This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DS80PCI800 Preset test fail

Other Parts Discussed in Thread: DS80PCI800, DS80PCI810, DS125BR820, DS80PCI800EVK, DS80PCI810EVM

Dear Sir 

Below is our use DS80PCI800 to test "PCIe Preset" result 

Just only P4 Preset test is Pass 

Is it correct test ?

28 inch trace length => DS80PCI800 => 3 inch trace length => test point 

  • Hi Kai,

    We have seen this occurrence in testing on our side as well for the DS80PCI800.

    Officially, the Tx preset test should be measured with a compliance board as shown in Fig 1 of App Note SNLA227, so there should be minimal trace from the transmitter and test point (instead of 28" + 3") to get a measure of whether your system passes the PCIe Preset preshoot and and de-emphasis requirements specified by PCI-SIG.

    Ultimately, the DS80PCI800 will only be able to pass P4, where there is no preshoot and no de-emphasis. This is due to non-linearities present in the repeater channel which may improve the output signal quality, but will do so at some expense to the preshoot and de-emphasis margin. For the presets that have a specific level of preshoot and de-emphasis (P1-P10), the non-linearity present in the CTLE and output driver results in overall preshoot and de-emphasis values that are out of the required margin limits. At the same time, however, if you run SigTest to check for signal quality, you will often see that the DS80PCI800 can work with the ASIC Tx to produce a clean output eye that will pass the signal quality test.

    In accordance to PCI-SIG, it is a requirement for any add-in card or motherboard processor Tx to produce presets P0-P10 within the required margin in order to meet their Integrator's List requirements. However, it remains a topic of debate within the PCI-SIG community about whether the repeater should be subjected to the same Tx Preset requirement, since it does not produce presets of its own and is rather intended to recover good signal quality for the receiver to resolve. The DS80PCI800 has been used successfully in many different PCIe 3.0 applications due to this understanding. The receiver is able to resolve an error-free link, even though all the preset margins are not necessarily met.

    If you require that the repeater is able to pass the Tx ASIC signal and still meet the preset margin requirements, I recommend that you take a look at the new linear DS80PCI810. This repeater is a modified design from the DS80PCI800 in order to preserve preshoot and de-emphasis values through increased linearity. We have observed success generating all 11 presets (assuming that the Tx ASIC before the repeater can generate all 11 presets correctly) through the DS80PCI810.

    Out of curiosity, is it possible for you to provide us with the endpoint device you are using to generate the Tx traffic? It is important as a baseline test to ensure that the endpoint device Tx (either add-in card or motherboard) can produce all of the desired presets without the help of the repeater.

    Thanks,

    Michael

  • Dear Michael

    Thanks for your reply

    How much preset margin that SigTest will be showed Pass with Preset test?
    I refer to your application note ,can 0.7dB pass preset margin test ?

    Thanks
  • Hi Kai,

    According to PCI-SIG, the combination of add-in card Tx + repeater (or motherboard Tx + repeater) must fall within the allowed margin specified in the Base Spec, which I have passed from Table 11 of the app note:

    Therefore, it is important that not only is the endpoint's Tx capable of meeting the preset limits, but the repeater also does not disturb the preset limits. For example, in the app note, you can see that without the DS125BR820, P7 preshoot = 3.43 dB and de-emphasis = -6.04 dB (see Fig. 14). This is well within the specification shown above, where P7 preshoot = 3.5 +/- 1 dB and de-emphasis = -6.0 +/- 1.5 dB. When the DS125BR820 is added in, the resulting P7 is still within the requirement: With the DS125BR820 added, P7 preshoot = 3.12 dB and de-emphasis = -5.36 dB.

    To answer your second question, there will always be some amount of change to the Tx preshoot and de-emphasis when the repeater is added in the signal path. Whether a difference of 0.7 dB from the endpoint Tx is enough to pass the preset margin test depends on whether there is > 0.7 dB margin available at the add-in card or motherboard Tx. In the case of the graphics card used in the SNLA227 app note, there is more than 0.7 dB margin according to measurements with the graphics card alone, so SigTest declares a pass with our repeater in the signal path.

    Thanks,

    Michael

  • Dear Michael

    We use DS80PCI800EVK to verify "Preset test " on previously pass " Preset test " slot
    Do you have experience use EQ or DEM of DS80PCI800 to pass more "Preset test"

    Below is our result
    Without DS80PCI800EVK on Pass slot 

    With DS80PCI800EVK on Pass slot (EQ = 00 / DEM = 00) 

  • Hi Kai,

    It is possible by adding additional de-emphasis and CTLE EQ that youmay approach the desired numbers for preshoot and de-emphasis, but ultimately I have not experienced a situation in which the DS80PCI800EVK can be used to pass more than the P4 Tx preset.

    However, we are able to pass the Tx preset test successfully with the DS80PCI810. I recommend obtaining the DS80PCI810EVM and set EQ = 0x00 (Level 1), VOD_DB = 000'b, and VOD = 110'b (Level 6) to pass Tx preset testing in the PCIe slot.

    Thanks,

    Michael