Snubber/RC Filter

This section concerns itself with the application of RC filter on large layout with long bus wires and high current boosters (3A) or more.  This section has the following topics.

1) WHAT IS A RC FILTER?
2) WHAT ARE THE SIGNS YOU MAY NEED A RC FILTER ON THE TRACK BUS?
3) WHAT ARE THE TYPICAL LAYOUT DESIGN CONDITIONS THAT MIGHT REQUIRE A RC FILTER?
4) WHAT IS THE MEASUREMENT DEFINITION OF TRACK BUS LENGHT?
5) WHAT DOES A RC FILTER DO?
6) WHERE DO I INSTALL THE RC FILTERS?
7) CAN I INSTALL MORE THAN ONE?
8) WHERE SHOULD I NOT INSTALL A RC FILTER?
9) DO DCC MANUFACTURES REQUIRE IT?

10) WHERE CAN I FIND INFORMATION ON HOW MAKE A RC FILTER?
11) IF I AM PLANNING A NEW LAYOUT, WHAT CAN I DO?  (Twisting the Track Bus Wires)
12) HOW MUCH POWER DOES THE RC FILTER DRAW?

13) HOW DOES THE RC FILTER FILTER THE NOISE?
14) WHAT OTHER OPTIONS OTHER THAN A RC FILTER ARE THERE (Tranzorb)?
15) HOW WAS THE RC FILTER DESIGNED?
16) REFERENCES/NOTES


1) WHAT IS A RC FILTER?

It stands for Resistor Capacitor Filter which refers to the two electronic parts that make up the filter circuit.  The term "Filter" all by itself has a very broad meaning in the engineering world.  The act of "Filtering" means to remove unwanted signals (Noise, Voltage Spikes, you name it) out of the signal you want to keep (the DCC signal).

Unfortunately there are Nicknames that are technically incorrect for our needs DCC application needs that are used in open forum discussions. They are:

Snubber:  (Power Supply and Electrical Mechanical Engineering)
Terminator:  (Network Engineering)
RFI Filter: (Radio Interference Engineering)
Signal Filter: (Communication Engineering)

Although there is engineering agreement as to what needs to be done with the DCC signal, these terms unfortunately can create confusion for the non technical person and/or force INCORRECT application rules for the RC filters installation on the layout.  IF you want to learn more about why I choose to use the term RC Filter for DCC applications,  Go here: RC Filter Name: Why?


2) WHAT ARE THE SIGNS YOU MAY NEED A RC FILTER ON THE TRACK BUS?

Do you experience the following?

1) loss of train control but not running away. To learn more go here: Loco Runaways?
2) Runaway engines (Take off) : To learn more go here: Loco Runaways?
3) Loss of decoder programming.  To learn more go here: Decoders lose programming?
4) Decoders blowing up. To learn more go here: Decoders blowing up?
5) Any combination of above.

Then you should consider installing RC filters on the track.

Background information.
DCC is designed to be rugged by re-broadcasting the same DCC packets relating to train speed control over and over.  Brute Force Redundancy.   Unfortunately the DCC standard of testing is a 4x 8 layout which is the most common size layout in the world.  However, when the layout grows in size, the parasitic wiring issues grow with it.  People often like to think in black or white (Yes or No) terms when they have a problem.  It is human nature to do this.  We do not want to think about complexity.   The fact is this problem comes in "shades of gray".    Why?  Because there are a lot of variables involved.

Wiring Variables:  The root of the above list of problem depends on layout's wiring interacting with the decoders used along with the current rating of the booster.   No two layouts are wired the same way.   Every layout is electrically unique as a fingerprint.   This is why there is inconsistency as to who seem to have problems and those that do not for the same layout size.  

Decoder Variables:  There is no way to know how a given decoder's microprocessor looks at the DCC signal because it depends on how it is designed.  Different designs can do it different ways.  This is beyond our control.  What I do know is a Microprocessor can use "edge detection" (AKA interrupt mode) to look for a DCC signal transition and this is independent of the speed of the microprocessor.  Edge detection, without any filtering, can see 10's of nanosecond edges.  I am not saying they are doing it this way but my point is that we the layout owners cannot rule out what a micro can or cannot see with respect to the DCC signal.   We buy different brands of decoders all the time and what time has shown is that some decoders are more sensitive to the quality of the DCC signal than others.

