Hi Abhirup, thanks for the good question. The 1tap DFE is not sufficient for equalization if there's no FFE. On the other hand, 1tap DFE + 14tap FFE might be sufficient for equalization for a less discontinuity (or smooth) channel characteristic. But, that should not be sufficient if the loss is > 40dB with lots of discontinuity. So, it depends. :)
Thanks CC for the message. I gave it a thought and seems DFE can still help in not doing noise amplification while a FFE does noise amplification. So it seems to be a compromise between two types of equalisation, using FFE and DFE. Having long FFE and 1 tap DFE seems to be a very good idea!
@@abhiruplahiri1 Hi Abhirup, thanks for the additional information & thoughts. I agree with you that the 1-tap DFE without noise amplification and FFE could help other residual ISI & reflection, without adding too much noise amplification; therefore, that's a good compromise/optimization.
Timing and Complexity, in conclusion.😮
Hi PARKER, nice to meet you. Correct. :)
Hi CC, why does the reference you shared additionally uses a 1tap DFE when 14tap FFE is used? Is the FFE not sufficient for equalisation?
Hi Abhirup, thanks for the good question.
The 1tap DFE is not sufficient for equalization if there's no FFE.
On the other hand, 1tap DFE + 14tap FFE might be sufficient for equalization for a less discontinuity (or smooth) channel characteristic.
But, that should not be sufficient if the loss is > 40dB with lots of discontinuity. So, it depends. :)
Thanks CC for the message. I gave it a thought and seems DFE can still help in not doing noise amplification while a FFE does noise amplification. So it seems to be a compromise between two types of equalisation, using FFE and DFE. Having long FFE and 1 tap DFE seems to be a very good idea!
@@abhiruplahiri1 Hi Abhirup, thanks for the additional information & thoughts.
I agree with you that the 1-tap DFE without noise amplification and FFE could help other residual ISI & reflection, without adding too much noise amplification; therefore, that's a good compromise/optimization.