I have a few K4H511638B-TCB3 left if anyone needs.
I have a few K4H511638B-TCB3 left if anyone needs.
Hi!
Here I have two Hynix HY5DU561622ETP Chips (16Mx16) from my bricked D-Link DNS-323. As I read in this thread one chip from that type is used in the v2.
Can I use these two chips for an upgrade from 32 to 64 MB? Or where do i get the recommended chips best? Neighter Hynix nor Samsung provide any free samples on their website like other companies do (TI for example) :(
Greeting, Sally
my waranty is almoast gone, so I was thinking of upgrading the memory too..
I can get the transcent memory, for 11 euro only:p
now I'm wondering, I'm not really super pro at soldering, but I've done it before with a point soldering torch.
How do I get these chips from their wat is it... sockets?
those pins aren't made of tin is it?:p
If I look at these images: http://www.wl500g.info/showpost.php?...0&postcount=30
then I see that in the 2nd image all the pins are melted away, including some parts of the chip... I mean, it looks a bit rough and very dangerous to do:p
If someone can explain me how to get these chips off properly without burning them away, I would be happy:D
The best way to solder/desolder such chips - is to use hot air gun or hot air station. It may be professional (ERSA, PACE), hobbiest (AOYUE-6031 Sirocco, AOYUE-908, etc.) or even good workers hot air gun (Makita HG650 for example) with small nozzle.
Some people upgrade router with an old soldering torch only, but this is not so accurate and require much time.
I strongly recommend you to train on something else before hacking router, for example - old PC expansion cards, motheboards, etc.
Anyway, you have to use good paste Flux (for example, http://www.efd-inc.com/Solder/Flux/ - good, but expensive) and always check quality of soldered pins. Most comfortable is to use stereo microscope, but magnifying glass is acceptable too.
thanks for your advice:D
I'll do some practice first yes.
never used flux on sutch a small thing tho, I used it for soldering big metal things together:p
I also found a good instruction video:
http://curiousinventor.com/guides/Su..._Soldering/101
it looks easy, as long as you have a steady hand I guess:)
need to get myself a new soldering iron, one with a temperature regulator:rolleyes:
Hi,
i tried to upgrade my router with two NANYA NT5DS32M16BS-5T, but it seems that one of the two chips is damaged. When I enable only 32MB of U3 everything works fine, also memtest shows no errors, but when I enable U4 and the whole 128MB the router shows some errors during boot:
The soldering seems to be fine (I have quite a bit of SMD soldering experience), and I also added the cap to UDQS of U4.Code:SQUASHFS error: lzma returned unexpected result 0x1
SQUASHFS error: Unable to read page, block 4000e, size 4769
SQUASHFS error: lzma returned unexpected result 0x1
SQUASHFS error: Unable to read page, block 4000e, size 4769
SQUASHFS error: lzma returned unexpected result 0x1
SQUASHFS error: Unable to read page, block 4000e, size 4769
Segmentation fault
/linuxrc: /linuxrc: 35: cannot create /proc/sys/kernel/hotplug: Directory nonexistent
Unable to handle kernel paging request at virtual address 43fc6034, epc == 8002d700, ra == 8002cdac
Oops in fault.c::do_page_fault, line 192:
$0 : 00000000 10008800 43fc601c efffffff 810b1dd4 00000060 83fc60e0 83fc6020
$8 : 10008801 00001f02 000043d1 00003d4a 2aaa8238 2abb24a8 00000000 00000000
$16: 810b1dd4 00000000 810b1ddc 000000f0 810aeba8 00004985 810fa2d0 802d0000
$24: 00000000 00453d88 803c2000 803c39d0 803c3bc8 8002cdac
Hi : fffffff9
Lo : ffffffa0
epc : 8002d700 Not tainted
Status: 10008802
Cause : 00000008
Process insmod (pid: 13, stackpage=803c2000)
Stack: 803c39d0 803c39d0 8103176c 8002750c 00000000 00000c00 00000400
00000000 8002cdac 802e64ec 00000001 00000014 810fa2d0 8002fe3c 80039870
00000000 00000000 810aeba8 80025dbc 810e5740 00000000 800399e0 80016548
803c3ac8 802e5120 00000014 810aeba8 00000001 00000400 00000050 00001f02
8003c048 811fb120 00000001 00000002 00000014 802e5120 00000014 00000400
8003c2bc ...
