HOW TO SMOOTH OUT FEEDBACK CONTROLLERS

[paul 27]

I have two Gaugemaster DF  feedback controllers, which I believe rely on the quality of the electricity supply,

lower cycles cause problems is there any way to smooth out as I have problems with some motors buzzing effect.

[CourthsVeil]

Paul,

the manufacturer itself says "A degree of motor heating and noise may be experienced with sustained low speed running." 

Afaik these controllers require AC supply and will not work at all with smoothed power.

Hth

   Armin

[AdamsRadial]

I looked back through the Roger Amos book on Model Railway Electronics for his comments on how the feedback controllers he designed worked, because I built one of them nearly 20 years ago for running n-gauge models.

 

The power supply used is a standard mains to 12V transformer feeding a bridge rectifier, but with no smoothing capacitors across the bridge output. This is essential to allow the motor back-emf to be sensed, because it can only be sampled in the periods when the un-smoothed DC supply is dropping to or just rising from 0 Volts.

 

I then remembered I also had a very early Gaugemaster controller for n which from my recollection could run on either AC or DC. I managed to find my notes I had made when playing with it (not the instructions sadly), but I had recorded that "Mnfr states running on DC supply will not best control, must be run on AC". What they had on the input circuit was a bridge rectifier, to which you could supply AC 16V OR DC 12V, the bridge effectively took the DC and passed it through the diodes to give the required output polarity.

 

Roger Amos has published details of a quite sophisticated controller which uses a 555 timer to create quiet periods during which an OP-amp samples the track voltage to determine the back emf, which it then uses to vary the output voltage according. However, the 555 timer is effectively creating the same on-off duty cycle that the OP thinks is causing the motor noise, so although there would be some scope for varying the sample frequency to try and reduce the buzz, it could never quite be eliminated.

 

It would seem you cannot smooth out the waveform and have feedback at the same time.

[knitpick]

AdamsRadial is right.  Basically, a motor is also a generator.  Which is to say that if you turn the motor by hand you will generate a voltage across its brushes.

 

A feedback controller uses this capability.  It briefly interrupts the voltage supplied to the track - either by using unsmoothed power and picking the time when the voltage drops to zero or for a smoothed track supply it deliberately interrupts the voltage supplied to the track.  As the motor spins under inertia during this zero volt stage, the feedback controller measures the voltage that the motor (now acting as a generator) is creating.  [this is the "feedback" also known as back emf.]  The controller then adjusts the voltage supplied to the track with the intent of keeping motor speed (and thus train speed) constant.  It is this power interruption that causes the buzzing noise from the motor.  Interrupting a smoothed track output will cause greater buzzing.

 

PS - the Amos book is quite good.  If you understand a bit about electronics and can solder - it is IMHO well worth getting. 

[davefrk]

Having had an email conversation with the OP, he is actually asking if there is a way of operating a loco on feedback that previously stutters and buzzes, something like (IIRC) fitting a car type 5W bulb across the feeds to the track. I heard or read this many years ago but never needed to try it. I think it was suggested it would put a bit more load on the feedback when using high efficiency motors like Portescap and nowadays the modern Hornby motors.

 

Dave Franks.

[Suzie]

What you have is old low-frequency technology that relies on motors having a large mass to smooth out the running. Modern feedback controllers operate at ultrasonic frequencies which cannot be heard and offer much smoother control.

[AdamsRadial]

One thing people tend to forget here is that square waves contain harmonics at quite considerable multiples of the basic frequency, so if you use a 50-Hz square wave you are actually supplying harmonics up in the tens of Khz range. It could be these which are causing the noise and buzz the OP has complained about. The feedback controller I built from the Roger Amos book was using a 50Hz sine wave, which contains far fewer harmonics (if any at all).

 

(ETA) Putting an inductance in series with the supply to the track might act as a simple low-pass filter to chop off some of these harmonics. I'm not sure what the inductance of a filament bulb is but I doubt it's significant here.

 

The OP has posted a similar thread in the Motors section where he suggests that his feedback controllers give less noise when driving 3-pole motors than 5-pole. If so, I would be interested to investigate this further, because of two things: it might suggest a better frequency to use for the chopped DC power, and it might be a way of running older models at smooth speeds, (which is something I also would like to do).

