Part Number: DS320PR822
Tool/software:
Hello team,
What is maximum input level or output level that can be accepted by DS320PR822 with free of distortion?
Is it the LINEARITY-DC and LINEARITY-AC specification?
If above is true, it is typ spec but we can consider it is maximum level?
if my understanding is wrong, please let me know definition of the spec or provide any good document which explains it?
Best regards,
Hi Taketo-san,
For example, with setting of EQ=19 for about 20dB gain at 20GHz, must its input be less than 70mVpp to keep within output to be 700mVpp for linearity operation? or output can swing more without distortion in this case?
The dB gain given for the EQ settings is different from the dB gain of the output driver. The EQ amplifiers are specifically designed to amplify at high frequencies for the purpose of signal conditioning, and their dB gain measurements are taken at PCIe signal frequencies such as 16 GHz and 8 GHz. The output driver is the last amplifier on the device signal path and is intended to have a flat response, which is why its gain does not need to be associated with a frequency.
The linearity measurements are only taken for the output driver, hence why I said "output driver set to the default 0 dB gain" (there is no EQ setting with 0 dB gain at signal frequencies). Therefore when interpreting the linearity data we need to carefully review the relationship to EQ.
The DC linearity specification is measured at 100 MHz to represent low frequency signals, we can consider that the EQ (main effects are in GHz range) is basically unrelated. The AC linearity is measured at 8 MHz and therefore has a potential relationship to the EQ: if the input signal to the redriver were to be a perfect 8 GHz sine wave, the EQ will directly increase its amplitude by the gain at 8 GHz, and it is possible that the amplification would lead to the expected output swing exceeding the AC linearity specification. In this theoretical situation I think there would be output distortion. However in practical applications the input would not be a perfect 8 GHz sine wave, instead there would be components at other frequencies that contribute to the total signal swing, so the actual relationship between the EQ setting and the AC linearity bounds is difficult to predict.
In case it helps to know, I have not worked on a customer project where it was necessary to perform fine calculations of linearity; the very high EQ settings such as EQ 19 that would be likely to cause linearity issues on the output are rarely ever used in practice, and input signals to the redriver that require signal conditioning should already have degraded content from insertion loss, so their AC and DC levels are not very likely to exceed the specifications.
Best,
Evan Su
