Installing my DCC buss wires now. Is it beneficial to twist them together or run them separately? My search comes up with most everything except the answer I need.
Thanks,
oldline1
In reality it depends on the length of run, you have to have a large layout before it maters.
Thanks. Longest run will be about 30'.
My layout includes a 162' double maimline. I have never twisted the bus wires.
Rich
Alton Junction
I think it has been more-or-less agreed in the hobby that for the very heavy majority of applications, twisting the two bus wires around each other a la DNA helix is not necessary. For very long runs, say 40'+, it might help, but really...the rails are not wrapped around each other on long runs. They run parallel, and close to each other in the small scales...don't they? Those rails carry approximately the same voltage, and the same DCC signal, and everything works.
I got copious amounts of 12AWG for a song. I just unspooled 200' of each colour and stuck one end in the drill and wound it up in seconds. My thinking is to be proactive on this, I do not want to have to redo all my feeders and buss because the "wind the wire crowd" was right all along.
Brent
"All of the world's problems are the result of the difference between how we think and how the world works."
I twisted mine, not tight, just enough to hold them together, just seemed easier at the time, instead of making 2 runs. It was easy enough to seperate them for track feeders.
Mike.
My You Tube
mbinsewiI twisted mine, not tight
I twisted mine about 8-10cm, easy to work with.
Mark Gurries (Wiring for DCC) and Dave Heap (NCC fourm administrator) both believe in twisting. 3 turns per foot.
Me? I am just a guy at the end of the bar.
Henry
COB Potomac & Northern
Shenandoah Valley
BigDaddyMe? I am just a guy at the end of the bar.
Pass the chips, please.
I do find the recent snubber discussion interesting, but I am keepign my mouth shut. The article they mentioned in that recent thread showed Digitrax boosters as having the worst signal quality out of all of them. Which is funny, because I have NEVER had a decoder get fried or scrambled on either of my layouts (both likely too small to actually have much issue) OR at the club - which absolutely has bus runs that hit that upper limit. Now, the bus runs on the club layout ARE twisted, but literally all twisted together - the eat main and west main are on different power districts, yet through each module, both sets of bus lines are all twisted together. Along with the general power bus to run structure lights, and sometimes a third DCC bus for a branch line. Been like that for years, no one gets their locos scrambled or fried. And no loss of control (except when someone lets the batteries die in their radio throttle).
Digitrax does not recommend snubbers. NCE does - to the poitn of now actually selling them. I sort of got chewed out by Mark for suggesting this means it's NCE that has an output driver issue, not Digitrax. The topic of locos loosing control at the far ends of the layout comes up ever now and then on the NCE list, and the answer is always to install snubbers. I rarely if ever see anyone report such issues on the Digitrax list. There may be actual scope traces showing Digitrax has a less square square wave, but actual end user experience seems to be just the opposite of what you might infer given the scope traces.
Twisting is probably very helpful. But don't twist after any block detectors.
--Randy
Modeling the Reading Railroad in the 1950's
Visit my web site at www.readingeastpenn.com for construction updates, DCC Info, and more.
Too many general rules in this hobby when many do not apply to layout unless they are very large or very power hungry. Sure if you have a massive layout you might want to look into such things but never met any that needed them including the clubs I have seen.
I mentioned earlier in this thread that I have never twisted bus wires, and I have a 162' double mainline.
As for snubbers, I have an NCE 5 amp system. For years, I never heard of snubbers. Then on the NCE-DCC forum, snubbers became all the rage. Gotta have them. So, I built my own snubbers and installed them on the bus wires. Didn't notice any difference.
That was on my old layout. So, on my new layout, I haven't used snubbers. No problems whatsoever. Snubbers? We don't need no stinkin' snubbers.
For the theory, look up twisted-pair wiring, probably best explained in conjunction with 8-pin Ethernet cabling.
The effect on the signal strength of DCC modulation is small, at best. This is not a differential small-signal modulation, as might be the case if a complex waveform were imposed on a carrier as in HDSL or AC powerline modulations. In addition, I think the clock rate chosen for DCC is relatively immune to jitter or other effects involving conductor length.
Winding the wire has two costs for model-railroading purposes, though: the added cost of the wire, and the additional 'resistive' or 'impedive' length represented by more conductor in the current path. While, as we've discussed, these are of comparatively small magnitude at typical drop lengths, substantial 'winding up' of the wires would exacerbate any trouble seen with straight wire in terms of absolute voltage drop. Which is a primary purpose of DCC power.
One thing the wrap does do is put some reserve wire in each circuit, should something have to be resoldered or if 'any wire cut to length is too short'. It will save a great deal of unhearable language under the layout to have it there without cigarette-splicing and soldering additional lengths uphand. It's also neater to have wires twisted in pairs to typical hot-and-ground DC devices -- it keeps them isolated to be clearer where they go. Coding them in pairs by device can also be helpful. (Note that there is little electrical gain in signal integrity by wrapping 3 or more wires in binary modulation together, but a great deal of potential sense in labeling feeds to complex devices, sometimes feeds comprised of multiple twisted-pair couples "cabled" together...
