high speed connection instead of super speed on CX3 based board

[leda_289486]

Hi,.

we have a batch of CX3 boards that don't connect in super speed but only in high speed. what is to be checked out first that could prevent cx3 to connect in super speed (power supplies, capacitors, connector, etc.) ?

Until then, all of these previously built CX3 boards where connecting in super speed without issue.

thank you for your help,

Best Regards

[KandlaguntaR_36]

Damien,

Please check the question 3 in this KBA: Trouble Shooting Guide for the FX3/FX3S/CX3 Enumeration - KBA222372

Regards,

Sridhar

[leda_289486]

Hi,

thank you for your information.

For  check, I have some questions related to document :

CHECK HARDWARE

2) Are the given value (20 mV for AVDD ,U3TX, U3RX and 100 mV for VDD) peak-to-peak values ?

2bis) have you a typical waveform of observed noise on some of your boards ?

CHECK FIRMWARE

1) Could you precise the software to be use to test this issue ?

1bis) Is the firmware suitable for a CX3 based board ?

With this software check, what results (debugprint,etc.) are to be expected and how to use it to correct issue ?

Otherwise, we have checked all the points you gave in the document but we cannot find any difference between a "no problem, always SuperSpeed" board and "always High Speed" board ?

Thank you for your help,

Best Regards,

[YashwantK_46]

Hi,

In CHECK HARDWARE

2) The given values are for peak-to-peak noise level for the given power supply lines.

In CHECK FIRMWARE

1) Please extract and  load the attached firmware into the CX3 board that you are currently having issue with and please send the UART debug logs that you will retrieve after the process.

1b) The firmware for CX3 is attached with the post which can be used for getting debug prints.

Regards,

Yashwant

[leda_289486]

Hi

thank you for information related to peak-to-peak amplitude. that's helpful.

I have extracted firmware but it doesn't work on our board. There are no debugging output and either no connection usb also so I can't not send you any information related to this.

Have you some oscilloscope traces related to this ?

Are you interested in our own oscilloscope trace of power supplies noise ?

Thank you for your help,

Best Regards.

[YashwantK_46]

Hi,

I have made some changes to the attached firmware. Extract it and check if you can get the debug prints and send them.

Also, how many CX3 boards do you currently have with the same issue?

And can you check if the boards respond the same with other Host devices or if there is any change that is observed.

And check all the boards with the firmware to see how many respond and how many don't and please share the result with me.

The CyU3PUsbSetTXDeemphasis() API should not be altered with and is to be used in its default state.

Also, You can try to change the voltage Swing using the SetTxSwing API (Enter Tx swing values by trial and error and find out which one has least effect of noise). Refer to API guide on how to use this API .

The debug log looks something like the photo attached below.

Regards,

Yashwant

[leda_289486]

Hi,

with your modifications, the debugprint is working. The only modification I made is to set serial port @ 921600 bauds instead of 115200.

here is the obtained output file while trying to connect a "high speed only" board with your firmware software.

Hope this will help to understand.

thank you for your help

Best Regards,

[leda_289486]

Hi,

additionnal information.

Here is the result with a working board, same firmware.

I've checked out that the board was connected in superspeed during this test.

Best Regards.

[leda_289486]

I have tested several values for TxSwing from 90 to 127 (max). This doesn't change the connection result.

Best Regards.

[leda_289486]

I forgot to tell you that 4 boards have this issue for now.

Best Regards.

[leda_289486]

Also,

what do you think of the datagrams ? From your advice, is the observed noise correct or not ?

Thank you for your help,

Best Regards,

[leda_289486]

I test on another usb controller on different pc but result is exactly the same.

"working boards" are working, "high speed" boards stay high speed.

Best Regards.

[leda_289486]

I have also tested several kind of cables of different lengths and different brands.

Behavior stay the same.

Best Regards.

[leda_289486]

Hi,

What are the tolerances of values of 0.1 uF capacitors on SSTXP and SSTXN lines ?

Are C0402_100NF_10V as suitable as C0402_100NF_16V  ?

Manufacturer reference muRata GRM155R71C104KA have a +/- 10 % tolerance value (100NF/16V) so are they ok ?

Thank you for your help,

Best Regards.

[YashwantK_46]

Hi,

Can u please remove the component DLP11SN900HL2L- common mode choke from the USB 3.0 lines and test if there is a change in one of the board?

Regards,

Yashwant

[leda_289486]

Hi,

it is quite difficult to remove these common chokes and replace them with something else correctly. Furthermore these boards are production not test boards.

What seems to be relevant is the capacitors nature on SSTX lines. What are the tolerances on these ?

Thank you for your help,

Best Regards.

[YashwantK_46]

Hi,

I currently don't have any idea about the tolerance values. Will update you as soon as i get them.

Also, The descriptions of capacitor used on the superspeed explorer kit are as follows:

Also, can you please explain what behavioural change did you notice when you replaced the capacitors in your board?

Regards,

Yashwant

[leda_289486]

Hi,

thank you for this information.

When changing capacitors form 100nF/10V to 100nF/16V and everything else stay identical, superspeed connection re-appears systematically on previously non working boards.

Could you explain why you select 25V capacitors ? These capacitors are on USB lines where voltage is supposed to be 5V only. Could this explain why our 10V capacitors didn't allow superspeed connection ?

Best Regards,

[YashwantK_46]

Hi,

There's is nothing wrong with the 10V 100nF capacitors that you originally had on the board and there was no need to replace them with 16V 100nF or 25V 100nF as well.

The board should ideally work at 10V capacitor as well( line voltage-3.3V * 3(times)) is usually taken.

Could you please provide the part number for C0402_100NF_10V capacitor?

Can you please check if you can replace the old capacitor with a 10V 100nF capacitor from any other vendor and update me?

Regards,

Yashwant

[leda_289486]

hi

100nF/10 V capacitors are yageo

CC0402KPX5R6BB104

For now I don't have other brands for these 100NF/10V. They have all this reference.

Thank you for your help,

Best Regards,

[leda_289486 · Accepted solution]

Hi,

in fact to secure the superspeed connection we have chosen to use 0402/100nF/16V/ 5% capacitors.

As there are just 2 capacitors of this kind required this is not a problem for cost aspect.

yageo CC0402JRX7R7BB104 are OK for this.

Best regards

[YashwantK_46]

Hi,

Thank you for the information.

Regards,

Yashwant

[leda_289486]

I saved the datagrams for observed noises on the different power supplies pins.

U3RX :

U3TX :

AVDD :

VDD :

Hope this can help to identify the reason of this "not superspeed" behavior.

Thank you for your help,

Best Regards.

[leda_289486]

one word.

can CyU3PUsbSetTxDeemphasis parameter have an influence on this behavior ?

We use default value for this parameter until now.

Are there others api functions that can adjust also this ?

Thank you for your help,

Best Regards.

[leda_289486]

Hi,

regarding oscillator 19.2mhz that is one point to check.

I have exchange oscillators between one working and one non working board, but the change didn't change the behavior of both boards.

So I suppose the issue is not related to oscillator 19.2Mhz.

if you have noise oscilloscope working traces for power supplies, I take it.

thank you for your help,

Best regards,

[leda_289486]

Hi,

have you some new information related to answers we sent ?

thank you for your help,

Best Regards.