MYSTERIOUS BEHAVIOUR OF VIESSMANN 5211 ACCESSORY DECODERS USED WITH ECoS 2 A journey of discovery by Walter Dorman October 2017

Transcription

1 MYSTERIOUS BEHAVIOUR OF VIESSMANN 5211 ACCESSORY DECODERS USED WITH ECoS 2 A journey of discovery by Walter Dorman October 2017 BACKGROUND My layout uses 3 different types of accessory decoders: Märklin 6083 (K83), Home built replicas of Märklin 6083 and Viessmann I use these to control points with Märklin point solenoids which have circuit safety cut-off at the end of the throws to prevent burning out of the coils. The layout is powered via ECoS Boost (ESU 50011) and controlled from ECoS 2 (ESU 50200). I also use RR&Co to run trains under computer control. INITIAL OBSERVATIONS Occasionally I could hear a point throw without any obvious reason for doing so. This occurred mostly at powering up of the layout, but sometimes it occurred at other times even if nothing was running. I was never able to identify which point was misbehaving. This was particularly so as RR&Co would set the points as required for the routes used in the schedules being run. There was one particular point to a rarely used siding which caused occasional derailments. At first it was a sticky point mechanism but once fixed trains would then occasionally go the wrong way even under RR&Co. On closer inspection the point had changed direction but the graphics on both the ECoS and RR&Co had not changed. The initial assumption was that a stray signal was being sent from the ECoS. On one occasion immediately after a train went down the wrong route I checked the point. I moved the lever to see if the mechanism was still moving freely. As soon as I attempted to move the lever I could feel the electromagnetic force resisting. As I let go it would reach the safety cut-off and stop. However the power to the coil was still on. This particular point is typically set to curved (i.e. red). Somehow it changed to straight (i.e. green) without any obvious reason. Then, as randomly as it started, it would stop misbehaving and work ok. INITIAL TEST This particular Viessmann 5211 decoder for this point was set to addresses 93 to 96 and at that stage only the fourth port (i.e. address 96) was connected; the other three being vacant and reserved for future use. I swapped over the 5211 decoder with another and the problem would not reoccur with this other decoder. This indicated that it was an issue with the decoder and not the ECoS (with or without RR&Co running). I then set up the defective decoder as a duplicate with a spare point solenoid next to the ECoS (in observation range). I also attached and light globe to 96 green in case I did not hear the point throw when trains were running. This way I was then able to get visual confirmation that the point had thrown even if I did not hear it. I was then able to definitely confirm that the point on the layout attached to the new decoder was working correctly while the spare point solenoid attached to the suspect decoder would misbehave. However soon after this test setup I uploaded a new firmware version to the ECoS and the mysterious behaviour stopped. Optimistically I thought that it was a software issue now fixed and left the swapped decoder in place and put the other one away as a spare. Little did I know this was not the end of the issue. W. Dorman Page 1 of 7 Oct 2017

2 SUBSEQUENT OBSERVATIONS & INVESTIGATIONS Sometime later I started installing semaphore signals to be switched via K83 s. The old 3-aspect Märklin semaphore (7041) has a peculiarity in that it is not possible to change from yellow to green without first going to red. When setting the signal up in the ECoS I was a pleasantly surprised to discover that the ECoS intuitively know this and therefore sends 2 signals to make this work automatically. (Incidentally RR&Co does not have this automatic feature.). All this was working fine with the 2 consecutive addresses that is needed for 3-aspect signals on a different 5211 decoder. Later, as I was installing and testing some 2-aspect signals (7039), but this time using the previously suspect decoder, I noticed that the signal would change aspect unexpectedly. At first I thought it had something to do with ECoS somehow being locked in thinking the signal was 3-aspect and therefore sending the two pulses. Nearly every time I set it to red it would then set itself to green. I tried editing the entry as well as deleting & re-entering the signal in the ECoS but to no avail. I then noticed that the signal was critically overheating. On examination this was because the power to actuate to green remained on instead of the expected pulse. Because these old signals do not have the safety cut-off the coil was then continually being energised and the coil was moments away from burning out! Co-incidentally this signal was connected to the fourth port on the decoder. At that point I realised that the same problem had returned. Or more correctly, not been resolved by the software upgrade, as previously assumed. I therefore concluded that there is definitely something wrong with this particular This particular signal had been allocated the address 88 and that is also a fourth port. The other unusual observation was that when signals were connected to the remaining three ports (i.e. addresses 85 to 87) any changes the aspects of these signals would also trigger a green change to the signal at 88 and it would stay powered on. This had not been observed earlier because in the earlier configuration ports 1 to 3 were not in use. Seeing this was the only accessory decoder mucking up I deduced that it had to be something to do with the circuit board and/or the components on this unit alone. I suspected that it might be a cold joint or damage to the traces on the board but examination under a magnifying glass did not reveal any problems. All the components and IC s also looked intact and nothing looked like it had failed by overheating. I was wondering if anyone else had this problem, and perhaps how it got fixed, so it was time to do some research. RESEARCH On searching the internet I discovered a post on the Viessmann forum [1] where someone had a similar problem. He was told that early versions of 5211 s were not fully compatible with the ECoS under Motorola protocol and that he could return the 5211 to have components changed which fixes a timing issue which is at the root of this problem. Unfortunately it did not say precisely what components needed to be changed. Since I bought my first batch of 5211 in 2006, and my most recent ones in 2013, it occurred to me I could compare the boards and find the differences. But first I had to identify which components were associated with timing. I looked up the data sheet for the MC chip and it indicated that there were two RC timing configurations. R 1C 1 was connected to pin 6 and 7 and R 2C 2 was connected to pin 10. [Fig. 1]. W. Dorman Page 2 of 7 Oct 2017