The detailed List of Problems in order of severity:

Remember I said the wiring problem comes in shades of gray.....here are 5 levels of problems.

1) First thing lost is the % of DCC packets decoders "decode" successfully go down.  You cannot see problem by staring at the layout running trains if this is not a big problem.  You cannot even observe the problem without special tools along with good test procedures.  When running more trains on a given booster, there is a definite measurable decrease in DCC packets being decoded reliably as you move farther and farther away from the booster.    

2) The first potential visible sign of problems comes in the form of temporarily losing control of a train on some specific portion of the layout but the speed remains the same.   It no longer responds to the throttle or it does not see the function command.   No horn or bell when you ask for it.   Most people may not even know they lost control because there is was no need to adjust  the throttle when the train was in trouble.  If they do notice something is wrong, by the time they get suspicious, they got control back since the train has now moved to another area of the layout were the wiring situation is better.   Latter the problem gets dismissed especially since other brands of decoders seem to work fine and the problem is forgotten.   They layout is still viewed as not having any problems.

3) Next visible sign is the runaway issues directly related to CV29 Analog bit = On after a short circuit of some kind.  Most are intermittent....like when you see sparks under the engine when it goes through a switch.   The DCC signal at this location is so corrupted and distorted when the short takes place that the decoder thinks it is seeing something OTHER than DCC packets.   The decoder thinks is seeing DC and takes off at full speed thinking DCC's constant full track voltage is a full speed DC voltage.  Fortunately the decoder has a fix for this in the form of turning the analog bit = OFF.   Some people correctly and proactively make sure all thier decoder have CV29 Analog Bit (DC mode) = OFF set from the get go.   The runaway problem may be eliminated but your decoder still sees the DCC signal electrical chaos.   The electrical problem is not fixed but the layout is still viewed as not having any problems.

4) Closer to the worse problem is the decoder loosing its working memory values and requires at a minimum of a momentary removal from the track power to correct the problem.   More commonly seen are actual corruption of random CV values which will require a trip to the programming track to see and fix.  This is a voltage spike issue at play but it was not strong enough to blow up the decoder.  (See Note 1 at bottom for references)

5) Finally the worse case come when the decoder smokes.   Voltage spikes have now gotten so strong, they blow up the decoder electronics.   Anger and frustration sets in and you do not know what caused the problem.  Some time it is diagnosed to something shorting out the motor leads to the track power.   But other times you have no idea why and it is not a motor short.  All you know is you need to get a new decoder.  Replacing the decoder solved the problem, possibly with a different decoder changing the situation.  It is considered a one time problem, and the layout is still viewed as not having any problems.  Only when people have blown up more than one decoder and it seem to happen over and over or at the same place on the layout do they get suspicious something else is wrong.

Ignorance can be Bliss on large layouts.

Summary
It is possible to measure poor wiring performance before it become visible to the operator in terms of train control.  This topic and MORE is the reason why this website exist:  Wiring For DCC
The understanding of electrical environment on the track along with using solid electrical engineering principles that apply complete with testing have lead to a set of recommendations found on the "Wiring for DCC" website and this one.  Stated another way, if you follow the recommendations, you will avoid the common electrical problems and get consistent reliability in operations regardless of the size of the layout.  You can build your mega layout with full confidence it will work right the first time and every time.


3) WHAT ARE THE TYPICAL LAYOUT DESIGN CONDITIONS THAT REQUIRE A RC FILTER?

5Amp and UNDER Boosters or Starter Systems MIGHT if:
6Amp and ABOVE Boosters or Starter Systems WILL if:

1) The track bus that is approximately 30 Feet (10 meters) or longer.
2) The track bus wiring is not twisted into a tight pair.  In other words, you have loose or unorganized track bus wiring.

Why is there a difference based on booster Current?

