End Decoder Problems With A Snubber (243) 

Absolutely right! The correct term for this circuit element is a _terminator,_ not filter or snubber. The issue is that an open ended transmission line reflects signals, and the very fast DCC edge signals when reflected add to the incoming signal producing those narrow spikes seen on the oscilloscope trace. Terminating the line with a capacitively coupled 100 ohm resistor (equal to the approximate characteristic impedance of the transmission line), prevents these reflections and eliminates the spikes. I need to be convinced that you need as much as 0.1uF though, it may work with 0.01uF with less loss, but I need to experiment to be sure.

Snubbers are used for supressing voltage and current spikes resulting from switching - there is no switching here causing the spikes so I suspect Snubber is a customer friendly term for the end result. Rather than matching the track characteristic impedance, another way to look at the same thing could be that it is a filter designed to attenuate higher frequency components which comprise the spike. As you will know, the cutoff frequency of a filter using the given components is 16kHz at the -3dB points (ie power of the waveform on the track would be attenuated by half if the waveform is at 16kHz). The DCC signal is at about 8kHz fundamental and so would be slightly rounded off (perhaps very rounded off) as the higher harmonics are attenuated, but the spike would be completely removed as it is about 1/50 duration of a single cycle of the DCC, so comprises many high frequency components. The cap value of 0.1uF seems high (overdamped) and I think your suggested starting point of 0.01uF is probably about right. I wonder if manufacturers have copied each others' values ?

Thanks very much for replying to my comment. It's nice to see other people's work to get idea to improve my Channel. Take Care DD.

How did it go? Did this fix what you hoped?

Yes, that combination of components was selected by the folks at NCE who strongly recommend its use. Jim Scorse is a great EE so so I trust his choice.

Works great!

at least the U25 realistically weathered itself!

Check your track voltage if the control losses are occurring in the same areas. You may need to beef up your track feeds in those locations.

Ok, I never thought of that. My feeders are 22 gauge. I don't have as many feeders on the same section which is on an upper loop coming off my helix whereas a bus wire would be visible so I have soldered all the fishplates for better contnuity.

Rails have much higher resistance than copper wires so voltages can drop off fast in them With code 83 I use feeders every 6-8’, every 3’ with code 70 rail.

Ok, good to know. I'm using Atlas code 100 rail. I try to have feeders about every 3 to 5 ft

In their manuals Digitrax says not to use capacitors alone, but don’t say anything about the RC filters. That said I did it back before I replaced my old wires with zip cable. Matter of fact the lead photo I used was at the end of a 40’ bus on the layout with an 8 amp booster. As always you should only use these if you are experiencing the symptoms I listed.

Exactly, I just slip a knife in there and open it enough for a suitcase connector.mThe great thing about it is it keeps the two conductors the exact same length which doesn’t always happen with twisted wires. I have seen some that basically were one straight wire with the other one twisted around it. If you order it make sure to get the OFC (oxygen free copper) vs the CCA (copper clad aluminum). This is the one I get: www.parts-express.com/Audtek-SKRL-14-100-14-AWG-OFC-Speaker-Wire-100-ft.-100-021 They also sell other lengths and diameters.

Sounds like you cooked it. Did you try a reset by programming CV8 to a value of 8? If that doesn’t work you may be able to get it repaired by Digitrax, check their website. The fact that it runs on DC simply means that track power is passed through the decoder to the motor and your DC throttle is controlling the voltage getting to the motor. With DCC power on the track at constant voltage of about 14V the motor is getting full voltage and taking off.

At the most basic level the command station creates a low voltage/amperage version of the DCC signal I showed in the oscilloscope scans. The magic with the boosters is that they literally boost the DCC signal to the voltage/amperage desired. Some systems just bring it up to whatever level they get from the power supply while others can be set to a specific voltage but the amperages are usually specific to the booster. These also have their own proprietary designs that do the magic of creating a stable, and hopefully noise free power bus. They also have built in ability to sense a short circuit and shut down. If you’re interested here is a link to a website with a 3 amp booster circuit that may be more than the basic info you were looking for.

