[K3]+[DXG] Un-demo information

[geekmaster]

If you can login with user=fiona, passwd=mario, then you can cat /etc/shadow, cut and paste your ssh buffer to a "shadow" file on your host pc, and use John the Ripper with the fiona wordlist to crack any fionaXXX or fionaXXXX password "instantly".

[megacoupe]

I'm trying to repeat the procedure you used but I have a few questions:

Originally Posted by stangri

So following the DXG serial access thread I popped the plastic cover, got the USB/RS232 TTL adaptor plugged into my Mac and I could get into the login prompt once connected to Demo DXG.

1. Did you modify the USB/RS232 TTL adapter in any way? Can you describe which terminals/wires you connected to which plates on the serial port? (My current setup, from left to right, is: 1=5V 2=TXD 3=RXD 4=Ground)

2. How do you get to the login prompt mentioned above? I'm assuming you're using Terminal, but what program?

I'll probably ask you about the rest of post as I get to those steps.

[knc1]

Originally Posted by megacoupe

1. Did you modify the USB/RS232 TTL adapter in any way? Can you describe which terminals/wires you connected to which plates on the serial port? (My current setup, from left to right, is: 1=5V 2=TXD 3=RXD 4=Ground)

• Do not try to power the adapter from the Kindle!
• The Kindle interface is 1.8 volt and will not tolerate 5.0 or 3.3 volt interfaces. Translation: Do not be surprised if the serial interface port is now fried.
• Option your adapter to take its power from the USB side of the cable if it is not already configured that way.

Translation:

Connect ONLY the Tx, Rx and Gnd leads!

.
Inspect the board at the pad areas.
Using a very bright light to penetrate the green protective masking helps.

What you are looking for is the pad which is connected to the very large area "shadow" - THAT pad is the "Gnd" connection.

Now you are down to two wires and three pads.

On the SoC in the DX and DXG only the USB downloader mode uses:
57500,8,n,2

On the DX, DXG both U-Boot and the kernel serial console uses:
115000,8,n,1

All other Kindle models use only 115000,8,n,1

Using ONLY the lead marked "Rx" from the adapter (which you now have optioned for 1.8 volt operation - see above) as a test probe ;

With your terminal emulator running, already configured for the proper line speed conditions ;
(Doing this first prevents any of the setup command and initialization commands from reaching the Kindle.)

Connect your "test probe" lead to one of the remaining three pads ;
Restart the kindle ;
Do you see readable characters in the terminal emulator?

• NO: Try the next un-tried pad - repeat the "Restart Kindle" step.
• YES: That is the proper connection point for the "Rx" lead.

This leaves you with two pads and one wire.
Have just verified the connection points for the first two wires.

Without connecting the last wire, type characters into the terminal emulator ;
Do you see the characters on the emulator screen?

• YES: Disable the "character echo" in the emulator.
• NO: Continue to next step.

Of the two unidentified pads left, try connecting the remaining lead to the pad immediately adjacent to the pad which has the "Rx" lead connected to it. That one is most likely to be correct.

Now (with the Kindle running normally - not asleep or halted - restart it if not sure) :
Type on the terminal - Do you see the characters echoed on the screen now?

• YES: That is the proper place for the last of the THREE wire interface connection.
• NO: Try the last remaining pad.

If the last remaining pad does not show the characters echoed, yet you can see the messages being sent from the Kindle port - -
Then your SoC serial port is damaged, put it back together, send it back to Amazon - the device is not repairable.

Note: I have over 60 years of hands on experience with electronic hardware, the above will work if nothing has been damaged.

[geekmaster]

If the DX has the same series 1K resistor as the K4/K5, and if the serial adapter has an LED powered sinking current through the kindle TxD, you may need to remove that LED from your adapter and replace it with a 22K resistor like I did. The kindle was not able to output a low enough voltage on the external side of the 1K resistor with the current load of that LED to appear as a logic zero.

In other words, you need some level-shifting circuitry, which you can purchase, or you can make one with resistors and diodes as shown in the level shifter circuit I posted to a thread (look in the tools index).

