InsideTronics

New Nintendo DSi teardown

2008-11-12T01:13:00.005+01:00

We have seen the first teardown photos of the brand new Nintendo DSi games console, courtesy of Bunnie.

In the mainboard, we can see the following circuits:
uPD 46128512AF1 Its a Mobile RAM of 128Mbit, 16 bit, as we can see in the NEC web:
46 Mobile RAM
128 128Mbit
512 16 bit
A Revision A
F1 Plastic BGA (Wire bonding)
There is even a datasheet of the 64Mbit part.
Its a DDR RAM but with a SRAM interface. This simplifies the ARM CPU, that doesn't need a dynamic RAM controller.

The NAND chip also includes the controller. It is a SAMSUNG MOVI NAND circuit, that presents a MMC interface to the host, also simplifing the ARM CPU. The capacity is 256Mbytes and the bus width 8 bits.
Doing part number decoding we have:
kmapf0000m-S998
KM MOVI NAND
A NAND + MCU
P 256 Mbytes 8bits bus
F S3C49VCX02

In the bottom left corner we have a chip without chipmaker identification. It is marked as: PAIC3000D. The name is similar to the TI codecs: AIC3, and it is near the earphone connector. I suposse that it is the sound codec chip.

In the right corner the chip from Mitsumi (MM3317A) seems to be the power supply + charger circuit. There is no clues in the Mitsumi web page.

ipod nano 4g and ipod touch 2g

2008-09-13T01:02:00.006+02:00

Apple has launch new models of his well known music players: the iPod Nano 4G and the iPod Touch 2G.

iFixit is again the first to give us insight photos of the new models: iPod Nano 4G & iPod Touch 2G.

In the new models the circuit board is smaller and with less components. For example, in the Nano, the DDRAM is located on top of the processor instead of using a separate package.

In the iPod Touch 2G, iFixit found a Bluetooth + FM chip from Broadcom, the BCM4325, but it is not to give Bluetooth or FM funcionality but for the Nike+ ANT shoes interface.

As usual, well engineered products, with pretty good design and incredible shrinking PCBs.

More info: Nikkei Electronics Teardown

2008-07-19T01:01:00.002+02:00

Another teardown of the iPhone 3g comes from Japan, by the Nikkei Electronics Teardown Squad.

Following the link you could see more images of the iPhone 3G teardown process.

New iPhone 3g teardown

2008-07-13T19:48:00.004+02:00

We have the iPhone 3G teardown sooner than ever, with the ifixit guys going to New Zealand to be the first to open the device.

Later, Semiconductor Insights put their expertise decapping the chips, to know what chip is inside every Apple mark.

Fist analysis:

The processor circuit: the processor is the same in the previous iPhone. By the marking: 8900B, the processor is the S5L8900B (Samsung marking in the chip). An ARM11 processor found in its antecesor. The surprise is that the processor is not the new S3C6410 from Samsung, with the same peripherals but in 65nm instead of 90nm of the S5L8900. Probably too early to complete software debugging. It is a POP construction as its antecesor, with the DDR-SDRAM over the processor in POP format. There is also a serial 1MB flash from SST.

The phone circuit: The UMTS baseband is the PMB8878 from Infineon, with an Infineon UMTS transceiver, and the Infineon SMARTi power management unit. The power amplifier is from Skyworks SKY77340, filters from Murata, and an amplifier for each band from Triquint. The NOR Flash for the baseband processor program is from Numonyx (the new company created by ST Microelectronics and Intel integrating their memory divisions).

The touchscreen controller now uses only one chip instead of three. It comes from Broadcom: BCM5974.

The LVDSL serial interface for the display is again from National: LM2512AA.

The GPS chip is from Infineon, the great winner in this design.

The NAND flash memory is from Toshiba although it also could be the Samsung part.

The audio codec is again from Wolfson. Semiconductor Insights says that is the WM6180C but that is not a normal Wolfson code.

Now the design comes in only one, bigger board, instead of two. With the components almost only in one side. It is a challenge to design the board not being able to put capacitors behind the processor. I suppose the board must be at least of 10 layers.

