Dual DCC system on a large layout?

[C855B]

Some thinkin' out loud...

The GC&W is going to be too big to have everything - trains, switches, signals - on the same DCC bus. So I'm thinking about what it will take to have two systems - one for controlling trains, called the "operations bus", the other for controlling stationary stuff and handling track detection, the "CTC bus". Assume JMRI for everything on the CTC bus, and that it has no connection to the operations bus. Also, Digitrax will be used as the system core, with 3rd-party turnout controllers and (probably) TCS for signal controllers. Undecided yet on detection, probably Digitrax.

At first glance, I lose a couple of things that happen to be low priorities for me. First is loss of switch control from the throttles. Controlling switches is a little awkward from the hand-helds, so the current plan is local control panels in addition to the JMRI-based CTC. The other loss is the possibility of Digitrax transponding for location detection. I'm not enthralled with this system because to be truly effective you need transponders in everything... I'm blessed with a liberal budget, but not that liberal!

Anyway, can anybody think of other pitfalls splitting it this way?

[jagged ben]

Why do you think that it will be too big?  Not enough addresses?

For perspective, we are running our club layout on a single command station (multiple boosters and power managers), and it is, well, one of the largest N scale layouts in the world.   Some 12 scale miles of mainline, about a third of which is double tracked.   We have I think 40-something mainline blocks and a couple hundred turnouts (not all of which are yet DCC controlled, but they're all in the plan to be).  And we are in the process of implementing detection, signals, and turnout feedback.

As far as pitfalls, I'm not an expert on Digitrax detection but have you looked at whether it requires being on the same DCC bus as the track?

I also recommend looking at RR Cirkits Tower Controller system for signals and detection.  Generally cheaper, very flexible.

[John]

I have 1 DCS, 2x DB150s, 3 SE8Cs, 2 BDL168s, 4 DS64s, about 12 or so UPs, radio, IR, etc ..  one DCS will be enough .. put your other boosters in strategic places, keeping the power bus runs to around 30 feet or so ..

Loconet should be 1 network for boosters, with booster ground between all the boosters ..  and one network for throttle .. if you want to get really fancy, run another network for all the BDLS and turnout throwing things .. but I think that is overkill .. there have been some pretty huge NTrak layouts in the last few years - bigger than anything you are planning -   check the north carolina ntrak club for some of their docs ..   I think the biggest thing you need to make sure is you don't loop the loconet back on itself .. you should be fine ..

I do like the RR-Circuits stuff .. Dick Bronson makes good stuff, and good support .. I use his old detectors, but have heard nothing but good things about his equipment ..

[mmyers]

Ask your question on the Digitrax Yahoo group. I seem to recall that some one has already set up a two command station layout similar to your idea. He has managed to link the two. There is also at least one person who uses NCE to power his trains and Digitrax for detection and signalling.

Accessory address commands are only sent once on loconet so they don't bog down the band width. DCS100 will handle up to 120 addresses. We had 100 going at the Derby City Express.

Martin Myers

[C855B]

Why do you think that it will be too big?  Not enough addresses?

Command latency, mostly. I've seen just enough in the way of performance issues when the address stack fills-up with multiple m.u.'ed consists to make me want to keep CTC control off the ops bus. Also, the CTC bus with independent addressing will allow guest operation without having to think too hard about address conflicts. I think by the time all is said and done (10-year plan, you know) I will have just shy of 1000 devices on the CTC bus, so clearly there are capacity issues in the making. It's easy enough to keep this stuff on its own system without any crippling loss of function I can think of, so I guess design it that way now and avoid the rush.

I think (?) Digitrax detection uses current sensing with a toroid. Or an optoisolator. Regardless, it's electrically isolated from track power, so doesn't have to be on the same power bus. RR Cirkits uses toroids, that I know.

> ...pretty huge NTrak layouts...

I hope you're not referring to the Louisville NSC show. I was there, and fortunately not operating, or, rather, attempting to operate. I got to enjoy the show. (No offense intended, Martin!) I'll check out the N.C. group. Thanks.

