FX3 ESD Failure

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
DeAl_292241
Level 3
Level 3
First solution authored 10 replies posted 10 questions asked

Hi,

Some customers returned their FX3 devices with a dead USB port on it (always SS section). The measured DC impedance on the SS diff pairs is completely different on a dead device from a working device. Our layout for ESD seems to be very optimal to me and we are using this ESD IC:

SP3010-04UTG Littelfuse Inc. | Circuit Protection | DigiKey

I would like to have some feedback from Cypress about this.

By the way, the case of the USB connector is not directly connected to GND of the PCB. The GND of the USB connector is connected to the PCB GND through a passif filter as we saw on many designs.

Thanks,

Denis Alain, Eng.

0 Likes
1 Solution

Hi Jayakrishna,

The resistors on the R_USB3 pins have 200 Ohms in series (measured on the bad unit). I removed both 22 Ohms on the USB2 bus and the device is not detected with a USB 3.0 cable. The VBUS is at 5V on the bad device.

I need to release my ECO next week. It sounds very difficult to find the cause of this issue. If you don't have any last recommendations, I will release the PCB and we will probably send you back the next defective device if any, since we have improved the way the ESD is soldered on the PCB. It could correct the issue (the quality of the solder joints is almost impossible to test in production!).

Thanks,

Denis Alain, Eng. 

View solution in original post

0 Likes
11 Replies
JayakrishnaT_76
Moderator
Moderator
Moderator
First question asked 1000 replies posted 750 replies posted

Hello,

Please share the complete schematic so that we can understand the problem better.

Best Regards,

Jayakrishna

Best Regards,
Jayakrishna
0 Likes

Hi Jayakrishna,

I can send you my sch directly, since this is confidential information. Can I have your contact information please?

I can share a snapshot of the routing / placement here (U10 is the ESD):

pastedImage_0.png

As you can see, the ESD GND is directly connected to the USB connector and SS traces passes through it. I cannot place this ESD closer. It's already a problem in inspection.

Note: the SS traces should probably be routed on bottom to decrease the stubs of the connector (more a signal integrity issue that we never experienced with this routing).

Thanks,

Denis Alain, Eng.

0 Likes

Hello,

I have sent my email ID. Please send the schematic so that we can review it on our side.

Best Regards,

Jayakrishna

Best Regards,
Jayakrishna
0 Likes

Hello,

Please answer my questions below:

1.What exactly do you mean by port dead. Is superspeed and high speed not enumerating or only one is enumerating.

2.How did you measure the DC impedance on differential SS pairs? Are both the SS pairs of the dead device showing difference in DC impedance? How much is the difference in DC impedance?

3.Please confirm whether you are getting 1.2V VDD fine. Also please let us know the ripple on that supply.

4.Please share the clock waveform with the voltage levels.

5.Similar to the layout image shared before, please share all the planes below that part of the plane.

Best Regards,

Jayakrishna

Best Regards,
Jayakrishna
0 Likes

Hi,

1) Dead port means that the USB 2.0 is working, but not the USB 3.0 link (both are tested in production).

2) We are using a simple multi meter (set in impedance). We saw that the DC impedance of a good board has about the same DC impedance on both + and - signals for each SS pairs, but not on the dead SS port. Also, DC impedance are "about" the same level between each board when we look at a specific SS diff pairs. This is not the case for a dead board.

To be more specific, I have a dead unit here (SS port is not working, but USB 2.0 is fine): the DC impedance on SS_RXp is about 2.7MOhms like other devices, but about 10 MOhms on SS_RXn. Of course, this measure depends the orientation of the multi meter probes (plus vs minus probes), but it just a good hint that something inside the FX3 is been damaged.

3)  1.2V VDD is fine (1.197V). The ripple voltage with noise is about 25mV (measured at full BW on scope and with a good GND while the device is working on a good unit).

4)  By clock waveform, I assume you are talking about the OSC of the FX3. Here are the values on a good board:

a) XTAL_IN: -0.04V 1.28V

XTALIN.png

b) XTAL_OUT: -0.34V to 1V.

XTALOUT.png

5) The plane just below is a full GND, except we have a cutout under the DC caps of the SS pairs. I can send you the other layers by email if necessary.

pastedImage_3.png

Thanks for your support,

Denis Alain, Eng.

0 Likes

Hello,

Please share all the planes below the part of the layout shared before to my email.

Also, all my questions in reply 4 are related to the board that is not working properly. Please confirm whether the information provided about the 1.2V VDD, ripple and the clock waveform are of the defective board and not of the working board.

Best Regards,

Jayakrishna

Best Regards,
Jayakrishna
0 Likes

Hi,

The measurements were done on a good unit. I retested it on a dead SS port unit and I have very similar result in term of VDD, VDD ripple/noise level and OSC levels (no difference).

I will directly send you the layout under the USB-FX3 section of this board.

Best Regards,
Denis.

0 Likes

Hello,

Apologies for the late response. Please try the following and let me know the result:

1. Remove the 22 ohm resistor on the DP and DM (usb 2.0) lines coming from the USB receptacle and use a USB 3.0 cable to connect to the host. Now check whether the VBUS is still present or not.

2. Please confirm that the resistance between R_USB3 and GND pin of FX3 is 200 ohm with a tolerance of 1%.

Best Regards,

Jayakrishna

Best Regards,
Jayakrishna
0 Likes

Hi Jayakrishna,

The resistors on the R_USB3 pins have 200 Ohms in series (measured on the bad unit). I removed both 22 Ohms on the USB2 bus and the device is not detected with a USB 3.0 cable. The VBUS is at 5V on the bad device.

I need to release my ECO next week. It sounds very difficult to find the cause of this issue. If you don't have any last recommendations, I will release the PCB and we will probably send you back the next defective device if any, since we have improved the way the ESD is soldered on the PCB. It could correct the issue (the quality of the solder joints is almost impossible to test in production!).

Thanks,

Denis Alain, Eng. 

0 Likes

Hello,

Please try the following and let me know the result:

1. On a good and bad board, try loading USBBulkSrcSink example project which comes along with FX3 SDK and share the USB logs. This can be used to understand what exactly is happening on the USB side.

2. If feasible, swap the FX3 chips from a good board and bad board. That is take out the FX3 chips from a good board and bad board. Then swap them with each other. If a chip from good board works on a bad board, then it is the problem with the chip. If a chip from bad board works on a good board, then it is the problem with the board.

Also, please let me know the ratio of boards that are failing.

Best Regards,

Jayakrishna

Best Regards,
Jayakrishna
0 Likes

Hi Jayakrishna,

As expected, the good board is working at both USB speed. The bad one is only working at USB 2.0 speed. I cannot easily swap the FX3 as suggested. On the bad board, the device is not detected at all on the USB bus at SS speed (no enumeration tries). I'm using a tool to see the enumeration process.

At this point, I need to close this discussion, since I need to release the board. This event happens in very rare occasions. I will directly communicate with you if any other device still failed in the field with the improved footprint for the ESD protection and we will see what to do next.

Many thanks for your support,

Denis.

0 Likes