Primary conclusions: Good, neat design. I'm afraid too power hungry, not being able to use the new processor from Samsung.

iPhone 3G

2008-04-14T00:12:00.004+02:00

There were a lot of rumors about 3G iPhone. Now, in the latest beta firmware, Zibri has found a proof of 3G chipset in the design. In the current design, the GSM/EDGE chipset comes from Infineon: SGOLD2 (PMB8876). In the code found by Zibri, there is a trace of a SGOLD3 chip.

The SGOLD3 comes in two flavours: PMB8877, an GSM/EDGE chip, and the more powerful 3G HSDPA SGOLD3H PMB8878. All people likes to see the SGOLD3H as the 3G solution, pointed by the code. There is no need to change to the SGOLD3 PMB8877 because the new features are not used by the iPhone, with its own multimedia Samsung processor. That's why I think that the SGOLD3H could be the chip in the already coming (June) 3G iPhone.

I think that all the delay in the launch of the 3G iPhone was related to power consumption and battery life. The iPhone has a lot of processors inside to make it modular and fast to market, but that leads to power hungry designs. It is not the best possible design. With 3G, it is even worse. It seems that the solution comes from the display and the 3G iPhone will come with an OLED display, that consume less power than conventional TFTs, and Apple has finally get the correct numbers.

Nowadays, hardware designs are more modular, and less optimized. It is the same that happened to software design a lot of years before. To get to market on time with complex design we relay in already working modules. That leads to big, plenty of processors, power hungry designs. Continuous improvement in circuit integration makes that work, the same way that continuously more power hardware computers make possible to use non optimized software in the past.

The important thing is to get to maket on time, but sometimes it is not so easy, and the long road could lead you to destiny first.

New Zune 2 teardown

2007-12-04T11:18:00.000+01:00

We have got images of the new Zune 2 80GB player from Microsoft, courtesy of Rapid Repair.

We see a better designed product, not just a reference design transcription like Zune 1. The processor is now an i.MX32 instead of the i.MX31. There is very little information of this processor in the Freescale Web, only a block diagram. It seems to be very similar to the i.MX31, but with SRAM memory inside the chip and a better hardware multimedia player. As the i.MX31, it is an ARM1136JF 532MHz with floating point processor.

You can see the main board in this photograph:

- U1 is the dynamic RAM memory, in this case a Mobile DDR 512Mbit (64MByte) 32bit bus, from Hynix (old Hyundai). You can find the datasheet here.

- U2 is a Hi-speed USB Transceiver from SMSC, the USB3316B-CP. You can find the datasheet here.

- U3 is a DC/DC boost converter 1.8A 3.3V from Texas Instruments, exactly the TPS63001DRC. You can find the datasheet here.

- U4 is a 250mA Low dropout regulator from Texas Instruments, exactly the TPS73201. You can find the datasheet here.

- U5 is the i.MX32 ARM1136JF 532MHz processor. You can find the block diagram here.
- U6 is a power manager with audio codec from Wolfson, the WM8350. You can find a data brief here.

- U7 is a Flash memory from ST. There is no more information about it.

These are the only identified integrated circuits till now. If we get better images we can identify more circuits.

See you.

New Samsung ARM Processor User Manual

2007-10-03T23:58:00.000+02:00

The brand new Aplication Processor from Samsung is ready to get into production. Samsung is now sampling it to certain customers.

For those of you that can't wait to read more about this powerful chip, follow this link and get the full user manual/datasheet:

Rapidshare download link

Although in a lot of websites, you could read that is the processor in the iPhone, that is not true. The footprint, the peripherals and the memory map are different. For example, the S3C6400 doesn't have the PowerVR MBX 3D processor.

It is not the same chip that goes in the iPhone but is very similar: same core and some common peripherals. It has a powerful ARM1176JZF-S at 533MHz with H.264 video aceleration, TFT and TV output, USB OTG 2.0 HS and a NAND Flash controller able to boot directly from NAND Flash, among its features.

