TUSB1004: Compliance test Fail

Hi ttd,

Can you share the compliance test result? I'd like to see the waveforms and parameters that fail.

For TD.1.4 I believe you are correct that the eye height is too large. Are you able to lower the output swing on your USB host? This would be the most straightforward way to lower your eye height. 

ttd said:

Then should I change the VOD smaller with VOD_CTRL Register (Offset = 32h) [Reset = C0h]?

Are the CRX/CTX pins on the TUSB1004 connected to your test-port or are the SSTX/SSRX pins connected? The reason I ask is that the VOD setting will only change the output swing for the SSRX1/2 transmitters. If these are not connected to the port under test, then it is unlikely that this setting will affect the test.

For TD 7.40 I'll look into whether we've seen this before. Can you provide the test log that details the failure?

Best,

Shane

Hi ttd,

1. From the compliance result you've provided I suggest lowering your EQ on the SSTX path.

2. I'll start putting together the TD 7.40 test here with our TUSB1004EVM and see what result I get. Let me follow up next week on this point.

3. I see you're also having issues with USB2 compliance. Looking at the eye diagram, there is excessive pre-emphasis on the signal that is violating the upper and lower bounds:

I recommend lowering your host pre-emphasis/overshoot to pass compliance.

Best,

Shane

Hi Shane,

Thank you for your reply.

1. I'm going to set lowering your EQ on the SSTX path and have a compliance test.

2. Thank you for tring to reproduce with TUSB1004EVM.

3. OK,  I'm going to set lowering your host pre-emphasis/overshoot to pass USB2.0 compliance.

I have additional question.

Fixed Equalization for CRX1/2, USB3.1 is recommended with fixed EQ and USB 3.2 is recommended AEQ mode.
My system is for USB3.2, so I guess that AEQ mode is better based on datasheet. 
Current situation USB3.2 compliance test didn't pass.

Are there advantages and disadvantages to enable/disable(fixed EQ) for AEQ?

Best Regards,

ttd

Hi ttd,

Using Fixed/AEQ does not depend on the USB spec version you're using (3.1 or 3.2). From the standpoint of EQ, there is no difference between these two protocol standards.

Fixed EQ means you're using a set EQ value for any channel connected to the type-A receptacle, regardless of the actual channel length. The drawback is that this EQ value is often not the best setting for any connection that is made, as it needs to compromise between short cables and long cables. AEQ fixes this by dynamically setting the EQ value based on the connection on the Type-A receptacle.

I recommend using AEQ to give your system the best chance at passing RX JTOL.

Best,

Shane

Hi Shane,

Thank you for your reply. I will use AEQ for recommendation.

By the way, according to the datasheet "Full Adaptive Equalization" , there are some combination of register setting.

Can you tell me if there are any common combinations to try for starters?Defatul settings are better? 

I also confirm when are additional add and sub necessary with OVER_EQ_SIGN  and OVER_EQ_CTRL?

FULLAEQ_UPPER_EQ =8h(AEQCFG="F")

OVER_EQ_SIGN = 0h(positive)

OVER_EQ_CTRL=0h(0 or -8) ← no  addition for EQ value of AEQ

 I don't understand follwing description especially. Does "in front of TUSB1004" mean yellow circle on block diagram?

ーーーーーーーーーーー

The TUSB1004 also provides the ability to add or subtract some over/under equalization to compensate for channel in front of TUSB1004 by programming OVER_EQ_CTRL field to a non-zero value.

ーーーーーーーーーーー

Best Regards,

ttd

Hi ttd,

ttd said:

Can you tell me if there are any common combinations to try for starters?Defatul settings are better? 

- Set the CTLSEL and USB_MODE registers to enable USB

- Set register 0x1C to F3

ttd said:

also confirm when are additional add and sub necessary with OVER_EQ_SIGN  and OVER_EQ_CTRL?

This is only necessary if you want to tune the AEQ higher or lower. I recommend keeping these registers at their default values to start.

ttd said:

I don't understand follwing description especially. Does "in front of TUSB1004" mean yellow circle on block diagram?

Yes that is correct.

Best,

Shane

Hi Shane,

Thank you for your answer.

I'm going to try to set 0x1C to F3. It means FULLAEQ_UPPER_EQ= Fh for sweep EQ wider range than 8h(default).

I think that it have a chance to adjust the appropriate EQ setting. Is that extend time for polling retry test? (TD 7.40 Fail?)

By the way, if you have the application document such as TI design for compliance test pass including the HOST and TUSB1004 and compliance test measurement equipment.

If it has TUSB1004 settings like EQ and board information like distance related to pass the test method. It is essential to refer for my system. 

Best Reagards,

ttd

Hi ttd,

Changing the FULLAEQ_UPPER_EQ register should not affect the time in your polling retry test.

ttd said:

By the way, if you have the application document such as TI design for compliance test pass including the HOST and TUSB1004 and compliance test measurement equipment.

We have passing TX and RX compliance results on the TUSB1004 EVM. Here are the results:

USB Transmitter compliance: 4861.USB 3.2 TX Compliance.pdf

USB Receiver compliance: USB 3.2 RX JTOL Long Channel.pdf

These were taken with the TUSB1004 EVM connected upstream to a USB host PC via type-A to micro-B cable. I see that the host used was an intel host and this test used SSEQ setting #2 and CEQ setting #5.

Do you notice any performance improvement after lowering the EQ or switching to adaptive EQ?

Best,

Shane

Hi Shane,

Thank you for sharing complicance test doc for TUSB1004EVM.

It will take some time as I'm planning to conduct a compliance test with an external certification. I will share the results when these are available.

Best Regards,

ttd

Hi ttd,

Sounds good. I'll keep this thread open for replies.

Best,

Shane

Hi Shane,

Thank you for your supprt. I pass the compliance test for USB-A I/F with TUSB1004.

but there is a USB-A I/F without Re-driver on my board and it failed the  Long Channel and pass short channel.

AEQ settings for TUSB1004, I assume that the Long Channel test is for long cables and the Short Channel test is for short cables(like USB memory), is this correct?

If so, what is the specific length of the long/short cable?
In general complicance test doc, it seems to be 3m cable + long host PCB for long channel and short channel means (No cable + short) host PCB.

Best Regards,

ttd