Megapoint Controllers

[AndyID]

It is very hard to define the start-up conditions of a microprocessor based servo driver. Unless you arrange for the power to be applied to servo after the microprocessor is running there will be an undefined state on the servo control wire until the processor has completed boot and the firmware has started generating the pulses. These initial pulses may be appear very long or very short to the servo and cause it to twitch at power up. The only way I know to mitigate this is to make the controller with a nano-power processor that can be kept in sleep mode when power is removed so that it does not have to perform a reset sequence and can therefore generate pulses straight away - but this is an added complexity that can have knock on side effects.

 

Just using a digital servo gets around this 'twitch at startup' problem because the signal processing in the servo ensures that it ignores any initial out-of-range or wildly different sized pulses until a stream of similar length pulses start to be generated when the servo will start to respond only after the controller has settled. Any spikes induced in to the control wire will be ignored too. Digital servos have a lot of advantages.

 

Yes digital servos still have the potentiometer, but the output of this is digitised to make a series of steps, and the servo only moves when it perceives that the incoming pulse has changed sufficiently to move to the next step - and it does this with a distinct tick. I don't see the tick as being a big problem - careful adjustment does appear to minimise it.

 

Digital servos do appear to be just as cheap as their analogue equivalents now - it is hard to tell the difference in price between the standard analogue TowerPro SG90 and the digital with carbon fibre gears but otherwise equivalent TowerPro SG92R. 

 

I think it's worse than that (with the analog servos anyway).

 

The ones I have (SG90) twitch at power up even with the input tied hard to ground, so there isn't anything the external controller can do about it.

 

If the digital servos don't have that problem, that would be a good reason to use them (as ScRSG suggested).

 

The I/O pins of the micro-controllers I'm familiar with default to input on power-up. I think it should be possible to initialize the controller without it sending spurious signals to the servo. If the controller updates EEPROM with every state change it should also be possible to restore the outputs that were present before power-down. I can see I'm going to have to get some digital servos to find out if that's really true

 

EDIT: In my case the state memory is determined by the change-over switches I use to control the turnouts. The controller is a single micro-controller programmed to produce two pulse trains. Change-over switches direct either of the pulse trains to the servos. The two pulse widths produce a large swing angle at the servos - approaching 180 degrees. Throw adjustment is purely mechanical.

 

Using a large servo angle essentially locks the switch blades against the stock rails. The tie-bar has almost no mechanical advantage over the servo.

Edited October 9, 2016 by AndyID

[Gordon H]

There is at least one digital servo that uses a rotary magnetic encoder rather than a pot, specifically the Hitec HS-M7990TH. I used one a while ago on a works project to prove to some doubters that commercially available servos could do the job. 

[Gordon H]

The ones I have (SG90) twitch at power up even with the input tied hard to ground, so there isn't anything the external controller can do about it.

 

Evidence points to a pull-up resistor to supply being more effective than a pull-down to ground in the majority of cases.

[AndyID]

Evidence points to a pull-up resistor to supply being more effective than a pull-down to ground in the majority of cases.

 

Thanks for that Gordon. I just tried it and it works (sample of one of course). There is still a very slight "tick" at power-up, but nothing like the major excursion when the input is held low.

[AndyID]

I've been observing my servos quite closely, and they do sometimes "tick", but it appears to be position dependent. I did a bit of experimenting and at one nominal extreme (2 ms pulse width), they do tend to dither slightly. There is no appreciable movement of the horns, but I can see the motor rotating through a few degrees every few seconds. At the other extreme (1 ms pulse width) I don't detect any dithering (ticks) at all.

 

It's likely that the at one extreme the servo's internal monostable has enough jitter (variation in pulse width from cycle to cycle) to make the servo dither. It might not be impossible to modify the servo to increase the "dead band" and prevent this, but I don't think anyone would want to attempt that.

 

In my case, because of the method I'm using to drive the servos, there isn't much I can do about it, and it doesn't particularly bother me. I believe the MERG servo controller can be configured to stop sending pulses after an interval. That should eliminate the "ticks". I don't know if the Megapoints controller has a similar feature. The disadvantage associated with interrupting the pulse train like that is, should the servo happen to be triggered by spurious noise, it will move to an indeterminate position and stay there until the servo starts receiving pulses again. Also, depending on the nature of the load, the servo may not stay in the desired position.

 

(Note: The above only applies to analog servos - specifically the SG90 type.)

[Enterprisingwestern]

There is at least one digital servo that uses a rotary magnetic encoder rather than a pot, specifically the Hitec HS-M7990TH. I used one a while ago on a works project to prove to some doubters that commercially available servos could do the job. 

 

Presumably, for model railway purposes, that one doesn't cost thirty bob?

