john barry

Here is v2. I have taken both your replies and put in what seems right. I have placed all the components but some unnconnected. C1, C2, C3 are OK just 5V bypass. I think R24 220K will be pull up or down on the link IC-1 to IC4-5 as in the points decoder. Check needed. IC5-1 probably has a pull up resistor to 5V Still missing IC2-9, IC2-10 IC1-3, IC1-10, 11, 12 IC8-3, IC8-4 Area around T4, D37, D38, C5, C6 which will be power supply. -john

Yes Keith. D36 cathode was the missing link I couldn't see. There must be some more connections around T2 and T3 to make a current source. Also so far I haven't any idea where T4 fits in. Updated schematic coming soon. Frederic have you decompiled the code yet? Just kidding! -john

Hi, Frederick, from post #27 you have IC1 pin 1 to anode of D34. Could you check that because D34 seems to be in the isolated R-S section. I would use the Fluke on the lowest normal ohms range and only count connections less than a few ohms. The diode/continuity setting may turn semiconductors on. 30 years ago I worked in a production facility with a 'bed of nails' automatic board tester made by Genrad (near Boston). It was an impressive thing capable of testing the naked board and then the assembled board including tests of each component right up to processor chips. It pretty well eliminated any need for diagnostic fault finding. I often wonder what might be possible now with modern computing power. Pleasing to see that the R-S section is pretty close to the one in the on line description I posted before. Its always easier to draw the circuit if you know what to expect. As promised a quick summary of my method: 1. If lucky start with parts list and silk screen. Otherwise make a big enough picture, assign names to all components, mark on picture and make up parts list. 2. Designate all component pins (eg R1-left, R1-right, IC3-1 etc) and add to parts list. The photocopier is your friend here, keep unmarked copies for error recovery. 3. Generate a 'netlist'. This is a list of groups of things that are connected together. As each pin is added to a group it is crossed off the parts list. This is the hard part...especially from photos! Sometimes it is necessary to just look for continuity from a pin to every other pin to be sure. COntinue until all pins are crossed off the parts list. 4. Draw circuit from (a copy of) the netlist, crossing pins and groups off as you go. This often requires restarts and revisions as the picture emerges, re-arranging the layout for clarity. This is pretty foolproof but requires patience. I have never tried it before without having the actual board. I once tried putting the netlist into pcb drafting software to get a schematic but the results were disappointing. -john

The RS8 Schematic story so far. Hard to get much further without an RS8. I haven't worried about the power supply.

Of course I can wait Frederick. (it is 14:00 Sunday here!). A couple of detailed pics of the bottom half of the component side would be good. If you feel like a little multimeter work I would like to know what is connected to: IC9 pin2 IC6 pins11 and 13 IC7 pins 9, 11 and 15 IC1 pins 1, 4, 8 and 9 D36 anode I worked out a methodical way of doing this years ago. I will share it here later. Johnsy, not really off topic. See above about why LDT DCC things seem to be excessively sensitive to J-K polarity. I haven't had the same problem with Lenz decoders or home made Arduino ones. Loco decoders would be useless if this happened! I suspect there is no answer as their (LDT's) firmware would be based on the input timing caused by their dodgy opto circuit so it couldn't easily be fixed by modification. I mentioned (on the now vanished TC thread) about losing the settings. One of the improvements in 1.2.3 is the ability to export and restore the DR5000 settings. Which reminds me...Keith and Frederick have you messed with the 'LDT timing' check box? Might solve something. Might also make it impossible to use LDT and Lenz modules on the same bus. Who knows! -john

Hi Guys, Making good progress on the schematic although a clear picture of the component side tracks would help. Can someone with a multimeter (and an actual RS8) tell me what IC9 pin 2 is connected to?

I can't find one Frederick. I used to specialise in repairing (undocumented) assemblies of about that complexity by first reverse engineering them but it is a bit hard with only your top view picture! I presume IC1 (Z86E...?) does the work using the multiplexer to get the inputs and then dealing with the RS Bus. This is the best description I have found http://www.der-moba.de/index.php/RS-R%C3%BCckmeldebus unfortunately (for me at least) in German. I made the effort to Google translate it a while ago RS Bus description.docx but it is a bit rough. john

Sorry I meant rise time I think. On the J-K question this could become a problem if using the Digikeijs boosters which have auto reversing of the polarity. Presumably the actual DR5000 output is always the same way. -john

As promised this is the input circuit of the LDT S-DEC-4 turnout decoder. Strange way to use an optocoupler! The fall time of the output would be horrendous. I bet the circuit of the RS-8 is the same. -john

Hi Keith, Frederick, I got here at last. This is pretty much what I last wrote on the TC forum. I found LDT turnout decoders were sensitive to J-K polarity. I reverse engineered one ages ago to convert it to a signal decoder and I found the DCC input circuit via the CNY17 optocoupler was weird and barely seemed to work. I will post it if I can find my sketch. I mention this because looking at Frederick's picture it looks like they use a similar arrangement to get DCC from the top RHS (IN2 circuit?) down to the PIC processor chip. This would explain the J-K polarity problem and why Frederick needed to connect IN2 to get anywhere. And I could speculate that maybe their PIC firmware requires the presence of J-K to operate even when not programming. Hard to see a reason but who knows. Frederick, from looking at the board there seem to be two groups of 4 channels with series diode current detection feeding into two LTV844 AC input optocouplers. Very similar to my various S88 boards, LDT and homemade. They should be pretty rugged and isolated from the PIC and RS Bus side and the RS Bus is again isolated by two more optocouplers. I would guess that if you have damaged the input somehow it would only be the diodes and resistors around that side. The only way without a deliberate effort would be a sustained track short (or a connection mistake with the same result) which would put track power across the back to back diode pairs and when they gave up it could be across the optocoupler input if there was no limiting resistor(surely not!). You could check the LTV844 input and output sides with a multimeter to see if it reflects the occupancy pattern you are seeing on the DR5000. The input pairs should have ~ 1V across them when occupied (DCC waveform) and in that state the corresponding output should be closed ie close to zero volts. The LTV844 pin out/block diagram is easy to find with Google. If you mail it to me I will fix it for free (only you Frederick) but the time and mail costs might not be worth it. -john