The higher current boosters WILL generate a larger voltage spike than lower current boosters.  Assuming all else is 100% same (Wire lenght. Wire Gauge,  Wire Routing, type of short circuit, ect ), the energy in the voltage spike is directly proportional to the booster current rating.  So if you you double the booster current rating you double the voltage spike energy.  The higher the voltage spike energy, the more destructive potential it has.  (There is an electrical engineering equation governing this relationship.  It is a well understood property)

If you are using boosters with Amps >5Amps, also see section 11. You will need to do some wiring changes.
 
Can a smaller layout need them?   It is possible that a smaller layout might need them if the wiring is really bad and/or using a booster with excessive current relative to the layouts needs.  But generally not.


4) WHAT IS THE MEASUREMENT DEFINITION OF TRACK BUS LENGTH?

A Track Bus consist of two wires.  One for each rail.  From the booster, they carry the current out to the track and back.  The length of the track bus cable:

1) STARTS at the track terminals of the booster/command station.
2) FOLLOWS the wire routing of the track bus cable even if it is passing through a DCC circuit breaker.
3) ENDS at the farthest distance away from the Booster.  The end of the line so to speak.

NOTES: 

A) There can be multiple track bus runs all starting from the same booster/command station.  They are electrically considered independent of each other.

B) DCC Circuit Breakers may create additional track buses but they do not interfere with determining the length of a track bus.


5) WHAT DOES A RC FILTER DO?

A RC filter can be thought of as a "Band Aid" to address electrical signal problems after the fact.  A RC filter offers the following benefits.

1) Waveform fidelity preservation in digital communication buses such as Cab Bus, Control Bus and Track Bus.  Addresses the waveform high frequency ringing.  To see what bad DCC waveform looks like go here:DCC waveforms

2) Filtering out Noise and Voltage-Spikes created by motors brushes, intermittent wheel to rail contact and intermittent short circuits.

The root cause of these electrical noise problems come for the properties of the wires interacting with the AC signals that DCC brings to the layout.   AC invokes properties of wire that were never an issue with plain old DC which is why they were never mentioned.

Even the NMRA did a study: "A Case for Using High Frequency Filters on DCC Layouts"
Page 17 of the Feb. 2008 "Scale Rails", a publication of the NMRA, written by Didrik A. Voss, MMR.

There is no downside to installing RC filters if used properly.


6) WHERE DO I INSTALL THE RC FILTERS?

Put the very first one on a given track bus at the FAR end of long track bus run.

Repeat for every track bus you have.

Why at the FAR end?   The booster already has some RC filtering capability built in which covers the starting point of the track bus.   The RC filter you install will provide the same capability at the opposite FAR end.   Together they work on filtering the middle of the track bus.  A divide and conquer approach to the problem that has it roots in engineering principles.


7) CAN I INSTALL MORE THAN ONE RC FILTER?

Yes.   If you find after installing your first RC filter at the far end of the track bus that you still have some problems, then install an additional RC filter at the location where you observed the problem.   There is no limit on the number of RC filters you can install but additional ones should only be installed to solve the problem.


8) WHERE SHOULD I NOT INSTALL A RC FILTER?

The only situation where installing a RC Filter can cause problems is if one is using "current based occupancy detectors". 

If you do not know what an occupancy detector is, then chances are you do not have then and hence no problem to worry about.  Install the RC Filter and skip this rest of this section.   If not, read on.

Rolling stock such as Locomotives, lighted passenger cars and caboose consume track current.   Current Based Occupancy detectors use diodes or transformers (current sensing coil) to monitor the track current of a given block of track to see if rolling stock is present on the track.  They are a part of a dispatching system to track train locations on the layout and optionally becoming part of a signal system.

Current based Occupancy Detectors have a input (Booster Side) and a output (Track Side).   They ONLY sense current flowing on the output side.

Why is this a problem?

Because the RC Filter draw current.    Placing them output side of a current based block occupancy detector will result in a false positive indication of occupancy from the detector.  You must place any and all RC filters before the detector as in the input side and not after. 

