LOLbooster Update: Overcurrent Protection Issues

Having abandoned the track-level voltage regulation, my recent testing has centered on the short-circuit and over-current protection. These tests are fun, as they involve loading down the booster as heavily as I can (a combination of 12V bulbs, LEDs, and non-DCC motor cars can get me to about 2A, according to my multimeter). I’ve found that the present configuration is not very tolerant of inrush currents, refusing to light more than about eight 50mA bulbs at once (since my multimeter is very slow, I can only guess that represents about 4A of inrush current); five if there are any motors idling on the track (non-converted locomotives draw a not-inconsiderable current at rest). The solution has thus far been to place a simple RC circuit on the current sense line to add a kind of filter/delay. My calculations suggested a 50–100nF cap and a 100ohm resistor should do, but I’ve also been experimenting with a 10uF cap as well (because I had one handy).

The results have been mixed. I haven’t been able to fit a proper heatsink to the LMD18200, because the breakout board that I’m using has no provision for fitting anything that’s very large. So, running a bunch of lights and trains at 2A for a few minutes gets it nice and warm. And by warm I mean hot (but not overheating). And when it’s nice a warm, something very weird happens. When it is cool, putting a quarter on the tracks under any conditions immediately trips the over-current circuit. When it is hot, this does not happen. Instead, if there is a considerable current being drawn, the quarter causes some kind of oscillation, the source of which I haven’t been able to identify, that makes everything behave quite odd (bulbs dim, motors stop, but still buzz), and reduces the current consumption to about 100 mA less than before the quarter was placed on the track. Needless to say, this is no good. Moreover, once hot, under normal loading conditions, the booster will randomly shut down because of phantom short circuits. Also no good.

Since the current sense line on the LMD18200 has a very high output impedance, and the RC circuit I’m using as a buffer probably does not, I’m going to guess that some kind of impedance mismatch is the issue, and stick a voltage buffer between the LMD18200 and the rest of the overcurrent protection circuitry and hope that fixes the issue. Tests to resume tonight, with luck.

3 Comments

  1. KenS says:

    Circuit breakers are typically thermal devices (larger ones may be magnetic or thermo-magnetic). If it’s in a hot environment, its going to trip at a lower current, as heat doesn’t dissipate as effectively (the likely source of your false positves).

    The oscillation is a bit odder, but DCC circuite breakers are known to fail to trip, or to trip slowly, when the load current is close to the trip point. I’m not sure of the exact mechanism here, and I’ve never been able to find an explanation. My suspicion is that the rate of change of current has something to do with tripping the breaker (and I could see that having an effect in a solenoid, but not in a pure thermal breaker), and when you’re already close to the limit, the rate of rise to maximum output in a short is slower. Further, undersized bus wires (excess resistance) can prevent a quarter test from working because they limit the maximum current the supply can produce. A lit bulb has high resistance, and could cause a similar effect, so your test rig may be part of the problem (or an excellent simulation of a badly wired layout). Testing with a pure resistive load might be an interesting comparison.

    Some combination of “slow to trip” and the LMDs attempt to reset once the problem clears could be the source of the oscillation. The LMD data sheet does mention “cycling in and out of thermal shutdown” as an effect of a short that doesn’t clear.

    There’s an interesting comment on the data sheet about a need to heat-sink the VCC pin with a square inch of copper on the PCB in the current Limiting section of the data-sheet; if your breadboard didn’t provide that, you could be having thermal issues there, too.

    • Don Goodman-Wilson says:

      No circuit breaker here: I’m reading the voltage off of the current-sense line, and feeding that to a schmitt-triggered monostable multivibrator that trips when the voltage reaches a certain level indicating 3A-ish. The mutivibrator is an IC, and it’s kept nice and cool, so the “breaker” is not the issue. The LMD18200′s sense line is, I think, acting strangely when the chip is overly warm.

      I do wonder about the capacity of the breadboard and jumper wires, and whether they’re hindering testing, or if I should simply consider myself as testing in border-line conditions.

      But it’s interesting that you found the line about cycling in and out of thermal shutdown. I wonder if that might be it…although why the thermal-sense line isn’t being asserted in this case is a little worrisome. Anyway, although I haven’t had a chance to do the testing yet, I did discover that I’d wired the RC delay up wrong, and that could account for at least some of the problem.

  2. Don Goodman-Wilson says:

    Did I mention that the weirdness goes away when I remove the RC network? It does. But then the inrush current of a bunch of bulbs and nonDCC motors is too great, and trips a short circuit.

Leave a Comment