USB2 Loopbacks and duty cycle setting

[DanB]

Need to exercise all systems ports (serial, parallel, network, USB, RAM, hard drives, etc) simultaneously for EMI testing. Ideally we'd like to complete a single pass as fast as possible (running in continuous mode).

Had problems with version 4.0 related to serial port errors. Decided to upgrade to version 5.1.

Seeing a problem in version 5.1 related to USB Loopbacks and duty cycles. Testing (5) USB 2.0 Loopbacks.

With USB duty cycle set at or below 48, 7 cycles passed in 4+ hours.
With USB duty cycle set at or above 49, 20 cycles passed in 5 minutes.

Note that in version 4.0 the USB duty cycle setting appeared to work "properly". We also saw similar problems in version 4.0 for the RAM duty cycle setting- duty cycle 99 and the test crawled; duty cycle 100 and the test ran fast. Note that the RAM duty cycle setting appears to have been corrected in version 5.1.

[Ian (PassMark)]

By design, the duty cycle setting is not linear with the USB 2.0 testing. This was changed in V5 on request of a number of customers, that had USB problems and wanted to be able to introduce much greater delays to help their investigations. In your case, it sound like you just want a Duty cycle setting of 50 or more.

To be more precise, the following is a description of the duty cycle delays inserted in V5.0 and V5.1:

Duty cycle delay 100:
No introduced delays.

Duty cycle delay 1 – 99:
Between every 8000 writes/reads, a delay of at least x ms is inserted.
Where x = (100-dutyCycle)*70

Duty cycle delay 1 – 49:
Additionally, between each write/read cycle at least x ms delay inserted.
Where x = (50-dutyCycle)/2

Duty cycle 0 - 9:
Additionally, at least 1 ms delay inserted between write and read.

We will update the help file.


FYI, we did change the RAM duty cycle settings in V5.x to make it more linear.