voltage is clamped to 6.5V and the rest is lost in resistors )  ) , so if you have 50 of them ( as I have ) thats a big draw. the Analog IP drops this to 5mA . 

 

If you are using the " magic sleeper" approach , then the simple sprung steel rod is the best 

 

agree re startup , 

 

 

The high current may be the case for original motors (easy to identify because they have 8 terminals and no external markings.

 

This is a Cobalt Omega Classic with voltage selector set to 12v.

35mA at stall.

( This Omega does have 9 terminals - it's just that the I managed to break the last one!)

 

 

 

 

[melmerby]

21 hours ago, Junctionmad said:

 

teh issue is for the Classic cobalt is the stall current  can be very high , 90mA on 12V for example, ( a dirty little secret of Cobalts, as the internal voltage is clamped to 6.5V and the rest is lost in resistors )  ) , so if you have 50 of them ( as I have ) thats a big draw. the Analog IP drops this to 5mA . 

 

 

They didn't like being driven by NCE RS-8 either, the motor could fail, which is odd as the RS-8 can't give out anything like 90mA at 12v (it self limits)

The two I bought as a trial both ended up sticking at the end of travel, so I went back to the totally reliable Tortoises.

 

[AndyID]

I may have spoken a bit too soon. It does work, but you have to change the gear ratio.

 

That's not as hard as you might think. You simply remove one of the compound gears and lock both the gears on the potentiometer shaft together. I think the reduction ratio is then something like 12 to 1. That lets you back drive the motor from the horns with very little force.

 

I powered it from 1.4 volts DC via a resistance of 5.8 ohms for a stall current of 140 mA. Stalled motor power dissipation is 82 mW. That produces a nice steady torque which you can easily back drive - feels a bit like it's spring loaded which is exactly what you want.

 

It should be practical to drive it from a higher voltage as long as you use a larger resistance value and the stalled motor power is low enough. You will have to account for the greater power dissipation in the resistor.

 

 

[AndyID]

This is one way you might use it (there are many others).

 

The switch on the left is the control panel switch that controls the points. + and - voltages are not equal. + will not pull in the relay (RL) but - (minus) will pull in the relay (probably something like -5 volts). Power dissipation is minimized when the points are set in the "normal" (low voltage) position.

 

There are two resistors which are switched by one set of the relay's contacts. The other set of relay contacts control the frog polarity.

 

"0" is zero volts relative to the plus and minus supplies at the control switch. On DC layouts it could also be the track common return.

 

[Ian_H]

Hi,

 

Just reading through this thread and it reminded me of a thread that I bookmarked years ago, it may be of interest;

 

 If you read through the thread there are several ways of operation that may be of interest, one particular method of operation is the stripping out of the electronics in the servo and simply using a DPDT switch with a couple of diodes and two micro switches. The following drawing was proposed by  Bertiedog a good number of years ago.

 

 

I hopeyou find the above useful

 

Ian

[AndyID]

On 29/03/2020 at 09:01, Ian_H said:

Hi,

 

Just reading through this thread and it reminded me of a thread that I bookmarked years ago, it may be of interest;

 

 If you read through the thread there are several ways of operation that may be of interest, one particular method of operation is the stripping out of the electronics in the servo and simply using a DPDT switch with a couple of diodes and two micro switches. The following drawing was proposed by  Bertiedog a good number of years ago.

 

 

I hopeyou find the above useful

 

Ian

 

Hi Ian,

 

That works but it's not really a stall motor.

 

That said, while I can get some servo mechanisms to work in stall mode, a lot of others are reluctant to work that way. I think the main reason is due to variations in the gear train - understandable considering how small and inexpensive they are.

 

I think the most reliable method is to use the motor and gearing with limit switches like the version you have linked. My way of doing it is even less expensive. It only requires the servo, a home-made cam and one micro-switch.

 

The cam is made from a piece of 2mm PVC - formerly a square downspout - manufactured with tin snips, "precision" files and a hand drill.

 

 

 

 

The last pic is the cam profile. I print them and glue them on to the PVC for profiling. The only important section is the bit between 3 and 5 o'clock. The radius has to be fairly constant there.

 

It operates via a single pole changeover switch from 1.5 volts. There is no current drain when the motor reaches the limits which means you could drive a large number of points from a single AA battery.

 

Cheers,

Andy

 

 

EDIT: I forgot to mention it's quite simple to add another micro-switch to control frog polarity.

Edited March 31, 2020 by AndyID
moinfo