It is understood the some layout have occupancy detectors centralized near the boosters which is wrong place to put them when using DCC.  If this is true, it is unfortunate.  You must make a choice between relocating the occupancy detector as close as possible to the block they monitor or not use RC filters.  It might be possible to use semiconductor devices called a Transzorb as discussed in section 14.


9) DO DCC MANUFACTURES REQUIRE IT?

Yes and No.

NO:  There is no requirement that a RC Filter of any kind needs to be installed.  The reason is simple.  The recommendation assumes a typical home layout size such as a 4 X 8 layout.  The wiring is short enough that the AC properties of the wires do not have much of a chance to interfere with DCC operation.   There is no need to make a recommendation for something most layouts will not need. 

YES:  When you get to large layout, these same properties of wire start to have their effect on the operational reliability of the layout.  Some DCC manufactures do recommend the installation of RC Filters/Terminators for this very reason. (NCE for examples shows it in some of the manuals).

Remember, the DCC signal waveform sent down the track is standardized as defined by the NMRA DCC standards.  So if any DCC manufacture is not recommending a RC Filter for large layouts, the reason is not based on engineering but on strong marketing ones.


10) WHERE CAN I FIND INFORMATION ON HOW TO GET OR MAKE A RC FILTER?

A complete parts list of the parts (there are only two parts required) is found here: RC FILTER parts List

It consist of a 100 Ohm 1/2W resistor and a 0.1uF ceramic or film Capacitor that can support 25V or more for HO.   A 1 Watt version of the same resistor is needed for larger scales.


11) IF I AM INSTALLING A NEW TRACK BUS, WHAT ELSE CAN I DO?  (Twisting the Track Bus Wires)

Installing a new track bus gives you the opportunity to implement some "preventive medicine" so to speak in the wiring to minimize the inductance problem with two long lose track bus wires.  Installing a RC filter is a "band-aid" for the wiring problem.  It does not directly address the root cause of the problem created by the wiring itself.  The practical solution for the wiring is to twist the track bus wires into "wire pairs" preferably color coded.  This will reduce but NOT eliminate the wire inductance.  For more information go here:  Twisted Pair Wiring

High Current Booster Users (Amps > 5Amps):  You will really need both RC filters and twisted track bus wiring TOGETHER to keep the voltage spikes under control for the same bus lenghts.  The appearance of Damaging Voltage Spike is less of a question of IF and more of a WHEN as in a time bomb waiting to happen.

The goal is to place the two bus wires as close to each other as physically possible over the track bus run.   The effective inductance reduction directly varies with the quality of the tight spacing between the wires.  The tighter the spacing the better.  The best way to accomplish this goal with two lose wires it to tightly twist the two wires together with a minimum of 3 twist per foot.  The act of twisting keeps the wires tight against each other as you use and install the cable.  The more twist per foot the better but there are practicality issues. 

Thicker wire insulation will reduce the effectiveness.    The fraction of millimeters in spacing count.  Example, Zip cord, although it keeps the two wires perfectly evenly spaced side by side over the cable run length such that twisting is not required accomplishing what we want to do, unfortunately it is not as effective as quality twisting to loose wires together.  Why? Zip cord add extra spacing between the wire in the center were you rip the zip cord apart.   However using zip cord still beats lose wiring.

We can never eliminate the wire inductance because the wire insulation keeps the two wire conductor to far apart to achieve it.  (Trivia, only a coaxial cable practically eliminates the inductance but it is unsuitable for DCC needs)

What about the track feeders?

Hopefully your track feeders are very short (under 18”) such that their length is negligible compared to the overall track bus length.

The goal is to have at least 90%+ of the track bus length wire twisted.  The benefit of twisting is not lost if you have very very short section of untwisted wire.  Besides, you have to be able to tap into the track bus with your track feeders and to do that you have to untwist the track bus wire at the point of the track feeder connection.  Just keep the track bus wires twisted as PRACTICALLY as possible.  Do not stress yourself over this twisting in trying to achieve perfection.

How to twist the track bus cables:

Using a quality twisted track bus cable, you can:

1) extend the track bus length by 50% to 45Ft/15 meters.
2) in combination with a RC filter at the far end, double the
track bus length to about 60Ft/20 meters.

