TUSB4041I: How to check behavior.

Kuramochi Tadahiko

Part Number: TUSB4041I
Other Parts Discussed in Thread: TUSB4041PAPEVM

Hi,

My customer's system including the TUSB4041I is occurring USB communication defect(Host MCU doesn't detect USB devices).

There are both correct sets and incorrect sets.

On incorrect sets, the defect occurs not depending on USB device and ports. The set can't communicate all USB device on all ports.

Though I would like to know if the TUSB4041I has any problem, is there register I should check?

I couldn't find the register on the datasheet.

Best Regards,

Kuramochi

over 5 years ago

0 David (ASIC) Liu over 5 years ago

TI__Guru*** 139985 points

Kuramochi-san

Does the host correctly recognize the TUSB4041?

Are the downstream devices bus powered or self powered? Please verify VBUS operation.

If VBUS is ok, do you have a USB Protocol Analyzer? Please use a protocol analyzer to capture the USB enumeration between correct and incorrect sets, then we can do further analysis. This would debug the design at signal/protocol level.

Thanks
David

0 Kuramochi Tadahiko over 5 years ago in reply to David (ASIC) Liu

Guru 28715 points

Hi David-san,

Thank you for your information.

The host seems to not be sometimes able to recognize TUSB4041I.
Probability not recognizing USB devices is about 7% when power supply turn on and off.

When communication is fault D+ is not pulled up at enumeration start.
D+ is pulled up when communication succeeds .

I think that the reason why D+ is not pulled up is TUSB4041.

Have you ever seen such a problem?

What should we check as TUSB4041I side?

If you need schematic, please send me(kuramochi@fujiele.co.jp) email.

I can send the schematic.

Best Regards,
Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Kuramochi-san,

Please check the reset timing to be sure the hub is being held in reset after the power rails have ramped. Also, check that the ground pad connection of the device has sufficient solder.

Regards,
JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

Thank you for your advise.

>Also, check that the ground pad connection of the device has sufficient solder.

I would like to check the intention of your question.

Do you mind thermal problems?or pattern impedance?

Best Regards,

Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hello Kuramochi-san,

The thermal pad acts as the main ground connection for the device, so the device may fail to operate or operate intermittently if there is a poor thermal pad connection.

Regards,

JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

Thank you for your support.

The customer checked what soldering a thermal pad is correct.

About the miscommunication sample, miscommunication is improved by toggling GRSTz.

So I guessing that this miscommunication is caused by start up failure.

Though I think that power sequence follows requirement, could you check below attached?

TUSB4041I powersequence1.xlsx

Best Regards,

Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hello Kuramochi-san,

The power on reset / GRSTZ timing looks correct. Can they check that the clock is starting and VBUS goes high when the system is not connecting?

Regards,

JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

Hi JMMN-san,

Could you check the attached file?

1323.TUSB4041I powersequence1.xlsx

Best Regards,

Kuramochi

0 Kuramochi Tadahiko over 5 years ago in reply to Kuramochi Tadahiko

Guru 28715 points

JMMN-san,

This symptom occurred with EVM also as attached.

TUSB4041IEVM.xlsx

Could you check it on your EVM?

Best Regards,

Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hi Kuramochi-san,

Please check the crystal frequency, it should be 24 MHz not 25 MHz.

Regards,
JMMN

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hello Kuramochi-san,

It looks like when the EVM fails to attach the 3.3V rail is not given sufficient time to discharge, this would result in an insufficient power on reset leaving the device in an undefined state.

Regards,
JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

Thank you for your information.

I understood that VDD1.1V is not related to this symptom.

Is it correct?

And what voltage is VDD to active power on reset?

About crystal frequency, my customer says that the cause of 24MHz is resolution of oscilloscope and true frequency is 25MHz.

I'll check it, again.
Best Regards,

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hi,

Did you mean to say:
"About crystal frequency, my customer says that the cause of 25MHz is resolution of oscilloscope and true frequency is 24MHz."

I just want to confirm that the frequency is 24 Mhz.

Please look at the timing of 3.3V / 1.1V and GRSTz to determine if a proper reset is provided to the hub during the power cycle test. Reset timing is described in the device datasheet.

Regards,
JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

Sorry for my late reply.
We check crystal frequency in this week.

And I capture waveforms when enumeration doesn't begin on TUSB4041PAPEVM and my PC.
I think that VDD and GRSTz doesn't have any problem.