 

In the back of my mind also are some experiences on the Trawlers when a ship with a direct-driven propeller had to be run with a 4-blade prop instead of the cheaper and more usual 3-blade prop because of serious vibrations set up by harmonics between the blades and engine. I can't help wondering if the greater number of magnetic events per revolution of a five-pole motor might be somehow resonating with a 50Hz or 100Hz chopper range, where a 3-pole set of events would not have the same exact multiple ( 5 into 50 or 100 gives an integer, unlike 3).

 

So, questions for Paul:

1) Are you able to post Youtube videos of a noisy 5-pole motor so we could possibly analyse the sounds and work out what the frequency of the buzz is?

 

2) Do you have two similar models fitted with 5-pole and 3-pole motors so that comparative testing could be done?

[34theletterbetweenB&D]

...The OP has posted a similar thread in the Motors section where he suggests that his feedback controllers give less noise when driving 3-pole motors than 5-pole. If so, I would be interested to investigate this further, because of two things: it might suggest a better frequency to use for the chopped DC power, and it might be a way of running older models at smooth speeds, (which is something I also would like to do)...

Simply, the 'better frequency to use' is in a band well removed from the mechanical resonance frequencies of the motor, and above the upper limit of human hearing. The work has been done, and it is why typical DCC decoders output at 25kHz or more. Use a well specified decoder with a lot of adjustment (Zimo or CTE would be my top suggestions for the purpose of taming old motor designs) and it will extract smooth running, to whatever the physical limitations of the motor are. It won't be able to reliably obtain a smooth creep into and out of motion, if the motor torque is insufficient to overcome the mechanical drag of the gear train at a small enough power input. Nor will it be able to do much about intrinsic mechanical noise of the mechanism: visually running smoothly but still making coffee grinder sounds is nothing unusual.

[paul 27]

Simply, the 'better frequency to use' is in a band well removed from the mechanical resonance frequencies of the motor, and above the upper limit of human hearing. The work has been done, and it is why typical DCC decoders output at 25kHz or more. Use a well specified decoder with a lot of adjustment (Zimo or CTE would be my top suggestions for the purpose of taming old motor designs) and it will extract smooth running, to whatever the physical limitations of the motor are. It won't be able to reliably obtain a smooth creep into and out of motion, if the motor torque is insufficient to overcome the mechanical drag of the gear train at a small enough power input. Nor will it be able to do much about intrinsic mechanical noise of the mechanism: visually running smoothly but still making coffee grinder sounds is nothing unusual.

I  use Analogue not DCC   and the noise is a  buzz humming from the motors, might try dismantling one any tips about stripping down a Hornby 5 pole motor.

[34theletterbetweenB&D]

I  use Analogue not DCC   and the noise is a  buzz humming from the motors...

 Fully aware of that. But here's the thing: no amount of poking around with the equipment you are using will have significant effect. It needs a high frequency drive, not mains frequency to achieve what you want. You can get that from both modern designs of DC controllers, and from DCC; with the latter the more effective of the two in a number of respects. You might like it if you try it....

 

... any tips about stripping down a Hornby 5 pole motor.

First suggestion: don't bother, you'll likely wreck it.

 

The 'secret' if you must do it, (common to most can motors) is having pulled the worm or drive coupler off the shaft is then to remove the brushes and springs before releasing the armature from the can. With that you have the armature assembly alone. Unless this has loose wires in the windings or is unbalanced there's nothing to do to it, likewise unless the shaft bearings are slack or the shafts damaged, nothing to do there. I have had a couple apart in the past ten years, one to stick a loose magnet back in place, the other to find and repair an open circuit winding.

[AdamsRadial]

Before you go as far as dismantling it, I'd suggest removing the motor from the chassis and running it just with a couple of wires going from the controller to it. See if you can feel the buzzing, see if it changes when you tilt the motor around through different angles, see if it is worse in one direction rather than the other.

 

Another thing to consider is, are the magnets weak in one or two of your motors? This would cause them to run hotter, demand more current, and maybe upset the feedback control by not giving as much back-emf at the same rotational speed as a better motor.

 

You could also assess all your motors in this way, if some are worse than others you might get a clue as to why by close examination.

 

Given your desire to try 3-pole motors from another thread, you could also get one off ebay and try some tests to see if they really do buzz less.

[brigo]

Many years ago I made a number of thyristor feedback controllers for my then local Club N gauge layout and for the late Don Jones model of New Street Station. I had investigated a number of commercial controllers, Digitol, ECM, AMR etc and I found one of the biggest problems to getting smooth running was the use a standard 16 volt AC transformers such as made by H & M.

