Do I need a filter/snubber/terminator on my accessory bus?

[St Enodoc]

I have RC filters/snubbers/terminators on each section of my track bus, which are fed from an NCE Power Pro through separate NCE EB1 circuit breakers.

 

I can't find a clear answer as to whether or not I should also fit one to my accessory bus, which is fed from a separate NCE SB5 booster in dumb mode (also through an EB1).

 

The reason I ask is that when I first installed the accessory bus I fitted an RC filter/snubber/terminator but it seemed to emit an audible high-frequency hum or almost a whistling sound, which didn't seem right, so I removed it. Now I have had a few problems with DCC Concepts ADS-8fx accessory decoder failures (which are still being investigated), so I am considering refitting the RC filter/snubber/terminator.

 

Can anyone advise definitively whether or not I should do this please?

[St Enodoc]

If you require filters on the DCC bus (which-ever one) it's worth investigating the source of the trouble, i.e. the controller. See if you can borrow a different DCC system and check if it solves the problems. These problems usually stem from insufficiently steep signal flanks, caused by the output stage of the booster. This then may cause the DCC signal to "bounce back" at the end of the line, like a wave on a key-side.

 

To cut a long technical story short: no, you shouldn't need to use filters in the first place, if the equipment is up to scratch. The requirement for filters strongly indicate said equipment requires investigation whether it meets MOROP/NMRA DCC spec's.

All my DCC equipment is NCE which I am led to believe is fully NMRA-compliant.

[Phil S]

Toplogically, any 'dcc bus' is the same as a model of a 'transmission line' ----  unlike the 'usual representation' of an electronic circuit, in which it is assumed that the connecting lines have zero imedance/resisitance, and the  symbols for each device: resisitor, capacitor, transistor etc etc, have all the physical factors affecting the circuit,

So a 'transmission line' is sometimes represented on a drawing as a repeating set of resistance, insuctance and capacitance ... repeating all along its length  (in the same was as '75ohm coaxial cable'  has the same characteristic behaviour along all of its length  [except where corrupted by tight bends or soldering]

 

On your layout - this physical distribution is by 2 'topologically parallel' conductors - be that nickel silver track, or varying sizes, or coppr wire -  and in addition, to complete the circuit,  at intervals,  'devices' in the form of decoders  or dumb light bulbs or LED lighting are added..

The additional trouble and expense of adding another piece of circuitry to 'correctly terminate' the transmission line on each stub too) is uneccessary at the low frequencies of the dcc signal and its harmonics  UNLIKE the rf waveforms which are typically passed through coaxial cables from aerials, or for video between equipment - where mis-termination will result in a reflection, which in analogue processing would appear as a ghost in the image, and when the 'colour subcarrier' is further processed to a lower frequency, may have further visible effects.

 

The frequencies used in dcc are nearer 'audio' than video - and the audio industry has been 'mismatching  distribution wiring for many years without problems - in fact it saves them amplifiers!  Using a low source impedance to feed a series of high impedance destinations - instead of terminating them all at 600ohms  [ However - a LONG audio link would be impedance matched ... but this is for MUCH MUCH longer lengths than a model railway

 

As a caveat - because there can be a wide range of factors involved - such as DutchMaster has already hinted at with some controller models -

Differing scales result in differing gauges of brass/Nickel Silver Rails - resulting in a high resisitance distribution for N gauge, compared to larger scales, and the large cross section of brass rale in G Scale .... which people compensate for by running parallel copper wire busses with multiple connections to the track ...  [ not necessarilly to 'every' piece of rail - but that too depends on joints, rail lengths, gauge, material, and whether the track has been 'wet-ballasted' - pva being a good insulator of fishplates ]..   Your accessory bus is likely to be of copper wire .... and is a perfect transmission line model with the accessory decoders etc placed at intervals along it.     As with the track; you can always do the 'coin test' to ensuure that any short circuit on the distribution is sufficiently large to trip the controller's protection .... this being an indication that the resisitive element of the tranmission line is sufficently low.

[St Enodoc]

Toplogically, any 'dcc bus' is the same as a model of a 'transmission line' ----  unlike the 'usual representation' of an electronic circuit, in which it is assumed that the connecting lines have zero imedance/resisitance, and the  symbols for each device: resisitor, capacitor, transistor etc etc, have all the physical factors affecting the circuit,

So a 'transmission line' is sometimes represented on a drawing as a repeating set of resistance, insuctance and capacitance ... repeating all along its length  (in the same was as '75ohm coaxial cable'  has the same characteristic behaviour along all of its length  [except where corrupted by tight bends or soldering]

 

On your layout - this physical distribution is by 2 'topologically parallel' conductors - be that nickel silver track, or varying sizes, or coppr wire -  and in addition, to complete the circuit,  at intervals,  'devices' in the form of decoders  or dumb light bulbs or LED lighting are added..

The additional trouble and expense of adding another piece of circuitry to 'correctly terminate' the transmission line on each stub too) is uneccessary at the low frequencies of the dcc signal and its harmonics  UNLIKE the rf waveforms which are typically passed through coaxial cables from aerials, or for video between equipment - where mis-termination will result in a reflection, which in analogue processing would appear as a ghost in the image, and when the 'colour subcarrier' is further processed to a lower frequency, may have further visible effects.

