About the TD feed and S-class messages - documentation and completeness of coverage

[Aleksa Jakovljevic]

Hi all,

I've been having a hand at trying to process the TD feed and decided to try to include all the data from both the C-class (TD berths) and S-class (signals, track section occupation, point position, &c) messages in a nicely formatted page (i.e. explained simply, I'm using PHP to dynamically update CSS formatting on a custom font to show TD berths, track occupancy, &c).

As a learning experience I focused on a small area where I had good documentation (signalling diagrams and/or panel pictures, among others) and settled on the Wilton Jn. area just west of Salisbury station. I've attached a couple of screenshots from my implementation: sections for which I am pretty confident in the accuracy of the on-wiki mappings are in yellow, there's one section in green whose occupancy can be mostly deduced from the presence of a describer in the berth and the occupancy of adjacent sections, everything for which I have no data is in grey - in addition there are a few test bits at the bottom.

The TD berths all step-up as expected. However, I've tried following the decoding scheme from the wiki for the S-class messages but realise that this data - beyond some TCs on the panel which appear to have been combined into a single data bit - appears to be incomplete. I don't know if this is related to the fact that there are several bits which never seem to be updated either. I figure this probably isn't a priority use case of this data, but while I'm at it, has anyone encountered this issue previously? Does anyone know if there's more complete documentation available somewhere else?

Cheers and thanks for any answers which help clarify this,

AJ

[Tom Davis]

Hi Aleksa,

I'm also interested in the Salisbury area. I have tried running AI over my logs to see if I can deduce any correlation between berth steps and signalling updates. I've published the results here:

https://groups.google.com/g/openraildata-talk/c/FlG-ks_DAvg/m/MR_Rr63NAQAJ

Regards,

Tom

[Juhani Pirttilahti]

Hi Tom and Aleksa

I did the data analysis on Salisbury TD to create that documentation, but it was several years ago and things may have changed since. Please note that the documentation on wiki is uncontrolled due to various reasons.

In this area I actually had some panel pictures for comparison and the track circuit names are as was indicated on the panel years ago. There are automatic signals and the track circuit bits typically represent the track circuit just after the signal (or a combination of track circuits, in some cases). I think this has been discussed on the other thread here, too, and it is well explained.

When calculating correlations one should take into account that in some control areas signals change aspects when the train has *completely* passed a signal, i.e. last axle releases the track circuit. Modern systems replace the signal when the first axle has passed over to the next track circuit or passed the next axle counter. I think the correct UK term is last wheel signal replacement. It takes some time, depending on the train.

While working out the area I had my own correlation calculations that took the delay (we're talking from 5 to 30 seconds, shunting signals do take longer) into account. The calculated correlations helped me to connect bits to signals. I had no other documentation than the panel pictures and several months worth of data from the TD Feed.

I think there are similar locations with last wheel signal replacement under the ZB train describer. I remember doing some analysis for ZB, too.

Hopefully this clarifies some things.

Juhani