I tested a H & M transformer (from a Clipper) which was rated at 16 volts at 2  amps. Some figures,

 

No load voltage, 18.5 volts AC

At 0.5 amp load, 18.0 volts AC

At 1 amp load, 17.5 volts AC

At 2 amp load, 16.5 volts AC

 

Now these voltages are the RMS value which is about 0.707 of the peak value for an AC voltage. H & M in their old controllers used a plate Selenium rectifier which had a bigger voltage drop than normal silicon diodes or rectifiers so nowadays the following voltages will be higher. The H & M's DC output were as follows

 

No load voltage, 16.5 volts DC

At 0.5 amp load, 15.0 volts DC

At 1 amp load, 14.0 volts DC

At 2 amp load, 12.0 volts DC

 

I then put a 1000 microFarad capacitor on the DC output to smooth it

 

No load voltage, 26.0 volts DC

At 0.5 amp load, 21.25 volts DC

At 1 amp load, 19.0 volts DC

At 2 amp load, 15.5 volts

 

So using a 16 volt AC transformer you can expect a modern low current motor to be getting pulses of up to about 24 volts. This voltage at up to 100 pulses a second is rattling the motor armature in its bearings making the buzz. Lowering the transformer voltage will reduce the size of the pulses and improve running.  What did use with my controllers, I used a 9 volt AC transformer for N gauge and OO, while for Don because of the voltage drop outdoors we used 12 volt transformers.

Sadly for the OP I believe his controllers have a built in transformer so it cannot be changed.

 

Another point with feedback controllers with N gauge or lighter locomotives is they can have a detrimental effect on running. If we have a motor which is vibrating this can lead to the whole locomotive to vibrate and momentarily break contact with the track. The controller will then generate a bigger pulse because it thinks the motor has come under a bigger load because the back EMF has disappeared, this will cause jerkiness in running. I have placed a weight upon a loco that was jerking to improve track contract, and it solved the problem. Lower voltage pulses will again help the problem.

 

Brian

[AdamsRadial]

Would clipping the output of the controller to some lower voltage help here? I don't think the back-emf would rise high enough to be affected if the voltage across the rails were to be limited to a 12 - 14V maximum.

[CourthsVeil]

...

Sadly for the OP I believe his controllers have a built in transformer so it cannot be changed.

...

Brian

 

Gaugemaster DF controllers definitely have the transformer built in.

   Armin

[AndyID]

It's also possible to build a feedback controller (some might prefer to call it "feedforward" because it's a type of positive feedback) that does not sense the EMF directly while the motor is "flywheeling". There is no interruption of the supply and therefore no buzzing or motor heating.

 

The method consists of sensing the motor current and increasing the applied voltage to cancel most of the voltage drop across the internal resistance of the motor. That means the motor EMF is always close to the EMF (speed) demanded by the controller.

 

It actually works very well but there are a couple of snags: 1) the resistance in the circuit between the controller and the motor has to be fairly constant (no dirty track or big voltage drops in the wiring); and 2) the controller has to be set to match the internal resistance of the motor. In practice this might mean three settings for high, medium and low resistance motors.

 

Also, double heading might be a problem, but you can always turn the feedback off when required.

[paul 27]

Would clipping the output of the controller to some lower voltage help here? I don't think the back-emf would rise high enough to be affected if the voltage across the rails were to be limited to a 12 - 14V maximum.

How would you reduce the voltage at the track outputs.

[Suzie]

It's also possible to build a feedback controller (some might prefer to call it "feedforward" because it's a type of positive feedback) that does not sense the EMF directly while the motor is "flywheeling". There is no interruption of the supply and therefore no buzzing or motor heating.

 

The method consists of sensing the motor current and increasing the applied voltage to cancel most of the voltage drop across the internal resistance of the motor. That means the motor EMF is always close to the EMF (speed) demanded by the controller.

 

It actually works very well but there are a couple of snags: 1) the resistance in the circuit between the controller and the motor has to be fairly constant (no dirty track or big voltage drops in the wiring); and 2) the controller has to be set to match the internal resistance of the motor. In practice this might mean three settings for high, medium and low resistance motors.

 

Also, double heading might be a problem, but you can always turn the feedback off when required.

Sounds like a constant current type of drive which will give driveabilty similar to a variable resistance controller.

[AndyID]

Sounds like a constant current type of drive which will give driveabilty similar to a variable resistance controller.

