NCE DCC problems .. loss of control and consisting issues


#1

Hi, we at the O scale Model Railroad Club of Toronto ( www.modelrailroadclub.com ) have chosen to move to DCC. This was predicated by a move out of our long time location back in the spring of 2013. Our system is NCE, with PSX breakers and generally using LokSound XL or Select L decoders. We have found however some strange and frustrating behaviours … especially when locos are consisted together and/or the locomotive is under load. We have tried installing the TVS diodes to clip the noted voltage spikes, but over and over, control of locomotives is challenging and response can be unpredticable. Matt Herman from ESU has visited several times and even he is perplexed, and has Jurgen from Germany involved. Being another large layout like us using NCE DCC, have you experienced issues with consisted engines or seen issues with locos under load?? Any comments/suggestions would be most gratefully appreciated by us fellow O scalers in Toronto. btw… I commend you for your vision to create what will be the premier O scale layout in North America. With my kind regards, Dave MacLean


#2

Dave -

I’m saddened to hear about your club’s difficulties. On occasion we do experience weird behaviors, but so far not a total loss of control.

Like your layout we use multiple PSX breakers co-located with each 10A DCC booster. Boosters are distributed throughout the layout, in the center of each power district.

Right now there is only one diesel consist on the layout because I need to get my butt in gear to finish more locomotives. When I do an install I put a 1.5KE20CA TVS diode in the locomotive, directly across the V4L decoder inputs. The reason is that the layout is signaled so TVS diodes cannot be connected across the DCC bus between the detector and the rails. In such a location I do anticipate that arcing between the rail and wheels may be worse. My ABA F3s also have the large PowerPacks installed.

On these locomotives I have observed the following:

  1. The horn in only one particular A unit will spontaneously start blowing and continue to do so until I re-select that engine with a throttle. This has been observed while the unit was running and also when it was stationary and as far as I know not selected by any throttle. The stationary consist did not have sound playing, as it happened shortly after we powered-up the layout. Since only one decoder has so-far done this suggests a decoder problem, but it might be an issue with the CS02 command station. Or the stationary consist horn could have been the result of someone other than myself waking up a CAB04PR that was last used to address the consist. Since radio turn-on is done by pressing the HORN button, a double-press could have turned on the horn by accident.

  2. Two Overland diesel engines, when pulling heavy trains upgrade, seemed to no longer want to pull very strongly. I haven’t had time to dig into them to see if there is a drive train problem or it perhaps is something else.

  3. A couple times the B unit kept running at constant speed after normally slowing the train to a stop. We do use drive hold a lot and it is mapped to F9. In this case I suspect that it missed the F9 off packets (which are not periodically refreshed by the command station.) So I plan to move Drive Hold to something else, perhaps F4, which does get refreshed. Time will tell if this is a cure or if the problem lies elsewhere.

Sorry if this next bit is a bit geeky. I’m a retired electrical engineer. One thing I have been fighting is trying to consist with back-EMF enabled. I know that conventional wisdom says turn it off. However, I have a number of C&LS diesels that have no starting torque, so I want BEMF enabled.

Speed matching issues occur throughout the speed range. In particular, I don’t know what ESU implemented for a PI control loop in the slow speed range, but it clearly is not a standard control system design. Setting Islow to 0 with a small value in Pslow should result in an engine stall. However, motor current slowly ramps up until we hit full wheel slip. That’s what a nonzero I term in a conventional PI or PID control system does. And Pslow should not cause a ramp-up in motor current either. If the wheels are stalled and Pslow is small, then the motor current should be constant over time.

I will be on the lookout for any and all weird behaviors. Please stay in touch and let me know what you learn. I’ll send you a private message so we can correspond through regular email.

Good luck!
Bob

Note: edited to correct NCE throttle to CAB04P instead of 06.


#3

Hi Dave,

Thanks for the kind comment on the A&O. Not sure about the superlative, but we do have fun.

I don’t personally recall any such issues with our NCE, etc. The 5,000 sq.’ Colorado Model Railroad Museum near here uses NCE and don’t recall having heard, nor personally experienced, any problems. But you certainly have the attention of the gurus that should know. That being said, Bob’s thoughts on the issue may prove the key. He’s an amazing EE and figures stuff out around here routinely.

I visited your club probably 15 years ago or so, obviously in your previous location. Your group was kind enough to let me operate a train over the line and enjoy the moment. The walk-around throttles were model airplane radios if I recall. You were ahead of your time with radio!

I hope to get back to Toronto again and will enjoy seeing your new layout.

All the best,

David


#4

I have one locomotive with a Loksound V4.0 that will do blow the whistle every time the address is selected. I am using Digitrax, so my guess is it is the decoder if you have one doing the same on NEC


#5

Jeff -

Good to hear from you again! Actually, our horns don’t blow when we select an engine.

John Parker’s BNSF uses Digitrax and he runs many Loksound engines from Intermountain. I never hear a horn come on when I steal the engine at the start of my run.


#6

I brought this issue up to Matt last summer and did some rather extensive troubleshooting with no resolution other than the problem is the Select L, especially when advanced consisting. The thought was the issue was related to back EMF and speed matching, I honestly feel as though that is not the issue and is a F/W or hardware issue with the Select L as it happens on both my CVP EasyDCC system and a friend’s NCE system. I was recently told my another modeler that ESU acknowledged at the proto modelers meet near Philly there is an issue, if a resolution has occurred I’m not currently aware of it. But the problem is such that I have several locomotives I cannot currently run on my layout as I lose the ability to control the consist. Sometimes I cannot speed up a consist but the problem is most notably one or more locomotives in the consist will not fully stop or respond to even an emergency stop command.
I really like the ESU concept on their decoders, purchased the programmer and setup files to push into new locos and save my working configs, but, right now I have 5 locos that I can’t safely run, I hope they soon has a resolution.

