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 applor  
#1 Posted : 24 March 2019 02:09:25(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,653
Location: Brisbane, Queensland
Gents,

A couple of my S88 sensors in completely different areas/modules have unfortunately developed the most frustrating bug of rapidly flickering when I turn track power on.

I can rule out S88 module/bus problems, since the fault occur no matter which input/module I connect the track to.

What is interesting is how it only occurs when I apply power (my S88 bus is through PC S88 and not the CS2, so I can see sensor events with the CS2 off)
My previous S88 issues have been constant regardless of power and rather simple to identify as joins between sections (using piko plastic rail joiners).

In this case I have cleaned between the rails of the trouble sensor with no improvement and also the flickering only occurs with power applied.

Currently a bit baffled as to the cause. I know its mechanical but not sure how, since the rail join is open and it only occurs with power applied.

Could a high resistance bridge between power and the rail cause this?

My only though so far is to rule out the rail separation as the cause by cutting it clean open with a dremel.

Edited by user 21 April 2019 07:03:01(UTC)  | Reason: Not specified

modelling era IIIa (1951-1955) Germany
Offline applor  
#2 Posted : 24 March 2019 23:49:13(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,653
Location: Brisbane, Queensland
The join in question, looks clean. The other join can't be at fault since it connects to another sensor section so that sensor would be triggering if that was the case.

IMG_4274.JPG

No replies so I will assume no one has any better ideas - I will cut the above join tonight and see if the problem is fixed.
modelling era IIIa (1951-1955) Germany
Offline JohnjeanB  
#3 Posted : 25 March 2019 00:20:46(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi Applor

I don't know exactly what "S88" you use to collect information to the PC.
My system includes mostly various Märklin S88: 6088, 60881, 60882, 60883, DigitalBahn DS2010.
Until recently the bus was dealing with 0V+5V levels but under a very high impedence (10 to 100 kOhm) associated with clamping diodes to make sure the converted signal is within the CMOS circuit voltage range (the circuit that is passing the signal to the CSx or to the PC) and a resistor in series to limit the current to very low levels.

Sorry for the explanation. It means that long wires are subject to mutual induction coming from other wires (especially the power digital cables).
Also -of course- your PC-based S88 must share the voltage of your layout or use entries with opto couplers.

First step: disconnect the faulty entries at the S88 entry. If the signal stops changing it means you need to re-arrange the wiring which bringes the sensors to the tracks.
If it does continue changing you have a problem with the sensor systems (e.g.: with the old 6088, the cable may be too long or have problematic connectors.

I recently discovered that my Wifi router (connecting my PC to my CS2) was creating problems (causing erroneous train detections) All went back to normal when I switched it off.
Now I need to install the WiFi antenna not so close to the Layout. In the meantime I use a crossover Ethernet cable.

Cheers

Jean
thanks 2 users liked this useful post by JohnjeanB
Offline DaleSchultz  
#4 Posted : 25 March 2019 00:31:59(UTC)
DaleSchultz

United States   
Joined: 10/02/2006(UTC)
Posts: 3,997
my thought is also some sort of electromagnetic interference but I don't know what could be doing it. Did you recently add some more wires for unrelated stuff?
Try disconnecting other things nearby the s88 lines. I doubt it is a mechanical connection on the tracks itself.
Dale
Intellibox + own software, K-Track
My current layout: https://cabin-layout.mixmox.com
Arrival and Departure signs: https://remotesign.mixmox.com
Offline applor  
#5 Posted : 25 March 2019 03:14:11(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,653
Location: Brisbane, Queensland
Gents,

Thanks for the replies. I have Tams S88-3 modules connected to the LDT HSI-S88 over USB to my PC. The HSI-S88 allows 3 separate S88 bus links, which I have configured with 4 modules per bus.

This particular sensor is located on the 2nd module in the bus and as mentioned the fault moves with the track (If I swap the wires with a different S88 port, the flicking sensors follows the wire)
I have made no additional cabling changes recently either.

I will do some more testing and messing about when I get some time but so far the only constant is the track section so it stands to reason its the cause somehow.
modelling era IIIa (1951-1955) Germany
thanks 1 user liked this useful post by applor
Offline DaleSchultz  
#6 Posted : 25 March 2019 04:18:43(UTC)
DaleSchultz

United States   
Joined: 10/02/2006(UTC)
Posts: 3,997
Originally Posted by: applor Go to Quoted Post
the only constant is the track section.


and the wire running to it? or did you switch wires at the track end?

Dale
Intellibox + own software, K-Track
My current layout: https://cabin-layout.mixmox.com
Arrival and Departure signs: https://remotesign.mixmox.com
Offline applor  
#7 Posted : 25 March 2019 11:32:26(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,653
Location: Brisbane, Queensland
Originally Posted by: DaleSchultz Go to Quoted Post
Originally Posted by: applor Go to Quoted Post
the only constant is the track section.


and the wire running to it? or did you switch wires at the track end?



The wire is about 20cm long and nowhere near anything else so it is definitely not the cause.

I have recently added some new isolated track sections and had to run a wire 5m long and in proximity to other cables and they exhibit zero false triggers.

edit: I was going to try and fault find this for 30 minutes tonight, except I've powered everything up and now it's not doing it Blink ConfusedCursing
modelling era IIIa (1951-1955) Germany
thanks 1 user liked this useful post by applor
Offline JohnjeanB  
#8 Posted : 25 March 2019 12:05:16(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi

When dealing with this kind of problems I found the following:
- initially when painting the rail's side or ballasting with glue I have found false detections which disappear after a few hours drying
- some short section with "Dremel cuts" may slide and make contact

But in the long run, it is the cabling arrangement that is key (so changing the receiving TAMs entry will not change the problem.
Try instead to have a separate loose wire connecting the two points of the faulty of course having disconnected the initial wiring.

