Before we begin: this has largely been written as notes to myself. Some things may not be structured or flow well as a result; I just needed references to my thought process for when I come back and pick up this project again later. If anything doesn't make sense, please ask - I'd like others to know what my thoughts are, but realise that the way I've explained them may not be completely clear.
Some months ago, I replaced all of the external lighting in the KJ except for the fog and headlights with LED bulbs. I've done this on pretty much every vehicle we've had for the past decade or so, and in general, it's been a pretty simple affair - replace the bulbs, then replace the flasher relay with an electronic unit so that the indicators don't hyper-flash when you're signalling a turn.
The KJ threw me for a loop on that last part: I wasn't expecting the hazard switch to contain not only the switch mechanism, but also IC control of the turn signal / hazard flasher relays.
I'm not a fan of splicing resistors in at each indicator position to cure hyper-flash - one, it removes the power consumption advantage of going to LEDs in the first place; two, it introduces the possibility of another point of failure in the electrical system, which is something I'd rather not have to deal with in the future. Clean, simple solutions are what I generally prefer over 'just make it work', so started looking into options for doing exactly that.
First stop was to do a bunch of reading up on how others have approached this problem. It's not that I didn't find good ideas, just more that they were ones which left me wondering if it might not be possible to approach this in a slightly different way. Having finally made a long-delayed trip to the junkyard, I pulled a couple of switches and decided to start looking a bit more deeply into how they work.
So, for the rest of the discussion, here's a visual aid in the form of a (not real great) photo of the top of the motherboard inside the KJ hazard switch:
After doing some digging, I was able to determine that the Atmel 6431B IC is part of the
Atmel 643 family of automotive flasher ICs. The 6431B appears to be Chrysler-proprietary, but as the pinouts on the ICs in that family match across the board, I'm fairly certain that this one will too.
That shunt between the two relays and the header for the electrical connector (barely visible, but it's the thing inside the oval that's usually described as an 'M-shaped piece of metal' when people are talking about this modification) is interesting. From reading through the datasheets for the ICs, it appears to have two purposes: one, to provide a reference voltage to the IC for telling if bulbs are good or blown; two, to act as a safety valve for the IC in the event that a bulb or its wiring is shorted.
So, basically, the IC and the shunt are co-dependent on each other to know the state of the bulbs. It makes sense that when people shaved down the shunt, their indicators wouldn't hyper-flash - they were effectively lowering the amount of current that the IC was being told to expect the bulbs to draw, and, since LEDs use less current than incandescents, it follows that feeding the IC a value more in line with what the LEDs were drawing would solve that problem.
My concern with this approach: if, for some reason, you ever need to go back to incandescent bulbs (or just temporarily replace an LED with an incandescent), things could get weird. I'm not sure what form that would necessarily take, but given that the shunt is also designed to protect the IC, my guess would be failure to work from seeing too much current coming across the shunt.
However, I did come across something interesting in the datasheets: there are ICs in that family designed to take a 30mOhm shunt, and there are ones designed to take an 18mOhm shunt. There are also ICs in there that are labelled as, "Frequency Doubling Disabling" - which is what I suspect we're looking for.
From a quick read of the various datasheets, frequency doubling is exactly what it sounds like - the technical name for hyper-flash. These ICs are designed to work in such a way that below a certain reported current draw, they don't hyper-flash, which has me suspecting that they're designed to run both incandescent and LED bulbs at the correct flash rate. All of the ones with this capability use 18mOhm shunts, and I'm not sure what the value of the shunt in the KJ flasher is - my guess is 30mOhm, but without pulling and measuring, there's just no good way to know until I do.
There's also an added complication in that the ICs that disable frequency doubling use a different physical form factor to the one that's already in there. They're a much smaller package, which would mean having to fit an adapter onto the board to be able to install them. This wouldn't normally be an issue, but I'm having a hard time finding one that will fit where the existing IC is due to the close positioning of the relays. Given the amount of free space inside the case when the board is mounted, though, I think it may be possible to replace the existing IC with a socket, run standoffs from that to the adapter board to get it up above the level of the relays, and still be able to fit everything back in. Just
Speaking of the relays: I don't think that just swapping them out for electronic ones will solve the problem. Their pulsing is controlled by the IC from what I've been able to tell so far (haven't done a full chasedown of the circuit, but enough to believe that this is the case), so they're probably just best left alone. What has me suspecting this is that the turn signals will hyper-blink with LEDs installed when worked from the stalk, but not when worked from the hazard switch.
In any event, it looks like there might be an option on the horizon for anyone else looking to convert to LEDs. I'm not going to be moving on this quickly; it's something that'll happen as spare time permits, but I am hopeful that we'll have another method for doing this relatively soon.