SONY VGN-P series sub-notebooks INTERNAL

This page / file contains true hardware internas.


Disassembly instructions 1st Gen

The VGN-P is not designed for easy access, but accessible without destructive means. The following applies to 1st generation VGN-P, e.g. VGN-P21Z.

If you prefer video over text, search Internet or There are bad videos, too, needing a lot of imagination.

WARNING: I do NOT recommend the actions described here to be performed by anybody. You work at your own risk when you feel the need to take apart a VGN-P.

These instructions are for HDD / SSD access, replacement. May need to unscrew and unplug more when the target is different.

Close cover. Remove Battery. Below Battery approx at 1/4 and 3/4 of Battery width there are 2 screws, remove.

There are 2 tiny blocker-posts located 25 mm from each screw, to inner and little to back end, preventing the slide described some chapters down. Usually these open without help, but better insert some stiff thin plastic sheet to open these for sure.

Turn around, open cover to vertical position. Remove the two rubber distance holders in front left and right corner of keyboard to find screws below, to be removed.

Slide the whole top assembly surrounding and including keyboard approx 3 mm to the front. Lift it at the rear with great caution due to 2 foil cables in the left part of the front. Turn it into vertical position.

It works best to call another person for help now, to hold the top assembly in position, and later the display assembly, too.

Remove 2 screws near front. One approx 1/3 from left edge, the other at approx 2/3 and being the left one in the PCB bearing the Instand-On- and Power-switches and Power + Charge + HDD LED.

Remove 3 screws located 1/2 to the rear, one left, one right, one approx 2/3 from left inside a corner of a cable channel having a T junction here. May be covered by glue tape, or removal of glue tape covering cables may be necessary to lift cables to access screw.

There are other screws you should leave in place, although they bear the same imprinted symbol in case next to them.

Remove 4 screws holding the hinges of display assembly in main assembly, near rear, near left and right. DANGER: Careless handling may destroy cables in rear-right corner running above that hinges end, especially the one going to loud speaker. Rotating display assembly hinges requires significant forces.

Everything needed to be loose is unfixed now. Pay attention to small folded piece of metal inside the Kensington lock hole in front position of right side.

I suggest not to unplug keyboard and pointer.

How to proceed is a matter of personal taste and reliability of your helper.

Suggestion: Rotate keyboard assembly down in original position. Do not close device, because re-opening it needs great care, see DANGER above. Turn the whole device to the left into vertical position, resting on its left short side.

The bottom case can be removed while open-holding the two obvious snappers in front side of Battery bay.

The HDD is visible now, and its cable running over it. Unplug cable at its destination.

The remaining work is obvious for the qualified reader.

To assemble, work in reverse order. Pay attention to position of sliding switches WIRELESS and ON before sliding back the top assembly, and carefully hold all part of top assembly equally down, so little grappers (noses) meet pockets in keyboard metal base plate.

Disassembly instructions 2nd Gen

I have not done it myself so far. Just looked at the photos and collected the key points different from 1st gen.

There are 2 screws hidden under the rear rubber foots, too.

Hardware analysis

This is neither complete nor necessarily correct. It has been written as by-product while researching a Memory Stick issue, to serve as a personal reference, to reach any possible target faster the next time. Did another disassembly 2011-07-08.

Description is from left to right edge of VGN-P.

Audio stripe

A 15 mm stripe with some Audio or network stuff (guessed), e.g. a Realtek tiny TQFP chip bearing the Cancer symbol. Realtek High Definition Audio Driver is installed in WinXP. A small (less 10 mm diameter?) coin cell in shrink-tube is 2-side-glue-taped on top of Realtek, wired to connector on main board. This is clock backup power.

Main Board

The 68 mm wide mainboard. Sandwich at this location described vertically from bottom to top.

Card sockets sit on 1/2-deep board screwed below mainboard (screws from top, accessible after a lot of disassembly, taking mainboard out). Card detection switches open on insertion of card. M.S. socket is tyco brand. Only a few ceramic capacitors next to sockets. SMD LED between sockets as activity indicator. Starrflex-PCB connection to Comm board. The converter / controller to USB2.0 and power switches for socket power (but may be inside controller as well) are not on this board.