On some circumstances on some entries I have:
- reduced the sensivity of the S88 entry with a pull-up resistor (5 to 10 kOhms) between a +5 VDC and the entry
- reduced electrical noise with a capacitor between the entry (+) and the ground (0). In my case the noise source was a decoupler.
Cheers

Jean
thanks 1 user liked this useful post by JohnjeanB
Offline applor  
#9 Posted : 30 March 2019 06:56:56(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,653
Location: Brisbane, Queensland
I went to drive the trains today and the sensor fault was there and didn't change through several restarts.

If I move sensor position, the fault follows - this rules out interference on the module bus, otherwise the same position would be trigger.

As mentioned, the cable from track to S88 module is about 20cm and nothing else nearby - this rules out interference on the sensor wire.

I test resistance between sensor section and the track earth and its open circuit - so it is not an earth short on the isolated track section.

I thought the problem may be related to power supply for the HSI-S88 USB but I tried every PSU I have and the sensor problem stayed.

I'm going to have to try some of Jeans modifications.

Otherwise I feel I've hit a brick wall which now prevents me from running my trains!

modelling era IIIa (1951-1955) Germany
Offline applor  
#10 Posted : 30 March 2019 12:12:20(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,653
Location: Brisbane, Queensland
I have since had some time to think and determined I can run the following tests to help pinpoint a possible cause:

1) Test the HSI-S88 using their dicostation software to confirm the sensor events are not somehow phantoms in Rocrail

2) Use alligator clips and wiring to connect the isolated rail, to better rule out any issues with the wire or solder joint

3) Test for voltage between the isolated rail and the active and neutral of the track system. Possible low resistance to the active track connection.
modelling era IIIa (1951-1955) Germany
Offline PMPeter  
#11 Posted : 30 March 2019 18:11:19(UTC)
PMPeter

Canada   
Joined: 04/04/2013(UTC)
Posts: 1,273
Location: Port Moody, BC
I had the same problem with one of my home made contact tracks. I rebuilt and resoldered it 3 times to make sure I had no leakage. I finally cut the isolated section all the way through on both ends to make sure I was not getting any form of leakage from the "B" side below the K track and separated the gaps with styrene, all to no avail. I swapped out the standard LDT S88 module with an LDT opto isolated version and the problem went away and has not returned in over a year.

From a pure electrical logic point of view it just made no sense.

Peter
thanks 1 user liked this useful post by PMPeter
Offline applor  
#12 Posted : 31 March 2019 06:12:43(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,653
Location: Brisbane, Queensland
Originally Posted by: applor Go to Quoted Post
I have since had some time to think and determined I can run the following tests to help pinpoint a possible cause:

1) Test the HSI-S88 using their dicostation software to confirm the sensor events are not somehow phantoms in Rocrail

2) Use alligator clips and wiring to connect the isolated rail, to better rule out any issues with the wire or solder joint

3) Test for voltage between the isolated rail and the active and neutral of the track system. Possible low resistance to the active track connection.



Well I performed the above tests today, results were:

1) Unfortunately dicostation does not poll events, it can only check for the HSI and do firmware, so this test was not possible
2) No change, sensor still flickered
3) standard 17V AC, same as every other sensor. In fact I performed all manner of resistance and voltage tests and the trouble section was exactly the same as those others that work OK.

Originally Posted by: PMPeter Go to Quoted Post
I had the same problem with one of my home made contact tracks. I rebuilt and resoldered it 3 times to make sure I had no leakage. I finally cut the isolated section all the way through on both ends to make sure I was not getting any form of leakage from the "B" side below the K track and separated the gaps with styrene, all to no avail. I swapped out the standard LDT S88 module with an LDT opto isolated version and the problem went away and has not returned in over a year.

From a pure electrical logic point of view it just made no sense.

Peter


Thanks for your reply, you certainly saved me from going through the same process. As a result I then performed a new test that Fabrice lead me to via PM:

French_Fabrice wrote:
HI Eric,

IN the past, when I had constant flickering, it was related to a missing (common) ground wire connected to the S88 bus...

I know you know that, but it's the only thing I'm thinking about... Possibly a ground wire close to your section which has trouble ?
Flickering means Voltage reference (i.e. ground) is not stable or missing... Or an opto-coupler which is defective in some electronics ?

Good luck
Fabrice



So with my setup there was certainly no missing ground wire. I first tried removing the ground wire to the module with flickering sensor (R2). This made no change to anything at all.

I then thought, do I need every S88 module individually connected to earth if there was no change to working sensors or the flickering sensor?

So I then proceeded to remove the S88 module earth connection for all the modules in that bus (S88 modules R1 to R4)

After doing that, all the working S88 sensors were still OK but to my astonishment the flickering sensor was triggering about half the speed it was before! Blink

Certainly onto something here!

I then went through all the remaining S88 mobules and removed the earth reference connection (another 8 modules).

I then checked Rocrail and hey presto, the flickering on both the trouble sensors was now gone! BigGrin

Therefore the only earth connection is the one directly to the HSI S88 power input terminals (as per documentation).

My best guess is that there was some sort of earth current loop happening as a result of each module being earthed.

