Fixing a defunct PW1

[hjalfi]

Good news! I scored a free Kindle PaperWhite!

Bad news! It's free because it's broken.

I believe this is a PW1; the pictures of the inside certainly match. It's third hand from somebody who got it from somebody. The screen's frozen at the book reader page (showing Atlas Shrugged...).

When connected to power, the orange light comes on, but nothing otherwise happens. If I hold the power button down for 30 seconds the light goes off, to come back on again when I release the power; and sometimes (but not always) the computer I plug it into show USB error messages indicating that it's on the bus but is not accepting an address.

My assumption, therefore, is that it's getting partway through kernel boot and then panicking.

The next step is to hook up the serial port and see what it says. However, given the unutterable tinyness of the serial pads, and my not-very-good soldering skills (do I *really* need to solder to the *bottom* of that surface mount resistor? https://www.mobileread.com/forums/att...8&d=1352503423) I thought I'd better check first:

* Has anyone figured out a way to get the PW1 into USB Recovery or Fastboot mode without needing serial port access?

The light going off after holding the power button down suggests it's waiting for someone to press the magic key. Which, of course, the PW1 doesn't have. Are there a couple of pads to short, or something?

(Also, if anyone can suggest anything else to do before I wreck the thing, I'd be very grateful --- it's probably suffered hardware failure and is beyond hope anyway, but it'd be interesting to try.)

(Also, is that an unpopulated micro-SD card reader on the edge of the motherboard? Hmm...)

[dhdurgee]

Before you try anything more drastic, try simply leaving it on the charger for a few days and see if the behavior changes. If after an extended charge it still will not start normally then check out additional options.

Dave

[Branch Delay]

Honestly, soldering the serial was a slight pain in the ass even with a microscope and decent soldering station. My advice would be to sell it for parts on eBay then use the 20 or 30 you'll probably get to purchase a new PW3.

There currently isn't a way to get into any sort of bootloader/fastboot mode without soldering.

[hjalfi]

Yes, that sounds like a good idea.

On the off-chance, I booted up Windows and gave mfgtool a try. It looks like the device appears and disappears from the USB bus more or less at random (this is very obvious in Windows because of the chime as devices appear and disappear). When the device is present, mfgtool will connect and I can download a u-boot image into it.

However, I don't believe that it's starting correctly afterwards --- the diag image is supposed to make a USB drive show up, right? But it's hard to tell because it drops off the bus again shortly afterwards.

So, yes, let's leave it to charge for a while. I have had it on for about six hours in several bursts but a much longer charge won't do any harm and, as you say, may do some good.

[knc1]

Originally Posted by hjalfi

Good news! I scored a free Kindle PaperWhite!

Bad news! It's free because it's broken.

I believe this is a PW1; the pictures of the inside certainly match. It's third hand from somebody who got it from somebody. The screen's frozen at the book reader page (showing Atlas Shrugged...).

When connected to power, the orange light comes on, but nothing otherwise happens. If I hold the power button down for 30 seconds the light goes off, to come back on again when I release the power; and sometimes (but not always) the computer I plug it into show USB error messages indicating that it's on the bus but is not accepting an address.

My assumption, therefore, is that it's getting partway through kernel boot and then panicking.

The next step is to hook up the serial port and see what it says. However, given the unutterable tinyness of the serial pads, and my not-very-good soldering skills (do I *really* need to solder to the *bottom* of that surface mount resistor? https://www.mobileread.com/forums/att...8&d=1352503423) I thought I'd better check first:

* Has anyone figured out a way to get the PW1 into USB Recovery or Fastboot mode without needing serial port access?

The light going off after holding the power button down suggests it's waiting for someone to press the magic key. Which, of course, the PW1 doesn't have. Are there a couple of pads to short, or something?

(Also, if anyone can suggest anything else to do before I wreck the thing, I'd be very grateful --- it's probably suffered hardware failure and is beyond hope anyway, but it'd be interesting to try.)

