USB superspeed peripherals Forum Discussions

The datasheet says the transfer rate is guaranteed for iso endpoint. I do not really understand the guaranteed transfer rate. Is it mean the transfer rate will not be lower that a certain value or be a constant from time to time? In my application, FPGA  gets data from ADCs in a constant data rate.  FPGA write data to the slave fifo P port of FX3 with the constant date rate. FX3 send the data to the host computer. In this application, should I use iso endpoint to ensure the transfer data from FX3 to host be a constant and avoid the full of the in buffer? Is it better to use iso endpoint than bulk endpoint.

In another application. the host computer send data to FPGA through FX3. FPGA produce waveform according to these data. If the host send data with a varying rate, the endpoint may become empty and the waveform thus become incorrect. In this application, is the iso endpoint better?

Just to clarify one thing, Can we use SPIDMA mode and bit banging SPI in single program?

I have an requirement where i need to read data from NOR flash using SPIDMA and config FPGA using bit-banging SPI? Is this possible?

We use USB-IF compliance test tool(USB3V) to test our FX3 camera, but it failed in "TD 9.24 U1 U2 Test (Configuration Index 0x00) - Device State Configured", please see attachement. In our code, we use CyU3PUsbLPMEnable()/CyU3PUsbLPMDisable() to switch the LPM state.Please help me and give me some advice?

Thanks.

I download the firmware USBIsoSourceSink using control center. Before downloading the control center find Cypress FX3 USB BootLoader Device. After downloading, it find none of device. The GUI shows empty space. I think It suggest error occurs while firmware running. Is there any informtion on solving the problem. I have try many different firmwares on this board. It is the first firmware presenting the problem. I think my borad is ok.

Does the auto commit happen when max packet size is reached or one buffer is full? Is the buffer size and max packet size the same thing? Assume only one buffer is used for a DMA channel and this buffer is committed.  Could the external device write data to this buffer at the same time while host is reading?

Hi ,

I'm having a CX3 device which is programmed to act as a UVC-CDC device(based on the sample code Cx3UvcOV5640_UVC_CDC).

I need some way to identify the firmware version loaded into CX3. With CDC implemented I cannot use the Debug prints because they share a common UART.

What are the ways in which i can implement this?

Thanks in advance

Regards

Ajay

I wants to use Cypress FX3 to send the image from GPIFII input to UVC isochronous output to PC by USB3

I found FW sample code (...\Cypress\EZ-USB FX3 SDK\1.3\firmware\uvc_examples\cyfxuvcinmem), but it wasn't GPIFII input.

I added some gpif api into sample code (as attachment file), but it didn't work normally.

Could you teach me how to modify it ? Or provide your FW (gpif to UVC ISO) to me ?

cyfxgpif_syncsf.h is my GPIF configuration.

Hi All,

I'm having a couple of issues and will start a couple of different threads to discuss them.

In this thread I will discuss camera multiplexing on the CX3.  We have a CX3 design that uses an ON semiconductor FSA642 to multiplex the CX3's MIPI port between two different image sensors (currently two OV5647s).  There is also an SCCB multiplexer that connects the active sensor's SCCB signals to the CX3 SCCB bus.  For this hardware, I have pieced together some firmware examples to use the camera's brightness setting to switch the multiplexer inputs between the two sensors (by setting brightness to 0 or 1).

This configuration is functional and works with both e-camView and a simple OpenCV based viewer I wrote.  There is a rather large limitation I am trying to overcome: the time to switch from one camera to another is several frame times because the CX3 must be switched out and back into application streaming mode.  Ideally, we'd want to capture images alternating from the two sensors, but nothing I've done has been able to perform this "live" switching.

Any thoughts on how to reduce the time to switch between sensors is much appreciated.

Thanks,

Scott

We have a USB 3.0 camera and we configure bulk endpoint for image data transfers. Normally, we use CyU3PUsbGetSpeed() API to know whether configure the bulk endpoint for USB 3.0 or USB 2.0 and its size accordingly. But turns out this sequence doesn't work out for all the mother boards when the device remains connected during boot up.

1. In ideal case, after CyU3PUsbStart();, in the evenCB, we get the evType = CY_U3P_USB_EVENT_CONNECT, evData = 1 (USB 3.0) before we configure the endpoint and the initialization goes well.

2. In some mother boards, when the device is connected before boot and when the OS boots up, first we get the event evType = CY_U3P_USB_EVENT_CONNECT evData = 0 (USB 2.0) and at this point when we use CyU3PUsbGetSpeed(), it shows HighSpeed. We complete the endpoint configuration at this stage. After the OS is fully loaded, we again get the events, evType = CY_U3P_USB_EVENT_RESET   evData = 0 (USB 2.0)  & evType = CY_U3P_USB_EVENT_CONNECT evData = 1 (USB 3.0). But at this stage, since the endpoint is configured for USB 2.0 already, obviously there are problems in transferring data through that endpoint.

Turns out that this is due to a special mode in certain mother boards, (XHCI Mode configurable through BIOS settings), the port remains as USB 2.0 compatible during the OS boot up and after its completion, it goes in switches to XHCI mode.

Now, with default sequence provided by Cypress, this condition is missed and we found a simple logic to make this work.

If we wait until the evType = CY_U3P_USB_EVENT_EP0_STAT_CPLT evData = 0, and then carry on with the end point configuration, then it works fine since all the CY_U3P_USB_EVENT_RESET & CY_U3P_USB_EVENT_CONNECT are completed before this.

Need advice if this kind of behavior is common and if the fix above is recommended.