 

Mike.

[Grovenor]

Presumably, for model railway purposes, that one doesn't cost thirty bob?

 

Mike.

Its certainly not what I would use for a point motor!

Hitec HS-M7990TH - 7.4 volt servo

 

Price  £134.99 Each

Regards

[Jon Fitness]

I've been observing my servos quite closely, and they do sometimes "tick", but it appears to be position dependent. I did a bit of experimenting and at one nominal extreme (2 ms pulse width), they do tend to dither slightly. There is no appreciable movement of the horns, but I can see the motor rotating through a few degrees every few seconds. At the other extreme (1 ms pulse width) I don't detect any dithering (ticks) at all.

 

It's likely that the at one extreme the servo's internal monostable has enough jitter (variation in pulse width from cycle to cycle) to make the servo dither. It might not be impossible to modify the servo to increase the "dead band" and prevent this, but I don't think anyone would want to attempt that.

 

In my case, because of the method I'm using to drive the servos, there isn't much I can do about it, and it doesn't particularly bother me. I believe the MERG servo controller can be configured to stop sending pulses after an interval. That should eliminate the "ticks". I don't know if the Megapoints controller has a similar feature. The disadvantage associated with interrupting the pulse train like that is, should the servo happen to be triggered by spurious noise, it will move to an indeterminate position and stay there until the servo starts receiving pulses again. Also, depending on the nature of the load, the servo may not stay in the desired position.

 

(Note: The above only applies to analog servos - specifically the SG90 type.)

I've never had much luck with that function of the MERG boards. As you point out, servos will twitch as spurious signals are picked up (for instance, if a loco with poor pick-ups sparks it's way past!) but will stay where they twitch to until a new signal is sent (usually by operating the servo's switch). On our layout, if I didn't notice this going on, it would quite quickly tie a signal linkage in knots so I removed the option from the board. At least then, when the twitches occurred, the servos immediately twitched back to where they should be instead of trying to steadily rotate a signal through 360 degrees!

MERG boards and the SG90s seem more susceptible to this. I built a little O gauge test track with 4 servo operated points all driven by a MERG board. Any loco with dodgy pick ups has the points whizzing about all over the place, so the lot's coming off to be replaced with a Megapoints board and some white Hobbyking servos to see if there's an improvement...

JF

[AndyID]

I've never had much luck with that function of the MERG boards. As you point out, servos will twitch as spurious signals are picked up (for instance, if a loco with poor pick-ups sparks it's way past!) but will stay where they twitch to until a new signal is sent (usually by operating the servo's switch). On our layout, if I didn't notice this going on, it would quite quickly tie a signal linkage in knots so I removed the option from the board. At least then, when the twitches occurred, the servos immediately twitched back to where they should be instead of trying to steadily rotate a signal through 360 degrees!

MERG boards and the SG90s seem more susceptible to this. I built a little O gauge test track with 4 servo operated points all driven by a MERG board. Any loco with dodgy pick ups has the points whizzing about all over the place, so the lot's coming off to be replaced with a Megapoints board and some white Hobbyking servos to see if there's an improvement...

JF

 

Hi Jon,

 

I will be very interested to find out how you get on with the other servos. The problem with the SG90s is that they have lousy noise immunity when the signal input is low (it's much better in the high state.) That means any wheel contact arcing is liable to trigger a phony string of pulses that drive the servo potty.

 

I've successfully increased the noise immunity of the SG90s by means of a "trick". You might care to give it a shot before you scrap the whole lot.

 

The trick is to elevate the ground voltage of the servos relative to the controllers ground. It doesn't take much to make a difference. What I do is insert a diode in the servo's ground return so that it is about 0.7 volts higher than the controller's ground. In theory this might be a bad idea because it is reverse biasing the input to the controller chip in the servo, but I haven't managed to destroy any servos yet.

 

Andy

[Vecchio]

I've never had much luck with that function of the MERG boards. As you point out, servos will twitch as spurious signals are picked up (for instance, if a loco with poor pick-ups sparks it's way past!) but will stay where they twitch to until a new signal is sent (usually by operating the servo's switch). On our layout, if I didn't notice this going on, it would quite quickly tie a signal linkage in knots so I removed the option from the board. At least then, when the twitches occurred, the servos immediately twitched back to where they should be instead of trying to steadily rotate a signal through 360 degrees!

MERG boards and the SG90s seem more susceptible to this. I built a little O gauge test track with 4 servo operated points all driven by a MERG board. Any loco with dodgy pick ups has the points whizzing about all over the place, so the lot's coming off to be replaced with a Megapoints board and some white Hobbyking servos to see if there's an improvement...

JF

Looking forward to your comparison! keep us posted!

