So after my last rather frustrating LL GP-60 install, I've just finished the second. Whereas the first would just refuse to read-back on the Programming track at times - due to dirty wheels, this one (much cleaner wheels) occasionally reads back result plus zero, occasionally reads back result plus one for a CV value. So short address is either 23 or 24; other CVs track similarly. I've checked the electrical connections closely; I've reset the decoder twice (writing 0x08 to CV 8), but all that does is change the short address to 3 (or 4). </font> Anyone seen these symptoms before?</font> Any suggestions for how I can fix this?</font> I'm guessing from the fact that the reset brings the short address back to 3 (or 4) that this is a read-back issue, not a write issue. I've managed to get the engine up and responding, so it's not a problem until I want to reprogram it, but it's still a little worry and I'd rather it was cleared up. (After all, if this bit is broken, what else is wrong?)
The default address for decoders is 03 and not 04. If you reset the deocder to it's default value of 03, then you can reprogram the loco, clearing out all old values. If you want 4 digit addressing you must set Cv 29 to 4 digit addressing. I prefer the lenz 521W to the Digitrax DZ 143. Stay cool and run steam.....
Yes, I know the default address is 03; hence why I think the decoder doesn't read back right. I guess the decoder is just a little broken; occasionally giving the wrong answer to my questions. The question is, how/why is the decoder broken? I mean, if the motor was shorted to the shell (for example) it would just shut down, or melt the decoder; not give the wrong answer 50% of the time. <sigh> I think I'll stick to plug-n-play decoders. I just keep breaking the wired ones. Come to think of it; 0x03 is b'0011, and 0x04 is b'0100... so it's not a simple Stuck-1 bit. Time to read up on how DCC talks to decoders on the programming track.
I actually have exactly the same problem. when i try to programm another short adress on CV1 my LZV100 says there is no decoder on the programming track.... then no read back possible but the short adress is changed... unable to programm long adress even when programming CV 17+ 18 separately and CV 29 turned on for long adress.... I've never had this problem before with my previous DN1400 and DZ123... Maybe digitrax have had a badely built serie... Seb
Try programming the decoder with this formula: Programming decoders with four digit addressing Have fun... Stay cool and run steam....
Knowledge being the greatest weapon, I have proceeded to learn more. It would seem that my NCE base station supports "Paged Mode" or "Register Mode" operations - no Direct Bit / Direct Byte support. It also seems the actual "read a CV value" algorithm is contained in the base station firmware, not the JMRI software. The finally piece of the puzzle: it seems the only way to read a CV back is to ask the decoder "are you 0x00?" through "are you 0xff?" in turn, and in sequence: 0, 1, 2, 3, .... So CV 8, Manufacturer ID, 0x81 = Digitrax takes 129 guesses to work out! This decoder of mine reports a Manufacturer ID of 0x81 half the time, and 0x82 half the time. From this, I must conclude that the decoder is replying to the guess "too slow"; so the base station is asking... </font> are you 127 ? ... no reply</font> are you 128 ? ... no reply</font> are you 129 ? ... (decoder starts replying, doesn't arrive in time): no reply</font> are you 130 ? ... (decoder reply to 129 finally arrives): Aha! 130 it is</font> This theory is borne out by watching the LED's flash and the motor jerk as the base station gets close to the right guess. Leaving me with the question of: what have I done wrong with this install to cause the decoder to reply "late"? The only things I can think of are a short / near-short between something and chassis - causing current flow at the wrong moments. Or a bad solder joint hindering current flow at the right moments. On the upside, by resolutely ignoring the reply from the decoder, I have managed to give it a sensible long address and get it to orbit the test loop under DCC control. I can write to it, I just can't read it back.