(Also, is that an unpopulated micro-SD card reader on the edge of the motherboard? Hmm...)

Don't guess at the model, it does make a difference.

*1 As suggested above, leave it on a wall charger for a few days.
**days**
(It may be trying to re-cover a bad file system, which on eMMC can be really slow.)

*2 Try forcing the "DO_FACTORY_RESTORE" trigger name onto the USB storage.
Examples posted here recently for Linux,Mac,Windoze.

*3 No, you don't have to solder to the resistor, you solder to the pad very close to it.
If your hot soldering skills are not up to the task, see my recent "Paint and Play" thread.

= = = =

And since I am in this thread, anyway, I might as well hijack it now rather than fix my broken signature link:

Strange it is to be beside you
Many years and tables turned
You'd probably not believe me
If I told you all I've learned
And it is very, very weird indeed
To hear words like forever plead
Those ships run through my mind, I cannot cheat
It's like looking in the teacher's face complete
I can say nothing to you but repeat what I heard
That love hack is just a four-letter word

[hjalfi]

On the back it says it's an EY21. What's this?

(1) After some hours plugged in to the dumbest charger I can find the light goes off. I've left it plugged in, but I'm a bit less hopeful.

(2) It's not getting far enough through boot to show up as a USB mass storage device!

(3) That picture needs clarifying, then, because the arrow for RX is definitely pointing at the resistor, not the pad.

Re using conductive ink... that's very interesting. I had thought about trying to rig up some kind of jig using three needles pressed into the pads, with my TTL converter plugged in to the other end, but keeping the whole assembly rigid enough to make good electrical contact would be tricky. (I can just about solder a 0.1 inch connection, two times out of three.)

Hmm. How about a piece of blutack on one of the enclosure cans, supporting the three needles, with a big rubber band around the whole chassis to maintain pressure on the pads...

[knc1]

Just write-up anything that you find that works, there are a lot of people out there who lack skills at doing micro-miniature soldering.

The main point of interest to our audience is 'low cost' -
since one kindle, once in a lifetime is all they are going to be doing.
(A serial port connection is required to JB all 5.6.x firmware.)

That same connector pad layout has been used on all devices, starting with the K4.
Only the location on the motherboard varies.
(On newer devices, they have left a bit more room between the pad and the resistors.)

Other pictures (and related info):
KT2:
https://www.mobileread.com/forums/sho...20&postcount=3

PW2:
https://www.mobileread.com/forums/sho...99&postcount=6

[hjalfi]

Well, good news and bad news.

The good news is that leaving it to charge over the weekend fixed the appearing-and-disappearing-on-the-USB-bus issue; the battery must have been dead flat. It shows up as a 15a2:0052 "Freescale Semiconductor, Inc. i.MX50 SystemOnChip in RecoveryMode".

The other good news is that my bodged together needles-masking-tape-and-white-tack serial interface looks like it might work. I have some awful photos here. https://goo.gl/photos/imx1GnfZxqTAHJWn8

(The probes are sewing needles pushed into the sockets of a PL2303 serial converter. I place masking tape over one of the grounding cans so as to avoid shorting out the needles, rest the needles on the edge, push them as hard into the solder pads as I can manage, and then bond them into place with white tack. Also notice that I've cut away the corner of the orange plastic insulator which was covering the pads.)

The bad news is that I'm getting absolutely nothing out of the serial port whatsoever. I don't know whether this is due to my interface not making proper electrical connection, whether the Kindle simply isn't transmitting, or whether I've fried the Kindle's serial driver by connecting the wrong voltage of USB serial converter...

In exactly what circumstances can I expect to see serial output?

mfgtool works fine. It's a bit frustrating that I can happily download and run images on the device but can't see any output. (You know what would be awesome? A mfgtool image that contained a Linux kernel and initrd and which talked USB serial. Bet there isn't one.) However none of the available profiles seem to do anything when run. I get the impression that the diags profile should bring up a menu on the Kindle's screen, right?

