Welcome to the forum   
Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

Share
Options
View
Go to last post in this topic Go to first unread post in this topic
Offline French_Fabrice  
#1 Posted : 25 March 2020 23:08:18(UTC)
French_Fabrice

France   
Joined: 16/05/2011(UTC)
Posts: 1,474
Location: Lyon, France
Hello,

Today, I have solved (maybe with luck) a really annoying problem related to unstable S88 feedback.

I must warn you: This has been solved for my specific configuration. I cannot guarantee that if you have same problem, the same fix can apply !

Now a bit of explanation about the context.

My layout Trossingen III, which is still in progress, is now (almost) entirely wired with feedback information, except for a small sidings section which has still to be done.

My equipment is:
- Central Station 60214 (CS2) with 4.1.2 software (I know the last version is 4.2.9, but I will upgrade it later).
- One L88 (60883) connected to the CAN-Bus (CS2 internal Bus)
- None of the 16 feedback contacts of the L88 are used
- The L88 Bus1 is connected to five LDT feedback modules (RM88N), for a total of 16x5 = 80 contacts
- The L88 Bus2 is connected to five other LDT feedback modules (RM88N), for a total of 16x5 = 80 contacts
- The L88 Bus3 is not used
- The S88 plug available on the CS2 is not used
- The total number of feedback contacts in use is around 145

- I use Rocrail software on my PC (High end workstation running under recent linux distro), to drive the layout in automatic mode
- All feedback modules + L88 are joined together with CAT6 shielded RJ45 cable
- Each RM88N module is equipped with additional 'diode trick', to provide enhanced conductivity for small locos
- All feedback wires are separated from power wires. When a feedback wire crosses a power wire, it is done at right angle
- The average length of each S88 bus connected to L88 is around 8 or 9 m

A bit of history:

I'm now 1.5 year after the beginning of the layout, which is a 3-level layout.
A few months after the beginning, the base level of the layout was done and wired, and a test using automation was done with success (5 trains running automatically and driven by Rocrail). As far as I remember, I don't remind having erroneous feedback information during this test, but to be honest it could have been possible and if it was the case, it was not at a high rate...

Now one year later, after having build the helix, the ramp, the station, steam depot and parade tracks plus return loop, the number of feedback contacts has jumped from 40 to around 150, and last week the final junction was done... and the first (simple) automation test was planned on the top level.

Now:

Alas, when the 1st run started, the issues came on feedback... and it was really strange...

To make it short, when a contact section with address N was occupied (loco stopped on it), both address N AND address N+1 started to jitter !!! Not at a quick rate (let's say once every 5 seconds), but at random rate and still at a not acceptable rate for automation. Angry Angry Angry
This behavior was noted systematically on some contacts of feedback module 1,2,5 of Bus1, and also on some module -3- of Bus2. Strangely, module 4 & 5 of Bus2 never shown such behavior (these 2 modules were never used in the past).

Then I had a long talk with a friend of mine who's an electronics engineer. I explained him the concept of S88 feedback bus, all what I did in the past to discard spurious feedback (It's not the 1st time I cope with unstable feedback info). We also looked at web site S88N http://www.s88-n.eu/s88-timing-en.html explaining the timing infos, and the weaknesses of the S88 concept.

He suggested me a number of things to check and possible paths of investigations, but insisted that this strange behavior was probably due to a non respect of the timings...

And before starting important work consisting in swapping some feedback modules, I went to look at some parameters related to S88 and L88 in the CS2, especially the ones related to timings...

I noticed that for S88, 2 parameters (cycle time and bit time) were respectively 150ms and 330µs.
And for L88, it was 100ms and 167µs.

As these parameters were never changed, it never came to my mind that some would need tuning, and also documentation on meaning of these values is very brief or non existent ! However, this strange fact of having address N+1 randomly triggered after address N lend me to the conclusion that a possible cause was an out-of-sync phenomenon ...

...And I decided to simply change the 2 timing values of the L88 with the values shown for S88... and MIRACLE ! NO MORE SPURIOUS FEEDBACK !!! BigGrin BigGrin BigGrin

I then went for a simple test (only one automated train) during 30 minutes continuously and... 0 error !

Well, I don't have a complete explanation of the cause. What I'm thinking is RM88N modules have been designed a while ago when L88 didn't exist.
A possible other fact is Marklin has designed the L88 to be used with S88AC and S88DC, which could explain some different timings on the L88.

In conclusion, I would say that if you mix L88 (Marklin) and RM88N (LDT), and you have respected advice of separating feedback to decrease crosstalk, etc..., and if you still have spurious feedback like I had, look at the L88 timings...

Hope this helps.
Cheers,
Fabrice
thanks 13 users liked this useful post by French_Fabrice
Offline kiwiAlan  
#2 Posted : 26 March 2020 00:12:59(UTC)
kiwiAlan

United Kingdom   
Joined: 23/07/2014(UTC)
Posts: 8,067
Location: ENGLAND, Didcot
Originally Posted by: French_Fabrice Go to Quoted Post

And before starting important work consisting in swapping some feedback modules, I went to look at some parameters related to S88 and L88 in the CS2, especially the ones related to timings...

I noticed that for S88, 2 parameters (cycle time and bit time) were respectively 150ms and 330µs.
And for L88, it was 100ms and 167µs.

