Design for MOI Educational Program (Research) Design, Fabrication and Testing of a fully integrated.5 Hz Clock Data Recovery Circuit with Demultiplexer in 0.5 µm CMO Process Project submitted to MOI Fabrication process: 0.5 µm Prepared by: Jinghua Li*, (Clock Data Recovery) hanfeng Cheng (Demultiplexer) Prof. José ilva-martínez Texas A&M University Analog and Mixed ignal center Department of Electrical Engineering College tation TX, 7784-8 *Contact person (E-mail: jhli@ee.tamu.edu) Aug 7th, 00
Design Number 675 Design Password Design Name Technology Fabricated on Run Layout ize tudent involved cdrdemux CDR Demux CN4ME_UBM,lamda=0. T4C-AK 0x0 microns Jinghua Li(Clock Data Recovery), hanfeng Cheng( to 8 Demultiplexer) Advisor Dr. José ilva-martínez
I Description of the Clock Data Recovery Circuit The primary objective of this project is to design, layout and characterize an integrated clock data recovery circuit with de-serializer operating at a clock frequency of.5 Hz for OC-48 optical communications. II imulation and Layout Tools The IC is designed using Cadence EDA tools. For our application, the parasitics become of extreme importance at such a high speed faster than.5 Hz. III Microphotograph of the Test Chip FI MICROPHOTORAPH OF THE TET CHIP IV Characterization of the Chip In characterization of the test chip, a careful set up for measurement is needed. In high frequency the performance of the circuit become sensitive to both the resistive and capacitive loading of the measuring equipment; A printed circuit board is designed for the testing of the clock data recovery chip. The test equipment include, Rhode & chwarz
FEB0 0Hz~7Hz pectrum Analyzer, Agilent Infiniium 0MHz,sa/s Oscilloscope(Testing the reference clock). As a preliminary test, two chips are characterized, both works at roughly the same frequency. Due to the over-design of the inductors(mistakenly used only half of the inductor, two inductors should have been series connected instead), half of the inductor value is used in the LC tank VCO, so the frequency is higher than expected, the VCO oscillating frequency is around.65hz, matched to the design frequency times by. The main limitation is that we don t have exact modeling of the inductor. Now there is no available layout extraction software package for TMC 0.5um CMO technology. All this effects have resulted in the design mismatch. The test results are attached as the following: Fig shows the frequency spectrum of the VCO output. Fig 4 shows the phase noise performance of the circuit. From the phase noise diagram, the phase noise at MHz away from the carrier is 05dBc.
FI FREQUENCY PECTRUM FOR THE VCO OUTPUT
FI 4 PHAE NOIE OF THE VCO OUTPUT
V to 8 Demultiplexer Chip Testing Report by hanfeng Cheng. Testing chematic 4 5 6 A B C D 6 5 4 D C B A Title Number Revision ize B Date: -Aug-00 heet of File: D:\sfcheng\ic\mudem\PCB\mudem.Ddb Drawn By: 4 4 5 5 6 6 7 7 8 8 9 9 0 0 4 4 5 5 9 9 7 7 8 8 9 9 0 0 4 4 5 5 6 6 7 7 8 8 vdd 0 vdd 5 49 d db 5 vdd_5 5 clk 5 clkb 54 r 55 6 6 56 56 57 57 58 58 59 59 60 60 6 6 6 6 64 64 40 4 5 6 7 8 9 q7 4 q6 4 q5 4 q4 44 q 45 q 46 q 47 q0 48 QFP64 6 6 DEMUX DEMUX C 47n R9 00K R R R R4 R5 R6 R7 R8 C 00n C 00n C 0n C4 00n C5 00n C6 00n C7 00n C8 00n C 0n C 0n C0 0n C9 0n Q0 MA Q MA Q MA Q MA Q4 MA Q5 MA Q6 MA Q7 MA CK MA D MA VDD_5 VDD 5 POWER JUMPER PWITCH JUMPER VDD_5 VDD 5 CV CORNER_VIA CV CORNER_VIA CV CORNER_VIA CV4 CORNER_VIA C5 47p C4 47n C6 47u CKB MA DB MA ADJ VOUT VIN VOUT 4 LM7 LM7_MD R0 k X C7 0.u X X POT k. Testing PCB
. Test results We use full differential input for both clock and data. For both input clock and data, the differential ohm input matching resistors are built on-chip. o there are no resistors on the input side. Only DC-block capacitors (0n). For the output side, there are 8 output streams since it s a to 8 demultiplexer. The output signals are all single-ended with a peak-peak swing of 400mV. Each output pin is connected to a ohm pull-up resistor, then connected to a DC-block capacitor and then connected to ohm termination (oscilloscope probe) by way of MA connector. Due to the limitation of the operation speed of the signal generator in our group (Analog & Mixed ignal Center of Texas A&M University), the highest input data rate is 660Mbps and the corresponding clock frequency is 0MHz for proper demultiplexing.we were able to verify the function of this chip but not able to measure the highest performance of this chip (this chip can accommodate an input data rate of 5bps). The full chip characterization will be done soon at the XILINX facilities in Austin. The function of the chip is verified to be correct by the following method. We input a fixed -bit length input data pattern (i.e., the pattern repeats every bits). We ve observed that all the outputs are repeating every 4 bits exactly as expected. What s more, when we reassemble the 8 outputs, the assembled pattern is the same as the input pattern in a cyclic manner (i.e., 4567 is considered the same as 4567 in a cyclic manner) Minimum Input Data wing (Vpp) 0.8V Minimum Input Clock wing (Vpp) 0.4V Verified Working Frequency Range 0MHz 660MHz We are going to redesign a better board and test with better equipment in XILINX to further characterize the highest performance of the demultiplexer. We will submit another more detailed report when we get new results.