NOOB FAIL: soft power frustration

[andyturk]

For the last few days, I'd been expecting another envelope with prototype PCBs from  OSH Park. I need a power supply for a project gizmo and after some nosing around on the inter webs, I came up with this:



The idea is to take a cheap DC DC boost regulator (NCP1402) and set it up with a soft power switch. Closing SW1 applies positive voltage to the chip-enable pin of the convert. This powers it up and applies juice to the mcu which will then turn around and raise POWER_ON via a GPIO pin to keep the converter enabled. At that point, the power switch can be used as a normal switch by the mcu. Two diodes (D2 and D3) keep the power enable output from stomping on the switch input.

The NCP1402 doesn't have built-in reverse battery protection, so based on TI's Reverse Current/Battery Protection Circuits app note, I stuck an NMOS FET in the ground path of the battery.

JP1 was to make it possible to measure battery current under operation. R1 pulls down the converter enable pin to prevent it from turning on accidentally. C3 is there to debounce contact noise on SW1. The inductor, D1 and C2 are part of the boost circuitry.

Have you spotted the problem yet? I certainly didn't.

Anyway, the boards showed up tonight:



An important observation is that it takes a special kind of asshat to build a SOD882 package into a homebrew circuit. The problem is that everything looks "big enough" on the CAD screen--just zoom in more if you can't see it, right? Those things are frickin' tiny!


With enough tongue angle to lick my eyebrows, I did finally get the things soldered on without any shorts.

So back to the photo of the assembled board (above). You may have noticed the LED was lit. You may also have noticed that there was nothing connected to the power switch input and nobody's thumb was keeping the button pressed. So, why is the LED lit?

Because the circuit doesn't work.

Investigating a little, I found that even when the chip enable pin of the converter was grounded (turning it off), there was still 2.4v on the output. For a soft power application, you'd want the output to be at zero volts, so something was clearly amiss. The boost converter fired up OK and produced 3.3v when SW1 was closed or a logic high was applied to POWER_ON, so that part was fine.

The problem is that this particular boost converter has a nearly directly line from the battery to the output--there's only an inductor and a forward biased Schottky diode in the way. As it turned out, my battery voltage was 2.6v and the 0.2v drop was within what you'd expect for the diode in the circuit (MBR0520). It all made sense, unfortunately.

Prior to building this board, I'd looked a couple of different soft-power circuits. The central idea was having a way to control the "chip enable" of the boost converter. Then while digging through options on mouser.com, the NCP1402 popped up as the cheapest boost converter (around $0.80 each) that met the other requirements.

I never clued in the fact that the other soft power circuits used converters that completely disconnected the output side when the chip wasn't enabled. Noob fail.

But hey, at least I can solder a SOD882 now. ;-)

[TerminalJack505]

Dave would tell you that he's happy that your project didn't work from the get-go since you learned even more than you originally set out to.

[Jovian]

If you want to bodge it, why not disconnect the the gate of Q1 from the battery and connect it to the CE pin, might just give you what you want.

Regards,
Jovian.

[PA0PBZ]

If you want to bodge it, why not disconnect the the gate of Q1 from the battery and connect it to the CE pin, might just give you what you want.

Apart from the fact that SW1 needs the +V to enable the chip... catch 22 

[Jovian]

If you want to bodge it, why not disconnect the the gate of Q1 from the battery and connect it to the CE pin, might just give you what you want.

Apart from the fact that SW1 needs the +V to enable the chip... catch 22 

But you have the +V, the FET switches on the low side. As far as the FET is concerned all you have done is add a diode to its gate. I am talking about the Chip Enable pin not the output of the converter. So provided the FET will switch on at 3-Vd Volts, then it should operate.

Regards,
Jovian.

[PA0PBZ]

Ah yes, the open jumper in the diagram fooled me.

[andyturk]

Dave would tell you that he's happy that your project didn't work from the get-go since you learned even more than you originally set out to.

Yeah, probably. The bummer is that it's back to the drawing board on the power supply and another 3 weeks to wait for boards to play with. Maybe I'll make Christmas tree ornaments out of the failures.