Frog Polarity switching: Relay or Microswitch ?

[AndyID]

10 hours ago, Izzy said:

As these opto couplers/isolators seem to perhaps be the simple answer to a lot of users problems re signal interference could I ask, as an electrical numpty, how these are wired and the specs that should be used? For I see that forward voltages range from 1.3v onwards but that 6/8v ones are quite expensive. Are they used like resistors/diodes, just inserted in-line, or is it a bit more complicated than that? Polarity sensitive etc.

 

thanks,

 

Izzy

 

This works quite well.

 

https://www.rmweb.co.uk/community/index.php?/topic/121357-no-twitch-servos/&tab=comments#comment-2668509

 

[Junctionmad]

heres my actual circuit  R1 = 680R , R2 = 1K , R3 suits LED  ( this is to support a signal lamp Led) 

Edited August 5, 2020 by Junctionmad

[AndyID]

Just for giggles and grins I tried putting the optocoupler (4N25) right at my pulse source rather than at the servo end. I can't make the servo twitch at all with my extremely aggressive interference generator, and the servo is at the end of a 16 foot long, loosely twisted pair

 

So, if the manufacturers of servo driver cards for model railways would kindly insert some inexpensive optocouplers* into their designs I suspect most of the erratic servo behavior some people experience would disappear.

 

But don't take my word for it. Run the tests for yourself.

 

 

*4N25 -  $22.40 per 100

 

 

Edited August 6, 2020 by AndyID
more info

[Crosland]

On 03/08/2020 at 01:43, AndyID said:

 

Agreed. I think power sequencing has nothing to do with the problem. The issue is that servos are edge-triggered devices with very little noise immunity at their high-impedance signal inputs.

 

I would agree with that. Power control (removing th control pulse) is for curing chattering servos that are trying to drive against the endstop of whatever they are connected to, and draw a lot of current. This doesn't work for digital servos that continue to drive even when control is removed. It can also be cured by using flexible linkages such as omega loops.

 

Power sequencing can be useful at start up when a lot of servos may be moving at the same time. This can result in the supply being overloaded.

 

On 05/08/2020 at 15:53, Junctionmad said:

The opto wired to active pull up with a resistor for pull down , is the best option as the default unpowered position will be low input 

 

Ah, but 50% of MERG members will swear blind you need pull-ups to set the inactive level

 

15 hours ago, AndyID said:

Just for giggles and grins I tried putting the optocoupler (4N25) right at my pulse source rather than at the servo end. I can't make the servo twitch at all with my extremely aggressive interference generator, and the servo is at the end of a 16 foot long, loosely twisted pair

 

So, if the manufacturers of servo driver cards for model railways would kindly insert some inexpensive optocouplers* into their designs I suspect most of the erratic servo behavior some people experience would disappear.

 

Are there two schools of thought here? DId you keep the servos isolated, with their own supply, or common the 5V and Gnd as per junctionmad's solution?

 

[AndyID]

1 hour ago, Crosland said:

 

Are there two schools of thought here? DId you keep the servos isolated, with their own supply, or common the 5V and Gnd as per junctionmad's solution?

 

 

In the case where I have the optocoupler adjacent to the pulse source the servo supply is independent and floating relative to the pulse source supply. I'll reconfigure my rig  to power the servo from the pulse source end and see if I get similar results.

 

BTW, my pulse source is just an ATTiny44A programmed to produce long and short pulses continuously. Servo position is determined by selecting which pulse by a switch. I'm not powering down the servo at all which I think is a more severe test. The optocoupler's output pulls servo input to servo zero volts and a pull up resistor pulls it high to servo +5 supply.

[AndyID]

I think my interference source was a bit more conducted than emitted which isn't entirely fair. I sorted that out and I get much better results with the optocoupler at the servo end rather than the pulse source end.

 

I tried three configurations.

 

a) Local power source at the servo.

b) Power fed from the same PSU as the pulse source over two wires (total of four wires including signal and return).

c) Power fed from the same PSU as the pulse source but using one wire of the twisted pair to supply +5V to the servo (total of three wires - twisted pair and a separate 0 volt wire.

 

Bottom line - it didn't make any difference

 

a, b and c all seem to be equally robust. Despite running mains feeds from "Mr Sparky" (my fluorescent tube bench light) in close parallel with the cables to the servo I could not make the servo do anything unexpected. But I could make it twitch when I moved the optocoupler to the pulse source end.

