Another CMDA phone this time a Nokia 6275i phone from late 2006 /early 2007 (10 years old).
This one has a ( pretty decent for the year) 2M pixel camera. Around this time Nokia was reputedly the worlds largest camera vendor. By virtue of the fact they were the number one phone vendor and the cell phone market dwarfed the stand alone camera market.
After whipping the battery out and spudging the two halves separate (Is spudging a word now :-))
This is a well engineered phone. With a sturdy board holding the pressure sensitive keyboard.
Unscrewing the main board from the frame. Unlike the Samsung SPH-A660 I looked at recently this one has RF shields covering the several circuits.
Lots of metal lids!
- Texas Instruments 4377177 – Digital Baseband Processor
- Samsung KAK13V006D – DRAM
- Texas Instruments 4376371 – Analog Baseband Processor
- Philips TEA5761 FM Tuner
Lots of chips with no data sheets that I could find, and mostly TI chipsets.
- STM XTV0984N image sensor processor
- HE09 Image Sensor (Which I believe is a Toshiba sensor, I was expecting STM as the processor was from STM)
- CSR 41-B14 Bluetooth Transceiver
- RFMD RF3448 – Power Amp (Or LNA – cannot find documentation and figure out if this is RF receive or transmit area)
- RFMD RF3449 – Power Amp (Or LNA)
- TI 438021 – Frequency up-converter (Or downconverter)
- AC7710 – LNA (Or PA)
- TI 4380033 BATMAN – Frequency downconverter (Or up-converter)
- Nat Semi LMX2310 Low power frequency synthesizer
A 10 year old phone is not quite sufficiently vintage for me, the digital chips will be too dense to see anything interesting with my optical microscope. I failed miserably looking at the GaAs RFMD chips. So I chose to look at the Philips FM Tuner
Looking at the small die on the rear board located with the big processors, it looks like it is a naked die with no packaging (The back looks like shiny silicon with laser etched markings). After taking it off the board (With a hot air gun) that is indeed what it is.
You can see the die covered with some protective underfill material (Polyimide maybe?) And some solder balls (Most have come off during removal from the board). This type of packaging is pretty uncommon. It seems elegant to me and I wonder why it is not used more.
Stripping the underfill material took considerably longer in the acid than depotting a normal plastic part to get a clean die, which was a surprise for me. Also I chipped the die handling it in the acid/solvents with my tweezers.
Measuring 3.5mm x 3.5mm it’s an interesting die to look at, with a small digital section down the right side, and the bulk of the die is analog circuitry with a very large amount of on chip capacitors.
The die marks do not match the TEA5761 part number. Searching on the VN53404 did not yield anything.
Looking at a (focus stacked) image from the dense logic area, I can see polysilicon gates that measure as a little less than 1 μm, and the metal tracks I measure at 1.2 μm. I think this is fabricated on a 3 metal 0.5 μm BiCMOS process.
Throughout the analog area there are vast amounts of capacitors with resistor networks associated. Presumably these are creating RC filter elements involved with the tuning of FM signals. The capacitors are a bit strange as they are covered with via’s all over the top plate which I can see no reason for.
They are a little odd – I figure they are not test structures since they have no metal tracks contacting them. It is not unusual for designers to throw some spare transistors into free space in case they find a problem on first silicon and can re-spin by routing in the spare transistors saving some masks in the re-spin (In theory, but I never saw this work in practice). However these are nearly all at the device edges so pretty impractical for using anyway.
From the bottom of the image to the top: you can see resistors with a third terminal, so I think these are pinched resistors, used to get high value resistance. The second and third rows are Bipolar transistors. I will make an educated guess the second row are NPN as you can see the blue/green polysilicon, so I think they are polysilicon emitter NPN. Thus I would expect the third row to be PNP transistors. The top row looks like a pair of MOSFETS.
National Semiconductor LMX2310
The second die I am going to look at is the Nat Semi LMX2310, a very small die at the centre of the board.
This is a ultra low power frequency synthesizer. It takes a VCO (Voltage Controlled Oscillator) signal based around the board crystal, and generates a stable, low noise oscillator signals for the RF/IF up and down conversion. A datasheet is still available from TI. I was a little surprised when I found the chip function, as by 2007 I thought this would be done with on-chip PLL in the up/down convertor chips. In the Samsung SPH-A660 I looked at the Qualcomm RFR6000 and RFT6100 include on chip local oscillators.
The main die marking is pretty ugly.
The other die markings include 2001, a designer initial (TE) and a small silicon doodle! (Is that a Christmas tree?)
This is the N-divider circuit.
And this is one of the eight blocks in high resolution – you can just make out some white diffusion regions. The yellow and pink traces are metal 1 and 2 layers.
And this is the charge pump area.
Once again zooming in on the region you can see it is made up of large multi-gate transistors:
And lastly this associated to the clock pin from the die corner. It is an enormous meandering resistor.I measured that as 1.5 μm wide and 3900 μm long ie. 2600 squares (The value of the resistor is equal to the # of squares x the sheet resistance of the layer). This is a polysilicon resistor so sheet resistance could be 2-200 Ω/square (Depending on doping level and whether it has silicide layer) so no way of calculating the resistor value just by looking at it.