Power Supply

Main board bottom-front nothing on PCB, room for cards to slide in. Bottom-rear the switching power supply components: larger coils (3* 3R3, 2* 1R5 smaller, 1* 4R7, 1* 100, 1* 1R2), some electrolytic (may be non-fluid) capacitors (e.g. 270 near input, near other side 2* 220, 1* 330), some transistors, chips MAX8785E, MAX8796G, MAX17017GTM, ceramic capacitors, ... .


Top-front are BGA SDRAM, covered by some thin copper-plastic(?), 8 pieces, BGA, NANYA NT5TU256T8DY-3C. 2 chips per package. 8 Bit data per package, means no chance to remove 4 of them to get lower Standby current, for applications where 1 GByte RAM is sufficient but power matters. Current during computer Standby is probably the "self refresh current" parameter given max 20 mA per package at 1.8 V, means sum for 8 at most 0.288 W, or approx 40 mA drain from Battery at 7.4 V. Unfortunately no typical current value given. An experiment (see main html) said 40 mA drain from Battery, but we have to subtract the OFF current of approx 9 mA and still do not know how much needed elsewhere, but can say RAM does not eat more than 31 mA Battery or 0.23 W. A VGN-P610 with 1 GByte only can be expected to have half Standby current drawn by this sub-circuit, if same RAM chips are used, with only 1 chip per package.

##### NOTE: There is a very interesting feature in data sheet, for sure not used by WinXP, and probably not used by Win7, either. It is "Partial Array Self Refresh", permits to keep only part of RAM data valid, in steps of 1/8 of full capacity. ##### COULD BE TARGET FOR a nice low-level software project.

For comparision, the Micron MT47H256M8 has max self refresh current max 12 mA in standard and 8 mA in L grade, means significantly less. It is single chip for 2 GBit. It has no "Partial Array Self Refresh" as far as I see.

Samsung does not list any DDR2 2 GBit devices on web 2011-07-10. I checked the 1 GBit. There are simply no mA values given in Rev 1.2 data sheet of K4T1G084QQ. 2011-07-10 in the Mobile section they announce as new the 1 GBit chips, which have Partial Array Self Refresh, but all DDR2 are still listed as Under Development, no data sheets on web. Same no data for simple DDR 1 GBit, all Contact Us.

LPC partner

25H RV
chip in square BGA is the other side of LPC (Low Pin Count interface) of US15W, seen F2117 in disassembly photos of other SONY notebooks, in Viliv N5 and others. There is a small metal case 6pin next to it, probably quartz.

It is a H8S/2000 series microcontroller by Renesas, Still unclear which chip exactly matches, but the family features are matching the application here. UNCLEAR whether it scans the Keyboard itself, or uses 1 of its 2 PS/2 interfaces to talk to dedicated Keyboard controller.


SST 25VF080B in 8 pin flat SO (like PowerSO8 of MOSFETs) is FlashROM, serial interface, 8 MBit == 1 MByte, stores BIOS. SPI up to 50 MHz, 3.3 V powered.

Clock generation

ICS UMS9610CL in 48pin is clock generator.

CPU + US15W + others

Top-rear CPU + US15W + others. The CPU and US15W are covered each by an individual soft thermal pad. Above it is a black not too thin plastic sheet, covering the whole ALU heat spreader. The following heat spreader is below 1 mm tickness, probably not very efficient. Above it the notebook frame made of plastic, then keyboard with its thin (less 0.5 mm) alu base plate.

Comm board

Next 60 mm are the GSM board, serving as interconnection for various buses, too. Connected to motherboard by 2 wide starrflex extensions of it. Hosts sockets for cables to HDD, VGA/LAN socket (has some lines to internal Display and USB socket, too), SD/M.S. board (may be more on other side).

SD-Card + Memory Stick controller

SMSC USB2256AHZS in 64pin package, attached to USB. Looks like it has its own SST39LF010 45 BGA memory as neightboor. No data sheet for 2256 on web. Other family members are in other package. Anyway, family has a 8051 CPU core and allows firmware update over USB. Some GPIO lines usually present, some family members have CompactFlash interface, too. Mainly designed for external card readers. Some have 4 internal power switch transistors for sockets, up to 200 mA per socket. External transistors may be used.