It is completely baffling how only 2 random sensors on completely different S88 modules and S88 bus happened to be affected and none of the others.


I just want to close off by saying a big thanks to everyone who replied and especially Fabrice who happened to lead me to the solution.

Hopefully this thread will be of use to others who may end up with the same problem.
modelling era IIIa (1951-1955) Germany
thanks 5 users liked this useful post by applor
Offline applor  
#13 Posted : 06 April 2019 06:11:02(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,653
Location: Brisbane, Queensland
Can you believe I started up my trains today and one of the two sensors decided it wanted to flicker with power on again?

It only flickers once a second (previously 50 times/second) and the other sensor isn't flickering at all.

I definitely had a solution here but it seems that it is not quite enough to completely resolve the problem.

There is only the main sensor earth left, so there is nothing else I can really do I think.

Thankfully the sensor playing up is in a pass through block so I have just left it disconnected.

I think the only way to resolve this one is to replace the S88 module with an opto-isolated version.
modelling era IIIa (1951-1955) Germany
thanks 1 user liked this useful post by applor
Offline blid  
#14 Posted : 06 April 2019 12:27:41(UTC)
blid

Sweden   
Joined: 02/01/2012(UTC)
Posts: 228
Location: Stockholm, Sweden
I once had a similar problem. On the left bus I had a mix of LDT-RM-DEC-88-Opto, Marklin 60880 followed by two LDT RM-88-N-Opto. I had that mix on the previous layout with no problems. The only way to get rid of the flickering was to replace one of the RM-88-N-Opto for a RM-DEC-88-Opto from another leg and move the last RM-88-N-Opto to the middle leg.
I should have worked as it was, but it didn't.
All my units have the reference connected (some are for turnout position hence reference is power). All 23 LDTs are Opto and have the suggested diod and the reference reversed.
OneGauge Marklin and MTH, ESU ECoS 2.1 on LGB tracks. MTH 3-rail 0-gauge, DCS on GarGraves tracks. Z: Rokuhan tracks, analog or DCC+TC Gold.
thanks 1 user liked this useful post by blid
Offline applor  
#15 Posted : 11 February 2020 03:00:42(UTC)
applor

Australia   
Joined: 21/05/2004(UTC)
Posts: 1,653
Location: Brisbane, Queensland
I am posting on this old thread again as I have further knowledge to share in the hope others avoid these problems.

Things had been mostly trouble free except for one section that would intermittently decide to flicker, including one new section where I had added a sensor by cutting the rail.

I found that where I cut the rail was causing a flickering sensor because there was a tiny flap of metal below the rail where the cut had been made.
It wasn't too hard to identify and remove the small metal shard which resolved the problem completely.
It made me think that these problems must always be caused by a high resistance contact between S88 section/rail and the power rail underneath the track.
They flicker because of the AC current which only returns occupancy for one half and they are intermittent due to the high resistance - which is too high to cause a short but can cause detection because S88 is so sensitive.

I had a track section which had been playing up very rarely over the last few months and was unable to completely resolve it, because I had been working on the end rail connection thinking the problem was there (which wasn't the cause) but the movement of the track when working was enough to sometimes 'fix' the S88 detection for a period of time.
After my realisation with the new sensor and the cut track I decided to pull this whole track piece out (not easy) and inspect it. Cleaned the solder joins and what not.
I then found a chunk of something, perhaps rail plastic, that was stuck in the gap between the rail and the power rail below. Removing it resolved the problem.

As it happens I then had two other sections last weekend also flickering again. One piece of track I removed but could not find a clear cause for flicking. I removed the rail from the track (2205 flex) and cleaned everything. Fixed.

Last flicking sensor was a very short section as it was a single curve track piece (2231). This one I could not resolve with cleaning and unlike 2205 flex tracks, you can't easily remove the rail from a 2231.
I didn't give up though and decided to cut the power rail underneath the rail where the solder join to the S88 rail was located as it seemed the most likely location of the problem.
Since the power rail runs both sides of the track, you can actually cut one side off and it will still feed the centre studs.
As it happens I found more of this plastic/rust gunk below my solder join:

IMG_6725.jpg

However unlike the previous track piece, cleaning that stuff out from below the rail did not resolve the problem! The power rail had been cut regardless, so it couldn't be the cause anymore (or not the only cause).

With no other option left, I decided to remove one side of the power rail from the entire piece of 2231 track - that is the piece below the S88 rail but also leaving the two end pieces intact as they hold the connection tabs.
After cutting away the power rail I could see absolutely nothing that could be causing the flickering, however with it removed the flickering stopped as expected.
The end sleepers where the track connection tabs are located are solid plastic and don't have a gap, so no way the high resistance join could be there.

I know in this discussion people have had success with opto isolated S88 modules. Without really knowing their differences I am guessing that it separates the S88 from the power feed earth and so isolates S88 from power that causes the flickering.

What I am saying is that this issue does have a cause and that cause is high resistance between S88 rail and the power rail below the track.
Unfortunately every sleeper except the end two has a big gap between the rail and the power rail below, so there are so many places where the high resistance leak can cause false S88 detection - which can make it an absolute PIA to fix.
If Marklin had made the plastic sleeper frame a solid piece (with no gap) then I suspect a lot of these issues would never have occurred.
Of course K track was never designed with S88 in mind back in the day. Not sure about C track but I don't believe it has this issue with the solid roadbed, so this problem typically only affects K track users.
modelling era IIIa (1951-1955) Germany
thanks 3 users liked this useful post by applor
Online Ross  
#16 Posted : 11 February 2020 07:15:02(UTC)
Ross

Australia   
Joined: 25/09/2006(UTC)
Posts: 868
Location: Sydney, NSW
Hi Eric/All,

I have had the experience using flex track where the contact rail was making contact with the metal centre rail below.
I found that the manufacturing process had embedded metal particles in the plastic that would make contact between the rail and the centre rail.
By removing the rail I used a multi meter to bell out the top of each sleeper with the centre rail and once I found the problem sleeper I was able to cut away some of the plastic so the rail no longer made contact.
There were so many problems with the length of track I discarded it for a new piece.