 

It's quite different - more like constant speed. The track voltage increases as the current drawn increases. You can run into the buffer stops and the wheels just keep turning.

 

If a DC motor had no internal resistance the speed would be dictated entirely by the applied voltage, regardless of the load. This circuit effectively takes account of the voltage drop across the motor's internal resistance and increases the applied voltage to cancel out the drop. The voltage compensation is proportional to the current.

 

You have to be careful not to get too greedy. If you apply too much "feedforward" the system gain exceeds unity and it becomes unstable.

 

BTW, this is a really old idea. It's been around for about 50 years.

[paul 27]

Is it just a case of fitting a resistor in line with the feed to track terminals of the controller bearing in mind it is a feedback controller.

[AndyID]

Is it just a case of fitting a resistor in line with the feed to track terminals of the controller bearing in mind it is a feedback controller.

 

It's hard to say without knowing how the Gaugemaster controller works, and they don't reveal much. I suspect a series resistor won't help at all but I don't think it would do any harm to try it and see what happens.

 

What type of motors do your trains have? Are any of your trains equipped with DCC decoders?

[kevinlms]

It's hard to say without knowing how the Gaugemaster controller works, and they don't reveal much. I suspect a series resistor won't help at all but I don't think it would do any harm to try it and see what happens.

 

What type of motors do your trains have? Are any of your trains equipped with DCC decoders?

Wasn't a lamp wired across the motor the way recommended to swamp the feedback? I think the suggested lamp was that sometimes used for loco lights. Although I suspect you'd need to be careful of potential heat & of course the light from 'escaping'. This was for the Portescap motors. I think I'd prefer to keep the feedback, rather than those options.

[paul 27]

It's hard to say without knowing how the Gaugemaster controller works, and they don't reveal much. I suspect a series resistor won't help at all but I don't think it would do any harm to try it and see what happens.

 

What type of motors do your trains have? Are any of your trains equipped with DCC decoders?

As I have mentioned in previous posts the only problems I have had are with 5 pole motors using DC  feedback no decoders.

[DCB]

Just wondering why you want a feedback controller ?  

 

Unless you have really bad track on a micro layout In would suggest a Voltage controlled power unit and a Relco.

I have a diode based controller which cost about £2 as well as Hammant and Morgan and OnTrack power units which deliver a set voltage depending on the control know setting from 0 - 14 volts.  Feedbacks usually deliver pulses of around 20 volts, great for starting tired mechanisms lumpy badly adjusted mechanisms which really need a service but noisy, whereas voltage control units are silky smooth and quiet

 

My 28XX Hornby Dublo chassis can motor 50 to 1 gears runs at constant speed (wheel rotation) give or take a few RPM uphill down dale whether light engine, pulling 20 coaches or crashed into the buffer stops on voltage control power, and Chinese Hornby /Bachmann are pretty similar.   

[81C]

Just wondering why you want a feedback controller ?  

 

Unless you have really bad track on a micro layout In would suggest a Voltage controlled power unit and a Relco.

I have a diode based controller which cost about £2 as well as Hammant and Morgan and OnTrack power units which deliver a set voltage depending on the control know setting from 0 - 14 volts.  Feedbacks usually deliver pulses of around 20 volts, great for starting tired mechanisms lumpy badly adjusted mechanisms which really need a service but noisy, whereas voltage control units are silky smooth and quiet

 

My 28XX Hornby Dublo chassis can motor 50 to 1 gears runs at constant speed (wheel rotation) give or take a few RPM uphill down dale whether light engine, pulling 20 coaches or crashed into the buffer stops on voltage control power, and Chinese Hornby /Bachmann are pretty similar.   

 

Relco's had a habit of burning holes in Hornby Locomotive wheels, I use the same controllers with Gaugemaster HF track cleaners with no issues on all types of motors warning  don't use coreless motored loco's with the HF cleaners they will cook.

Have you thought to contact Gaugemaster to see if they have a solution they have always been very helpful.

[paul 27]

Relco's had a habit of burning holes in Hornby Locomotive wheels, I use the same controllers with Gaugemaster HF track cleaners with no issues on all types of motors warning  don't use coreless motored loco's with the HF cleaners they will cook.

Have you thought to contact Gaugemaster to see if they have a solution they have always been very helpful.

Yes they have been back to Gaugemaster  and been serviced as they have a lifetime guarantee I asked if they could do any

thing as  it is only with certain motors but they said they could not adjust the settings or feedback.