1. Optimizing Storage Electronics for Small Form Factor Drives
Duncan Furness
Agere Storage Product Marketing

2. This market has huge potential!
CE Growth
CE HDD Unit Volumes by End Application
200
Digital Video/Camera
180
Converged Device
Portable Audio
160
Gaming
PVR/STB
140
Upside Case
120
Unit Volumes (in millions)
100
Portable
Mobile Phones
Audio MP3
PDA 80 60
Priorities
Digital Video Cam
Portable Video
Digital Still Cam 40
1. Power
2. Cost 20
Handheld Gaming
Notebook
3. Capacity
2003
2004
2005
2006
2007
2008
This market has huge potential!
9/18/2018

3. Data Rates & Capacity Requirements by Format
100 GB
Video
1.8”
Audio Music
10 GB
Pictures
1 GB
1’ & 0.85”
100 MB
So, now that we know what the rates are, what the capacity points can be, clearly, we realize that there may be some gaps we need to explore. These gaps include the potential pockets where the overlap does not occur. We’ll explore these further in conjunction with other factors.
What is critical from our RC IP, will be 3 items: Speed, SNR and feature set. Understanding the form factor, targeted content format, and the capacity point of each will help drive these 3 requirements, for all of the Ip we develop, RC, PA and MC.
10 MB
Voice
1 MB
0.01
0.1 1 10
100
1000
Data Rate MB/s
9/18/2018

4. CE SFF Storage Requirements
Small form factor:
1”, 0.85”
Removable drives 32 x 24 x 2.1 mm
Priorities:
Low-power consumption:
Battery powered
Extremely cost sensitive:
$65 to $85 price with erosion
High capacity:
Replacement of Flash
Achieve a threshold of performance
9/18/2018

5. Sample HDD Power Budget
Electromechanicaland Mechanical mW
Electronics
400
Almost ½ is Electronics!
A large opportunity.
10% !!
300
Camera
Storage
200
Display
Talk
Battery
Limit
100
SOC PA
SPIN UP
SPIN SS
SEEK
Cell Phone Power Budget
Mechanical
Manufacturing
HDD
Components
Electromechanical
Architecture
Electronics
Power Mgmt
9/18/2018

6. Manufacturing Processes Advances Enable Lower Power
Reduction in operating power through process shrink:
Beyond 65 nM, improvement diminishes.
Leakage current increases, but is three orders of magnitude smaller than operating power:
Future concern
Can be mitigated through Silicon on Insulator (SOI) and through triple-well CMOS design
Motor controller design can benefit through use of SOI FET’s to reduce leakage power
9/18/2018

7. Lowering Power through Feature Partitioning
HDD
Other IP:
WiFi, UWB, LCD + Audio Amp Drivers,
GPRS/GSM, UMTS/3G
Video/Audio Codecs 1 PA
SOC
RC, HDC
+ IP
Increasing feature
density through SOC MC
HDD CE 2
Mem
ADC
Int
Repartitioned
HDD
DSP
(HDC,
uC,
A/V,
RC) PA
Int
Pwr
Mgmt
WLAN MC
Combination of packages reduces I/O buffer requirements and reduces overall system power consumption.
9/18/2018

8. New Architectures and Requirements
READ CHANNELS
Repartitioning enables fewer I/O buffer requirements:
Because new interface requirements are serialized, fewer I/O buffers are needed reducing power.
New Algorithm enabled by lower data rate for multimedia applications:
CE device file formats are multimedia-based requiring reduced throughput on the host side, allowing for a more compact detection algorithm.
New COMA modes for quiescent operation in CE devices:
Sub-10 uA leakage enables competitive quiescent operation for HDD.
HDC
Serial interface:
Parallel interfaces provide excessive bandwidth for SFF
Reduce amount of I/O drivers
Lower core voltages
Processor speed optimization
Power management:
Turning off power rather than just clocks to idle blocks
Need to accommodate wake-up requirements
The Read Channel and HDC comprise the critical functions of the SoC
9/18/2018

9. New Architectures and Requirements
PREAMPS
Lower voltages enable lower power at architecture:
CE devices can do with lower voltages as a result of lower data rate requirements reducing power.
Reduced data rate enables compact architecture:
Reduced slew rate and capacitance requirements change architecture reducing transistor count.
9/18/2018

10. Power Mgmt: Advantage HDD
Power Consumption/GB (for files >8 MB)
Peak Current Drain mA
mAh/GB
300
0.60
250
0.50
200
Flash (8 MB/s)
0.40
Flash (4 MB/s)
150
HDD (1”)
0.30
HDD (sub 1”)
100
0.20 50
0.10
Current
Architecture,
Flash in 130 nM
Process Node
Distributed HDD
Architecture,
Flash in 65 nM
Process Node
Current HDD
Architecture,
Flash in 130 nM
Process Node
Distributed HDD
Architecture,
Flash in 65 nM
Process Node
Lower overall power consumption in distributed architecture enabled by data buffered into SDRAM.
Lower frequency operation in HDC, COMA modes and additional savings in RC data detection algorithms to reduce power further.
9/18/2018
Source: Semico, Agere Marketing, Dec ‘03

11. Balancing the Need for Low Power and Cost
Native Speed
Application Speed
Select the minimum size buffer to meet the power goal and minimize cost.
HDD
MP3 or
MPEG
Buffer
FAST
SLOW
Average HDD PD for Streaming Applications
200
MPEG4 video
180
MP3 Audio
160
110 mW Cell Battery
140
120
PD (mW)
100 80 60 40 20 1 2 3 4 5
MB of Buffer
0.25 MB MP3
1.7 MB MPEG4
9/18/2018

12. Cost Reductions SFF are moving towards serial interfaces:
Reduces #pins:
SD IO, MMC – strong direction
USB
Integration of buffer memory into SoC:
Pin reduction, package reduction
Process migration when it provides a benefit:
Masks are approximately 2X for 130 nM to 90 nM
Die size savings need to offset NRE
SoC feature partitioning:
Number of packages and PCB’s reduced
9/18/2018

13. Cost Reductions Further HDD integration steps
Potentially enables a single IC approach for SFF
Eliminates cost of PCB
Reduces cost of head flex
Allows very thin packaging of HDD
Challenges:
Noise on head signals
Power devices on SoC
Not trivial, but attractive rewards PA
SOC:
RC, HDC
I/F MC
SOC:
RC, HDC
I/F
+PA
+MC
9/18/2018

14. Channel Leadership is Key!
Capacity Drivers
Advanced channel techniques:
Proprietary algorithms and signal processing IP targeted at boosting effective SNR:
Iterative decoding
Reverse ECC
Perpendicular recording with decision based baseline compensation
Architectures that allow scaling from enterprise to microdrive
Channel Leadership is Key!
9/18/2018

15. Takeaways Huge emerging CE market opportunity in SFF HDD’s.
Lower power enabled through component architectures, power management and manufacturing process.
Average power in HDD can beat flash, with system innovation.
Capacity improvements driven through read channel leadership.
Cost reduction driven by partitioning, process and partnership.
9/18/2018