SDVO to VGA converter

Crontel CH7317A-BF in 64pin QFN package, exposed pad. RGB outputs at VGA side have only a C-ferrite-C PI-network, but no 75 ohm termination to GND. This is not really necessary at start of line, but common practice with VGA controllers, to improve signal quality, to absorb reflections at start of line. It needs significant energy. Data sheet gives typical "Full scale output current" of 35.3 mA, sufficient to drive double-terminated 75 Ohm lines, and mentions them explicitly. ###### may try adding 75 Ohm to improve situation ### BUT THIS would cause driver to always detect connected monitor ####

Data sheet gives average power of 0.674 W (3.3V and 2.5V summed) at 1600*1200*32bpp*60Hz 162 MHz pixel clock, which should be same like 1920*1080*32bpp*60Hz.

Data sheet contains no register description. Is probably in SDVO spec by Intel, which is given under NDA only. There is this statement at Intel web: Chuck De Sylva (Intel) May 1, 2006 4:29 PM / PDT SDVO developer information / We do not distribute any information on SDVO to anyone that has not signed RS-NDA with us. There is no longer any external enabling effort, drivers for these parts are all done internally. There is no user accessible interface, the SDVO opcodes are not public, nor will they ever be.

GBit Ethernet controller

M (logo) 88E8057-NNC2 in 64pin QFN package, exposed pad. May mean "NN02"? Only Product Brief of web, no Data Sheet / Book / Users Manual. Asked sales rep by email 2011-07-10. What should be fixed? High power consumption without Ethernet cable plugged in, without connection to link partner, at least WinXP, see main text.

More components

There is a polyfuse 0603 or 0805 size X02.


4 pieces of tiny 8 pin plus backside contact (emerging at left and right) devices, probably backside GND, pin8 VCC, Pin1,2 unprotected side, Pin5,6 protected side, pin4 control, JUST GUESSING. Probably combination of EMI filter and ESD protectors for USB.

More tiny components, some looking like MOSFET transistors for power switching.

other side

The tiny Bluetooth module by Alps sits in a 10 pin socket and is double-side tape glued to Commboard to hold it in place. It had 2 different color paper stickers on it, green on top.

Larger WWAN (GSM+UMTS) module Qualcomm designed by Option sits in a socket.

Smaller WLAN module Atheros sits in a socket.


1.8 inch HDD follows. The HDD sits in 4 small rubber shock absorbers, which probably only reduce high-acceleration hits due to their limited size.

HDD height is 5 mm, but rubber block height is 8 mm. This means a 8 mm HDD may fit, or just misfit by less than 1 mm in height, because between the rubber blocks there is a little less than 8 mm height clearance.

THIS IS GOOD NEWS FOR HDD UPGRADE to large capacity. All provided there is no PATA(Host)-to-SATA(device) adapter in the starrflex cable to HDD. Anyway, it might be bad idea to press-fit a 8 mm rotating HDD, because it might receive any hit by typing on keyboard directly. At first glance, may look like only little acceleration, but human feeling is easily fooled in this field, would have to consider (compute, measure) carefully to decide.


At the very right edge, covering the front-rear area of [Fn] key (little more), there is one of these special tiny mobile phone antennas on an extra patch of starrflex PCB. Connected to Comm board by thin coax cable, more exactly to a tiny module, which is for sure Bluetooth. Some more coax cable go from Comm board to display assembly.


Access to the flexible PCB inside, which extends to the outside as strip of flexible clear plastics, plugging directly into connector, is only possbile by destructive means. The alu base has been fixed to plastic top by melding down and widening a large number of plastic posts.

This makes any idea to exchange right Fn and Shift keys by patching the flexible PCB (which is difficult on its own) too difficult or expensive to try.

Some more details

The hole for Kensington lock has internally a small metal part fixed with one of the screws to give it more strength.

Electric shield in plastic bottom case is an alu foil, but is not connected to device ground, as far as I see.

Most screws do not go into plastic, but into a metallic insert in plastic. Good.

Hardware analysis 2nd Generation, the VPCP...

As of 2011-07-10, only a few short words: It is the same concept, similar but not necessarily the same components, internally re-organized, means PCB look differently, have different interconnects, parts are attached at different locations / boards.

Standard Battery is VGP-BPS23, small mechanical differences to 1st Gen, without any real need for it. Looks like intended incompatibility. May refuse work even when electrically connected, not tried so far.

WWAN / GSM / UMTS is Gobi2000.

Keyboard changed, not only layout / some key sizes, but mechanical interface, too.

Author: Harun Scheutzow
Last change: 2012-02-20