Some serial adapters work with no level shifter, but others need part or all of it. It helps to have an oscilloscope so you can see peak voltage levels you are sending or receiving serial data.

You can harvest diodes and other components from the circuit board in a dead compact fluorescent lamp.

Also, the serial GND does not seem reliable in some kindles. It helps to attach to a known ground like a metal shield in the kindle, or via the USB cable shell by connecting the USB cable to the same computer as your serial adapter.

[knc1]

The photo shown on the page linked to showed one of those "general purpose" adapter boards, which from the silk screen markings looked to be a 5.0 / 3.3 volt board.

I requested the Make/Model of the board but am still waiting for that answer.

[geekmaster]

Originally Posted by knc1

The photo shown on the page linked to showed one of those "general purpose" adapter boards, which from the silk screen markings looked to be a 5.0 / 3.3 volt board.

I requested the Make/Model of the board but am still waiting for that answer.

3.3 is not enough if it has an unbuffered LED lit by a logic zero on the kindle TxD. Both 3.3v and 5v TTL use the same logic zero/one switching zone, and the kindle cannot output below that with too high of an external pull-up resistor.

And 1.8v TTL uses lower voltage ranges for switching, so needs a logic zero on its RxD even closer to 0v. I made my k4 and k5 use "open collector" inputs with a series diode, so either side can pull it to GND, and both sides pull their end up to their preferred logic one. And I removed that pesky serial status LED on my 3.3v serial adapter that interfered with kindle TxD output.

It is easier to use a real 1.8v adapter, but they are hard to google because they come in a 1.8m standard length, which confuses the search results. Cheaper to hack a level converter (if needed) for a cheap 3.3v serial adapter using the "super simple" serial level shifter I published...

But only testing will tell if you need it, and there is that "unreliable GND" issue on the serial connectors...

[megacoupe]

Originally Posted by knc1

Inspect the board at the pad areas.
Using a very bright light to penetrate the green protective masking helps.

What you are looking for is the pad which is connected to the very large area "shadow" - THAT pad is the "Gnd" connection.

Based on that description, it looks like pad #1 (pictured below) is the Gnd connection:

On the DX, DXG both U-Boot and the kernel serial console uses:
115000,8,n,1

All other Kindle models use only 115000,8,n,1

The above are the options I went with.

Using ONLY the lead marked "Rx" from the adapter (which you now have optioned for 1.8 volt operation - see above) as a test probe ;

With your terminal emulator running, already configured for the proper line speed conditions ;
(Doing this first prevents any of the setup command and initialization commands from reaching the Kindle.)

Connect your "test probe" lead to one of the remaining three pads ;
Restart the kindle ;
Do you see readable characters in the terminal emulator?

• NO: Try the next un-tried pad - repeat the "Restart Kindle" step.
• YES: That is the proper connection point for the "Rx" lead.

This leaves you with two pads and one wire.
Have just verified the connection points for the first two wires.

This is where I'm stuck. I've tried connection the "Rx" lead to all 4 pads and restarting the DXG in each instance, yet still no output.

Strangely, I've noticed that once the DXG is fully booted, if I touch the lead to one pad or another, I see a character that looks like a diamond with a question mark inside. Sometimes one of these characters appear, sometimes two, three or four. Occasionally, I got some other strange characters. Below is a pic I took after touching the lead to different pads many times:



What does it mean that I can't get a regular output but these weird characters show up?

[knc1]

From that picture -
Pin 4 is ground,
Pin 1 is the (unused) power connection -

Now just sort out Pin 2 and Pin 3
One is TX, other is RX

[megacoupe]

If neither produces an output in minicom when the DXG is rebooted... what to do?

I don't know if it's relevant, but instead of cutting up the wires that came in the package with the adapter (they were very short), I simply ripped some out from an old computer case. That wouldn't be preventing an output from showing up in minicom, correct? After all, crossing the Rx and Tx wires and typing on the keyboard produced letters in minicom...