 

The frequencies used in dcc are nearer 'audio' than video - and the audio industry has been 'mismatching  distribution wiring for many years without problems - in fact it saves them amplifiers!  Using a low source impedance to feed a series of high impedance destinations - instead of terminating them all at 600ohms  [ However - a LONG audio link would be impedance matched ... but this is for MUCH MUCH longer lengths than a model railway

 

As a caveat - because there can be a wide range of factors involved - such as DutchMaster has already hinted at with some controller models -

Differing scales result in differing gauges of brass/Nickel Silver Rails - resulting in a high resisitance distribution for N gauge, compared to larger scales, and the large cross section of brass rale in G Scale .... which people compensate for by running parallel copper wire busses with multiple connections to the track ...  [ not necessarilly to 'every' piece of rail - but that too depends on joints, rail lengths, gauge, material, and whether the track has been 'wet-ballasted' - pva being a good insulator of fishplates ]..   Your accessory bus is likely to be of copper wire .... and is a perfect transmission line model with the accessory decoders etc placed at intervals along it.     As with the track; you can always do the 'coin test' to ensuure that any short circuit on the distribution is sufficiently large to trip the controller's protection .... this being an indication that the resisitive element of the tranmission line is sufficently low.

Thanks Phil. I'll do a coin test on the accessory bus tomorrow. I've done that successfully on the track buses. Distilling the rest of your post, I think you are saying that I don't need a termination on the accessory bus - which is indeed copper figure-of-eight 79/0.2 mm 2 speaker wire, twisted to about 3 twists/foot.

[Phil S]

I forgot the 'final part' of my 'Caveat': 

From Zero-1 Days, when I had 130 points and signals operated off  30 decoders - which was probably over the expected number !! (as 25 decoders x 4 outputs > the 99 address limit)  ..   as I added the final few decoders, I found myself having to SHORTEN my bus distribution  (which was 2 sets of 2.5mm copper - one for track, and 1 for accessories - separated even in those days! ), AND remove a Capacitor which Hornby had placed accross each decoder's input, in the last few decoders added - in order to maintain the waveform shape (which I was monitoring on an oscillosocope)  ....  but this had loaded the system over and above its designed maximum number of decoders (because they said you could overlap numbers.  

 

I have also worked in installing Coax-Data-bus (broadcast engineering) systems where reflections in a miss-terminated stub DID produce identifiable NULL points in the cabling - and if a receiver was placed in one of those, then it found no useable signal !!  But at a frequency much above that of dcc!

 

I will also concede, that whilst a spec might say '1 km' is no problem - and when the cable is coiled in a lab, on a drum, it is no problem - but when sent a physical 1km past and adjacent to a series of other interfering signal sources --- it may have some reduced reliability, if it works at all !

(it was 20 MBs-1 in that case) .... again well outside the range of dcc .... so if you are having problems, and the basic wiring is okay - as Dutch Master suggests - its time to look at your DCC Controller/Source - and determine WHAT aspect of it MAY BE causing the problem - and try another type to see if clears the problem. It might be that a 'low output' type has a poor output stage - and driving it via a Booster might be the cheapest solution!  [My experience is of 3A- 10A  output models]

[GoingUnderground]

Richard Johnson, the proprietor of DCC Concepts, is a regular contributor to another forum, and on there he did at times recommend fitting snubbers to fix problems. But to be fair to him, he does provide the specs of the required resistor and capacitor and the wiring diagram so folks could buy the components and assemble the snubbers themselves. I think Brian Lambert's website also has the specs and wiring diagram for DIY snubbers.

[Crosland]

You can't really treat DCC as a transmission line since the characteristics are continually changing as the loads (the locos) move around the layout.

 

The components that are often recommended are not terminators to match the impedance of the DCC bus. They are simple high pass filters that attenuate any high frequency noise. They can be useful on any system, regardless of manufacturer.

 

The noise has more sources than just the edge rate of the signals from the booster. Slower rise times are beneficial, not the problem. That's why the NMRA specs have a maximum slew rate.

[St Enodoc]

Richard Johnson, the proprietor of DCC Concepts, is a regular contributor to another forum, and on there he did at times recommend fitting snubbers to fix problems. But to be fair to him, he does provide the specs of the required resistor and capacitor and the wiring diagram so folks could buy the components and assemble the snubbers themselves. I think Brian Lambert's website also has the specs and wiring diagram for DIY snubbers.

That's all correct as far as track buses are concerned  but for accessory buses the information seems to be harder to find and, when I can find it, it seems to be more ambiguous. I should add that I have always had good advice and service from Richard and his team both with regard to sales and technical support.

[St Enodoc]

You can't really treat DCC as a transmission line since the characteristics are continually changing as the loads (the locos) move around the layout.

 

The components that are often recommended are not terminators to match the impedance of the DCC bus. They are simple high pass filters that attenuate any high frequency noise. They can be useful on any system, regardless of manufacturer.

 

The noise has more sources than just the edge rate of the signals from the booster. Slower rise times are beneficial, not the problem. That's why the NMRA specs have a maximum slew rate.

Thanks Andrew. It is indeed the simple RC filters that I am talking about.

[St Enodoc]

Thanks Phil. I'll do a coin test on the accessory bus tomorrow. I've done that successfully on the track buses. Distilling the rest of your post, I think you are saying that I don't need a termination on the accessory bus - which is indeed copper figure-of-eight 79/0.2 mm 2 speaker wire, twisted to about 3 twists/foot.

Today I did the coin test on the accessory bus - no problems there. I have not refitted the RC filter to the accessory bus.

 

I also replaced a malfunctioning ADS-8fx with a new one, which at the moment is working exactly as it should.