Help: Basic Electronics

[dmaze]

This wound up getting sketched out on my whiteboard at work, but it seems simple enough to me that I can build it on a breadboard. I assume some of you fine readers have basic electronics clue, maybe even to the point of having actually applied your 6.002 knowledge recently. So, if you wouldn't mind reviewing the following and telling me if it makes sense...



V1 is 12V, V2 is 5V, but that's not terribly relevant. The idea is that if the switch S1 is closed, then the motor M1 will draw current through the transistors Q1 and Q2, with how much being determined by resistors R1 and R2. If current is flowing through Q1, then it's also flowing through the optoisolator, lighting the LED.

(Original version here; current flowing through Q1 also pulls current through Q3, and thence the relay, lighting the LED..)

Does the basic idea make sense? I assume that then I can put arbitrary digital junk on the right-hand side of the relay. I think I have my transistors in the right direction too. I can figure out actual values for parts later.

Edit: [14 Dec 2006] Updated schematic after commentary from nonnihil.

Comments

[rjpb.livejournal.com]

My electronics clue is too rusty to confirm this works, but I have to ask: Is this for driving model trains? :)

[dmaze]

Umm...maybe? :-) The important, interesting part to me is the right-hand side, and I've been thinking on-and-off about hacky ways to connect my existing left-hand side to the right-hand side for a while. If I can in fact put arbitrary digital junk on the right-hand side (I can't see why I couldn't) then I could have multiple bi-color LEDs lit on the basis of multiple switches S1.

[nonnihil.livejournal.com]

Been a long time since I've looked over this sort of thing, so take this with a grain of salt.

First off, you're using PNP transistors, which IIRC are generally a bit less efficient than NPN and require more current in cutoff. Consider switching to NPN.

It looks generally workable to me, although I wonder a bit about having two large, different inductances (the motor and the relay) there. It seems like a recipe for sparking at the switch and possible a bit of weird motor jerking. You might consider a largeish capacitor somewhere in the general vicinity of M1/Q2/U1. Shorting M1 with a capacitor will reduce sparking, while shorting Q2 with a capacitor might also reduce the chance that motor load jitter burns out Q2. Or just get a really huge Q2.

Oh, and a fuse. Always a fuse for anything driving a motor.

[dmaze]

Sparking at M1/S1 is definitely a problem...but it does make some intuitive sense to me that putting a capacitor across that should help issues. I just need to make sure that putting a capacitor and inductor in parallel doesn't create unwanted small-signal noise. Got it.

Maybe instead of the relay I actually want to use an optoisolator, though. I don't expect to be running any particular current on the LED/digital side of it, ind I do see loosely how multiple parallel inductance sources could be an issue. I should try to draw the result up tonight.

[dmaze]

Better? I guess there's no fuse. And I already know I need a really huge Q2. Also, I shall profess my dislike for gschem as a schematic editor; it is a worse UI disaster than most other Linux programs.

[iabervon.livejournal.com]

What's the left side that you already have? It seems to me like the right general strategy is to put a resistor between the ground track and ground (instead of having them connected directly), and do a thresholded voltage measurement across that resistor. If you can get an appropriate mosfet (so that its min V_GS * the range of currents you want to supply isn't too big a fraction of 12 V), it should work with a pullup resistor to produce your desired signal for a high-impedence input (so not an LED directly, but logic driving one). Or measure it with an analog comparator; ICs are cheap these days.

[dmaze]

As rjpb surmised, the actual application is my model railroad. I have a 25-year-old analog controller for it that works fine. But somewhat to my surprise you need to be a somewhat geeky model railroader to go for signals. I've had the "put a small resistor in-line and measure the voltage across it" strategy mentioned before, but I don't really know how to do that from first principles and have no idea how to discover which parts exist much less are useful.

The other exciting detail here is that I might not want variable-voltage DC across M1; there's an alternate standard where I can use +/-12V to send digital signals. That simplifies one aspect of this (I can send "don't move" across the tracks and still detect block occpancy, where now no voltage implies no current implies no detection). It complicates one aspect (the train can respond to the controller by manipulating its current draw). I can buy a DCC system off-the-shelf, but it's pricey, nice ones are overkill for the layout I have, and it still doesn't do signals.