NovelPanda
Junior Member level 3
Dear All:
Due to limitation on measurement facilities, I need to design an on-chip 28Gb/s PRBS to generate data stream for testing. To minimize area and achieve high speed, I have designed a full-rate PRBS-4 which is composed of 4-DFF, 1-XOR and 1 additional DFF for re-timing. The DFF is master-slave that has no re-start control. The input 28Gb/s CLK can be easily generated by a clean external source. In transient simulation (post-layout with C+CC extraction, APS, conservative, Error preset= conservative, Transient noise included), the 28Gb/s output eye and DFT are shown as figure 1 and 2 below. However, when there is no CLK coming in, the PRBS still generates data stream with eye diagram (period=2/14G) and DFT shown as figure 3 and 4 below. What later surprised me is that, when I set the APS (in High-Performance simulation) Error Preset to "do not override" (figure below), there is no data output, but the PRBS still work well in 28Gb/s when CLK comes in. So is it a normal phenomenon for PRBS or an ill design, or the setting error of Specre APS? If it is a wrong design, what further steps should I take to correct it? In addition, I have no idea what is the meaning of "do not override" in APS. Compared to "conservative", which one is more likely to generate a more trustful result?
Thanks in advance!
Due to limitation on measurement facilities, I need to design an on-chip 28Gb/s PRBS to generate data stream for testing. To minimize area and achieve high speed, I have designed a full-rate PRBS-4 which is composed of 4-DFF, 1-XOR and 1 additional DFF for re-timing. The DFF is master-slave that has no re-start control. The input 28Gb/s CLK can be easily generated by a clean external source. In transient simulation (post-layout with C+CC extraction, APS, conservative, Error preset= conservative, Transient noise included), the 28Gb/s output eye and DFT are shown as figure 1 and 2 below. However, when there is no CLK coming in, the PRBS still generates data stream with eye diagram (period=2/14G) and DFT shown as figure 3 and 4 below. What later surprised me is that, when I set the APS (in High-Performance simulation) Error Preset to "do not override" (figure below), there is no data output, but the PRBS still work well in 28Gb/s when CLK comes in. So is it a normal phenomenon for PRBS or an ill design, or the setting error of Specre APS? If it is a wrong design, what further steps should I take to correct it? In addition, I have no idea what is the meaning of "do not override" in APS. Compared to "conservative", which one is more likely to generate a more trustful result?
Thanks in advance!
Last edited: