I would plan for ”normal blocks”. However, using a PC software like TrainController Gold, I only use 2 sensors in such blocks as follows:

ENTRY: Sensor 1 after 0 cm
BRAKE START: Sensor 1 after X cm
STOP: Sensor 2 after 0 cm

Also, my PC software takes care of your second challenge; Long trains. You can define train length and block length in the software. The software then reserves the blocks and routes (turnouts) required to fit the train!

Thanks both for your answers:

Follow-up questions for Jean...

- Not sure to understand "staging block is adjacent to other blocks"
- do you mean a configuration where the staging block & normal block are converging into a switch?
- RR will always send the train from the normal block first and the staging block after in automatic mode, correct?
- Does that mean that, ideally, only staging blocks should converge into each others??

- I seem to remember one of your post saying that in case of short-circuit or down electricity, the staging blocks must be manually cleared and reloaded?
- Can you expand on that please - seems like a pain, does it happen often in practice?
- Will this be different for normal blocks?

- My 8 meter length is a dual track (one up, one down) transition between layers, so not a 'long term storage' area as the trains will just travel through it. My goal is to minimize the time/delays between trains in the transition.
- Seems like Staging Blocks are more efficient (can pack more trains if enough sensors), but slower because of coasting between the stop points + it can handle 'long' trains.
- Normal Blocks will mean that I have to account for the 'Long' train, meaning longer blocks so less efficient & maybe the higher speed between blocks might not help because further apart.

Looks like I am leaning towards Staging Blocks - anything else you think I should know to avoid pitfalls?
Thank you again so much!

Hi Roland,

It helped me to understand the logic - thanks for your answer!!!

Thanks Jean!

Helps me a lot to understand all this!

So, to better answer my own questions and get some definitive answers on the above, I decided to create a simple layouts to better understand the relationship between RocRail, sensors type and locations, and locomotives.
What I've found is (I think) worth sharing as I was not able to find this information upfront.

This is my test layout, trains are only going counter-clockwise.
The CS3 is connected to a 6-Boosters IoTT Silver Hat (in the center) which is powered by 2 x 18.5V switching mode power supply.
I only use 2 out of the 6 for this test (one powering the left side and the other powering the right side).
The turnout switching is done through servos (that's why the whole track is raised a bit), except for the crossovers that are using the normal Marklin electro-magnets.
The PCBs on the right side where all the wires go is an IoTT Green Hat that controls up to 16 servos and gather information from 32 sensors (isolated C-Track rail).


This is the Wintrack screenshot (green is "enter", cyan is "pre2In" and fuchsia is "in") for the different blocks.


This is the RocRail screenshots showing the various blocks (relating to the above sensors).


Here is what I found:

For the block B01A1 using 3 sensors (the fuchsia middle one is the 'pre2in', I am able to get very accurate stop using a 94mm track (FB3) for 'in'. This is based on the 'pre2in' (FB2) sensor being 172mm long with another 94mm between them. The 'pre2in' reduces the speed down to minimum and the 'in' stops it quickly. This works well for the MFX and FX decoder that have a F4 function to cancel the deceleration delay, but I found out that some FX decoders actually do not have this F4 function available, hence they are difficult to stop accurately. Also, there are some visible and abrupt speed change when the loco is going through the sensors - this is not an issue for me in hidden yards, but I'd like to get smoother deceleration when visible like stopping at a station.

Blocks with only 2 sensors ('Enter' and 'In') can also be relatively accurate for the stop if using BBT where RocRail calculate the deceleration in 10 steps based on the available block length. The fact that it 'learns' and adjust the stop and interval time each time it is being used is also nice, but it takes a few passes to get it right. I will be using this for the visible part of the layout as it presents a very smooth deceleration, but also need to find a solution for those FX without F4.

Finaly, I also tested a Staging Yard and I am not sure to yet understand everything about it and how it exactly works, namely that right now, the train briefly stops at the last sensor (F14A1) but then directly moves out to go to another block automatically. This can be stopped by 'closing' the Staging Yard but it then needs to be re-opened manually... need to do more reading around how to get to the bottom of it. As Jean mentioned, I can see that RocRails prefers to send a train towards the normal block B01A1 first if empty, and only to the Staging Block S01A1 if the previous is occupied - it does not seems to be random between the two.

Other things that I learned in the process:

- Do not place a sensor near/before a turnout (remember, I use slow servos to switch them) as some time is by RocRail to organize all the turnouts before the train gets there. Either I have to plan some distance or add some block starting delay to avoid running over the 1st turnout while it moves.

- The direction of the loco is important to be sure that the wheels without the rubber bands are in front. If not, the sensor using the isolated rail does not see a short-circuit till the back of the locomotive is over it. This leads to not stopping in the right spot and running over, especially with short sensor rails near turnouts (FB7 & FB9 in my case).

Hope this helps someone else though their own discovery.
Thanks for reading.

Edited by user 08 August 2025 02:39:46(UTC)  | Reason: Not specified