Call Trace: [<8002750c>] [<8002cdac>] [<8002fe3c>] [<80039870>] [<80025dbc>]
[<800399e0>] [<80016548>] [<8003c048>] [<8003c2bc>] [<80022374>] [<80022300>]
[<8000a8fc>] [<80039308>] [<80092934>] [<80095a68>] [<8000c43c>] [<80026468>]
[<80020228>] [<8000c43c>] [<80095f30>] [<8001684c>] [<800b619c>] [<800b5d44>]
[<8001670c>] [<80046974>] [<80016548>] [<800b6154>] [<800160cc>] [<80002134>]
[<80030238>] [<801a120c>] [<8002562c>] [<80030238>] [<800256d0>] [<80027428>]
[<8002750c>] [<80022258>] [<80022300>] [<80020770>] [<800225b4>] ...
Code: 00031080 00471021 00650018 <8c440018> 2402ffff ace40014 00001812 10820012 00c33021
Segmentation fault
Segmentation fault
Segmentation fault
Segmentation fault
/linuxrc: /linuxrc: 54: cannot create /proc/sys/kernel/hotplug: Directory nonexistent
Segmentation fault
Segmentation fault
Can someone tell me, if it's possible to activate the whole 64MB of U3 whitout activating U4, or if its possible to activate only U4 without activating U3?
BR Peter
Ok, i managed to enable the the 64Mb of U3 without enabling U4 with sdram_init=0x0013 . Infos about the sdram configuration can be found here: http://oleg.wl500g.info/sdram.html .
edit: Good news! After enabling the 64Mb of U3, I ran a memtest, which found no errors. So I changed U4, resoldered it, set sdram_init to 0x0011 and the router booted! Now I'm running a memtest on both chips.
For everyone who wants to upgrade the memory, I suggest to set sdram_init to 0x0013 after changing U3 (and before changing U4), and running a memtest on the chip (with ./memtest 32M).
Thanks to Oleg for this nice mod and his great firmware!
Br Peter
Well... I think I finally got my stuff ready, except for the memory module, since I get confused a bit:o
allright:
the samsung chips are 64MB x 64bit each, so you get a 512MB total in one module...
the transcent chips are however only 32MB x 64bit each, so you only have a 256 module...
now I guess I can not upgrade the router to 128MB with the transcent chips since there can only be 2 chips placed on the routers mobo:confused:
transcent: http://www.transcendusa.com/Support/...64V3M_1180.pdf
samsung: http://pdf1.alldatasheet.com/datashe...24BT0-CB3.html
or am I completely wrong here:p
the samsung ones are not really easy to be found somewhere here in europe.
I have been searching for a memory module too and found out that the chips on the Kingston KVR333X64SC25/512 that is listed in this thread are not usable. They are 32M x 8-bit (8M x 8-bit x 4 Bank) DDR333 SDRAM sealed in uBGA packages. See http://www.valueram.com/datasheets/K...64SC25_512.pdf.
Maybe they used to have the right chips and Kingston changed them...
Better check the datasheets before you buy one.
The Transcend module seems OK: http://www.transcendusa.com/Support/...3M_1184_s1.pdf
well... still it's 64MB
or I should say this: because the processor is 32bit, it will be able to adress only 32bit of the 64bit lane on the RAM module, so you get double space and it becomes 128MB:rolleyes:
but that seems to be really theoretical:confused:
I just received the Transcend module.
It has Hynix chips nowadays so if you are looking for Samsung or Elpida chips don't buy this module.
If you do buy this module, you can 'harvest' the 22ohm resistors from R1-R8 from the chip-side of the module or R9-R16 from the back-side. To get the clock timings right I will try to transplant one of the capacitors C26-C29 from the back-side the way NotebookKiller suggested.
To Wpte: each chip has 32M addresses storing 16bits (2 bytes, one word) of data each. That makes 2Bx32M=64MB per chip so the two chips will get you 128MB. For reading/writing a single byte the chips have UDQS and LDQS pins to activate the upper or lower byte only. See the datasheet http://www.hynix.com/datasheet/pdf/d...TP(Rev0.3).pdf
My WL-500W router was dead after I enabled U4. I used Hynix chip. I already soldered a 5pf capacitor at pin 51. It looks like the CFE isn't running. I couldn't see any output from the serial console.
Is there a way to disable U4 without booting into the system?
To Pim Borst:
Do you know which lines the capacitor should be soldered? It isn't very clear on the photo attached in NotebookKiller's post. There are two lines that are very close to each other.
Hi Andrew,
The two lines that are close to each other are the clock lines indeed.
I removed some of the PCB mask and measured that they are connected to CK and /CK of the chips (see the datasheet in earlier post).
Wednesday a colleague of mine is going to help me with the upgrade. He has professional hot-air soldering equipment. I hope it will work. Good luck with your WL500, I hope you get it working again!