3 Fig. 1 MC Functional Block Diagram and Chip Pin Assignments I opened the 5211 and located the components and identified their values: R 1 = 12kΩ and C 1 = 3n3F (Red circles); and R 2 = 270kΩ and C 2 = 3n3F (Yellow circles). [Fig. 2] Fig. 2 Viessmann 5211 Circuit Board Circa 2006 I then opened the most recently purchased 5211 and noted that it was indeed a newer version as it had surface mounted components. This was good news as it would reveal the newer values mentioned in the link. The bad news was that while I could check the values on the SMD resistors (which were the same values as those on the old board) there were no markings on the SMD capacitors. This made it impossible for me to identify the changed capacitor values. [Fig. 3] W. Dorman Page 3 of 7 Oct 2017

4 Fig. 3 Viessmann 5211 Circuit Board Circa 2013 I then searched for revisions to 5211 schematics on the internet hoping to find the changes but there was nothing which accurately indicated the changes. Various schematics had different values and these were also different from the old board I had. Source Brand/ Model R1 C1 R2 C2 Zondervan Viessmann kΩ 3n3F* 270kΩ 3n3F* Bogobit / Pallund? Viessmann 5211? 12kΩ 2n0F 100kΩ 2n0F Elector Electronics May 1989 EDiTS 12kΩ 1n8F 100kΩ 3n9F Ross Stewart / My own Märklin 6083 Replica 12kΩ 1n8F 270kΩ 3n9F * Same as my problematic unit. Only the schematics from Zondervan matched my problematic board. Not knowing what to try I resolved that the values on the Märklin 6083 Replica were the best one to try for because none of those units showed any incompatibility with the ECoS. On this basis, and because the two resistor values were identical, all that needed changing were the two capacitors. I took further comfort in finding supporting documentation [2] that this had fixed the problem for others. I became curious as to how many of my decoders were the old type and therefore would need fixing. On opening the cases on all decoders I discovered that I in fact have three different versions of The third one is an earlier version which also has surface mounted components. [Fig. 4] Unfortunately the capacitors again have no markings, so that was not much help. Fortunately beside the troublesome 5211 I discovered that only two other decoders were of that early version of the W. Dorman Page 4 of 7 Oct 2017

5 Fig. 4 Viessmann 5211 Circuit Board Circa RECTIFICATION I took the punt to replace the two 3n3 capacitors to 1n8 and 3n9 as used by Märklin. By luck I had some of these left over from building the home made Märklin 6083 Replicas. I had not ruled out that the MC IC could be faulty and causing/contributing to the problem but I did not have one to swap over. Besides, they are harder to replace than two capacitors anyhow. So I replaced the two capacitors and tested performance. Success! It was very satisfying to see that the old 5211 with the changed capacitors now functions correctly like all the other units. REFLECTION As I have 3 of the old 5211 s. I was wandering why the other two did not ever show up any problems on the fourth port. The first is the suspect unit discussed; the second one has a point connected to the points 4 th port for a siding which is very rarely used. Further, unlike the troublesome unit which was normally curved (i.e. red) the points connected to this unit have the typical setting to straight (i.e. green) so any unexpected switching to green would have gone unnoticed. The third unit does not yet have any devices connected to it. I have not removed the two other 5211 s to test and confirm whether they also behave exactly as the troublesome unit. The test would only be of academic interest as the way forward is to replace the capacitors in those units having now identified what the cause and fix to the problem is. Whilst I managed to identify and solve to the problem, the exact value of the capacitors on the newer 5211 boards still remains unknown. It would be interesting to know. Another thing which is untested is whether this issue is unique to only early version Viessmann 5211 decoders used with ECoS 2 (50200). W. Dorman Page 5 of 7 Oct 2017

6 K83 CIRCUITS Old Viessmann 5211 Board - Source: Zondervan [3] Viessmann 5211 (?) - Source: Bogobit /Pallund [4] W. Dorman Page 6 of 7 Oct 2017

7 EDiTS Source: Elector Electronics May 1989 Märklin 6083 Home Made Version Source: Ross Stewart [5] LINKS [1] [2] %3D4523.0&prev=search [3] [4] [5] W. Dorman Page 7 of 7 Oct 2017