Ken


#7

Thanks for posting! The plot thickens… If the problem turns out to be a firmware bug, at least the Select and V4 can receive firmware updates without opening the engines.

My ABA F3s have V4L decoders and they are in an advanced consist. When the B unit doesn’t respond to speed commands, after turning off Drive Hold, it does respond to emergency stop. Because those units ran well with NCE 408 non-backEMF decoders, I think I will try turning off BEMF so they don’t fight each other and see if the control loss happens again. If it recurs, then we know it is not a BEMF issue.

I have been following up with Dave and the Toronto club. I was able to watch a video of an oscilloscope connected to the DCC bus. The DCC signal was completely corrupted and showed a very large resistive voltage loss. One cause, perhaps not the only one, is that some DCC bus runs are 140 feet or longer from the booster to the end rail drop. Conventional wisdom suggests DCC bus runs no longer than 30 feet, but the booster can be located in the middle of a 60 foot long bus.

Please stay in touch and let us know when you learn more about what is happening.


#8

Hi,
Thanks everyone for your interest and thoughts. The intent here is that hopefully the good folks at ESU can use our discussions as a way to identify root cause(s) of the problem.

Just an update from Bob’s recent post. We at the Model Railroad Club of Toronto have taken Bob’s advice and moved all of the power supplies and associated boosters out of our Dispatch Office. They have been relocated to the layout areas themselves. This has reduced our longest feeder run to 40’ in yard areas and about 80’ in the mainline section. The latter will be further reduced in the near future. What we have seen is a reduction in the loss of control issues, but not the elimination.

Thanks again,
Dave


#9

My buss runs are, and have been, pretty much within lengths suggested and occur across all power districts. My power districts are mostly 12ga with one using 10ga heavily twisted buss runs with soldered jumpers no more than a foot long using 18ga rs232 cable I left in the shielded PVC jacket. The DCC system is earth grouned. The track was clean and no other decoders were having any issues including ones from QSI, MRC, and NEC. I was told to check if CVP could send out multiple packets of the command as NCE can do. I worked briefly with CVP and used the troubleshooting tool to see that packet was in fact sent but could not send out multiple retries of the same packet, which I’m not a fan of anyhow as that makes the buss that much more busier., besides, CVP won’t send extra packets.
What I do have is a DCC architecture that is almost textbook to using snubbers. Now I know the idea of needing snubbers to many is a complete waste of resources, however working in the communications business for many years filters at the end of a line does help balance long runs, remove reflections of signals, and helps to terminate the line at the end of the run. I have not tried using any snubbers to date as this problem occurred as stated before on two different DCC control systems. I am quite certain the installation in the locomotives is fine, even with my P&D units I use connectors between them to ensure good electrical contact so all decoders are picking up from all 3 locos. The other two are dummy Atlas units I powered using Weaver components and all wheel pickup. Mechanically I have installed 18t lower gears in all of the locos and they speed match rather closely, two locos use the double ended Pittman motor P&D supplies and the other 3 use the Pittman that came in the Weaver locos, I think off the top of my head they are 9513 and 9514 series motors. In two of the locos I had previously installed NCE408’s and they worked fine. I honestly don’t believe this is an NCE or CVP problem. ESU had released a newer F/W and I did upgrade the decoders but didn’t help, I have had other layout issues to work on and haven’t touch these locos for nearly 6 months now.


#10

I attended the SONC in Rockville MD the past few days and was able to meet and have some lengthy discussions with Matt Herman from ESU. I was as frustrated as anyone over the loss of control issues as can be seen in my earlier posts on this thread. What I found in Matt was a guy genuinely concerned on getting the issue fixed and willing to go the extra mile to resolve. Prior to the national I tore out a section of the layout and installed a new yard and while I was at it removed the old 12ga buss and installed a 10ga stranded buss with twisting per NMRA best practices. I also installed a snubber at the end of the buss. Still the same results with loss of control or not stopping both engines in advanced consisting. I updated Matt on Saturday, he even offered to stop at my home on his travels home, we only live about 90 miles apart from one another. However, he asked me to again try something since I mentioned having variable voltage power supplies to move up to 18V, I was running on 16. Well, I now have 5 working locomotives! Since changing the supplies to 18V the decoders started behaving correctly. We also discussed breakers and where automatic short protection was, with the CVP it is built into the “zonemaster” power distribution. I was tripping the protector which is also variable, I lowered the dwell time it will hold a short before going into protection mode but still reacts well. I have found between these two changes I can run the consisted locomotives throughout the layout!
Now I know there are allot more people out there far smarter than I who will say that might not be a solution, and maybe it isn’t, I fully understand voltage and amperage are inversely proportional and depending on who’s supply you’re using upping the voltage might lower the amperage, I believe on the one’s I’m using the deliver the full 7amps regardless of the voltage setting.
I did offer and Matt accepted that if they need any testing my layout is open to them and look forward to a basement teaming with full throttle sounds, the drive hold and sounds are incredible.
Also, 4.7.2 programmer is available as a download on their site and will install new firmware to your decoders.

Ken