[hjalfi]

...hey, look, the PW uses 1.8V serial! Which is probably why I wasn't getting anything; even if the driver survived being connected to a 3.3V connection it won't have the oomph to change the levels enough for the bits to register. Be nice to have found that out *before* I hooked it up.

Rumour has it the driver on the Kindle is reasonably robust?

[knc1]

I only see two leads connected in those pictures, you need three.
Connect up the ground lead.

And yes, you should see something on the serial port line.

Even if the kernel never starts, you will see u-boot messages on that port;
And even before that, you will see messages from the SoC's internal start-up code (such as when it starts to run build-in services, which you know has happened).

So yes, you should be seeing something on the serial port line.

Hook up that ground lead and recheck all of your work.

[knc1]

Originally Posted by hjalfi

...hey, look, the PW uses 1.8V serial! Which is probably why I wasn't getting anything; even if the driver survived being connected to a 3.3V connection it won't have the oomph to change the levels enough for the bits to register. Be nice to have found that out *before* I hooked it up.

Rumor has it the driver on the Kindle is reasonably robust?

Yes, it probably survived.
It has for other people.
I don't **recommend** using anything but what is in the Freescale technical manuals (because of habit, and I don't want to be buying a lot of burnt up kindles).

I tried to do my part:

Wiki 'Hardware' index:
https://wiki.mobileread.com/wiki/Kind...ex#Serial_Port

Forum index:
2012:
https://www.mobileread.com/forums/sho...ghlight=serial

2014:
https://www.mobileread.com/forums/sho...20&postcount=3

https://www.mobileread.com/forums/sho...0&postcount=14

And a whole lot on the subject over the last three years here.

All I can do is write on the subject, index the posts, and tell people how to find things:
https://www.mobileread.com/forums/sho...d.php?t=237083
(which is linked to from the 'sticky' "where do I start" first post)

After all of that, if the horse fails to drink and then
NMP

[hjalfi]

*Any* horse would find it hard to drink if it had to do so via frickin' forum searches. It's totally not your fault, but finding current information in any forum is genuinely really hard --- it's one of the least effective ways possible to distribute data.

At least this here is better than the xda-developers forum, which I've basically just given up on, and for that I thank you.

[knc1]

That is an unsolved problem in general.

Anything 'in print' ages, in technical fields it ages very fast.

But pick a format that might keep up with day-to-day changes and that results in a lot of disconnected information impossible to organize.

Twobob and I thought that a 'structured wiki' might be the answer.
We made a start at one:
http://wiki.knetconnect.com/twiki/bi.../TWiki/SiteMap
a couple of years ago, but it never picked up much interest.

(Note: I will probably have to put back up the 'leech guard' panel RSN - that site seems to attract DOS attacks.)

[hjalfi]

Oh, I know. (And just to make it absolutely clear: I'm totally not complaining at *you*; you've been very helpful.) It's just a hard problem.

Anyway, I'm trying to scrounge up a 1.8V serial adapter. I'll report back when it's ready. (And I totally intend to write this all up, if it works.)

[knc1]

Originally Posted by hjalfi

Oh, I know. (And just to make it absolutely clear: I'm totally not complaining at *you*; you've been very helpful.) It's just a hard problem.
- - - -

In addition to it being a "structured" Wiki - -
I added the ability for anyone to file an "Action Item" against the page they are looking at.

Each page has the ability to display (under a folded text - like our 'spoiler' tags here) a list of the "Action Items" filed against that page.

Each user's personal page has the ability to display all "Action Items" that are assigned to them or a "User's Group" that they are a member of.

The idea was to help combat "aging information" in the Wiki.
Turning every viewer into a 'proof reader'.

The prototype of what the user's personal page would be like:
http://wiki.knetconnect.com/twiki/bin/view/Main/MikeTwo
(Of course, the user can customize it to their liking.)

- - - - -

At about the same time I created that system -
TWiki "went commercial" -
So I lost interest in contributing it back to them.

If I ever tried to use that site (again) -
I would have to first replace TWiki with FosWiki (a non-commercial fork).