From: Matthias Wächter <matthias_at_waechter.wiz.at>
Date: Tue, 17 Sep 2002 13:22:52 +0200 (CEST)

On Mon, 16 Sep 2002, Stuart Tedford wrote:

> Excellent work.

Thanks!

> This mod should save a lot of battery power. Could you post the results
> when you get round to doing a test?

OK, here are the first test results.

0) The hardware

AJB Player 6000, updated to a 40GB IBM drive (4200 rpm).

a) Modification of the firmware (using ots-rockbox-1.3)

-> searching for half an hour within the various .h and .c files for the
2MB limitation, finally finding player.lds and modifying two lines of
code. Looks like the correct place.

b) The test data

A VBR encoding of a classical piece by Anton Bruckner: Symphony Nr. 4,
Psalm Nr. 150

-> 97.1 MB (101891241 bytes) in 5 tracks and a total of 72:16/72:33/72:38
minutes (Rockbox total/Rockbox actual/Winamp total). Total byte count
divided by Rockbox actual tells an average bitrate of 187257 bps.

c) The results

-> 7.81 MB buffer free in the menu. Wow!

-> The HD is spinning (red LED as indicator) at the following play times:

Track 1 (18:03): 0:00- 0:07 (HD was already spinning at start)
                 5:38- 5:48
                11:28-11:39
                17:07-17:20
Track 2 (15:43): 5:27- 5:37
                11:50-12:01
Track 3 (9:31): 1:57- 2:06
                 7:53- 8:05
Track 4 (20:31): 4:02- 4:13
                 9:47- 9:57
                15:24-15:35
Track 5 (8:45): 0:51- 1:02
                 6:34- 6:46

Makes a total of 13 spin-ups and a total of about 2:18 drive-up minutes.
Not bad for nearly-192kbps material within 72 minutes, is it?

d) Quality

Listened to the first track, no unexpected noise, then unplugged the
headphones and just noted the HD-on-times.

> I can't wait for the details on how to do this mod. I hope it isn't too
> hard to do, or to get the part.

Well, you name it: First get the part. I got a HYUNDAI (www.hynix.com)
part from a rare-part-reseller from Germany (www.bimex.de), packed as five
(so I currently have four spare parts left since I didn't make it to blow
the first part as expected :-)). I received the box at my post office
today by paying about 64 EUR MLP (Money Leaving Pocket). The part is
called GM71VS65163CLT5 (50 ns), and the data sheet is located at
http://www.hynix.com/datasheet/pdf/dram/GM71V(S)65163C(CL).PDF

Second is to un-solder the old part (in my case, the ISSI 2MB DRAM). We
did this with two heating fans, set up for 400°C and directed to the pins
of the device. About 20 seconds later, I could move the part with a small
screw-driver and drop it on the table.

After removing the chip I cleaned the solder pads with braid wire.

Then I took the new part and bent two pins (A10, A11) upwards - this is
because all other resoldering is done on previously NC (not connected)
pins except for these two pins.

I soldered the part by first mounting catercorner pins 1 and 26 and then
soldering all other pins with _enough_ tin-solder and possibly soldering
two or more pins together. Afterwards I used the braid wire again to take
all the excessive tin-solder from the gaps between the pins, so that no
pins are wrongly connected any more.

After becoming aware of the schematics error of my AJB (SH1's A1 is
connected to RAM's A0 and so on) I noted that the address pins A0-A9
(DRAM's view) have to be connected to A1-A10 of SH1's address bus, but the
order was unimportant, so I didn't swap the matching address pins. This
way, only 4 address pins had to be swapped/added by additional wires. The
other 7 wires were needed to move VCC, /WE, /RAS, VSS, /LCAS, /UCAS and
/OE to their 'new' places. Don't expect easy soldering here - you need a
really fine solder tip and eventually a lot of braid wire to avoid
short-circuits. I used varnished transformer wire and used some of the
vias on the board for easier soldering. I will post a picture of my
AJB-crime soon - I expect a digicam for birthday... :-)

After re-assembling the device I put in the batteries and - it started
Rockbox without any additional patch or correction! Of course, the info
menu only stated a buffer size of 1.81 MB, but there was no patch at this
time, of course.

At home I patched the firmware, updated the archos.mod via USB and started
the device. It's like in the fairy tale, it worked at first! It showed
7.81 MB buffer and drastically enhanced the play time per disk spin-up. I
don't know yet if the play time per battery charge is drastically enhanced
but it was a lot of fun anyway!

Sehr Wus,
- Matthias

-- 
                    "To get control over people, make them trust you.
                                            To make people trust you
                      don't try to tell them the truth about history
                 but make happen what you told them about the future."

Received on 2002-09-17