I use them with my Digitrax system, people have been using these for years without any problems. Just be aware that the lights and any other accessories you power using your track power takes that my away from your locos.

@@TheDCCGuy Thanks for the quick reply

Yes, the BD20 instructions tell you how to desensitize the device so you won’t get false detects. However as I said in the video if you aren’t seeing any symptoms you may not need the RC filers in the first place, especially if you have twisted your wires or are using zip cord.

Yes they do and that is why the first oscilloscope scan I showed had such a perfect waveform. However its the wires, motor and wheel arcing, intermittent shorts outside the command station/booster that interact with the DCC power/signal and create the problems.

The command station will generally have circuitry (diodes) that send spike voltages back into their own power supply so their output is relatively clean. As the signal passes along the track reflections from the far end bounce back and at result in ringing. Close to the command station, the diodes within the command station will (along with any other filtering present) remove most of the spike. Further away the ringing may be present at significant levels.

Yes they do draw a small current and their wattage as I said is 0.25-30 watts so use ohms law to calculate the amperage based on your voltage. Also as I said in the video they go at the end of the bus first then can be added at intermediate points. They would be useless near the booster as most already have some type of noise suppression filters built in.

I’ll see what I can do.

It does not seem to matter.

You should always use the correct size wire for the scale, rail code, and length of the run. As I ad in y video pit one at the end of the power bus and maybe a second midway if symptoms remain in some areas.

Thanks for that feedback. Not surprising since there is nothing especially secret about these. They probably don’t get much promotion since they likely don’t have a big profit margin and it is so easy and cheap to make your own.

I don’t own an oscilloscope.

If you can find it in large enough sizes. I have stripped twisted pairs out of old Cat5 cable to use with connecting my NCE Button Boards and Switch8 accessory decoders to switch machines.

I remove them from all my UK models. The only US models I have seen them on are Bachmann. Some of the UK instruction sheets actually point out that it may be necessary to remove them from the motor leads for good DCC operation. They can interfere with BEMF and create poor performance although most of the better decoders now apparently can deal with it. I did a video, #79, “Are your decoders being choked” on this subject.

That’s right, and as I said in the video if yours is a small layout you might even get away without twisting.

I have never seen anyone suggest they all should have the same orientation. And if you think about it it doesn’t matter since there is power on only @A@one rail at a time. Current flows from the on rail to the off rail through the RC filter not both ways at the same time as would occur if power was on on both rails all the time.

Not likely unless so e CV was reset that impacted on the Keep Alive function. That’s usually the type of behavior you see without a Keep Alive installed! I suggest you recheck the connection and if that is not the issue then call the TCS techies, Dan knows these well.

Thank you for the quick response. I have emailed Dan as suggested and will get back if there is a solution. To confirm, this is not a spiking issue?

I wouldn’t think so.

Sounds good.

The thickness of the wire mainly affects how much voltage loss you get as you get far from the base station. With your 10 ga wire, you will get very little loss but adding RC filters at each end a good idea as is cheap insurance.

You should have few problems with that arrangement. However if you see any of the symptoms I mentioned then try the RC filter bandaid.

I think the BDL168 uses diodes for sensing occupancy in a block and since these draw some current they may result in false detects. With current sensing based detectors there is a work around but I don’t know of any with the diode based ones. Since it sounds like you don’t have long wire runs you may not have any problems, do you see any of the symptoms I mentioned? What size wire was used? What amperages are being used?

I was guessing the big black lumps near the hook up board were CT coils. Guess not. We have had some strange CV settings, and a couple blown decoders. Wire size is 14 awg. The feeder wires leave the BDL168 and go just a few feet up to almost thirty feet.

