Now, from reading forums and reviews online before hand, I knew what . . . → Read More: Augen Gentouch 78 Android Tablet]]> I picked up an Augen Gentouch 78 Android tablet the second that I heard they even exist from, of all places, K-Mart.  I guess I was lucky to find one in stock on August 10th since they were apparently in such high demand.

Now, from reading forums and reviews online before hand, I knew what I was getting myself in to.  The thing has its share of problems, although most of the annoyances are currently incomplete software.  Considering that I was wanting to build my own Android tablet/in-wall PC for over a year, I jumped on this because of it’s low price.

I decided to crack the case open to see exactly what was on the inside, and to see how easy it would be to hack should I decide to add functionality to it.  After struggling with those pesky, fragile plastic clips, I finally got the back off (click any picture for large size):

As can be seen, I guess I got lucky and received a 2100mAh battery.  With the version 3 Augen update, my battery indicator no longer goes to 100%, but it still seems to hold a charge for a good 4+ hours of moderate use.

The main board appears to be partitioned (by ground rings) into 4 sections:  Wireless (top-left), Unknown (top-right), Power Supply (middle), and CPU (bottom).  I tried to take notes of the markings on all the major chips.

First is the WiFi chip in the top-left.  It appears to be labeled “ROCm / ATHEROS / AR610ZG-BMZD / E022410-51C /00880068 /1009 PHIL”, clearly indicating an Atheros WiFi chipset.

Next, in the power supply section, is an 8-pin IC near the power button labeled “RT8272 / GSPCCW03″.  Searching indicates a high voltage buck converter.  Some people indicate that the USB port is incapable of charging the 8.4V battery because USB is limited to 5V/500mA.  This buck converter’s datasheet indicates an input voltage rage between 4.75V – 24V and a configurable output range of 0.92V to 15V.  I didn’t do any testing (perhaps I should have), but it’s quite possible to speculate that they could be boosting the USB voltage levels here to (attempt to) charge the battery.  My suspicion is this:  If the battery typically only lasts 4 hours on a 2100mAh battery, that tells me the unit probably draws about 525mA during operation.  That’s clearly over USB’s limit of 500mA, which would explain extended battery life, but not actually charging the battery.  One test that should be done is to see if the battery indeed charges via USB while the unit is powered off or in sleep mode.

There is another 8-pin IC on the middle of the power supply section near the LCD connector labeled “020L1/ 8057Q”.  I couldn’t find any info on this chip.

The far right 32-pin IC near the LCD connector appears to be labeled “ALC5621 / A4K62S1 / GA18C” which is an I2S/PCM Realtek Audio Interface Codec.  The datasheet link on Realtek’s website didn’t work, but I found another copy and uploaded it here: ALC5621.  The datasheet clearly shows Microphone inputs, Line inputs, and Aux inputs, so it should be possible to add those features to this tablet (might require a driver/Android change in software to use).  In particular, microphone inputs should be on pins 32, 1, 2, 4, and 5.  It’d be interesting to trace these on the PCB to see what connectors they come out of (if they even route anywhere), but with careful soldering, you could probably just tap directly to the pins.

Next up is of course the CPU, which we already know is a TeleChips TCC8902.  I deciphered the additional markings as “0BX / AN2D6K-CN / 1023″.

The NAND flash chip is a 16Gx8 labled “SAMSUNG 013 / K9GAG08U0M / PCB0 / YCC687AW”, which should be the advertised 2GB of flash in total.

There are two 1Gb DDR2 SRAM Hynix chips as well, labeled “hynix   KOR / HY5PS1G1631C / FP-25   927A / NTHA8Z83MQ1″, which confirms 256MB of RAM total.

One chip that caught my attention was the 8-pin IC near the RAM.  It is marked as “AT88SC / 0104CA / U   1016″ which indicates an Atmel Secure CryptoMemory chip.  Could this be utilized by a future update from Augen to lock us out of custom Android images via the bootloader?  Is it intended to allow for a SIM card to be added?  Is it intended for some form of DRM?  It’d be interesting to see how this chip is utilized now, and in the future with more updates.  As of now, Augen has been quite generous at keeping this device developer-friendly and hackable.

The three crystals by the CPU I couldn’t figure out.  Their markings are, from left to right:

• “2700P
  H78LA”
• “DA7263″
• “NSK L
  120JH”

As can be seen below, the bottom side of the mainboard just contains passive components, nothing interesting.  Also, there is another PCB that is not utilized whatsoever.  This PCB is labeled “MID 810-LCD-V4″.  It even appears that the resistive touch interface was patched to bypass this board.

And finally, I took a photo of the sticker on the backside of the LCD.  The sticker reads “HLY070ML245-12A / 10072901719″.

