01-06-2012, 01:09 PM
Proposals for Memristor Crossbar Design and Applications
[attachment=23846]
Desirable Manufacturing Goals to Ease
Adoption of Memristor Crossbars
• Design should be compatible with standard manufacturing
techniques to facilitate wide use and experimentation by
many participants (i.e. universities, research labs, existing
fabs, etc.) without a significant investment in new equipment.
• Design should be easy to integrate with standard electronics
components and materials.
• Design should incorporate materials capable of RHIGH>>RLOW.
• Design should be robust to both temporal and spatial
variation of memristance.
• Design should avoid internal feedback current paths in
crossbar which can limit speed and ability to read resistance
states reliably.
• Design should allow ease of reconfiguration of resistance
states.
Desirable Goals for Memristor
Crossbar Array Applications
• Complement (not conflicting with) existing
technologies and markets to achieve ease of
acceptance
• Identify uses compatible with smaller emerging
markets with potential for high growth (e.g.
FPAAs , commercial robotics, neural interfaces)
• Solve problems for which conventional electronic
hardware and software do not provide efficient
solutions but which memristors can (e.g. pattern
recognition, traveling salesman problem)
Problems with Conventional Drive
Waveform Circuits
• In many electronics applications variation of circuit
parameters due to temperature change, aging, etc. require
adjustment of drive waveforms (e.g. LEDs may require a
higher amplitude voltage drive over time to produce a
consistent light output).
• Waveform adjustment is also desirable for mode adjustment
in various applications (e.g. inkjet printheads changing
resolution or drop size often involves timing or amplitude
adjustment of drive signal for heater or piezo.)
• Timing modulation and amplitude modulation circuits
implemented in hardware can require complex circuitry and
have limitations in adaptability and the range of possible
waveforms.
• Software based solutions require a microprocessor which can
be difficult/expensive to miniaturize for several portable
electronics applications
[attachment=23846]
Desirable Manufacturing Goals to Ease
Adoption of Memristor Crossbars
• Design should be compatible with standard manufacturing
techniques to facilitate wide use and experimentation by
many participants (i.e. universities, research labs, existing
fabs, etc.) without a significant investment in new equipment.
• Design should be easy to integrate with standard electronics
components and materials.
• Design should incorporate materials capable of RHIGH>>RLOW.
• Design should be robust to both temporal and spatial
variation of memristance.
• Design should avoid internal feedback current paths in
crossbar which can limit speed and ability to read resistance
states reliably.
• Design should allow ease of reconfiguration of resistance
states.
Desirable Goals for Memristor
Crossbar Array Applications
• Complement (not conflicting with) existing
technologies and markets to achieve ease of
acceptance
• Identify uses compatible with smaller emerging
markets with potential for high growth (e.g.
FPAAs , commercial robotics, neural interfaces)
• Solve problems for which conventional electronic
hardware and software do not provide efficient
solutions but which memristors can (e.g. pattern
recognition, traveling salesman problem)
Problems with Conventional Drive
Waveform Circuits
• In many electronics applications variation of circuit
parameters due to temperature change, aging, etc. require
adjustment of drive waveforms (e.g. LEDs may require a
higher amplitude voltage drive over time to produce a
consistent light output).
• Waveform adjustment is also desirable for mode adjustment
in various applications (e.g. inkjet printheads changing
resolution or drop size often involves timing or amplitude
adjustment of drive signal for heater or piezo.)
• Timing modulation and amplitude modulation circuits
implemented in hardware can require complex circuitry and
have limitations in adaptability and the range of possible
waveforms.
• Software based solutions require a microprocessor which can
be difficult/expensive to miniaturize for several portable
electronics applications