Eye diagram test

Question

I am trying to perform a eye diagram test on a device to check its USB performanance.

I am using a board "SQIDD" in order to perform this test. The following figure shows my test setup,

I am using EHCI HSETT software in my PC to perform the test. Pc is connected to the one end of the Sqqidd board and to the other end device is connected.

I am using Tektronix DPO7254 Scope to perform the eye diagram test. In the software i configure as following,

• far end test
• upstream
• D+ and D- are present on the board so the Differential probe is connected on the board.

Initially when i am doing the test the switch in the board is on INIT and once the TEST PACKETS are configured on the PC i toggle the switch to the TEST mode. Doing this i get quite satisfactory eye diagram, as follows,

What I was wondering is why is it necessary to toggle switch from INIT to TEST? What does this actually do?

So after this I perform the test putting the switch on the INIT the eye diagram test measured is as following,

Why is the eye diagram so disturbed? Is it really necessary to use the sqidd board to perform the eye diagram test? Can i also perform as following?

So in the above diagram I am thinking of directly connecting the device to the Host PC and measuring D+ and D- directly on the device. Is this possible? Is this also a way to perform eye diagram test on a device?

Answer 1

Regarding your question when you leave the SQIDD board in the init mode you will also be measuring the reflection of the cable connected to the host PC and the reflected packets from the host. The Test mode simply flips a switch where your device will be "directly" connected to the scope.

So in init mode you can look at the SQIDD as an op switch just connecting the DUT to the HOSt and when in Test the SQIDD terminates the DUT either in to the Sma probes into the scope or in to a 45Ohm termination on the board.

You can measure the USB signal directly I guess but the SQIDD really is the best choice you just NEED to use the test mode SQIDD, and measuring directly makes little or no change compared to the Test mode SQIDD do keep in min that your cable needs to be as short as possible (4" is typically used)

Going further into this do you use SMA probing or do you have active probes attached to the SQIDD.

a bit more info: http://testusb.com/HSEYE.htm

Answer 2

Have you read through and understood the test setup indicated in the manual?

http://cp.literature.agilent.com/litweb/pdf/N5416-97007.pdf

If you're following the correct procedure, then I'd guess that you either have triggering set up incorrectly, or you're trying to measure a high speed (480mbps) signal while the SQIDD only supports up to full speed (12mbps).

You could remove the SQIDD and test directly from the board, but the probes introduce unwanted effects on the signal, so you wouldn't be able to trust most measured parameters from the signal.