Saturday, April 14, 2018

First Use Of New Tools


Using my new tools to fix my Droid MAXX, lying open front and center.

Sometimes life just amazes me. Three days after receiving my Adafruit tools, I needed them. Twice.

This isn't the first time that I've learned something completely new out of the blue, and almost immediately needed it. It's just incredibly serendipitous.

First Problem

Yesterday, Friday, I worked from home. The project I'm working on is embedded system firmware to control a daughter board. The main smart chip on the board was shutting down unexpectedly, which I could tell initially from the timeouts reading responses from it, and then from its status output LED going off.

There's a power enable line for the chip in the cable pigtail that I control via software, and scouring through my code assured me that no code path should be shutting it off during the scenario I was exercising.

Yet the LED was going off and the chip was unresponsive. Maybe it was a bug in the chip, and my commands to it were triggering the shutdown.

Or maybe something besides my code was affecting the power enable. So I pulled out my new meter, opened up the cable to the daughter card, and clipped the meter into the ground and enable lines, switched to VDC.

With the code in the state where power was enabled, the LED was on, the chip was responsive, and the meter showed 3.2V. With the code in the state where power was disabled, the LED was off, the chip was unresponsive, and the meter showed 0.0V. Ok, good, correct setup.

Running my tests, with additional debug logging verifying the code was NOT running through the parts that shutoff power, I saw the LED go off and the meter drop from 3.2 to 0. So something unrelated was indeed dropping the power enable.

After more testing, I noticed a pattern in the logging, unrelated to my code, that correlated with the shutoff. Now, correlation does not imply causation. But it's a hint about where to look.

So running a couple more debug commands that didn't exercise my code path at all, I was able to confirm that the correlated activity did indeed result in disabling power to the chip.

Yes! Phew, it's not the chip crashing on me, and it's not a problem in the code I just wrote.

Monday, I'll get together with the hardware guys and chase the problem down. First, I'll try it with another unit.

For all I know, my unit is damaged, since it's all pulled apart on my desk to connect the daughter board. Open frame hardware is always vulnerable.

That's an example of diagnosing a problem and ruling out two large potential root cause areas.
Monday update: Another unit showed the same problem. So it wasn't damage in my unit. 
I discussed the evidence in the logging with the hardware engineer. We identified some register accesses used to control GPIO (General Purpose Input/Output) pins on the processor that were affecting more pins than intended. These were causing the GPIO pin controlling the power enable to be cleared.
This shows one of the challenges inherent in embedded systems. You're working at a much lower, more direct level in the hardware, with lots of opportunities to create strange conflicts between different parts of the system. 
Second Problem

Today, Saturday, I was down in the basement, and as I put my phone in my pocket, it slipped and fell flat on its face on the concrete floor with a sharp thud. DOHH!

It's a Droid Maxx in a case that goes in a holster. The case protected the screen and phone from overall damage, but the power had shut off (yeah, another power problem). I held the power button in, but it wouldn't power on, even after several attempts.

I plugged the charger cord in. The display came on, showing 0% power. It tried to boot, then quickly shutoff. It just repeated this cycle continuously. So there was some level of system function, but no power.

The shock of impact had apparently knocked something loose inside, which I figured was most probably the battery connection. Dead phone. Stupid reason. Words were said.

And this is a sealed phone, not a removable battery like older models. More words were said.

I emailed my wife who was visiting some friends that my phone was dead. Ugh. I look forward to going to the cell phone store the way I look forward to going to a used car lot (although I'm being unfair, my last several experiences with them have been excellent).

But Internet to the rescue, I Googled "access battery in droid maxx" and found a page showing how to do it, plus someone in the comments had pointed to a YouTube video they liked better.

I read the page and watched the video. Yep, I can do that. Worst case, the phone is already dead, I can't make it any worse.

Since I didn't have a spudger or plastic pry tool as shown in the tutorials, I improvised one from a piece of 1/8" thick padauk, which is a very hard tropical wood. I split off a strip about 1/2" wide and beveled the end of it with a block plane.

Between that and a small flat screwdriver, I was able to get the case open. The metal driver tip did break a couple bits of plastic off the edge of the back, but no major damage.

This also shows why I ordered multiple tool sets (I've added an iFixit set to the shopping list on my Adafruit post, because it has pry tools). I used the smallest flat screwdriver from one set for lifting tabs, the T5 from another set to remove most of the screws, and the even tinier flat tip from a third set to get into the points of a T4 screw.

I used the compartments in the base of my Panavise to hold all the loose parts.

With all the screws removed, I separated the board from the display, and sure enough, I could see immediately that the tiny connector on the flat battery cable was loose on the back of the board. This wasn't the first time I had dropped it, so it probably loosened up progressively. Everything inside the phone is secured in place pretty tightly.

I pressed on it and it clicked positively into place. I pulled it back off to look at it, then clicked it back in. So I was confident that it was secure.

As I was handling the phone with everything folded together loosely, I accidentally held the power button in. I noticed the display come on, so I let it continue. It booted up (SIM removed), so power appeared to be restored.

I shut it down and reassembled the phone. The tweezers were a big help dropping the tiny screws back into their holes.

Popped the case back together, reinserted the SIM, and held the power button in. Voila! It powered on, connected to the network, and I was back in business, with a couple of quick tests to verify that things were working right and it was staying on. I texted my wife that I had fixed it.

That's an example of digging in and not being afraid to play around and see what's going on. It's all a learning experience!

No comments:

Post a Comment