The best way to accomplish the construction of a twisted cable it to do the following.

1) Measure the desired track bus run length as it would be place under the layout. 
2) Add about 15-20% more to the length you measured.  
3) Cut two wires, color coded STRONGLY recommended, to the same measured length.  Does not have to be perfect yet.
4) Locate some space on the floor were you can lay the wires side by side in a straight line over the length of the cable.
5) Install a large eye hook into a stud of a wall or someplace secure that is accessible by the wires.  The Eye must be large enough to fit two wires in the hole.
6) Take the end of both wires closest to the hook and pull it to the hook tying through the hole with a knot.  You do not want the wires to slip out of the hook.  Do not worry about damaging the wire.
7) Starting from the Eye hook end stretch and pull on the two wire as you walk down the cable until you reach the other end.  The goal is to make sure the two wires are equally tight against each other and not one wire is sagging or longer than the other wire nor is one wire twisting over the other.  They should be side by side.
8) When you reach the other end of cable, pull on the two wire so they are equally tout and then cut off any excess wire length so that both wires are now exactly the same length.
9) Using a variable speed drill, insert BOTH wire into the chuck and tighten the chuck.  Do not worry about damaging the wires in the chuck.  The goal is to be able to pull on the wire while using the drill and they not slip out.   IF done correctly, the two wire will have exactly equal length side by side as you pull up and back against the wire cable with the drill.
10) Using a very slow speed, turn on the drill and allow it to twist the two wire together.  Take your time.  If you go to fast, the twist will not spread down the cable evenly and concentrate at one end.  It takes some time for the twisting to work its way down the length of the cable.
11) As you twist, the wire will get shorter start to pull you toward the eye hook end.  That is OK as long as the cable remains taut during the twisting process.
12) You need to twist until you get about 2x the desire twist per foot.  So if you want a final 3 twist per foot, you need to over twist to 6 twist per foot.
13) When you get a cable that is evenly over twisted, loosen the chuck.  Watch out, it will slip out and spin wildly untwisting itself stopping at the desire twist per foot.
14) Cut off the damaged wire ends that were on the Eye hook and in the chuck.  You now have a nice twisted cable to install.

Although this sounds complicated, it actually goes pretty fast and is easy to do.

12) HOW MUCH POWER DOES THE RC FILTER DRAW?

The Track RC filter (terminator/Snubber) Booster loading effect is minimal.  Assuming one is using a 0.1uF ceramic capacitor and a 100Ohm resistor and a track voltage of 14V, it will consume about 0.25Watts (1/4Watt) average.  The actual value will dynamically vary between 0.2W and 0.33W depending on the data pattern.   The power consumed will be dissipated in the form of a heated resistor.

Notes:
1) You must use a 1/2 watt resistor for all scales HO or smaller which has track voltage of 14V or less.   A 1/4W resistor will get to hot.
2) You must use a 1 Watt resistor for all scales larger than HO since the track voltage is typically 16V or higher.  
3) The capacitor will get warm from the heat coming from the resistor it is attached to.  It does not get hot by itself unless it is defective.

If you want to know background of how this answer was determined.  See section 15.


13) HOW DOES THE RC FILTER FILTER THE NOISE?

The secret is the properties of the capacitor.   When voltage is changing levels and/or polarities, the capacitor  will conduct current.  AC is constantly changing its voltage and polarity and DCC is a form of AC.

Once the voltage stops changing, the capacitor gets charged up and the current stop flowing.  
Hence the capacitor blocks current when things are NOT changing

Stated another
way, only in the presence of an AC voltage of any kind will current flow through the capacitor and into the resistor.  Only the AC signals will see the 100 Ohm load.

The capacitor can tell the difference between Fast AC and Slow AC or even DC.    More accurately, the capacitor selects with frequencies it will pass and which ones it will ignore.   The frequencies it wants to pass will see the 100 Ohm resistor as a load.  By selecting the value of the capacitor, we can have it focus on the frequency that represent the noise and voltage spikes and ignore the base DCC frequencies which we do not want to touch.