I am not sure whether you can adjust the sensitivity of the BDL168 like you can with a BD20. With those you can make onboard adjustments and cut down on sensitivity so the false detects are eliminated. So take a loo at the BDl168 manual and see if there is a similar way. Also those may not be sensitive to the current draw of the RC filter since it only uses a 100 ohm resistor. It would only take a few minutes to wire up a RC filter, install it, and see if the detector lights up.

I just read through the manual and it does tell you how to adjust sensitivity so you might take a look at that if the RC filter does give you false detects.

I would if the bus is longer than 30’. And yes you would put one at each end if each exceeds 30’. Generally though fir shorter runs my recommendation is to install them only if you experience the issues I listed.

I believe in a stepwise approach. First are you having problems? If not then do nothing. Some systems and wiring configurations are more prone to these issues than others. If you see the symptoms then try the RC filter. If that doesn’t take care of the problem then breaking the 100’ run into shorter ones however remember you still will have to count the full length of each wire run from the booster. If you can put your command station booster i the middle of the layout and fan out your buses that would be the best configuration. Or you could break it into 2 50’ runs with a booster in the middle of each, so your 50’ runs would in effect be subdivided into 2 25’ ones. Divide and conquer!

@@TheDCCGuy Larry I finally got time to sincerely thank you for the tips as I progress so does my, How-To knowledge all the while trying to remember "Now. Which end goes In The Cement?! 😆

I have never looked at this system so have no idea as to whether these would work. However the folks at DCC Concepts in the UK sell a unit similar to the NCE snubber and their electrical engineers may have considered this. I suggest you send them an email inquiry via their website dccconcepts.com .

Yes

@@TheDCCGuy Thanks !

Shouldn’t be a problem. I even have one on my 8’ long modular layout.

I would think a transient voltage supression diode would work well to remove the spike, but would need to be sized for the DCC voltage in use. These are used to clip superfast voltage spikes like this. Say if your track volatge is 14V, then an 18V 1W TVS would be worth trying (across the rails). Too low a voltage would cause a short all the time and the diode would fail. Too high and the spike will not be gone. This would be a good option along with the snubber, but I would use 100 ohms 1/2W in series with 0.01uF.

If they want to send me a setup I’ll be happy to give it a test drive, but I know nothing about their system.

@@TheDCCGuy Okay. Hopefully they'll send you a system to evaluate. Thanks for answering me.

I’ll send them and email and see what……

Interesting phenomenon that I have never seen happen in almost 30 years of DCC. The fact it occurred at the same place might suggest some external EM noise interaction, or it just might be the configuration of your layout each time. If it has never happened anywhere else ever before or since it would suggest the local issue. But if elsewhere as well then the snubber might help. These are so cheap and easy to make and install a test would be easy enough.

As I said in the video beginning at 18:55 they are connected between the two bus wires of your DCC bus.

Only if you have problems, which usually means with DCC power bus runs of 30’ or greater. That said poor wiring and other issues may complicate this guideline.

First, how are they powered and how are you controlling them?

@@TheDCCGuy They are powered from the DCC BUSS. We “talk” to them using the Digitrax throttles.

Has anything changed on the layout, for example a new command station or boosters? I assume all those boosters are powering individual blocks?

Hello again Larry, In a way things have changed over the years. It started doing this from the very beginning since the change from NCE system to Digitrax. I don't know if it did it under the NCE system as it was just before my involvement in Bobs layout. Through the years (6 or so) we have added boosters and went from DCS100 to DCS240.

Tom-With so many changes it makes it difficult to isolate a single problem but several, things come to mind. First I know that some products reportedly have trouble with the late model Digitrax command stations such as the DCS240. The NCE Light-It won’t work with their commands according to the manual or website. I really have not used throttle commands with my Switch8s so can’t comment there, I just use the Button-Board and pushbuttons. So it could be the change from NCE or it could be signal issues or it could be noise on the bus. The snubbers are cheap and easy to make and would be a quick test to see if something is going on with the square wave shoulders.