Ross
thanks 2 users liked this useful post by Ross
Offline Roland  
#17 Posted : 18 November 2020 06:09:52(UTC)
Roland

Canada   
Joined: 09/11/2013(UTC)
Posts: 332
Location: Toronto, Canada
I'm seeing a similar issue this evening where most of my sensors in RocRail are flickering non-stop while the CS3 is powering the track. The issue started happening after I placed six Doppelstockwagen (as in my avatar) each with interior lighting on both floors. I currently have three S88 (Märklin 60881) modules connected in series and only sensors on the farther two modules experience the flickering. As a test I connected the 3rd S88 directly to the 1st (dropping the 2nd from the series) and the sensors on that 3rd s88 no longer flicker. So the problem has something to do with my middle s88 in that chain.

The affected s88 is about 20cm from the power and ground bus powering most of the track in that area. Is that not a reasonable enough distance? Or could the cause of the interference be something else?
My Layout Build | Märklin CS3+ | K-track | Merkur | Viessmann | LDT | iTrain | Modeling DB + SBB
Offline JohnjeanB  
#18 Posted : 18 November 2020 14:06:45(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi Roland
Originally Posted by: Roland Go to Quoted Post
The affected s88 is about 20cm from the power and ground bus powering most of the track in that area. Is that not a reasonable enough distance? Or could the cause of the interference be something else?

If the 60881 is in good functioning order, then it may be much farther away from the other S88 modules thanks to a good RJ45 cable.
You need to verify:
  • If the cables bringing the signal to the 60881 are not too long / tightly tied with other cables
  • If the female RJ45 connectors on the 60881 are not damaged


Note: the contact tracks must be at least 25 cm long to avoid flickering or even have a non detection when a very long car passes over it. Of course clean wheel are also key.
Cheers
Jean

Offline Roland  
#19 Posted : 18 November 2020 16:51:12(UTC)
Roland

Canada   
Joined: 09/11/2013(UTC)
Posts: 332
Location: Toronto, Canada
Originally Posted by: JohnjeanB Go to Quoted Post
Hi Roland
Originally Posted by: Roland Go to Quoted Post
The affected s88 is about 20cm from the power and ground bus powering most of the track in that area. Is that not a reasonable enough distance? Or could the cause of the interference be something else?

If the 60881 is in good functioning order, then it may be much farther away from the other S88 modules thanks to a good RJ45 cable.
You need to verify:
  • If the cables bringing the signal to the 60881 are not too long / tightly tied with other cables
  • If the female RJ45 connectors on the 60881 are not damaged


Note: the contact tracks must be at least 25 cm long to avoid flickering or even have a non detection when a very long car passes over it. Of course clean wheel are also key.
Cheers
Jean



Hi Jean,

Apologies - I should have worded my question more clearly. I wanted to know the minimum recommended distance for a 60881 s88 feedback module from CS3 power/ground bus lines, to avoid interfering with the RJ45 cables. The RJ45 cables are brand new so they're in good condition.

Thanks,
Roland
My Layout Build | Märklin CS3+ | K-track | Merkur | Viessmann | LDT | iTrain | Modeling DB + SBB
Offline JohnjeanB  
#20 Posted : 18 November 2020 18:14:52(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi Roland
AFAIK You should simply avoid feedback cables to be tightly attached for quite a distance (1 or 2 m) to the B or 0 of digital power.
"Clean cabling" of long feedback cables with other digital or AC power is to be avoided. Best is short feed Back cables alone (I loosely staple (I use a DIY stapler with long staples not stapled all the way) the cables under the plywood base of my tracks.
Jean

thanks 1 user liked this useful post by JohnjeanB
Offline Roland  
#21 Posted : 18 November 2020 22:23:27(UTC)
Roland

Canada   
Joined: 09/11/2013(UTC)
Posts: 332
Location: Toronto, Canada
I decided to test connecting my S88 modules directly to my CS3+ which has an S88 Link built-in (rather than through the LDT HSI-88-USB and then to my PC). The result is no more flickering in either the CS3 or in Rocrail. I know that the standalone S88 Link (60883) has the option to supply 12V for the feedback modules which I understand is less susceptible to interference. I wonder if 12V is used by the CS3 built-in S88 Link as well.

My Layout Build | Märklin CS3+ | K-track | Merkur | Viessmann | LDT | iTrain | Modeling DB + SBB
Offline JohnjeanB  
#22 Posted : 18 November 2020 22:30:53(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi Roland
Originally Posted by: Roland Go to Quoted Post
I know that the standalone S88 Link (60883) has the option to supply 12V for the feedback modules which I understand is less susceptible to interference. I wonder if 12V is used by the CS3 built-in S88 Link as well.

All the early S88 (6088, 60880 and their competitors) were based on CMOS 5 V logic and there were limitations.
Märklin and others did improve the system:
  • by using RJ45 connectors and cables
  • by switching to 12 V logic which offered a better protection against phony detections.

