Moving from N to OO, L shape 12ftx9ft ideas

[aleopardstail]

Thanks for the vote Nearest B&Q to me is 15 miles so I use whatever I can find in my local hardware store, I had heard about the U-method, nice to see it in the flesh! I eliminated the micro-switches by using an extra pole on the panel mounted switch that operates the point, much simpler IMOP, sometimes if the point is far away I use a relay operated by the same method.

How are you getting on with servo-twitching etc, there are many lurid stories all over the net but I have not personally experienced any problems yet.

not having problems with them, largely because I've written my own software that makes adjusting the final positions quite easy, helped by the thinner piano wire and length it has to flex providing a nice spring. software is adjusted so when the point is thrown the servo can comfortable hold the position - note also it has had the horn clipped so it never binds on the channel. I'm using a Raspberry Pi Pico, largely because I wanted to learn how to work with one, a bog standard Arduino could drive the same controller board pretty easily.

I suspect the usual Adafruit type drivers could be adapted easily enough - the trick is to make it so you can adjust them all individually in terms of range of movement and have the two end points nicely defined. I've not tried the commercially produced point servo controllers but I think they all have the ability.

I suspect most of the issues people have with them are either down to electrical noise in overly long wires (better to keep the servo wires short and have a localised driver that can have a nice shielded line going to it, or to use shielded cable for the power and signal line), or them just not being quite adjusted right so the thing is constantly fighting itself.

I suspect not using the blade to stock rail contact for anything electrical is a good idea as well, then the servo really just needs to make the two contact lightly enough to work, not try and weld them together.


and yes the switch at the control source is a lot easier for things like frog switching, unless you have a slight delay (insufrog points don't care though)

[fourtytwo]

I also write my own software (bare metal assembly on PIC16F) using interrupt routines driven by timers, latency is tiny and there is almost zero jitter (I have never seen any).
There is a problem with noise immunity when long lines are used but I use a heavy gauge ground bus around the layout and a bit of local filtering in vcc for each servo to help (8R2 @ 1000uF) that reduces the peak current (in the all important ground wire) from 1A to ~250mA, probably overkill but I didn't want problems as the quantity of servo's grew (semaphore signals). I use the EEPROM of the PIC to store stroke positions and use a simple menu via a laptop USB port to set them up.

You can avoid momentary shorts during frog switching even with live frog points by doing the so called "dcc friendly" modification to the points (even if your using DC like me).

[aleopardstail]

I also write my own software (bare metal assembly on PIC16F) using interrupt routines driven by timers, latency is tiny and there is almost zero jitter (I have never seen any).
There is a problem with noise immunity when long lines are used but I use a heavy gauge ground bus around the layout and a bit of local filtering in vcc for each servo to help (8R2 @ 1000uF) that reduces the peak current (in the all important ground wire) from 1A to ~250mA, probably overkill but I didn't want problems as the quantity of servo's grew (semaphore signals). I use the EEPROM of the PIC to store stroke positions and use a simple menu via a laptop USB port to set them up.

You can avoid momentary shorts during frog switching even with live frog points by doing the so called "dcc friendly" modification to the points (even if your using DC like me).

amazing what these cheap as chips chips can do these days. noise in long lines is always a problem, Dingleberry is tiny but for the loft layout (if it ever happens) it will be localised controllers to keep the lines short with something like a RS485 bus to help with the noise issues

local filtering is also very good, when you say "heavy gauge" ground wire, what sort of wire are you talking about? last time I did a larger loft layout I used some spare cable from fitting a ring main in the house, probably complete and utter overkill but it worked

[fourtytwo]

amazing what these cheap as chips chips can do these days. noise in long lines is always a problem, Dingleberry is tiny but for the loft layout (if it ever happens) it will be localised controllers to keep the lines short with something like a RS485 bus to help with the noise issues

local filtering is also very good, when you say "heavy gauge" ground wire, what sort of wire are you talking about? last time I did a larger loft layout I used some spare cable from fitting a ring main in the house, probably complete and utter overkill but it worked

Absolutely, when I think back to my days of relays and primitive logic controlling signalling in the 60's and 70's I am amazed.

The wire is 1.2mmsq stranded, some USA stuff I happen to have hundreds of feet! Not that big but adequate for a bedroom sized layout, come to think of it I also used it in a large loft, something like 30ft square.

Anything balanced like RS485 is a good idea but remember it doesn't have infinite common mode voltages so it's still possible to glitch it, galvanic isolation is the best way to go like opto's for baseband or something transformer coupled if suitably modulated (FSK etc).

The most communicating processors I have ever had on a single layout was I think 15 or so using a high voltage (15V) single ended CSMA protocol (Ethernet like) with a re-transmission packet protocol to fix errors.