6Mhz square wave coming out on USB pins

ben2 · ‎2005-09-13

Posted on September 13, 2005 at 11:06

ben2 · ‎2005-02-14

Posted on February 14, 2005 at 10:54

I have been playing around with the usb interface on the str710 eval board, specifially the bulk usb demo that was posted on here, I have made a few minor changes to do with the data once received.

I have noticed 2 problems.

1. The chip seems to get itself into a state where it will not accept the address asigned by the host. This I think is probably a software bug. Looking at a USB analyser output it appears that no ack is being sent.

2. This is the most confusing problem. Once up and running if I unplug/insert the usb cable I eventualy get to the stage where the chip appears to ouput a 6Mhz square wave on both usb pins. This confuses the host no end. I have checked with a scope and it is the insertion of the cable that causes this problem, not the removal. Once the chip starts to generate the square wave it continues even when there is no usb cable inserted.

Is anyone aware of any hardware problems with the USB core on the STR710 chip ?

Has anyone else seen this ?

many thanks for any assitance.

mohamed23 · ‎2005-02-16

Posted on February 17, 2005 at 06:19

Hi ben_fnr,

There is no hardware problems with the USB core on the STR710 or STR711 chips, I suspect that you have a bug in your software

May be in ''usb_pwr.c'' and ''usb_prop.c'' files. I have seen similar

behavior in a previous software and it was a wrong initialization

of the USB analogue macrocell (Clocks/PLL) with suspend/resume events, or a wrong implementation of your Bulk Class ( Endpoints... ) in ''usb_prop.c''.

Regards, :-]

ben2 · ‎2005-02-17

Posted on February 18, 2005 at 06:05

Hi Rave,

thanks for the comments. I am pretty sure it is not a clock/pll problem as I do not change these bits. It is using an external 48mhz oscillator.

dont think it is a wrong implementation of the bulk endpoints as I can correctly transfer data (OUT host->device) not using any other endpoints., apart from endpoint 0.

When it goes wrong it happens before any endpoint transfers. The host cant even set the address of the device because of the 6Mhz square wave coming out of the STR7.

I am not sure exactly what should happen after the chip goes into suspend mode. As the cable has been removed wont it do a reset upon cable insertion and not a resume ?

The chip is being left in a running state at all times. No clocks/pll's are changed. The usb part goes into suspend mode as far as I can tell once the usb cable has been removed and 90% of the time starts again perfectly ok when the cable is inserted.

mohamed23 · ‎2005-02-17

Posted on February 18, 2005 at 07:51

Hi ben_fnr,

I think that I can resolve your problem, if you send me your firmware & software PC :o

Regards

nopti_albe · ‎2005-09-05

Posted on September 05, 2005 at 10:49

I have this exact problem... what should i do?

It's a hardware bug as far as i can see... if i put the scope on only one pin it will almost always get into the ''clock generator'' state. If i put 2 probes on both D+ and D- it will never happen.

[ This message was edited by: tudor on 05-09-2005 14:40 ]

mohamed23 · ‎2005-09-12

Posted on September 12, 2005 at 16:27

Hi all,

are you using the STR710-Eval boards ?

With regards,

nopti_albe · ‎2005-09-13

Posted on September 13, 2005 at 11:06

I use it... and it works well. I din't use CAN, lcd,big external flash. Apart of that everything works wonderful. For usb i have replaced the emi-removal chip and it works now.