Because most of this older stuff does not support 12 V logic, Märklin has installed a 12 V - 5 V switch
Cheers
Jean


Offline PeFu  
#23 Posted : 03 March 2021 23:55:01(UTC)
PeFu

Sweden   
Joined: 30/08/2002(UTC)
Posts: 1,208
I’m bumping this ”old” thread after having experienced flickering contacts for the first time. I have five LDT RM-88-N-Opto modules connected via RJ45 patch cables amd the Tam’s adapter to the CS2. They are all installed using the ”diode-trick”, according to the below drawing. Suddenly, running several trains, flickering contacts started to occur at some contacts on the mid, third, module only. Flickering started when most contacts on this module were occupied. I started on disconnecting the 16 inputs, one by one, but the flickering just moved to one or several other contacts. Just before pressing ”buy”, upgrading for a CS3+ testing the more interference protecting 12V S88 connection replacing the 5V on the CS2, I did a final test: I removed the ground cable serving all 16 BY251 diodes for the diode-trick. Success!

Smile

I guess one of the diodes are broken, leaking ground to the module?

Confused

However, I will probably replace them only if I find a need of the diode-trick later on...

2FF09D08-BDCC-4AF4-B1D6-6D0124E64C8B.jpeg
Andreasburg-Mattiasberg Bahn is inspired by Swiss railways |Forum Thread |Track Plan |Youtube | C and K track | CS2 | TrainController Gold V10
thanks 1 user liked this useful post by PeFu
Offline JohnjeanB  
#24 Posted : 04 March 2021 02:47:59(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi PeFu
Yes it may happen that one of your S88 modules got bad.

One way to check is to replace (swap) one S88 module with another you have and see if the problems remain with the location or if it follows the module.
If it follows the module then the module may be bad.
Otherwise it is a cabling (wiring) problem.

You mention the diode trick and it works perfectly well PROVIDED you don't use the S88 to connect the ground. It is a signal module made to handle small currents.
The diode must be directly on the rail so that when a derailment occur with possible high current, it does flow through the rail and the CS2 AND NOT trough the S88 module.

Another source of problems may be the wiring: if there is a long distance between the detection rail and the S88 (one or 2 meters) and the wire is tight with other high current wires.

Chances are your modules are OK but you may have had a high current flowing through your S88 system (grounding) This is the only place where you have no resistors (so high current can go through. Everywhere else (detection entries) the impedance is high (10 kOhm or so) making it impossible to have high currents (1 mA or so).

I just installed a CS3 but for 9 years I had a CS2 with old S88 modules (6088 and 60880 6 modules with flat cables for a total cable length of 3 meters) and a link 88 for a total of 120 incoming signals - 5 volt logic. The whole layout is operated with Rocrail (check the videos near my signature)

Now with the CS3 I changed the wiring (adaptation cable I made) and the sensor adresses and all is good
Cheers
Jean
Offline fkowal  
#25 Posted : 04 March 2021 04:19:22(UTC)
fkowal

Canada   
Joined: 01/02/2012(UTC)
Posts: 69
Location: Toronto
Perhaps there is a simpler explanation. The diode trick has two diodes, one to protect the S88 module (1N4003), and the other (BY251) is to help the locomotives as it permits the isolated rail to act as a locomotive ground for one half of the digital square wave without disturbing the S88. Diodes have current ratings, and if they are exceeded briefly the diode will fail. Perhaps you had a momentary short with a loco or wagon in your isolated track section. The BY251 is destroyed. What if a blown diode does not create an open circuit, rather the opposite, it is always closed? This will cripple the S88 as it is alway occupied (grounded). The trains will run fine though. Perhaps the electronic experts can comment further on how a damaged diode behaves.
thanks 1 user liked this useful post by fkowal
Offline PeFu  
#26 Posted : 04 March 2021 11:00:17(UTC)
PeFu

Sweden   
Joined: 30/08/2002(UTC)
Posts: 1,208
Thanks Jean and fkowal, you have added great input for my research.

Originally Posted by: JohnjeanB Go to Quoted Post

You mention the diode trick and it works perfectly well PROVIDED you don't use the S88 to connect the ground. It is a signal module made to handle small currents.
The diode must be directly on the rail so that when a derailment occur with possible high current, it does flow through the rail and the CS2 AND NOT trough the S88 module.


The change in connection for my ”success” is described below. As I understand, the only purpose of the BY251 is to provide enough ground to the contact rail, to improve engine grounding capabilities? Of course, after the change, this feature is not in place anymore, for this module.

I have arranged the diodes central for each module, according to the last picture below. It’s also indicated in the picture which cable I cut to bypass the described error on one module.

I believe I will replace the 1N4003 and BY251 diodes. I don’t have them at home now, so this will happen later. I’m anyway OK with having things running now...

Smile

PRINCiPLE BEFORE CHANGE (WITH ERROR)
2FF09D08-BDCC-4AF4-B1D6-6D0124E64C8B.jpeg

PRINCIPLE AFTER CHANGE (WORKING OK)
5D8DCB7E-3808-42F8-96C2-3B0053BA8726.jpeg

PIC SHOWING WHERE GROUND IS CUT (BLACK CABLE)
F56B8BD6-6C21-4167-8293-8E36E6E7F00E.jpeg



Andreasburg-Mattiasberg Bahn is inspired by Swiss railways |Forum Thread |Track Plan |Youtube | C and K track | CS2 | TrainController Gold V10
Offline PeFu  
#27 Posted : 04 March 2021 19:04:36(UTC)
PeFu

Sweden   
Joined: 30/08/2002(UTC)
Posts: 1,208
I found some 1N4001 diodes at home, replaced the 1N4003 and connected the ”diode trick” again. Everything works great! I seems as if the 1N4003 was broken and caused the flickering.