One extremely disappointing thing was that I saw no indication of a G sensor or accelerometer in the thing, despite the user manual indicating that one should exist for auto screen orientation.  If I hear of a new batch of these coming out that does include it, I’ll seriously consider exchanging my tablet for one that does, or by finding out where it should be attached and install it myself.  I’d also hope that version 2 has better soldering jobs and better quality control on the product.  I’m glad I opened mine because the WiFi antenna was actually getting in the way of the stylus from a very loose screw and too much antenna cable that was badly pinched.

Update: Thanks for all the comments below.  As of November, just before the K-Mart 90-day return policy expired, I returned the Tablet for a full refund.  Two support tickets with Augen that were simply closed with no attempt at resolution, lack of promised and much needed updates, and false advertising on the box and in the manual regarding features meant this tablet was not worth keeping.

Reelights work through magnetic inductance. Two magnets on the spokes of the wheel spin around and briefly . . . → Read More: Reelight SL-120 circuit theory and making it flash flasher]]> I noticed the other day that my front, white, Reelight SL-120 had a crack in the plastic. Since it was partially cracked and needed to be fixed I decided to take it apart to see how it operated.

Reelights work through magnetic inductance. Two magnets on the spokes of the wheel spin around and briefly pass by the light. The light has an iron core that meets the magnets as they pass by. The iron core is wrapped with enamel coated wire. As the magnets pass by the iron core, the collapsing magnetic field creates and electric current that travels through the wire. The electric current in the wire travels to a capacitor where the energy is stored. The energy in the capacitor is then used to power a 555 timer circuit to flash the two white LEDs.

Simple right?

But what if you wanted to make your Reelight SL-120 flash a little faster?

First we need to look at the LM555 datasheet to figure out all the elements of the circuit and how Reelights has implemented the circuit.

It looks like the SL-120 uses an Astable waveform. The frequency is 2.2 Hz (roughly – I counted the number of pulses over 60 seconds).
The astable equation looks like
2.2 Hz = 1 / T = t1 + t2.
t1 = 0.693(Ra + Rb)C
t2 = 0.693(Rb)C

Looking at the Reelight SL-120 I see that 560k is Ra and 56k is Rb but the cap is un-marked. Damn. Gotta do this the hard way…
2.2 Hz = 1/0.45
0.45 = 0.693(Ra + Rb)C + 0.693(Rb)C
0.45 = 0.693(616)C + 0.693(56)C
0.45 = 426.89C + 38.8C
0.45 = 465.69C
C = 0.000966 (roughly 1uF)

If we wanted to make the lights flash more often, we would need to change Ra.

For fun I threw a 100k resistor on top of the 560k resistor. What happens is that Ra is now 84.85 [1/(1/560 + 1/100)] so the frequency would be:
1/(0.693(84.85 + 56)0.001 + 0.693(56)0.001) = 7.33 Hz

To debug the printing problem and to add another, better layer, . . . → Read More: Timestamp in Linux using ftime for millisecond timing]]> I am working on an embedded project at the moment that prints debug messages to a serial console.  Interestingly, it seems that some of the prints are arriving out order from the actual flow of the software when they originate from the same thread.

To debug the printing problem and to add another, better layer, of debugging to the project I added a time stamp to a preprocessor definition for the existing printf() macro.

Here is the existing print macro:

#define debug_print(fmt, arg)          printf (fmt, arg)

To add a time stamp we need to use the timeb struct from sys/timeb.h.  Since a struct is used, we need to give it a focus. Otherwise we’ll get compile errors. It will need to be put in a false do while loop.

#define debug_print(fmt, arg) \
  do { \
    struct timeb tp; \
    ftime(&tp); \
    printf("%ld.%d: " fmt, tp.time, tp.millitm, arg); \
  } while (0)

Now, in your program you need to make sure sys/timeb.h is added and then use the new print macro.

#include <stdio.h>
#include <sys/timeb.h>

#define debug_print(fmt, arg) \
  do { \
    struct timeb tp; \
    ftime(&tp); \
    printf("%ld.%d: " fmt, tp.time, tp.millitm, arg); \
  } while (0)

int main() {
  while(1) {
    debug_print("This is dbg print and a number: %d\n", 10);
    usleep(10000);
  }
  return 0;
}

Enjoy!

That’s an error that can be displayed on a Toyota/Lexus MFD after changing the map region in a service menu.  I managed to select one that’s not for the disc I have.  This setting stays in flash in the navigation unit, which has a battery . . . → Read More: please insert correct map disc]]> Well, I made that error appear because… I’m an idiot.

That’s an error that can be displayed on a Toyota/Lexus MFD after changing the map region in a service menu.  I managed to select one that’s not for the disc I have.  This setting stays in flash in the navigation unit, which has a battery backup.  In order to ‘re-set’ the navigation unit, it must be dismantled and a board that has the battery must be disconnected.

My many thanks to htmlspinnr from priuschat who described the processes perfectly.

Pull NAV ECU:

• Remove rear plastic cover by pulling plastic retaining clips/pins. I found two flathead screwdrivers helped with this.
• Remove 4 8mm/philips screws, 2 on each side retaining ECU.
• Remove 4 harness connectors at front of driver’s seat.
• You’ll need to spread bracket just a tad in order to dislodge the ECU from the bracket.

