Fry me a river

January 22, 2012 4:08 PM

Electronics emergency! I have to install a panel tomorrow that I wired up with the AN6884 IC, it was working fine, but then I stupidly and hastily used an AC adapter without testing the polarity....

Here's the panel when it was working properly. The IC was immediately fried when I hooked it up, I quickly replaced it with a different AN6884 chip and it more or less functioned, but now one LED remains dimly lit in the third row no matter whether or not there's sound triggering it. I can't imagine why that one LED remains lit, I replaced the particular LED with a new one and the problem persisted. I suppose there's a current leakage of some sort? Could that be attributed to a capacitor in the circuit or something else? Could the new IC be wonky? Any idea what I need to replace?

I will try to give more info as necessary. Thanks.

posted by cloeburner to technology (6 answers total)

It's hard to speculate usefully without knowing more about the circuit. Is there a single AN6884 controlling the whole display, or is there one chip for each arm of the spiral? If the former, how are the ten LEDs at each level wired up?
posted by hattifattener at 5:05 PM on January 22, 2012

2 thoughts:

For each ring/row, how do you have the LEDs wired - series or parallel? The AN6884 is only rated to sink 18.5mA max per output, which is ~ a single LED. If you've got them wired in parallel, then the load (even if not nominally driven) may be too much for the output, and behaviour is uncertain. Couple that with variation in LED specs (e.g. you have one with a slightly low forward voltage - either from the factory, or slightly damaged due to, say, a quick reverse voltage ;-), then that particular LED may light dimly first (or continuously). Ideally, you'd wire them in series, then follow the standard method for calculating R with series LEDs.

(Which is all somewhat counter-intuitive - how can you drive multiple LEDs when I_sink is limited to 18.5mA? The answer is "LEDs are current-driven devices, but ...")

It's also suspicious that row 3 (pin 3) is the "0db" level. I'd suspect any supply filtering capacitors first (not shown on your diagram, but most circuits would include a 10~100μF electro between V_cc and gnd, followed by the electros on the input (pin 8) and output (pin 7). Hell, they're cheap, replace all of them…
posted by Pinback at 5:09 PM on January 22, 2012

Ok, I have run this circuit before with multiple LEDs, 50 to be exact. It has been plugged in for over a month straight and is still going strong, so I don't know if it's that.

I have them wired series/parallel. Forward voltage of LEDs ~3.4v, forward current 20mA. In that particular row I have 3 groups of 3 LEDs in series with a 100 ohm resistor wired at the end of the series of each of them. The one that is acting faulty is by itself [there are ten LEDs per row] with a 500 ohm resistor going to VCC.

It wasn't acting weird before I fried the chip.

There is only a single AN6884 running the whole thing.
posted by cloeburner at 5:23 PM on January 22, 2012

OK, thankfully I had one more AN6884 laying around and I replaced the last one and now that light is functioning properly. Must've been something with the chip?
posted by cloeburner at 5:40 PM on January 22, 2012

Problem solved, I guess— presumably that channel wasn't turning off completely, which is a plausible malfunction for a damaged chip. It makes sense that the one LED on its own string would be the one to light, since the Vf of an LED varies a lot with the current.
posted by hattifattener at 6:04 PM on January 22, 2012

I'd still bet it's because you're overloading the chip. Your series/parallel design totals up to 80mA per output (assuming a 12V supply, which lines up with your given resistor values), against the spec'd limit of 18.5mA per output. In all likelyhood the chip will probably not only survive, but keep functioning fairly well - analogue chips like that can be remarkably robust and tolerant of abuse, particularly when it's intermittent - but all bets are off as to the behaviour of any given chip.

I'd suggest the first replacement chip had a slightly leakier output when overloaded than previous or subsequent individuals. Leaky enough to just light up the one parallel string with the lowest V_F value (3.4V for the single vs 10.2V for the series string) - which, as hattifattener says, varies a lot with current.

That said, I'd also bet that the first replacement chip would have failed sooner rather than later when compared to the original & 2nd replacement. But they'll all fail sooner than they would if they were otherwise loaded correctly. Myself, I'd use a driver transistor (where's 2N2222 when you need him? ;-), or an array driver chip, on the outputs.
posted by Pinback at 11:05 PM on January 25, 2012