Smile
Andreasburg-Mattiasberg Bahn is inspired by Swiss railways |Forum Thread |Track Plan |Youtube | C and K track | CS2 | TrainController Gold V10
thanks 2 users liked this useful post by PeFu
Offline fkowal  
#28 Posted : 13 June 2021 18:19:47(UTC)
fkowal

Canada   
Joined: 01/02/2012(UTC)
Posts: 69
Location: Toronto
I am picking up this thread as it relates to S88 reliability issues. I have been following it closely as I have been experiencing a related problem. My sensor is not flickering, it is intermittent, falsely tripping every few minutes or so. At least once per hour.

My setup is a CS2 with a link88, and two S88AC modules connected in series on Buss1. I am running at 12 volt, and using the diode trick. The S88AC units and the Link88 are all connected with CAT5 shielded computer cables. My layout is using 4 block sections controlling 4 Bogobit brake modules so I may run 3 trains on the same track running in the same direction. All of the block sections are contact tracks of approximately 90 cm in length. The sensor is triggered when the train leaves the contact track. There are no circuit tracks used.

The system works well provided there are no false signals. I only noticed the problem when I was working on some scenery and a false trigger started a train running. I thought I was imagining things, and then it happened again. No trains were running, but I heard the Bogobit module's relays clicking due to the false trigger. The false trigger is always on the same contact track section.

I have tried the following approaches to try to get rid of the random sensor trigger:
1) I had ground connection to both of the S88AC modules. I removed them both, and added a single ground to the S88 link. System works. Intermittent trigger remains. No success.
2) I completely removed the diode trick from the relevant S88AC port wiring. No success.
3) I removed the wireless Wifi receiver/transmitter which was in the area of the CS2 and S88 link. No success.
4) I was about to dismantle my contact track as I figured I must have a leakage to ground somewhere. I figured it was the plastic rail tie as all of my other contract track sections have the steel rail physically cut. Just to be sure, I disconnected the contact track wiring from the S88AC. Problem with false trigger was STILL THERE.

Now, I was really upset as the fault is internal to the S88AC, Link88 or the CS2.

Now, I was fearful of having to go through component changes as my colleagues earlier in this thread have done. Time consuming and expensive especially if I had to employ a LDT opto S88 module which I would need to have shipped from Germany.

While checking out the Internet on possible solutions, I remember seeing a LDT schematic showing a grounding cable being soldered to the metal housing of the RJ45 jack housing on a feedback module. I feel foolish as I cannot find the relevant webpage now to share with the group.

So, I connected a ground (brown cable) to the RJ45 housing temporarily with a miniature C-clamp to the S88AC with the problem sensor. SUCCESS! I have now the layout running for 18 hours without a false contact.

My questions for the electronics experts in the forum:
1) does grounding the RJ45 housing make sense as a solution?
2) are the RJ45 housings robust enough not to be damaged if I solder a ground wire to it?

I am relieved and at the same time preplexed. If this is the problem, why is there no "factory built in" means to connect a grounding wire to the S88AC?
thanks 1 user liked this useful post by fkowal
Offline fkowal  
#29 Posted : 13 June 2021 18:31:51(UTC)
fkowal

Canada   
Joined: 01/02/2012(UTC)
Posts: 69
Location: Toronto
I just found the link where it is recommended to add a ground cable to the RJ45 connector.

https://www.ldt-infocent...i/doku.php?id=de:rm-88-n
thanks 2 users liked this useful post by fkowal
Offline JohnjeanB  
#30 Posted : 13 June 2021 21:24:29(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi
I am not a specialist (although an engineer in micro and linear electronics)
- you need to connect any point of the Link 88 (or its slave S88) but preferably only ONE point to avoid having large currents flowing though the Ethernet cables which may destruct them.
- having a parallel ground connection to the system brings the risk to have unwanted current if the grounding of the Link88 and its slaves is not fully independent of other parts of the layout (rails, switch decoders) except on one point.
- each entry is high impedance (10 kOhms or so) and noise immunity is good within the Link 88 and the S88 BUT it takes even a high impedance or resistance to trigger a ghost signal
- the signal may get polluted by WiFi signal or by other nearby digital current (such as track power) or a talkie-walkie (Few years back I was bothered by a nearby radar testing facility that would change the oscilloscope line into a thick wave).
- The worst case is to have a "nice" wiring with neat cable trees tight together (because of mutual induction being no negligible over long distances).
- of course you mention the metal residue when you cut rails in place with a Dremel-like tool. It will happen almost every time unless a thorough {vacuum} cleaning is made.
- While ballasting using glue on a section of contact rail (even a short length is enough to trigger a phantom signal (until the glue is fully dry (2 hours or so)

When facing a strange behavior (occupation signal changing without cause) I recommend cutting the wire near the contact rail and disconnecting the other extremity from the L88 / S88. Replace the line by a flying line (away from existing wiring. If then it is OK it means one of your cables is noisy (radiates E/M energy) and so a different cable path may solve the issue.