Looking for see it in new interesting devices for the beginning of this new year.

In 2009, the ARM Cortex-A8 processors and the new Intel, x86 based, PC-on-a-chip Moorestown platform, will fight for the new portable gadgets and the Mobile Internet Device (MID) category of devices.

Teardown iPod Nano 3G

2007-09-14T02:39:00.000+02:00

Just some days after the official presentation of the iPod Nano 3G, we have the first inside photos, courtesy of iFixit.

The design has less chips than previous versions of iPod Nano. The main processor is a Samsung S5L8702 (the iPod Nano 2G uses a S5L8701), probably with an ARM9 core.

The RAM comes from Qimonda (the spin-off from Infineon (the spin-off from Siemens) ). It is a 256Mbit (32MByte) Mobile 1.8V DDRAM, code HYE18M169CX75. Here is the datasheet.

There is also an audio codec, Apple branded, and a power manager chip also mark with an Apple code. There is no wheel controller; I suppose it is integrated in the power management chip.

On the other side of the board we can find the 8GByte NAND-Flash part, from Samsung: K9HCG08U5M, the same than in the iPhone. It is a multidie part (4 dies).

There is no NOR boot flash. I suppose that the processor is able to boot directly from Nand Flash.

S3C6400 processor

2007-07-22T19:56:00.001+02:00

The iphone processor is from Samsung and it is an ARM1176 core, similar to the general use Samsung processor S3C6400.
Here you have the first two pages of the S3C6400 datasheet, for you to have an idea of how the iphone processor could be:

New data about Samsung iphone processor

2007-07-11T02:27:00.000+02:00

The Samsung processors site has just changed. Now there is two interesting documents:

- The mobile solution brochure, with a bit more data than before about S3C6400.

- The SoC part number decoder, interesting information about Samsung processors part numbers.

Another interesting link shows the photograph of the S3C6400, that shows the marking on it:

Click to see it better. The marking is S5I3000X01. Following the part number decoder document:
S System LSI
5 Large clasification: MOS
I Small clasification: Digital core
3000 Serial number
X First version
01 Mask number

Following this clasification, the numbering in the iphone die is:
S System LSI
5 Large clasification: MOS
L Small clasification: optical?
8900 Serial number
B Version B
01 Mask number

Also, you could see in the photograph the two rows of balls, up and around the chip, where the PoP (package on package) memory to make a multichip package is soldered (see main post).

I think that, following the Samsung rules (the MCP for S3C2440 with more features is SC32442), the real part number of the processor if it ends as general purpose product could be something like: SC36402 (the MCP version of the S3C6402, which is the S3C6400 with little changes, like the PowerVR MBX 3D processor).

Inside iPhone (complete hardware analysis on progress...)

2007-07-05T00:12:00.000+02:00

Al least, we have the iPhone, and there was several sites in the Internet opening the phone and showing photos of the boards. At the same time, other sites are already working on the firmware code.

The best hardware work was done by the guys of Semiconductor Insights. They even decap the chips to see the die markings. iFixit show more photographs of the iphone progresive teardown.
There is additional PCB and chip dies scans in Microblog, the best quality ones.

Here, as an electronics engineer, I will try to pick all this information, and make a guess of the complete iphone hardware specification. This specification will evolve with uncoming informations and hints from other people. If you have additional information or have an opinion I would like to read you in the comments of by a private e-mail to mailto:InsideTronics@gmail.com.

iPhone Hardware Specs:

The main electronics are splitted into two boards.

- The main processor board:
The main processor it is a stacked die package with a processor marked as S5L8900 and two 512Mb DRAM dice (128MBytes of DRAM). This configuration is normal in a mobile. Normally the main processor has pads in the upper part of the sustrate to solder a PoP (package on package) multidie chip. The layout of the pads is a Jedec standard (JC63 and JC11) and this allows to save space but at the same time use the preferred memory supplier.The DRAM MCP (multichip package) is the K4X1G153PC-XGC3 from Samsung that, according to the Mobile DDRAM part number decoding from Samsung has the following characteristics:

DDR-SDRAM, 1Gbit, 8K blocks, 64ms, 1.8V core, 1.8V I/O, x16 bits bus, 4 banks, C generation, FBGA, Extended low temperature, speed 7.5ns@CL3.