I reserve judgment on snubbers until someone demonstrates to me how they work electrically, complete with 'scope traces. I'm too lazy to do the research myself. Did anyone ever successfully reduce SCSI termination to a science?
The math is actually straightforward (well, if integrals are straightforward....): V=L dI/dt Reduce the inductance of the transmission line by twisting, you reduce the induced voltage.
There's been a huge discussion on the NCE Group. Scope traces have been posted in the photos section there, but you need to be a member to see them.
WHat complicates the matter is that the NMRA specification requires a very rapid transition but only through part of the range. 2.5V/uS from -4 to +4. Not sure what they say for the part of the waveform above +4 or below -4. Something slower is acceptable? The faster the transition, the more ringing and the more harmonic spikes you will get.
The new scope traces posted show the same layout, same conditions, just one has an NCE 5 amp booster connected, the other has a different brand 5 amp booster connected. What the other brand is is not specified. Both traces look horrible, to be honest. Not very square at all, very slow rise and fall times. The one item is the trace showing the voltage the decoder sees. In the case of NCE, the peak is over 44 volts. In the case of the other brand, it's just over 32V.
Those are very short term peaks though - you need a REALLY expensive scope to actually qualify the peak as other than the absolute maximum value seen in a capture. Other previous posts in that same folder show the tiny little peaks above the flat top of the square wave without snubbers. There are many traces posted, but they all deal with the same equipment, just different numbers of snubbers on the bus and different numbers of locos, and running at different speeds. There's no direct comparison with one brand with and without snubbers to another brand with and without snubbers.
Supposedly, the older Digirtrax boosters used drivers with a slower slew rate, causing the vertical component of the square wave to be leaning over more than NCE which uses a faster driver. The newer Digitrax boosters (all current production) reportedly have the faster drivers as well now. With a slower rise/fall rate, there's less chance for ringing. So what once was true, that there is no need for any sort of snubbers, may not be true with the faster driver.
But, I've never had any problems, and I see little point in fixing what isn't broken. Once I have enough track up to make a difference, I can check my layout with my scope and see what I'm actually getting.
No need to solder under the layout if you use Posi-taps. Need a longer wire, Posi-connect. Cost more but you can buy in bulk or find a deal but once you buy you can use on every new layout you build and take care of any changes you need to do.
rrinkerWhat complicates the matter is that the NMRA specification requires a very rapid transition but only through part of the range. 2.5V/uS from -4 to +4. Not sure what they say for the part of the waveform above +4 or below -4. Something slower is acceptable? The faster the transition, the more ringing and the more harmonic spikes you will get.
The slew rate of the square wave is specified to make it as much of a 'logic transition' as sensible levels of power-DC slew can produce. My suspicion is that the settling time for clock recognition is orders of magnitude longer than the actual square-wave transition needs to be, and that what is effectively a high-frequency half sine wave would do as effectively as a 'rise' or 'fall' if PWM is in use as a fast transition with necessary ringing artifacts. I'd expect one of the effects of electrical 'snubbing' to be to ease or clip rapid state transitions (just as zeners or similar devices can clamp any transient overvoltage at the end of the square-wave risetime) and for them to work, width rather than 'edge' sensing is (at least I think) an important attribute.
A further consideration here is discriminating pulse noise of various kinds from actual data modulation. Here again, edge detection is not particularly your friend, especially if some of the 'noise modulation' is irregular carrier dropout/return or interference, or loss of power to digital circuitry, as might be produced by loss of contact or spark emission from things like microarcing.
As noted in the earlier discussion when someone familiar claimed 'square waves don't exist', the point of the NMRA 'standard' is to establish as-clean-as-possible pulses of known duration, with the transitions between them meaning little other than as indeterminate separating states. Peak-to-average wants to be as high as possible, but overvoltage even at nanoscale duration may endanger some circuit elements. So we have limits specified in terms of minimum risetime and decay to give clean pulses, but in themselves they aren't part of the waveform of interest -- not for power, and not for logic. To the extent "the waveform" causes problems, it ought to be safe to emend it -- via snubbing or even buffering and amending the datastream (anyone remember the 'overhead' on NuBus as applied to early Apple computers???) -- to extract the width-based signal logic as effectively as possible, maximizing signal integrity while minimizing other sources of 'interference'.
On the other hand, were edge detection to be important in decoding, the slew around zero-crossing would have to be fast, and this defined between some range of + to - voltage achievable at the detector (whatever that is). The DCC modulation has two types of 'zero crossing': those where the state goes 'rail to rail' across 28V potential difference, and those in the half-modulation from zero (or offset) up to 14V. Note the intrinsic difference in achieving practical fast slew in those two situations, when measurement over a fixed 'middle' voltage range is involved.What then becomes highly interesting is the most 'expedient' waveform generation that possesses fast slew +4 to -4, but changes more slowly as one approaches peak sustained DC voltage to minimize the various transients. While still producing the most detectable pulse wavetrain for decoding...
At least such is my present understanding of how the evolved Lenz DC-power modulation is designed to work, and what it is intended to accomplish.