There seem to be conflicting opinions about the compatibility of DCC and the old Twin-T detector systems. Why this is so? I have no idea. I have been using DCC in conjunction with Twin-T detectors for ten years and have had no problems. I use the ouput of the occupancy detectors to drive a fairly simple ABS signaling system. The home made fine scale brass signals use surface mount LEDs and a simple diode matrix (no transistor drivers) to drive the ABS logic. The whole system is benign and transparent. In fact the the low currents involved allowed my prototype layout to be powered up all the time. I have never had an LED or power failure.
My current layout is in Slovakia (lots of green mountain scenery ) and is proof of the designs flexibility.
If anyone is interested my E-mail address is -- plowry10@hotmail.com
railway pete
Who says? Granted, the ORIGINAL Twin-T design was unidirectional, using DC bias. With DCC you MIGHT exceed the reverse voltage limits on some components, but you really can;t get those olf germanium transistors anymoire, and the modern silicon replacements are much more robust. There were later Twin-T articles plublished with 'improved' versions that used AC bias current - those should work just fine with DCC.
Personally I just don;t like block detection systems where the track voltage passes through a diode drop or more before reaching the track. If every section of track is not detected you end up with speed variations, or else you have to insert back to back diode pairs for each section that will not eb detected. Transformer current detectors don;t have this problem, plus you cna locate the current transformer near the junction of the sub bus and main power bus, keeping the bus wires short, while the rest of the circuit is a some convenient point, as long as you don't do things like bundle the transformer wiring with the power bus.
--Randy
Modeling the Reading Railroad in the 1950's
Visit my web site at www.readingeastpenn.com for construction updates, DCC Info, and more.
Thanks for your comment Randy. When I first started using the Twin-T circuit I had a DC layout. True the DC bias was used to detect a track load when track power was switched off but as nothing was actually moving who cared which if one of te two transistors was condcting. Years ago I bought $10 worth of scrap IBM 1400 PC cards. Little did I know that the transistors were high quality germanium (2N16xx). While experimenting I used those with great success. The voltage drop in the track power circuit was 0.7 volts and didn't make much difference in the operation. I then designed a PC board which integrated the twin-t and a very flexible ABS signaling system. The germaniums still worked fine but evenualy I would run out of those and started using silicon power transistors. Again success. Then along came DCC. I don't have any power regions so track power was fully on all the time. I didn't need any bias anymore. Also the power rating of the transistors was not much of an issue. Due to the square wave nature of DCC power they were either fully on or fully off. Either extreme causes minimum power dissipation. So gone were the heat sinks. As far as the detector power drop is concerned, the decoders have some sort of regulation that takes care of the lower voltage. In any case most of them work within a voltage range of 14-18 volts.
There was one minor problem though. It seems that the DCC square wave which has a high frequency of 98K baud (yep Lenz used standard communications protocol) leaks its higher frequency harmonic components across the track. (there is a very small capacitive element across the rails of a long(ish) block of track). The Twin-T is sensitive enough to detect this and can give a false 'occupied' indication. The solution was to put a small cap across the Twin-T back to back PN junctions.
All in all I now have a working Twin-T system without any bias trickle current circuitry, using two silicon power transistors, one common collector resistor and a small cap. The ouput drives a small inverter/delay circuit which then drives a cheapy reed relay. The relay drives the block's signal diode matrix and provides approach communications to the adjacent block detector matrixes.
The point of all of this is the fact that I hear NMRA gurus spreading the tale that DCC and Twin-T are not compatible and I was wondering what their rationale was because, as with the bumble bee which pundits say is aerodynamically too unstable to fly, my system works like a champ.
As for the signals themselves I actually copied Odie Odegard's system of making them myself. He came up with the 'N' gauge parts etched on a black and white mask and had FOTOCUT in N.Y. etch them in 3 mil brass. He then assembled them at home. I just changed the scale and added more detail for the HO version. The cost is minimal compared to the commercial versions and the detail is awesome. If you are interested let me know.
Pete (aka Railway Pete)
rrinker Personally I just don;t like block detection systems where the track voltage passes through a diode drop or more before reaching the track. If every section of track is not detected you end up with speed variations, or else you have to insert back to back diode pairs for each section that will not eb detected.
Personally I just don;t like block detection systems where the track voltage passes through a diode drop or more before reaching the track. If every section of track is not detected you end up with speed variations, or else you have to insert back to back diode pairs for each section that will not eb detected.
Diode drop is not such an issue with DCC. Simply create a bus that is used to power non-detected track (e.g.: yards, industrial spurs, etc) that has the diodes at its head end.
Why should I add more stuff when I can just use the coils and not have the issue? I see no reason to use diode drop detection on DCC. And there is another issue with devices like the BDL168 - it encourages poor wiring practice because every section, up to 16 of them, needs to come back to one central point. That's a lot of (relatively) long bus runs. Instead, just break out a sub bus right by the detected area and run the connection tot he main bus through the current sense coil. Easier to add after the fact, too. Only long wire runs are some twisted pair phone line connecting the coils to the sense circuitry.