As these parameters were never changed, it never came to my mind that some would need tuning, and also documentation on meaning of these values is very brief or non existent ! However, this strange fact of having address N+1 randomly triggered after address N lend me to the conclusion that a possible cause was an out-of-sync phenomenon ...

...And I decided to simply change the 2 timing values of the L88 with the values shown for S88... and MIRACLE ! NO MORE SPURIOUS FEEDBACK !!! BigGrin BigGrin BigGrin

I then went for a simple test (only one automated train) during 30 minutes continuously and... 0 error !

Well, I don't have a complete explanation of the cause. What I'm thinking is RM88N modules have been designed a while ago when L88 didn't exist.
A possible other fact is Marklin has designed the L88 to be used with S88AC and S88DC, which could explain some different timings on the L88.

In conclusion, I would say that if you mix L88 (Marklin) and RM88N (LDT), and you have respected advice of separating feedback to decrease crosstalk, etc..., and if you still have spurious feedback like I had, look at the L88 timings...

Hope this helps.
Cheers,
Fabrice


I would suggest that the timings of the RM88N modules may be deliberately slowed down to minimize spikes from the operation of the trains from upsetting the s88 bus. This may mean that the shift registers do not get clocked properly when a faster clocking rate is attempted which may produce the symptoms you experienced. It is certainly a method I would look at using if I had problems with s88 modules (slugging the clock and data lines) to get around the problem.


Offline DaleSchultz  
#3 Posted : 26 March 2020 02:20:55(UTC)
DaleSchultz

United States   
Joined: 10/02/2006(UTC)
Posts: 3,997
what a relief! I am sure this will be useful to other users!

I have been using 6088 (and Viessmans versions) and currently have 13 modules connected (13 * 16 = 208). I am about to connect module #14.
I am glad not to have had such problems with the Intellibox, and I think I have left all its s88 defaults alone (apart from number of modules).
Most (or all) of my cables are shielded, but I do not cross power lines at 90°
Dale
Intellibox + own software, K-Track
My current layout: https://cabin-layout.mixmox.com
Arrival and Departure signs: https://remotesign.mixmox.com
thanks 1 user liked this useful post by DaleSchultz
Offline jvuye  
#4 Posted : 26 March 2020 23:55:13(UTC)
jvuye

Belgium   
Joined: 01/03/2008(UTC)
Posts: 2,881
Location: South Western France
Thanks a lot Fabrice!
Very valuable , and considering where I'm at in planning my layout, you "timing is impeccable" ! ( Pun fully intended!)
Suppose one would only use one single type of feedback module, would such spurious problems show up?
In that case, which one would you suggest?
Is the (polling?) timing of individual feedback modules something you adjust in the Rocrail program? ( Do you see how little I know still about all this? Laugh Laugh )
As you know I run an EcOS system.
Thanks again for the valuable tip!
Cheers
Jacques

Jacques Vuye aka Dr.Eisenbahn
Once a vandal, learned to be better and had great success!
Offline applor  
#5 Posted : 27 March 2020 06:30:16(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,652
Location: Brisbane, Queensland
That's interesting Fabrice and great to hear you sorted it. All of my S88 issues have come back to being caused by K track, as the S88 is very sensitive and will trigger with very high resistance joins that aren't visible.

Not sure if you read my thread but here it is:
https://www.marklin-user...ckers-with-power-applied

On my layout I have 12x Tams S88-3 which I built from kits but also have one additional Digikeijs DR4088GND. They all work fine together with my LDT HSI S88 USB.
I do like the Digikeijs DR4088GND it's S88-N spec and well priced.
modelling era IIIa (1951-1955) Germany
Offline French_Fabrice  
#6 Posted : 27 March 2020 13:34:30(UTC)
French_Fabrice

France   
Joined: 16/05/2011(UTC)
Posts: 1,474
Location: Lyon, France
Originally Posted by: jvuye Go to Quoted Post
...
Suppose one would only use one single type of feedback module, would such spurious problems show up?
In that case, which one would you suggest?
Is the (polling?) timing of individual feedback modules something you adjust in the Rocrail program? ( Do you see how little I know still about all this? Laugh Laugh )
As you know I run an EcOS system.
...



Hello Jacques,

I think the following (others may add their advice of course):

-Single type of feedback module : Use the type recommended in accordance with the brand of your Central Station (in your case, ESU). Normally, same spurious as I encountered should NOT occur. If any, then probably the cause is not the same as mine: crosstalk due to parallel feedback wires and power wires, mechanical problem not providing a constant contact, bits of metal below rails generating intermittently contact (as Applor had), use of long flat cables between modules (change to RJ45 !), high electromagnetic field (I think you don't need to build a Faraday's cage around your layout Laugh ), etc... (but the devil could stay in "etc"...Confused )
-Polling of individual feedback module: You can't (in fact I don't know but I've never seen that), because it is only one chain. The only thing you can adjust is available in the Central Station which drives the whole chain timings.
-Rocrail doesn' t have an adjustable polling time for S88. I think that in the past it was true, but Marklin (thus CS2 & CS3) seems to have modified from "polling" to "event-driven" method since version 2.x or 3.x of CS firmware

Cheers and have fun
Fabrice
thanks 1 user liked this useful post by French_Fabrice
Users browsing this topic
Forum Jump  
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.

| Powered by YAF.NET | YAF.NET © 2003-2024, Yet Another Forum.NET
This page was generated in 0.473 seconds.