Open NAV ECU, perform “fix”:

• Remove 8 screws from the bottom of the ECU.
• Cover is held in place by 1 tab on each side that fits around a small bump on the top cover. Using a small flathead screwdriver, spread the cover at the bump. The ECU cover should then tilt up and come right off.
• With the plugs at the front, the left board is the NAV logic, and the right appears to be the AVC-LAN interface and power supply. There is a ribbon cable separating the two. These boards should lift right out, as the cover screws are holding them in place.
• At whichever location is easiest, lift the retaining clip at one end of the ribbon cable connecting the power supply board to the “logic” board, then gently pull the ribbon cable free. This effectively disconnects the battery backup from the ECU.
• I also spotted a black momentary switch on the power supply board. This may also be a reset/backup power interrupt, however being a multi-layer board, I couldn’t easily follow the traces to double-check. For safe measure, I press and held this for a few seconds as well.
• Re-insert the ribbon cable, ensuring that it’s aligned correctly. Push the retaining clip back down to secure the cable.
• Replace the boards, ensuring that the boards fit onto the alignment tabs. The power supply board was a bit tricky here.
• Replace the cover. I found that securing the screws over the power supply board first helped to keep it aligned as mine wanted to pop out of place.
• Replace the ECU, reversing the steps of removal.
• Power on. You’ll see on the NAV screen: Loading application, do not power off.
• I then saw the “Agree” screen. Success!

Interestingly, when I re-applied power to the navigation unit, the default GPS location was 1374 Pennsylvania Ave NW.

I created a chart with the known data collected about the voltage at the thermistor for  a list of temperatures.  I wanted a trendline to show . . . → Read More: excel trendline formula is inaccurate]]> While trying to create a graph with trendline to predict what a thermistor reading for a Prius’s radiator/engine temperature would be at certain temperatures I started battling MSFT’s Excel.

I created a chart with the known data collected about the voltage at the thermistor for  a list of temperatures.  I wanted a trendline to show the calculation so that I could use it to calculate the resistance of the thermistor at those temperatures.  However, after creating a chart and looking at the trendline formula’s calculation of known data, something was completely screwed up.

After some Google searches there was a link to Microsoft’s Help and Support website that indicated that “the trendline formula should only be used when your chart is an XY Scatter chart.” Great.  Well, at least I didn’t waste too much time on it and it was an easy fix.  However, why didn’t MSFT’s Excel team not grey out the trendline option for non-scatter charts?!

While pressing the radio’s seek button like frogger to . . . → Read More: it's the word, the word, the word…]]> Last week I was disparately looking for something good to listen to on the radio while I was driving around Atlanta, GA.  The radio stations in Atlanta seemed to have taken a horrible step in the wrong direction in the last 10 years… But I digress.

While pressing the radio’s seek button like frogger to avoid the massive number of rap stations I heard the end of some 70s sounding song with “It’s the word, it’s the word, it’s the word…”  Damn, why does that sound so familiar?

Right!  It was a sample in track two of “the word” from the Swordfish soundtrack by Paul Oakenfold.

Again, my music friend pointed that the chorus was from the movie sound track to Grease [the original].  It was from Frankie Valli – grease.  It was the last six seconds or so…

Paul Oakenfold – the word == Frankie Valli – grease

I’m using Internet Systems Consortium DHCP Server V3.0.3 (ISC-DHCP).

To give a specific address range, we first . . . → Read More: Sending DHCP IP based on clients hostname]]> I’m working on a project where I need my DHCP server to send an IP address from a specific DHCP range if the hostname is a certain value otherwise it needs to send a normal IP address range.

I’m using Internet Systems Consortium DHCP Server V3.0.3 (ISC-DHCP).

To give a specific address range, we first need to detect if the DHCP client is sending the option 12 host-name and if so assign it to a class.

class “FooBar” { match if (option host-name = “foobar”); }

Once the client has a class associated to it we need to give it an address from a pool that doesn’t deny it.

subnet 192.168.0.0 netmask 255.255.255.0 {

Other DHCP settings here but make sure that the range is not specified; it overrides the pool settings.

pool {
allow members of “FooBar”;
range 192.168.0.90 192.168.0.99;
}

Otherwise we want the systems that don’t specify a hostname to get an address from the normal IP address pool

pool {
range 192.168.0.254 192.168.0.100;
}

Add more pools if needed

} # ending the subnet declaration

Pretty simple, hu?

UPDATE:

Thanks to a post I read by Bill Stephens of the PepsiCo Business Solutions Group, you can match a hostname using substr.  If your DHCP clients are named foobar-xyz you can use substr to match for foobar:

class “FooBar” {

match if ( substring(option host-name,0,6) = “foobar”);

}

(This would match the first 6 characters of the option 12 host-name)