 

So, putting optocouplers at the controller does seem to help but it's better to put them close to the servos if possible.

 

EDIT: I forgot to mention my cables are more like 32 feet long. I was counting the span between my hands as three feet rather than six - doh!

 

 

Edited August 7, 2020 by AndyID
more info

[AndyID]

On 07/08/2020 at 16:27, AndyID said:

Bottom line - it didn't make any difference

 

a, b and c all seem to be equally robust.

 

 

I spoke too soon.

 

I've set up my test rig with pluggable breadboards to allow me to more readily change configurations and really see what I'm doing. I have a test where the three wire configuration (version c) fails every time. The four wire configuration (b) and the local PSU configuration (a) never fail using the same test setup.

 

To be sure I wasn't kidding myself I went back and forth several times between b and c taking care not to change anything else and the results were always the same.

 

I'm not going to post any more results in this tread as this is all a bit "off topic". Future updates will be in the "No Twitch Servos" topic.

https://www.rmweb.co.uk/community/index.php?/topic/121357-no-twitch-servos/page/2/

 

[NFWEM57]

On 20/07/2020 at 17:49, DGO said:

OK so I've decided to come back to model railways after a short 40 year gap, I've also decided I'm going to switch my turnouts with servos though I've yet to decide on exactly how. But then I come to controlling the polarity of the frog. So as I see it I can either switch the polarity using a relay at the same time as I send the command to the servo (or use sensors to tell the system the servo is moving and change)  or I can stick one or two micro switches on the other end of the servo arm to the turnout and use that to switch the polarity, Are there any clear advantages or disadvantages to either method ?

 

Currently my thinking is that I'd use a pair of microswitches to control the feed, that way whilst the turnout is actually moving the frog will be isolated completely 

Tried microswitches with MERG mounts.  Always having to adjust. In the end opted for ESU servo modules with ESU extensions which have 2 relays per channel.  Delay in relay switch over built in and selectable.  No twitching issues. Dedicated supply for ESU modules.  Newer servo module has bounce effect for semaphore signals and 8 channels.  Expensive yes, but so is time when I could be doing other things on the layout or rolling stock.   In comparison to DCC ready  tortoise or equivalent at circa £30 a point, 8 channels of ESU control with metal gear servos and home made aluminium mounts is around £110, so £14 a point.  You can spend a fortune in time to save a few pounds. 

[melmerby]

8 hours ago, NFWEM57 said:

   In comparison to DCC ready  tortoise or equivalent at circa £30 a point, 8 channels of ESU control with metal gear servos and home made aluminium mounts is around £110, so £14 a point.  You can spend a fortune in time to save a few pounds. 

Or you can use MTB MP-1 point motors & Digikeijs DR4018 at £19 (or better) per point with frog switching included (8 channels and no home made mounts needed.)

[KingEdwardII]

9 hours ago, NFWEM57 said:

8 channels of ESU control

An alternative at a better price per channel is the Chesterfield Arduino based DCC Servo Decoder, which has 16 channels (i.e. up to 16 servos), for £28:

 

https://chesterfield-models.co.uk/product/servodecoder/

 

I use this for semaphore signal servos. Has semaphore bounce.

 

I have DR4018 and MTB MP1 motors for turnouts. Simple to mount and reliable in operation.

 

Yours,  Mike.

[WIMorrison]

I found this company and I know people that who have used them are very happy - especially as it is cheap.

 

ArCoMoRa

[melmerby]

44 minutes ago, KingEdwardII said:

 

I have DR4018 and MTB MP1 motors for turnouts. Simple to mount and reliable in operation.

 

Yours,  Mike.

I've used an Arduino Nano to drive some MTB MP1 motors, but it does mean making your own.

 

[Harlequin]

I've used Megapoints parts to control MTB MP1s. Very happy with it but that is not so cost effective as some of the combinations mentioned above, I think.

 

Of course, there's a lot to be said for a simple "mimic" panel with switches controlling the turnouts and indicating their lie. MTB MP1s are particularly well suited to this method of control and then the cost per turnout is very favourable.

 

It's just another option - and possibly a more relevant one in the (non-DCC) forum!

 

Edited December 14, 2021 by Harlequin