Pruning and board layouts

tieke always knew her circus skills would come in handy around the garden:

When not running around collecting dead tree, large chunks of the weekend were spent scratching my head, trying to fit all of this into the space normally occupied by one AA battery:

That’s a microcontroller, 4 Watt step-up power supply, accelerometer, LED driver, and switched output stage. Oh, and bonus connectors. Ended up with a design that looks like:

Three boards, to go on top of each other. There’s links between the boards that are hard to move around. There’s components on each board that can’t be moved. Getting everything to fit together was complicated. Didn’t I say I was going to do something easy this time?

It should all fit, and it should all work, but frankly, I’m just going to have to try it. Now, squeezing that much power into that small a space might mean that it all melts, but we’ll have to see…


10 thoughts on “Pruning and board layouts”

    1. Diptrace. I like. Admittedly, it can’t do more than one board at a time, the outlines above where drawn by hand, but hey, it’s free and pretty awesome.

    1. Yeah, that hole caused me hours of consternation, as did the fact that that transistors are 5 mm high when the rest of the components are 1 mm. Oh, and the output socket fits into the box, but not when the output cable is plugged in to the socket. Oops.

      Tonight’s job is printing all this out on card and checking the fit for real.

      Also, any idea why a 5V step-up is giving me a nice steady 4V? It’s a MAX1703 and the pin that sets the voltage is grounded, as it should be. Volts in are 2.8V off two batteries, switching the load from 600 mA of LEDs to a 100k Ohm resistor changes the output from 3.8V to 4.0V, but I don’t think I’m overloading the thing. I don’t have a good track record with step-ups.

      1. That’s weird. My first thought is to run it into something truly tiny, but you’ve done that and that indicates it’s nowhere near overloaded. Second thought would be to try messing about with putting in a resistive divider on the feedback pin and seeing if you can force it where you want it, but gah. I wish it were a controller rather than an all-in-one coz it’s really helpful to look at the PWM duty cycle under changing loads to make sure it’s working right. The datasheet makes what you’ve done look perfect.

        By the by, why did you go for through-hole transistors rather than sot-363’s? Power?

        1. Yeah, I’ll have to have a play with a resistive divider tonight to see if it does what I think it should.

          And for the through-holes, they’re peaking at 200 mA each and while I’d love a surface mount, three transistor array with bias resistors built in, I’ve yet to find one. And also, I’ve really no idea what I’m doing here, so I’ve even less idea about how to do it better.

          1. Ok, poking at it with the oscilloscope reveals spikes up to 0.8V on the feedback line, at the switching frequency. This could well explain the weird output voltage.

            This chip is on an adaptor and then plugged into a breadboard. The datasheet says a variety of things about how traces must be short, caps within 5 mm and so on, to which I mostly went “blah blah blah”, then that could well account for all this. (At least, that’s my hypothesis, which has the short-term benefit of explaining all my problems, for the long-term cost of not being falsifiable until I’ve ordered the proper PCBs, several weeks into the future.)

          2. And yes, could use FETs. My understanding is that they’re mainly used coz they’re more efficient, but I’m just driving LEDs here, so if I have a smaller voltage drop across the transistor, then I need a bigger voltage drop across the current-limiting resistor. So efficiency isn’t going to change?

          3. Well, one big plus of a fet (aside from not needing bias resistors) is that there’s no voltage drop across it when it’s in conduction, and voltage drops in switchers usually come right out of your efficiency. That’s why people use schottky diodes in front of the inductor rather than standard ones, and why synchronous switchers with another fet in place of that diode are so great from an efficiency standpoint (but suck from a design standpoint since you can get shoot-through if you’re not careful.)

            Layout is pretty important, especially getting the input and output caps as close as possible to the chip because there are enormous transients on those nodes, particularly between the inductor and the output cap. I try to use a ceramic backed up by an electrolytic (or if possible just a whole mess of ceramics) with a copper area/pour for the inductor/cap node, and yeah, under 5mm, preferably practically touching. That’s for a buck: I’d have to look at the datasheet for this one again. If I get a chance I’ll do some more looking at the layout and see if I can come up with anything good.

Leave a Reply

Your email address will not be published. Required fields are marked *