The best practice is to have L88 / S88 near the sensor tracks so that the length between the contact rail and the decoder is limited to 1 m maximum.
Cheers
Jean
Offline kiwiAlan  
#31 Posted : 13 June 2021 22:31:16(UTC)
kiwiAlan

United Kingdom   
Joined: 23/07/2014(UTC)
Posts: 8,082
Location: ENGLAND, Didcot
Originally Posted by: fkowal Go to Quoted Post

My questions for the electronics experts in the forum:
1) does grounding the RJ45 housing make sense as a solution?
2) are the RJ45 housings robust enough not to be damaged if I solder a ground wire to it?

I am relieved and at the same time preplexed. If this is the problem, why is there no "factory built in" means to connect a grounding wire to the S88AC?


My reaction is as follows: -
1, yes it makes sense, especially as you are running a shielded cable. Depending on the internal cable arrangement, and how you have the common connection of the s88 module attached, it is still possible to induce noise into the system, especially if the cable shield is not grounded anywhere. It needs to be grounded at some point, but preferably at only one point.

2. I would attach the grounding cable to the back of the PCB where there will be lugs from the metal cover soldered into the PCB. I'll have to look at mine to see how these are done to come up with a better answer. I would do a test with it connected to the s88 common connection if the only other means of grounding it is to run a long wire. Also consider the possibility of grounding it at the other end as an alternative.
Offline fkowal  
#32 Posted : 17 June 2021 02:09:52(UTC)
fkowal

Canada   
Joined: 01/02/2012(UTC)
Posts: 69
Location: Toronto
Jean, thank you for your suggestions.
My layout is an around the room (shelf layout) about 15m x 32 cm. At the ends of the dog bone it is wider to allow the rails to turn around.
The wiring is neatly bundled, along the viewer's side of the layout. The shielded cable is running at the back about 30 from the other wires. I do hope this is adequate as I cannot do better. I am using multiple S88AC units, using well below the 16 possible connections, simply to keep the wires running from the S88AC to the contract track sections as short as possible.
As one of my earlier attempts to solve the reliability problem, I have removed redundant ground connections and now have only one (1) on the Link88

Kiwialan,
Likewise, thank your for your feedback that my solution has merit. I have gone one step further and have a single RJ45 jack grounded at the Link 88.

The layout continues to be stable.

Frank
Offline JohnjeanB  
#33 Posted : 17 June 2021 11:32:03(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi Frank
I believe the issue is not with the RJ45 cable linking L88 and S88s but rather with individual wiring bringing back the information to the S88 / L88

I stayed / worked in your nice city for a few month lodging at the Marriott in Rodgers center (Fun to watch games from your room)
Cheers
Jean
Offline fkowal  
#34 Posted : 18 June 2021 14:06:32(UTC)
fkowal

Canada   
Joined: 01/02/2012(UTC)
Posts: 69
Location: Toronto
Hello Jean,

Yes, I quite agree with you. The ghost trigger was always occurring on the same contact track (1019), and I was about to disassemble and rewire. However, the thought occurred to me to do one last check. I simply disconnected the problem contact track from the S88AC. The false trigger on port 1019 (Buss 1, module 2, port 3) continued.

So now I had a false trigger on an S88 port that had no contact track wire on it. So, the problem was not with my contact track isolation or wiring. It was internal to the RJ45 cables, the S88AC, Link 88 or CS2. I researched what to do next. In earlier forum posts colleagues swapped out components, which provided a work around, rather than a proper solution. This time consuming, expensive, troubleshooting approach I wanted to avoid.

Therefore I tried grounding the RJ45 housing as suggested by Mr. Littfinski which was, and continues to be, successful.

Frank
thanks 1 user liked this useful post by fkowal
Offline fkowal  
#35 Posted : 11 September 2021 20:29:19(UTC)
fkowal

Canada   
Joined: 01/02/2012(UTC)
Posts: 69
Location: Toronto
An update on the unreliable S88.
The problem with the unreliability continued after grounding the RJ45 housing. I had the false impression it was stable. So, it seems not to make a difference.
Finally, I replaced the S88AC unit with a new one and the problem has been eliminated.
To have only 2 S88AC units and have one of them be defective is a cause for concern.
Offline DaleSchultz  
#36 Posted : 11 September 2021 21:04:16(UTC)
DaleSchultz

United States   
Joined: 10/02/2006(UTC)
Posts: 3,997
I have had individual ports on some s88 modules die or become erratic. I mark them to be ignored by my software and plug the wire into another port. It is annoying.
Dale
Intellibox + own software, K-Track
My current layout: https://cabin-layout.mixmox.com
Arrival and Departure signs: https://remotesign.mixmox.com
thanks 2 users liked this useful post by DaleSchultz
Offline fkowal  
#37 Posted : 23 September 2021 16:41:28(UTC)
fkowal

Canada   
Joined: 01/02/2012(UTC)
Posts: 69
Location: Toronto
Unsure
Well, I am embarrassed, as I have already posted I have found the solution to the problem twice, only to find out later it is not the case. Similarly, I have questioned Märklin's quality, in retrospect, without justification. Swapping out the S88AC was not the solution.

The ghost triggers did return after a few days with new rolling stock on the layout.

I started thinking out of the box. I had noticed ghost triggers when I switched the power back on after an Emergency Stop on the CS2. I thought it is a transient CS2 issue, but it may also be an electromagnetic interference related to the power surge at start up on my wiring BUSS. I relocated my Link88 farther away from the BUSS wires. Seems to help. No triggers when power is switch "ON".

The second aspect I looked at was the ghost trigger seemed to always at block sections which had inclines and curves. The ghost appeared when the train entered the contact section. As my triggers are to be activated when the train "leaves" the contact track section, a ghost trigger when entering is disastrous.