So the correct selection of the capacitor, the RC circuit attacks noise and voltage spikes that are superimposed on top of the DCC signal itself without harming the DCC signal.


14) WHAT OTHER OPTIONS OTHER THAN A RC FILTER ARE THERE (Tranzorb)?

There are semiconductor devices called TVS diodes or transzorb diodes.  They are harder to find and more expensive than a RC filters parts and someone needs to know what they are doing to use them properly.  Polarity issues and voltage rating being very important.  This is not a part you can find at Radio Shack.  The RC parts CAN be found at Radio Shack.

Chosen properly, a Transzorb will only attack voltage spikes.  They are normally in a off state and lay in waiting.  When a voltage spike comes along that exceeds the Tranzorb's clamping voltage rating, it will automatically "TURN ON" so to speak to attack the excessive voltage.  When the voltage falls below it clamping voltage, it will automatically turn back "OFF" .

A Tranzorb can only addresses the worst problem where potentially decoder damaging excessive voltage spike comes along due to the long lose bus wires and the coresponding high amount of inductance they will have.    In absolute terms, the Transzorb will do a better job than the RC filter in killing these voltage spikes.  No argument there.   Transorbs are very heavy duty solution to just one specific part of a bigger problem.  It is NOT a complete solution.

The advantage with the RC filter is that it is always on...as in always looking for any fast changing voltage be it a part of the DCC waveform or NOT.    The later being created by intermittent wheel contact or worse random shorts which can distort the DCC waveform including its transitions.   What ever the "electrical garbage" is, the RC filter will attempt to clean it up instantly by locally attacking the noise that will be weaker and faster than the boosters DCC signal.  The latter being very strong and relatively slow.  It is easy to tell apart what is the DCC signal and what is not.   Please do not get the idea that a RC filter can repair the DCC waveform.  It cannot do that.  It just gets all the proverbial mud off of it and what is left is what it is.  The decoder has to do the rest.


15) HOW WAS THE RC FILTER DESIGNED?

How were these parts. 100 Ohm and 0.1uF chosen?

The design goals were chosen based on the information about:

1) Highest DCC base operating frequencies.   We do not want to filter out these signals.
2) Booster Output Impedance.  We want the filter to be strong relative to the booster drive capability.
3) Understanding the nature of the noise and voltage spikes we are attempting to eliminate or get rid off.
4) Understanding that no two layouts are wired the same.  There is nothing electrically constant about the problem.

The goals of the design are:

1) Use parts that are easy to get.  Any electronics parts store should have these parts in stock and be inexpensive to buy.
2) Keep the heat dissipated low.  Do not want to burn something.

If you want to understand more about the detailed technical design behind the RC filter go here:


16) REFERENCES/NOTES

1) Corruption of decoder (MCU or Micro Controller Unit) memory.

Here is an application document that talks about how microcontrollers can become damaged due to electrical hazards.

http://cache.freescale.com/files/microcontrollers/doc/app_note/AN2764.pdf

Here is the key sentence that says it all starting at the bottom of page 6:

"For MCUs, performance degradation can take many forms. Common forms of temporary degradation include but are not limited to internal reset, latch-up, memory corruption, and code runaway. MCUs with internal reset circuits can generally resume operation without operator involvement if the fault is an unexpected reset or code runaway that is caught by a watchdog timer. External reset circuits may be required when internal reset circuits are not suitable. Recovery from latch-up and volatile memory corruption (RAM) requires cycling the power to the system. Nonvolatile memory corruption (FLASH, EEPROM) requires a more extensive process of re-programming the system, which can be viewed as a temporary MCU degradation if the system can be re-worked, or as a permanent degradation if it cannot be re-worked."

===The words RAM and FLASH corruption goes directly to answering you question.  OR Worse===

"Permanent degradation is typically due to silicon damage that can cause increased leakage current on input/output (I/O) pins or power pins. The damage can affect analog measurements, input impedances, and output drive strength. With increased leakage current, the electronic system may still operate within specification for a while, but it may ultimately fail due to damage from the transient stress. Other permanent degradation can be caused by melted or fused power traces and bond wires resulting in opens and/or shorts."

AKA you smoked the decoder.