And 5V was supplied from not an adapter but power supply.

I switched power on/off by the power supply.
Could you try it also on you EVM?

Best Regards,
Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hello Kuramochi-san,

GRSTz and power rail ramp looks ok. Can you zoom in on GRSTz and DP/DM. It looks like the hub is going into suspend mode. Also, you can enable port status outputs using the Additional Feature Configuration Register and see if HS-SUSPEND goes high.

Regards,
JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

Thank you for your advice.

Which bit should we confirm to check SUSPEND?

And bellow is zoom waveform.

Because AUTOENZ/HS_SUSPEND was dropped manually by turning off SW2(on EVM), Does this AUTOENZ/HS_SUSPEND means that the device doesn't enter suspend mode ?

Best Regards,

Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hello Kuramochi-san,

HS_SUSPEND is a status output so the setting of the pin does not affect the suspend state of the hub. I'm not sure why the bit description was removed from TUSB4041, it should be bit 4 to enable the status outputs. Obviously the SS_SUSPEND and SS status pins are not valid for TUSB4041.

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

I checked the register(0xF0) when communication does not succeed.

Then, 0xF0 was 0x00 as attached, so I think that the device didn't enter SUSPEND mode.

(Reading register is succeed because Vendor ID LSB Register<0x01> can be read 0x51.)

8037.TUSB4041IEVM FJE 20181011.xlsx

Could you give your opinion against this condition?

And please check this phenomenon on your EVM.

Best Regards,

Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hello Kuramochi-san,

The status change is not reflected in the register. The bit in the register, just enables status outputs. At power on reset, please set bit 4 in register F0. Then, HS_SUSPEND status will appear on AUTOENZ/HS_SUSPEND pin.

Regards,
JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

Thank you for your reply.

Could you tell me intention of your request?

If HS_SUSPEND goes high by writing 1 to 0xF0, what do we know?

Best Regards,

Kuramochi

0 Kuramochi Tadahiko over 5 years ago in reply to Kuramochi Tadahiko

Guru 28715 points

JMMN-san,

I checked that HS_SUSPEND goes high when settin bit 4 in register F0.

Best Regards,

Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Kuramochi-san,

Thank you for the data. To determine why this is occurring, I need to see a zoom in of VBUS, GRSTz, D+ and D- when there is a failed connection. I don't see any issues with the EVM in my lab as long as the host controller is given time to report the disconnect event before repowering the hub.

Regards,
JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

Thank you for your reply.

This symptom seems to occur only when SMBUSz pin is pulled down.

(My customer is pulling SMBUSz pin low.)

Then, frequency of occurrence is once a 20~30 times.

Could you confirm it with customer SW setting mentioned below attached file ,again?

And I attached waveforms you requested.

1200.TUSB4041IEVM FJE 20181016.xlsx

Best Regards,

Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hello Kuramochi-san,

When the SMBUSz pin is pulled low, the TUSB4041 will stay in programming mode until the SMBUS host set / clears the cfgactive bit in the TUSB4041 registers, that is likely what is causing the enumeration issues.

Regards,
JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

Thank you for your information.

I confirmed that CFGactive bit is 1 only when enumeration issue occurs.

I believe that this causes enumeration issue.

If there are not both EEPROM and SMBus host, how we should do to set 0 into CFGactive bit surely after releasing GRSTz?

Best Regards,
Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Hello Kuramochi-san,

If the customer is configuring the hub using pin configuration, then SMBUSz should be high.
If the customer is using an EEPROM to configure the hub, then SMBUSz should be high.
If the customer is using a SMBUS host to configure the hub, then SMBUSz should be low.

In the case where SMBUSz is low, the SMBUS host must write/clear the cfg active bit for the device to operate as a hub.

Thanks,
JMMN

0 Kuramochi Tadahiko over 5 years ago in reply to JMMN

Guru 28715 points

JMMN-san,

Sorry for my uncertain explanation.

I confirmed that enumeration issue occurs when SMBus is high and there is no EEPROM.
So I would like to know how to avoid enumeration issue in the case of pin configuration(not EEPROM and SMBus setting).

Best Regards,
Kuramochi

0 JMMN over 5 years ago in reply to Kuramochi Tadahiko

TI__Guru 60865 points

Does the SMBUS host access fail when the enumeration issue occurs? Is the CFGACTIVE bit written / cleared or not?

Thanks,
JMMN

Interface

Interface forum

TUSB4041I: How to check behavior.