0
\$\begingroup\$

I feel like there's an easy way to solve this, but I don't have the experience to even know what questions to ask or what research pathway to follow. I'm honestly not even sure I'm using the term 'bipolar' correctly.

I have an existing circuit with seven outputs (L,M,N and P,Q,R,S), corresponding to common-positive LEDs (green, red, green), then (red,green,red,green) respectively. Here's the relevant part of the diagram: https://tinyurl.com/ylczynz3 enter image description here The circuit takes a 50V supply to one of two relays (or neither), controlling seven LEDS L,M,N and P,Q,R,S. These are marked "UP" and "DOWN" to indicate direction of traffic on a model railway. There are about 20 of these circuits at the moment, with the possibility of a few more being added later.

This circuit works well, but now I need to repeat the functions to a remote location. The cable between these two sites does not have capacity for 20-plus sets of either three wires (LMN) or four wires (PQRS) to be repeated across, and I can't run extra cables because I've run out of space for terminals.

However, I do have access to a split potential supply (nominally +15V and -15V) at both the source and the remote location, so my idea is to have a single wire (call it T) running between the two sites. At the source end, it would take inputs from L,M,N corresponding to -15V, earth, or +15V on wire T. At the remote end, wire T would give LEDs P+S, P+R or Q+S connections to ground, respectively. Note green LEDs Q and S are mutually exclusive due to the relay logic.

I think the solution will involve some diode matrices, possibly optocouplers and chains of NPN and PNP transistors, but I'm not sure what the optimal arrangement would be. I'd like to try building a handful of LMNT and PQRST circuits with off-the-shelf discrete components as a test, but I'm open to surface mount ICs and screw terminals for the final design - both the L,M,N,T and P,Q,R,S,T boards. The goal for the latter in particular would be to minimise board surface area.

All LEDs within each location must share a common positive source, with logic between the LEDs and ground, so I can't just rotate half the LEDs 180 degrees or use bipolar LEDs.

So, 1) is this possible, and 2) where/how should I start?

(Separately, I'm considering replacing the two relays with transistors and the 50V supply with 12V, distinct from the above 15V. But I don't think that has an impact here?)

(Edit - thanks to whoever added the screenshot of the circuit :) )

\$\endgroup\$
8
  • \$\begingroup\$ You say the cable does not have sufficient capacity for 20 more circuits. How many wires are available? \$\endgroup\$ Commented May 18 at 13:59
  • \$\begingroup\$ Terms to search for are multiplexer and demultiplexer (MUX/DEMUX) or serialiser and deserialiser (SERDES). There are microcontrollers. \$\endgroup\$ Commented May 18 at 14:01
  • \$\begingroup\$ "20 sets" - Does one pair of relays control 20 sets of LEDs, or does each set of 7 LEDs have its own pair of relays? \$\endgroup\$ Commented May 18 at 14:18
  • \$\begingroup\$ Replacing the relays with transistors is pretty simple - and maybe more simple than you think. BU - light M is a problem because it is the only one that goes through two switches in series. Hmmm \$\endgroup\$ Commented May 18 at 14:21
  • \$\begingroup\$ @greybeard - thanks looks like I've got some googling to do :) \$\endgroup\$ Commented May 18 at 14:37

4 Answers 4

Reset to default
1
\$\begingroup\$

Oops, just realised: wires LMN run from the circuit above to LEDs geographically not too far from the additional PQRS set via one of the other 100wire cables, terminating on an adjacent terminal block, so I can just use those with four diodes to solve P and R. So I guess this whole thing was pointless. D'oh.

https://tinyurl.com/yr99p4uo

It might even be possible to only run L and N across the extension, not M, with logic P≠Q and R≠S, to further simplify the wiring.

With that said, thanks to @jonathanjo, @Analogkid, and @greybeard for their help :)

\$\endgroup\$
0
\$\begingroup\$

Does a scheme like this help?

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
1
  • \$\begingroup\$ For exclusive signals it should. \$\endgroup\$ Commented May 18 at 14:07
0
\$\begingroup\$

This will make the +/- 15V signal from the L,M,N signals.

schematic

simulate this circuit – Schematic created using CircuitLab

the inputs here (LMN) are current sources, so the will not cause the LEDs to glow, but will reverse bias them with 3V, or possibly more if the 12V goes away - it might be better to use the +12 used for the LEDs as the positive supply.

This will make an open instead of a ground when M is active (low) but it's still three states.

\$\endgroup\$
0
\$\begingroup\$

Here is a first-pass at an all-solid-state re-work of the circuit. Each "inverter" is actually a small power darlington transistor rated for 50 V and 1.5 A. There are seven of them in a ULN2004 Dip, and they make excellent logic devices for this type of circuit. Click on the schematic for a larger image.

enter image description here

U1E is replaced with the equivalent internal circuit to show how it functions. Note that with a little circuit re-work, D1 and D2 can be replaced by the two unused 2004 sections, but you would have to add two pull-up resistors so there is no net change in component count.

AND - I see that in your circuit, both relays switch various signals to GND. If you change the switch so that its pole is connected to GND instead of +50V, then things become much simpler. I haven't worked up that circuit yet, but I think it will reduce all of the control wiring to no relays, no transistors and just two diodes (plus D1 and D2).

UPDATE:

Here is the schematic of what I think is the same control scheme except that the control switches the two inputs to GND rather than to +V. No single- and double-pole relays, no transistors; just two additional diodes.

Note that this scheme requires the M LED has a separate connection to Vcc. In fact, LEDs N and L also should have independent connections for more stable operation.

enter image description here

\$\endgroup\$
7
  • \$\begingroup\$ LEDs P and R not going to work with only 0.6v off bias. \$\endgroup\$ Commented May 18 at 22:14
  • \$\begingroup\$ Yes, they will - been doing it since the 80's. That could change if we get more information about the LEDs, but until then . . . If it is an issue, then the two spare 2004 sections solve it. \$\endgroup\$ Commented May 19 at 4:15
  • \$\begingroup\$ huh? both LEDs will see the same voltage. darlington also has 0.6v drop \$\endgroup\$ Commented May 19 at 13:45
  • \$\begingroup\$ There are other interlocking contacts in the two relays that I didn't bother showing, but they effectively use three push buttons in lieu of the three-way switch I showed. LEDs on the various panels L, N, Q, S are green, M, P, R are red, all 3mm or 5mm off the shelf or from mixed bags. \$\endgroup\$ Commented May 19 at 15:43
  • \$\begingroup\$ In-field LEDs repeating PQRS might be the same off-the-shelf singles or tricolour with diodes to give a yellow aspect, but I want to move to BLMA/Atlas-brand signal heads with three internal surface-mount LEDs; specs according to the packet are "2.0-2.2 Volts; 25-30 Milli-Amps". This thread - atlasrescueforum.proboards.com/thread/3657/… - gives Vf Green = 3.00V, Yellow = 2.01V, Red = 1.98V at supply 11.65V with 750ohm resistors off each LED. \$\endgroup\$ Commented May 19 at 15:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.