Clocking
MIT Fall 2024
Please Log In for full access to the web site.
Note that this link will take you to an external site (https://shimmer.mit.edu) to authenticate, and then you will be redirected back to this page.
Assuming that each NOT gate has a t_{pd}=35\text{ ps}, answer the following questions:
Refer back to lecture 6. Starting with a 100 MHz clock, calculate a three-tuple of divider, multiplier, divider coefficients to generate 260 MHz. Make sure you keep your intermediate frequencies above 12MHz and below 1.2 GHz.
You're working at a company designing the jPhone. Consider this part of the system below where:
Registers R1 and R2 have the following timing characteristics:
- t_{pdR} = 2\text{ ns}
- t_{cdR} = 1\text{ ns}
- t_{setup} = 3\text{ ns}
- t_{hold} = 1\text{ ns}
The combinational logic m9 has:
- t_{pdm9} = 6\text{ ns}
- t_{cdm9} = 2\text{ ns}
and:
- t_{clk} = 15\text{ ns}
Uhoh it looks like your boss Jeve Stobs is having another one of his "issues" where he starts yelling and demanding things. Yeah the jPhone is revolutionary, but at what cost? Somebody should probably talk to HR about this, but in the meantime, you still have to deal with what he wants. After some redesign, the circuit above has been modified such that it does the next big thing, but you're concerned that this extra delay is going to break the design. Specifically Module m9 has been updated such that it now has:
- t_{pdm9} = 15\text{ ns}
- t_{cdm9} = 3\text{ ns}
and you're running the system at:
- t_{clk} = 8\text{ ns}
Since the system is being pushed so hard, there is now a positive skew of 2\text{ ns} in this system (meaning the clock edge arrives at R2 2 nanoseconds later than it does at R1).
Once you're done, camera time is next! Move on to the pixel reconstructor :D.