TUSB1002: Debugging

Hi Thomas,

Typically yes but this depend on which mode you are using. If you want to use a VNA to understand the frequency of the channel while TUSB1002 is active you can try using the PCIe Mode of the device. This takes away the USB dependencies but fixes the linearity and DC gain. Please see 8.3.2 Detailed Design Procedure for some additional details. 

I would recommend using TUSB1002A as it has basic redriver mode that could be used for this and is a more updated version of this device. Is this a USB application? 

RXP/N and TXP/N pairs should be routed with controlled 90-Ω differential impedance (±15%) What is the tolerance on your routing?

What key factors are you looking for to validate your system?

Hi Malik

It is a USB 3.0 system. Must admit that we looked at a reference system and did'nt pay enough attention to the design, besides the impedance. The impedance tolerance is below 5 % we used a MEGTRON builtup for the PCB.  

We know the IC has different modes and settings. But we wanted to see what these modes ans setitngs would infer on for instance S21, S11 and S22, for each of the individual lanes / lanepairs.

What is the difference between TUSB1002 and TUSB1002A ? I see they are pin compatible...

Hi Thomas,

With 5% tolerance the 85 ohm routing is okay. TUSB1002A has improved EQ and more up to date USB state machine for USB 3.2. As well as the added basic redriver mode.  

You can see an example of how the different EQ, DC gain and linearity settings change SDD21 when measured on the TUSB1002 alone. TUSB1002 will impact your measurements in a similar way. For example, higher EQ will shift the S21 (or SDD21) curve to show less loss in dB at higher frequencies. This would be the same for both lanes. When measuring return loss (S11 and S22) only major variation that would occur is when the RX termination is enabled vs disabled. Return loss is not typically measured at the transmitter. 

But doesn't the state machine impact the S21 measurements ?

Understanding correct, the termination is used to detect settings, how to verify that the USB state is correctly detected ?

Is the state machine different for USB3.0 / USB3.1 / USB 3.2 ? 

Hi Thomas,

In short, no the state machine has no impact once TUSB1002 is actively re-driving. TUSB1002 only has basic equalization and does not contain a DFE. The USB state machine is used to handle LFPS signaling, far-end receiver detection, enter and exiting of low power link states etc. Once the USB link has been established and TUSB1002 has detected actual high speed data then  equalization is applied. EQ settings are determined via pin strap on the configuration pins (non-high speed data pins) and sampled at reset or power-on reset. This occurs before USB state machine has started. 

Far-end termination detection is used to detect link partner. What "USB state" are you trying to verify.

There is a difference between the version as each newer version has increased functionality and considerations however they are all backwards compatible so TUSB1002 3.1 state machine is good for your USB 3.0 application. 

Thanks, now to the actual measurement. So we put it in PCIe / test mode. But since we only have a two port VNA, are there any good ways to measure then ?

Hi Thomas,

This can be difficult with just a two port VNA. You will have to measure each pin separately (EX: TXp --> RXp) while terminating the other pins with 50 ohm. Then combine the touchstone files to get the whole picture by channel.

Thanks, this was also how we planned to do it :-) Just wanted to hear if that should cause any for us unexpected behaviour from the IC, state machine etc ...