Answers Database
XC2000L/XC3000L/XC4000L: Estimating power consumption for 3.3V Devices (2000L, 3000L, 4000L)
Record #1520
Problem Title:
XC2000L/XC3000L/XC4000L: Estimating power consumption for 3.3V Devices
(2000L, 3000L, 4000L)
Problem Description:
Keywords: power consumption, xc2000, xc3000, xc4000, 3.3V
Urgency: standard
Description
This is a guideline for estimating power consumption for our
low voltage (3.3V) devices. For more information regarding
estimating power consumption in FPGAs please refer to the
1996 Databook.
Solution 1:
Since our 3.3 volt devices have the same architecture as
thier 5V counterparts and almost all power (except Icco)
is dynamic, the power changes with the square of the
supply voltage.
That means, at nominal 3.3 V, the power is 43.6 percent
of the power at nominal 5 V. If you use the mac
commercial values, 3.6 V and 5.25 V, then it is 47
percent.
Based on this information, the following figures may be
used to estimate possible power consumption of a design :
Device CLB Output Clock Longline
------ ------- ------- ------- ---------
XC2018L 0.048 0.349 1.40 N/A
XC3090L 0.052 0.349 0.87 0.031
XC4005L 0.065 0.392 1.74 0.044
XC4010L 0.065 0.392 3.49 0.065
XC4013L 0.065 0.392 5.01 0.078
where :
all units given in "milliwatts per million transitions
per second"
"CLB" = one CLB driving three local interconnects
"Output" = one output driving 50 pF
"Clock" = one global clock line connected to all flip-flops
"Longline" = one split Longline
"CLB" and "Output" are independent of device size.
"Clock" is proportional to the number of CLBs, i.e. to the
gate count.
"Longline" power is proportional to the square root of the
number of CLBs, or the square root of the gate count.
Note : For very low power watch out for the high Icco of
the XC3100L ( due to gate pumping ), and the ficticiously
high Icco of the XC5200L, due to anticipated future gate
pumping.
End of Record #1520
For the latest news, design tips, and patch information on the Xilinx design environment, check out the Xilinx Expert Journals! |