> RR Cirkits

I'll check them out again. Interesting stuff, although wrapping your head around the Tower Controller is not a casual read. I haven't grokked how it fits relative to JMRI, but it doesn't help that I'm a rank noob when it comes to the control side of JMRI... I'm a computer guy, I'll figure it out. Maybe even contribute code. Anyway, I'm currently leaning towards Tam Valley for turnout control.

The club I belong to has a pretty similar setup, John. IIRC it's 1 DCS, 3 DB150s, 4 PM42s, a UR91 and 92, a dozen or so UPs. We are not doing any turnout or other accessory control, so BDLs and DSs aren't in our plan... yet. Oh - I'm the "network termination policeman" in our club. It seems like every show we setup somebody tries to loop the Loconet. Sheesh.

I'll add the Digitrax group to my Yahoo read. I'm already on JMRI, but don't check in very often.

Thanks for the input!

EDIT: Ben, you probably haven't checked into the GC&W engineering thread. Layout concept starts in the last couple of pages. It appears that it will be roughly the same scope as your club's layout, the main maybe a scale mile or two longer by the time we're done.

[John]

Mike .. you're doing the right thing .. its best to research all this stuff before you run wires and such ..  keep the questions coming .. we may have already run into the problem and solved it ... I've got a pretty good handle on jmri .. and will be glad to help any way I can

[C855B]

Thanks, John. This is definitely one of those cases where I know just enough to get into some serious trouble. I guess my biggest worry is choosing a solution that works great in test settings and discovering that - after buying 100 of Product X - it doesn't scale-up well. I do that at work enough as it is.

That North Carolina N-Trak info was great, BTW. Their little tutorial on detection and signaling clarified the Digitrax interface scheme nicely. I'll be spending some quality JMRI time today, I guess.

[seusscaboose]

its best to research all this stuff before you run wires and such ..

AWW heck John... it is SOOOO much fun to go back and do it all again when you figured out what you should've done in the first place!  You're taking all his fun away

[jagged ben]

Command latency, mostly. I've seen just enough in the way of performance issues when the address stack fills-up with multiple m.u.'ed consists to make me want to keep CTC control off the ops bus. 

That's a nice thing about the RR cirkits Tower controllers.  Your CTC would all be on Loconet not on the DCC bus.  We're not experiencing any problematic command latency.

Also, the CTC bus with independent addressing will allow guest operation without having to think too hard about address conflicts.

Not sure what you mean.  Loco operation shouldn't conflict with anything else anyway.

EDIT: Ben, you probably haven't checked into the GC&W engineering thread. Layout concept starts in the last couple of pages. It appears that it will be roughly the same scope as your club's layout, the main maybe a scale mile or two longer by the time we're done.

Cool.   

[C855B]

> Loco operation shouldn't conflict with anything else anyway.

OK, then clarify this for me (or, I maybe I could go read the NMRA DCC docs... Zzzzzzzzzz...  ). I understand... I think... that turnout controllers have an additional bit in the address which creates another address group separate from loco addressing. However, my question/issue is stationary decoders, and whether they piggyback on the loco address range, or there is a separate accessory bit - maybe even the turnout bit - that keeps stationary decoders in their own address group.

At one time I knew this, but it has blurred in the interim.

Interesing on the RR Cirkits using Loconet. Certainly is going to have a higher bit rate, that's for sure!

[WBRRVP]

 Digitrax stationary decoders use a separate address range from the mobile decoders, so you still have 9,999 addresses for locomotive decoders and another range for detection/signals/turnout controllers so you can have an engine with address 4294 and a stationary decoder with "4294".

Keeping as much of the traffic on the Loconet vs the DCC/track bus will go a long way to alleviate any command latency.

[mmyers]

I hope you're not referring to the Louisville NSC show. I was there, and fortunately not operating, or, rather, attempting to operate. I got to enjoy the show. (No offense intended, Martin!) I'll check out the N.C. group. Thanks.

I am. The 100 slots is a reality and the system worked fine. Unfortunately it took until noon time Sunday to get there.  Don't get me started on the rest of that travesty.