[geekmaster]

You could still have voltage translation issues. You may need a level shifter circuit. You may have an LED on your serial adapter that is interfering. You may have a bad ground. You may also have the wrong baud rate.

[knc1]

Originally Posted by megacoupe

If neither produces an output in minicom when the DXG is rebooted... what to do?

Well, we could question if that J23 is really the serial port.

Maybe find some other reference on the 'net to confirm or refute the information in the article you are referencing.
(slim to no chance that it is wrong though)

I have not yet seen your post where you gave either the Make/Model of that adapter device or a link to the product page.
Without that, I can't access the technical details of that adapter device.

So the question about "did it damage the SoC port" will have to remain unanswered.

Also, that port is low voltage (1.8v) CMOS - easy to damage just from static electricity transfered by touching the contacts.

Translation: Keep open the idea that it has to be sent back.

Originally Posted by megacoupe

I don't know if it's relevant, but instead of cutting up the wires that came in the package with the adapter (they were very short), I simply ripped some out from an old computer case. That wouldn't be preventing an output from showing up in minicom, correct? After all, crossing the Rx and Tx wires and typing on the keyboard produced letters in minicom...

Correct - that shows the setup is working down to the wire ends.

Your other photo documentation shows that part of the setup should be correct.

[megacoupe]

Originally Posted by geekmaster

You could still have voltage translation issues. You may need a level shifter circuit. You may have an LED on your serial adapter that is interfering. You may have a bad ground. You may also have the wrong baud rate.

Here's a link to the adapter I bought: http://www.ebay.com/itm/180953407401

And here's a pic from user stranger_st of the very same adapter:



I don't know if you can tell, but there appear to be 3 LED's on the board: one that lights up when plugged into a USB port, and two others labeled TX and RX (which I've yet to see lit up).

As far as I can tell, two other users here (stranger_st and stangri) have managed to use this adapter without making any modifications to it. What was their secret?!?

[geekmaster]

Those clear things mounted at the RX and TX labels are LEDs. The one labelled RX turns on when the RXD line is pulled to ground. The problem is that the kindle 1.8v TTL cannot sink enough current to reliably pull that line down to a logic zero. On mine, I removed that LED and replaced it with a 22K resistor, and I put a diode in series between the RXD line and the kindle (with the band painted on the diode on the kindle side). That way the kindle TxD can pull the output to GND, or allow the 22K resistor to pull the adapter side up to 3.3v or 5v, depending on how is is set. See the "super simple" serial level shifter thread. You might not need the extra stuff that connects to the kindle RxD. Be sure to use a known good ground at the kindle, like the shell of the usb connector. The resistor needs to be about 20K or so to work properly, but some variance is allowed.

Here is the level shifter circuit:
https://www.mobileread.com/forums/sho....php?p=1897643

And its theory of operation:
https://www.mobileread.com/forums/sho....php?p=1898072

[knc1]

Pin 4 of J23 is such a "Known good ground"
You can see that the pad is part of the component-side ground plane in the earlier picture.

Finding the technical data sheet on the adapter is a bit more complicated:
http://item.sozhekou.com/5590509381.html
(Page down a bit, well, a whole lot)

Also:
http://onefrog.taobao.com/

That adapter board is intended for use from 1.8v to full 12v RS-232 levels with 16v tolerance.
http://prolificusa.com/portfolio/pl-...ge-controller/
The serial adapter's ttl communications are set by programming the on-board eeprom using the special Win-Driver that is available with the purchase of the adapter.
See the OneFrog distributor pages and the attached file.

I suspect you skipped that part of the procedure (programming the levels to 1.8v with the special Windows driver.)
That is probably how someone else could use the board and you couldn't - they did not skip the "program it to correct voltage levels" step.

The most likely situation now: Your serial port on the SoC is fried.
The Soc is a $10 part (minimum order of 10 units) - either figure out how your going to replace it without SMD re-work tools or ship the Kindle back home as "broke".

[twobob]

Debrick info Pm'd. Good luck.