Vecchio

[Vecchio]

Hi Jon,

 

I will be very interested to find out how you get on with the other servos. The problem with the SG90s is that they have lousy noise immunity when the signal input is low (it's much better in the high state.) That means any wheel contact arcing is liable to trigger a phony string of pulses that drive the servo potty.

 

I've successfully increased the noise immunity of the SG90s by means of a "trick". You might care to give it a shot before you scrap the whole lot.

 

The trick is to elevate the ground voltage of the servos relative to the controllers ground. It doesn't take much to make a difference. What I do is insert a diode in the servo's ground return so that it is about 0.7 volts higher than the controller's ground. In theory this might be a bad idea because it is reverse biasing the input to the controller chip in the servo, but I haven't managed to destroy any servos yet.

 

Andy

Andy, just to understand better - you are using the same power source for DCC and for the servo electronics or different sources?

Vecchio

[AndyID]

Andy, just to understand better - you are using the same power source for DCC and for the servo electronics or different sources?

Vecchio

 

Hi Vecchio,

 

I'm not using DCC, but that's not important. What matters is that the ground reference of the servos should be at a higher potential than the ground of the controller that's sending pulses to them. As I said, it is a "trick" to improve the apparent noise immunity of the servos without having to modify the servos themselves. It  seems to help in some cases, but I cannot guarantee it will help in all cases.

 

Andy

[Jon Fitness]

Well, I'll try it but it won't be for a week or so as I'm away from tomorrow. I won't be worried if I destroy an SG90 or 2!

I already have 2 megapoints boards on our main layout, operating a mix of SG90s and HK servos (micro and sub-micro) but they only control signals. I get a bit of a twitch on start-up (not me, the servos!) but no noticeable disturbance during normal running. Of the other 8 MERG boards on the layout only 2 seem to suffer badly from interference. It certainly lets you know which locos have c**p pickups!

All the points on our layout are solenoid operated and they don't seem to have any effect but I did have a relay operating from a microswitch on one of the points and that did spike the servos when it was de-energised! I had to fit a diode to that to stop the spike from the discharge.

JF

Edited October 11, 2016 by Jon Fitness

[leopardml2341]

Please read in a light hearted manner, perhaps we need a version of this for model railways?

 

https://www.google.co.uk/url?sa=t&source=web&rct=j&url=http://www.visus.pt/mirrors/VIVOTEK/MARCOM/2%2520Value%2520official%2520release/Official%2520Release/ProductAnnouncement_MD7530_MD7560/BS_EN_50155-2007.pdf&ved=0ahUKEwiT9MaZutLPAhXBDsAKHd3sBv0QFghRMAM&usg=AFQjCNH4TZjaNVhjXZJds0cUy2YWtuGXmQ&sig2=lvGl2czdytAq7uzGDihLBw

 

 

Now more seriously, there are some useful techniques that are used, try not to route signal level and power cables in parallel bundled routes.

 

Try screened cable for the servos, try a small value capacitor between the servo control line and return line.

 

Rgds,

Edited October 11, 2016 by leopardml2341

[Gordon H]

Its certainly not what I would use for a point motor!

 

 

I didn't say that is what I used it for, just that such a thing existed. 

I was using it as an actuator on a heavy duty camera panning mechanism.

[Revolution Mike B]

I saw Dave Fenton this weekend at Warley and I'm glad to see the poor bloke is now becoming overrun with orders. I've long been championing his system and as someone that's installed, dismantled and reinstalled again over the last year a fair few control boards and motors, it's safe to say that it does what it says on the tin.

 

The only reason I didn't use his stuff on Oak Road is that I wanted something I could plug and play quickly with no messing around fitting frog switches and extra wires meaning Cobalt IPs were they way forward. That said, I'd still have preferred to use Megapoints due to the quieter action and better adjustment.

 

We've found that if money isn't the driving force, the Auto Frog solution makes fitting the stuff a lot easier, especially for DC and now that Dave is working on some stuff to avoid the '12' board, things should get a lot more interesting.

 

If we could get a bracket that was like the Peco Smart Switch one then I think things would really take off because I'm not convinced with the ally channel.

 

Mike

[alibuchan]

I was never happy with the ally channel option, especially when on my first trip out all of the servos on the bottom board fell out, then this popped up in my Facebook feed.

 

http://www.alurailtech.co.uk/home.html

 

An aluminium bracket that the servo bolts on to for £2.50. Not the cheapest option but works out a cheaper option than trying to produce my own 3D printed thing. It also had channels to locate the micro-switches for throwing frogs or signals.

 

No connection other than a happy punter.