An incline with a curve in it always implies a minor twist in the track. Even though the relevant curved track is R4, the twist is small. Yet the train may rock between the rails sufficiently to cause the ghost signal.

I had already followed French Fabrice's advice and changed the Link88 timings to match those in the CS2 listing at
150 ms and 330 micro seconds. So I thought to change the length of time to trigger the Link88.

The new timings I have used are 150 ms and 750 micro seconds. I now have at least 30 hours operating time on the layout with various rolling stock without issues.
thanks 1 user liked this useful post by fkowal
Offline DaleSchultz  
#38 Posted : 23 September 2021 17:44:37(UTC)
DaleSchultz

United States   
Joined: 10/02/2006(UTC)
Posts: 3,997
good work!
yes 'chatter' on such systems always needs to be handled carefully. When we push a switch the switch actually may close and open a few times very rapidly as the contacts come together, or the human finger pushing the button jiggles a bit. When sensing such switches electronic detection systems have to take that into account and do what is called de-bouncing. S88 sections and a train are essentially switches that can open and close momentarily and in adjusting the timing, you have increased the 'debounce' period. So contact losses under 750ms are now ignored.
What a relief!
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 PMPeter  
#39 Posted : 12 October 2021 18:16:08(UTC)
PMPeter

Canada   
Joined: 04/04/2013(UTC)
Posts: 1,273
Location: Port Moody, BC
Well 6 more contact tracks created and 5 function perfectly and one triggers immediately. The 5 that are working perfectly were all created in flex track. The 1 that triggers on was created in a 2241 curved section. I have removed the track from the layout and removed the soldered wires and then used small alligator clip connections to rule out any interference. I have tried a different S88 port and get the same result.

I have cut out the connections below the 2241 of the "B" metal and that had some effect by clearing the trigger when the CS2 boots up, but once I turn the CS2 control on the trigger returns. So all I can assume is that the plastic ties are somehow contaminated. I have run a small file under the isolated rail in between the ties to make sure all is clean, but obviously cannot get under the rail where it joins to each tie. Not sure what else to try other than making a new contact track with another 2241.
Offline JohnjeanB  
#40 Posted : 12 October 2021 22:30:46(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi Peter
I had some issues with S88 sensors and noticed the following situations creating problems:
- a WiFi router located near the layout can cause phantom detection but when moving it away by a couple of meters all went fine
- cutting rails with a mini disk (diamond or equivalent) produces a very thin metal dust. It is invisible to the naked eye but enough to create some resistance (a few kOhms and it is already detected). Vacuum-cleaning is a MUST.
- having long sensor wires "neatly" tight with other cables (digital power) can create parasitic detection. Just do a temporary connection between the rail and the S88 entry AND disconnecting the neat cabling BOTH SIDES.
- when using the old generation M88 (6088 and 60880) with flat cable, the total length of all cables cannot exceed much more than 2 meters

S88 entries are very high impedance (approx 10 kOhms) so very little is enough to trigger a false signal
The S88 system is a floating system: it needs to be connected ONLY ON ONE POINT to the 0 of the digital power. Not doing so means that currents go through the S88 but AND IT IS NOT MADE FOR THIS (especially when there is a short circuit

I hope this helps
Jean
Offline PMPeter  
#41 Posted : 12 October 2021 23:24:20(UTC)
PMPeter

Canada   
Joined: 04/04/2013(UTC)
Posts: 1,273
Location: Port Moody, BC
Hi Jean,

Thanks for your reply. I have over 70 of these sensors on my layout and this is the 1st one that I cannot solve.

I always vacuum the gaps and the track piece after I make the cuts. As I already mentioned I also removed the soldered wires and moved the track piece next to the LDT module and made the necessary connections with short jumper wires to make sure I had no possible interference from adjacent wiring. It made no difference.

I am at the other end of my layout from the Ethernet switch (not WiFi) so that cannot be it. The WiFi router is on another floor diagonally opposite to the layout, so that can't have an influence.

I don't use any flat S88 cables and I only have the single connection to the O common.

In the meantime I have it positioned so that I no longer have a false triggering. However, if I adjust the track piece up or down or sideways I trigger the input. So I can only assume that there is a foreign piece of metal between the isolated rail and the ties which is causing this intermittent contact. Other than that there is nothing left to create this false indication.

Peter
Offline JohnjeanB  
#42 Posted : 13 October 2021 01:07:31(UTC)
JohnjeanB

France   
Joined: 04/02/2011(UTC)
Posts: 3,084
Location: Paris, France
Hi Peter
I forgot to mention that wet rails (when ballasting with glue) or humid fingers are enough to trigger the S88 signal.
I really had cases where the conducting dust was totally invisible (but still there) and a combination of vacuum-cleaning and brushing was needed.
After all old microphones were made based on this principle
Good luck
Jean
thanks 1 user liked this useful post by JohnjeanB
Offline PMPeter  
#43 Posted : 14 October 2021 15:46:59(UTC)
PMPeter

Canada   
Joined: 04/04/2013(UTC)
Posts: 1,273
Location: Port Moody, BC
Found it!

Since I was able to trigger the activation by slightly shifting or lifting the track section it had to be a piece of metal or foreign material. I had nothing to lose so I cut away the "B" metal below the ties of the isolated rail (the pukos are still powered from the other side) and sure enough there was a piece of rust in between the plastic and where the "B" metal was. Once I cleaned that all up it worked satisfactorily.
thanks 3 users liked this useful post by PMPeter
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 1.569 seconds.