Now the processor. This is the most difficult part. Semiconductor Insights says that the die number is S5L8900. You could even see the number 8900 in the chip marking. In the past, several Samsung chips has this kind of numbering. Last iPod processor is the S5L8701 (8701 in the marking), and in the Samsung web site you could see a ARM9 MP3 player processor named as S5L8700. I think it is clear that the processor is a Samsung part.

In Microblog, Nick Chernyy has decap the CPU chip and he has a clear photo of the die marks.

People has already found in the firmware strings, that the processor is a ARM1176. With this ARM core, Samsung has only one processor in the web site: the brand new S3C6400. I have found the S3C6400 datasheet in the Internet (it is not an easy task) and I have read it. This is the block diagram of this processor:

But I'm afraid this is not the real processor in the iphone. I think it is a propietary design for Apple. One of the reason is that the firmware interrupts (from some info found in the Internet) doesn't match with the datasheet, and the blocks declared in the firmware are similar but not identical to the S3C6400. In the firmware there is also declarations that shows that there is a PowerVR MBX 3D processor from Imagination Technologies inside the S5L8900.

I think that is a similar design, with a multiformat hardware video accelerator(MPEG4, H.264, etc.), PowerVR 2D and 3D accelerator, NAND Flash booting feature, USB OTG 2.0, and ARM1176J2F core, and no more DSP capabilities (with ARM11 cores nobody use a DSP anymore, except for video).

The third chip in this board is the NAND flash. It is a part also from Samsung: K9MCG08U5M
It is a 8GByte chip, with four 2GByte dies stacked. In the 4Gbyte iphone version, the chip is the K9MCG08U1M, that is a 4GByte, dual die part.

In the other side of the digital board there is a small integrated circuit from SST, marked as SST39WF800A. It is a 2MByte (8Mbit x 16) 1.8V NOR Flash. The processor boots from this flash memory.

The Intel NOR flash in the other board is for the baseband Infineon controller. It is not feasible to use the NOR flash through the connector between the two boards.

The fourth chip is the I2S voice codec from Wolfson: WM8758. A very good sound quality part, indeed, the same than in last iPod.

The USB battery charger, from Linear Technology LTC4066, is also a good part.

Other chip is a LVDS 24 bit driver from National (LM2512SM) to carry the signals to the display with less wires and without interference.

The remaining chip, from NXP (Philips before), I guess it is the power manager with the switching power supplies (see the coils near the chip, with capacitors and schottky diodes). In the firmware there is references to the PCF50635 power manager. Must be this part or similar.

- The radio board:
The radio board has the following circuits:

The baseband processor is an Infineon PMB8876 S-Gold multimedia engine with EDGE functionality. Comes with an Intel NOR Flash + SDRAM PF38F1030W0YTQ2 (4Mbytes NOR + 2Mbytes PSRAM) to run the baseband code (the capacities in the Semiconductor Insights article are wrong, see the datasheet link).

The RF GSM transceiver is an Infineon part, probably the M1817A11.

The bluetooth chip cames from CSR (Cambridge Silicon Radio): the Bluecore4.

The Wifi is a Marvell part: the 88W8686.

- Other parts:

The touchscreen is from German manufacturer Balda.

In the touchscreen there is a SPI multitouch controller from Broadcom, the BCM5973A.

The camera has a Micron 2Mpixel sensor, probably MT9D112D00STC.

The display is 320x480, can be a Samsung or AUO module.

The battery is a Li-IonPolymer 3.7V.

More will follow...

PD. Changes 6/7/2007: PowerVR 3D reference included, changed capacity values in Intel Flash/PSRAM.
Changes 19/7/07: Die photo of processor included.
Changes 21/7/07: Microblog scans reference added.
Changes 6/8/07: SST NOR Flash memory reference added.