 

Was a bit gutted that after having a chat to Dave yesterday, he said I couldn't have my relay driver board because he had sold so many of them. Just before he left he threw an extra box of them in the car but said he didn't expect to sell that many. I knew I should have pre ordered them. It is due to be posted out today so can't complain really.

 

Alistair

[ScRSG]

Out of interest, regarding the twitch - is it the board or the servo which picks up the interference from a passing poor pickup loco? If it is the board then would it not be sensible to move the boards, say, into your panel rather than under the track baseboards, is it not distance from the source of the interference?

[Bulldog]

 

I was never happy with the ally channel option, especially when on my first trip out all of the servos on the bottom board fell out, then this popped up in my Facebook feed.

 

http://www.alurailtech.co.uk/home.html

 

An aluminium bracket that the servo bolts on to for £2.50. Not the cheapest option but works out a cheaper option than trying to produce my own 3D printed thing. It also had channels to locate the micro-switches for throwing frogs or signals.

 

No connection other than a happy punter.

 

Was a bit gutted that after having a chat to Dave yesterday, he said I couldn't have my relay driver board because he had sold so many of them. Just before he left he threw an extra box of them in the car but said he didn't expect to sell that many. I knew I should have pre ordered them. It is due to be posted out today so can't complain really.

 

Alistair

Another alternative servo mount

http://www.ebay.co.uk/itm/0-00-009-N-Gauge-Servo-Bracket-Point-turnout-Control-Laser-Cut-Kit-Peco-/122128749218?ssPageName=STRK:MESE:IT

[AndyID]

Out of interest, regarding the twitch - is it the board or the servo which picks up the interference from a passing poor pickup loco? If it is the board then would it not be sensible to move the boards, say, into your panel rather than under the track baseboards, is it not distance from the source of the interference?

 

With the low-cost analog servos, it's the servos themselves, and it gets worse if the cable between the servo and the controller is extended.

[Vecchio]

if you are running a sound loco or you are at a show you will not hear the servos tick. If you are in your mancave where you hear the clock on the wall ticking (should you have one in there) you will hear it. I fly RC planes since I am 18 and the servos always tick. But again - once the bird is in the air you don't hear it )

Vecchio

 

P.S.: On my newer and bigger models I use digital servos - and surprise: they tick as well....

Edited November 29, 2016 by Vecchio

[Jol Wilkinson]

I've had considerable problems with servo operation of signals, DC control ans kit built locos (no built in suppression)  - albeit with another brand of Servo controller. Two important lessons I have learned is to use good quality servos with decent on-board  electronics (SG90's aren't really good enough) and that using a Tam Valley No Buzz Servo Quieter seems to provide an answer (don't you just love the way the Yanks name things) These are available from DCC Coastal in Ipswich.

 

http://www.coastaldcc.co.uk/products/tamvalley/accessories/

 

Long servo leads with the SG90's are a problem, which is why I tried to stick with the controllers I have (they also have a very good remote set up box). The problems are now sorted but it has been a long and sometimes frustration experience.

[Revolution Mike B]

I'd just like to confirm that none of the servos we have (the cheap Hobby King ones) twitch. They key to making sure everything works as it should is getting the servo lined up so that it can move throughout the whole of its (set) travel. We've used various sizes of extention leads from 300mm to 1m, linked together in a whole manner of ways and it does exactly what it says on the tin. We still get them chattering to each other, but as mentioned above, you'll never hear it in a busy hall although it may get a bit annoying at home......just live with it.

 

For the record, we're running 84 servos on a layout that's 18' x 9' (N Gauge) with 7 analogue controllers.

Edited December 1, 2016 by scoobyra

[ScRSG]

With the low-cost analog servos, it's the servos themselves, and it gets worse if the cable between the servo and the controller is extended.

We did some experimenting today on the new large group layout we are building. The (MERG) servo control boards are mounted in the control panel, so are a max. of 20ft. away from the furthest servo. Using the wiring we were used to, which was admittedly quite thin, the servos kept twitching , quite violently. So, to cut a long story short, we doubled the thickness of the wiring of the 5v supply to the servo and they stopped twitching! This suggests that there was a voltage drop on the original wiring which was cured by this simple solution. Two servos were on this line. Would adding further servos to this cause any further problems? Effectively we have created a "bus" for the 5v servo supply. Anyone care to add to this? The control boards themselves are OK, and connecting the servo directly on to the board, it works perfectly with no twitch, which seems to have confirmed the cable length idea.

[leopardml2341]

I'd wager that many modellers without some electrical engineering knowledge are caught out by the wire Cross Section Area issue. One only has to look how it's sold to modellers - mainly by reference to its current carrying capacity and not it's CSA; which in fact determines to a large extent its voltage drop due to higher resistance per unit length.

Edited February 6, 2017 by leopardml2341