Hello,

at the moment i am trying to do my first steps in developing software for the FX3.

I have an USB hardware analyzer, which i use to record traces of the communication between the PC and the FX3.

This analyzer can only record USB high speed, so i use the FX3 at a USB 2.0 port with transmission speed "high speed".

Already during downloading of an image i see a lot of "invalid frames" in the trace.

Downloading itself works, because i think, USB is doing a lot checks and repititions, so error are corrects.

Has someone a clue what it could be? I am using a FX3 with the VID=1480 / PID=0000 problem.

Could it be a chip problem? Layout problem (it is a custom board)? What else?

Has someone already taken USB hardware traces? Are there also "invalid frames" in your traces?

(I have attached a zipped trace, which can be opened with software from http://www.ellisys.com)

Best regards,

Martin

It seems the USB trace was filtered out, perhaps size too big (zipped ~3,3 MB). If you want it, just contact me.

Hi,

I don't see any invalid frames when using the firmware download or during firmware operation.  What OS are you using, I am using 32 bit XP with Renesas USB 3 controller.

Sodafarl

Hi,

Please try attaching the trace again.

If it doesn't work please create a tech support case (MyAccount -> MyCases) so that one of our engineers can analyze the same.

Regards,

Anand

I am using Windows Vista (64 bit edition).

I try again to attach the trace to this posting.

Attachment seems to be filtered out again. I have uploaded it on a webserver. It can be downloaded from here:

http://engelschall.com/~martin/ellisys_invalid_frames_bad_or_not.zip

I haven't seen anything like this on the firmware downloads I have been looking at. Try different USB cables, the shorter the better. As the Ellisys analyser will take two cables try and keep the overall length less than 5 metres for USB 2. It would be worth checking what the analyser records for a good USB device - will you see the same thing for example. Is this the Cypress software you are using to download the firmware. What USB 3 controller card are you using. Can you try this on a machine with XP as well.

> Try different USB cables, the shorter the better.

I will try this.

> It would be worth checking what the analyser records for a good USB device - will you see the same thing for example.

I think i have not seen invalid frames (at least in such a big amount) with other devices, but i will check this again.

> Is this the Cypress software you are using to download the firmware.

I am downloading via WinUSB (on a Vista 64 bit machine) and self written software, but i have seen the same thing also with Cypress software on a different machine (notebook with Vista 32 bit machine). All of them USB 2.0, USB controller on mainboard. Download of the FX3 image itself is always ok and software on FX3 is running, so i think all errors during download were corrected by lower layer of USB. Correct me if i am wrong, but I assume (but i am not sure) software (on PC side as well as FX3 side) has no influence on error correction at such a low level?

> What USB 3 controller card are you using. Can you try this on a machine with XP as well.

I am not using an USB 3 controller (PC side) at all at the moment, because my analyzer can only capture USB 2.0 HS at maximum, not USB 3.0 superspeed.

One idea: Would it be possible, either on FX3 side or on PC side to query error counters? Perhaps there is some statistic generated? "Invalid frames" or "Repeated frames" or so? So i can check this topic without the use of an USB analyzer, which removes the possibility of influence of the USB tester // longer / bad cables on one side and it easier to check for someone else without an USB tester on the other side.

Hi,

Are you connecting FX3 to host through a hub? Or is there a hub connected to the host

I'm seeing a lot of SSPLIT and CSPLIT transactions on the hub and they're related to hub.

Looking at the traffic before the SSPLIT and CSPLIT it looks like there are packets for which FX3 doesn't respond. It gives the feeling that there might be signal quality on the connection may not be good.

Regards,

Anand

In the above trace i used a hub. Environment looked like this:

PC(Vista 64 bit, WinUSB, USB-Connector on mainboard) <-> USB 2.0 Hub <-> USB-2.0-Analyzer <-> FX3

I just have uploaded another USB trace which i created also some days ago:

Can be downloaded from here: http://engelschall.com/~martin/fx3load.zip

Environment was this: Cypress Control Center on Notebook Vista 32 Bit with Cypress Driver

Notebook <-> USB-Analyzer <-> FX3

All cables were around 1m and shorter. Included in the trace is initial enumeration (VID=1480, PID=0000),

then the upload of the UVC example via Control Center and renumeration to VID=04B4, PID=0x00FA.

Can you please again have a look into the second trace?

There are lot of packets for which FX3 is not responding. Looks like it is not receiving the packet for some reason.

Please let us know the answer to the following questions

1. Are you using FX3 DVK

2. Is the cable USB-IF certified

3. Have you tried using a different PC

Regards,

Anand

> 1. Are you using FX3 DVK

I have a custom hardware, which perhaps makes the problems. I try to get a FX3 DVK to be able to retest.

> 2. Is the cable USB-IF certified

I think so. I tried different cables, longer and shorter ones. All of them have the USB logo on it, so i assume they are ok.

> 3. Have you tried using a different PC

At the moment tried 2 PCs (1 desktop, 1 laptop). Both times i see the problem of "invalid frames".

I can see the problem only with the USB tester. Without it i cannot see, if there are invalid frames.

But with different DuT (e.g. an usb stick) i see no invalid frames.

I must put it on hold on my side for now till i have the FX3 DVK to compare behaviours

(beside someone else has an idea what else i can test or how i can fix it).