There have been calls for this LJ to feature more hot n3kk1d pics of me and less geeking out. As is well known, I live life selflessly, purely for the benefit of others, so I will oblige:


Anyway, back to the geeking out. I spent the weekend pfaffing with the regulated power supply for the Mitochondrion.

Removing surface mount chips, the easy way!
Run a piece of thin wire behind the legs of your chip. Nichome resistance wire is good – thin, strong, solder doesn’t stick to it much. Anchor one end (I just squished it between board edge and vice). Pull at right angles to chip, so all the load is on one leg. Apply soldering iron to that leg, bingo, leg comes off. Repeat for all legs. This takes about five seconds per side and rarely screws up the board. I like. This technique was found on the internet, somewhere.

MAX710, MOSFETs, and how to break them
The new centre board for the Mitochondrion does all the things it should, but only for five minutes at a time. The power supply chip heats up and shots down. Bother.

I’ve had four MAX710 fail in exactly the same way – the internal MOSFET dies, slowly. With a new chip, the resistance across the drain of the MOSFET (LX pin) to ground is huge, as it should be. When a new chip is soldered onto the board, the resistance is remains huge. Running it for a few minutes and the chip stays cool. Running it for five or so, things start to get hot. The resistance is now between 30 and 3 Ohms, depending upon how much I’ve used it. Low resistance => high current => overheats. Conclusion – MOSFET is ded.

Given that this is a design straight from the datasheet (fig 2a), I’m a little disappointed. I’ve tried on both breadboard and PCB, both die in the same way. No idea what’s causing this. The only deviation from the reference design is the distance to the big fat capacitors. It should be <5 mm, it is 15 for one, 25 for the other. Apparently this risks "ground bounce and noise". Datasheet (and google) say nothing about overheating. So, question for the device geeks here, could the layout of the caps be causing this? Or have I screwed up somewhere else? Still, that might be academic. Current plan for getting the whole thing to work, where "work" means gives me a steady 5V, 250 mA from 4 NiMH. However, “combinations of cell count and output voltage are to be strictly avoided: … four cells converted to 5V.” – says Linear. Crap. So, I’m thinking of giving up on the MAX710, and trying Linear’s solution for this, an LT1372 and seeing if I can melt that little sod as well. Alternative plan is just to use a simple step-up converter. When battery voltage is above 5V, it won’t be regulating, but 4 NiMH will only give me just over 5V, not 6V, and Picaxes at 5.5V tolerant. Which is the better idea? I dunno, and I’m not sure where to look to find out…

17 thoughts on “

  1. It took me a while to figure out what body part that was.

    The FET thing is really strange. The main reason a fet heats up is because whatever is driving it has slow transitions, but if it’s integral with the chip, that’s lame. It’s possible the inductance of the traces to the output cap, is giving the same effect. Can you stick a high-value 0603 across the output/ground right at the chip itself? We’ve resorted to soldering onto the legs of chips on some occasions. With the output at the end of the chip, it shouldn’t be too horrible to squeeze in something small.
    That’s a neat chip. I wish we made stuff like that. (Of course, I also wish it worked for you.)
    Out of curiosity, when it’s on its way to failing, what does the switch node look like on a scope? I guess they call it LX. It’d be nice if they provided screenshots of what the reference design looks like in continuous and discontinuous mode, so you had something to compare to, but it would be interesting to see if the switcher is changing.
    As a point of comparison, while we recommend having the input and output caps close to the chip, there are plenty of designs, especially LED drivers, that specifically state that the output cap is entirely optional. Those are primarily buck rather than boost. Still, having a chip cook based on that seems pretty lame.

    1. and by the way I dunno if you saw my little boost LM2731.

      It’s the one on the far right, board dimensions about 1 cm on a side.
      2731_boost
      It’s switching at 1 MHz or so, but when I check the output at that frequency I can’t see significant ripple if it’s under load, and it’s a pretty simple, small layout. The inductor footprint is really the only size-limiting thing, and Coilcraft has stuff that’ll run at the chip’s current limit in 1210, I believe.

      1. Ooh, that’s dinky. And dinky enough, coz all of this has to fit into a 18 mm wide, not much height, space.

        If I’ve got a a boost converter kicking out 5.0V, what will happen when 5.3V gets fed in there? Kindly tell me that it’ll be perfectly happy and let something between 5.0 and 5.3V out..

        1. (Reading fail) you want a buck/boost. That’s just a boost. If I get some time tomorrow I’ll see if I can come up with a clever tiny SEPIC, which is something I’ve wanted to do for a while as a nice bike light/generator setup with high efficiency.
          If you feed 5.3V into the output of a boost, it’ll just kick the switching down until it gets to 5.0V, or to as close as it can get to 0% duty cycle. It shouldn’t hurt it.
          And, by the way, I forgot and posted the picture of the larger layout. I *have* done an under-1-square-cm layout for that part, but it used annoyingly small components.

    2. I’ll try bodging a cap in there, see what difference it makes. I’ve one more unbuggered MAX710, and it’s no use to me just sitting there in the packaging. Hopefully I can get onto it tonight.

      1. Ok, I’ve got two questions / things to try:
        – what’s the frequency of the ripple? The scope pics in the datasheet are showing 2.5kHz in high efficiency and 20KHz in low noise, roughly.

        – does the overheating happen in boost mode (if you put less or flatter batteries in)?

    1. Re: Victorian!Rhi sez

      You’re just jealous coz my ankle brings all the boys to the yard,
      And they state quite clearly,
      Your ankle is insufficient in comparison,
      I would provide you with the relevant education,
      But I fear my fees are beyond your limited means.

Leave a Reply

Your e-mail address will not be published. Required fields are marked *