Faulty USRP microprocessor?


#1

G’day,

today I discovered that my USRP appears to have a problem. The USRP
device is
nolonger detected when connected to a PC (i386) running WinXP, Linux or
NetBSD. I tried different systems and cables to no avail. This is with
GNU-Radio current, but also tried with the last stable release.

Power supply rail measures 3.3V after the regulator and the LED is fast
flashing, which normally would indicate that the processor is running,
but
firmware hasn’t been loaded. Signal lines from processor pins to
connector/cable have continuity. I’m now suspecting the USB interface on
the
microprocessor chip.

Has anyone had similar experience and what was the solution to your
problem?
If so, what was your solution?

I really would hate the idea of having to replace the microprocessor
because
of lack of spares and in particular its package size.

Any help is very much appreciated.

73, Berndt
VK5ABN


#2

G’day,

I haven’t made any progress in finding a solution to my problem. A
direct
email to Matt (Ettus) explaining my predicament didn’t receive a reply.
It
was hoped that the manufacturer of the USRP would provide some after
sale
techmical support for their products.

The USRP is no longer communicating when it was powered up after being
in
storage for the past 6 month. As mentioned in my previous email, see
below,
my suspect is the co-processor that hosts the USB interface, although I
can’t
be sure. Perhaps someone on this list knows a procedure that would lead
to an
accurate diagnosis.

At this stage I’m contemplating to replace the processor. What needs to
be
considered prior to replacing this processor, e.g. does it need to be
flashed? I’m also looking for a supplier for this chip.

Any help is very much appreciated.

73, Berndt
VK5ABN


#3

Berndt Josef W. wrote:

G’day,

I haven’t made any progress in finding a solution to my problem. A direct
email to Matt (Ettus) explaining my predicament didn’t receive a reply. It
was hoped that the manufacturer of the USRP would provide some after sale
techmical support for their products.

Berndt,

I am sorry to have not gotten back to you, but I get over 300 non-spam
emails per day, many with technical questions, and it can be hard to
wade through them. I try to go through the backlog at least once a
week.

The USRP is no longer communicating when it was powered up after being in
storage for the past 6 month. As mentioned in my previous email, see below,
my suspect is the co-processor that hosts the USB interface, although I can’t
be sure. Perhaps someone on this list knows a procedure that would lead to an
accurate diagnosis.

It sounds like the transceiver circuitry on the FX2 has gone bad, but
the processor is still running. This the most common failure mechanism
we see. We have static protection diodes on the recent boards, but it
does not seem to change things. The best course of action is to replace
the FX2 chip. We do not see this failure often, but when it does, it is
often when used with cheap aftermarket power supplies for the laptop, or
with a nonstandard USRP power supply.

At this stage I’m contemplating to replace the processor. What needs to be
considered prior to replacing this processor, e.g. does it need to be
flashed? I’m also looking for a supplier for this chip.

They do not need to be flashed. You should be able to get the chip from
Digikey or any other Cypress dealer. The exact part number is
CY7C68013A-100AC. They are about $15 in small quantities.

Matt


#4

G’day,

I’m glad to inform that replacing the FX2 chip resolved the problem and
the
USRP appears to be fully operational, e.g. usrp_benchmark and usrp_wfm
work
out of the box.

I since tried to connect the board using a USB2.0 cable of 5m in length
and
can’t establish communication with it. The cable length is within the
USB2.0
specifications. It may just be a poor quality or bad cable, but would
like to
hear if anyone else has been successful in using a long USB cable.

Using an oscilloscope, I can see a pulse train of about 0.5Vpp with
channel A
(D+) and channel B (D-) at the USRP connector. Does this sound right?

cheerio Berndt


#5

On Thursday 13 November 2008 09:12:37 Matt E. wrote:

I am sorry to have not gotten back to you, but I get over 300 non-spam
the processor is still running. This the most common failure mechanism
They do not need to be flashed. You should be able to get the chip from
Digikey or any other Cypress dealer. The exact part number is
CY7C68013A-100AC. They are about $15 in small quantities.

Matt

G’day Matt,

many thanks for your reply. You’re confirming my suspicions and I’m now
confident in tackling this problem head0on by replacing the FX2.

Hope to be back on-the-air with USRP and GNU-Radio soon.

73, Berndt
VK5ABN


#6

On Tue, 2008-12-30 at 12:07 +1030, Berndt Josef W. wrote:

I’m glad to inform that replacing the FX2 chip resolved the problem and the
USRP appears to be fully operational, e.g. usrp_benchmark and usrp_wfm work
out of the box.

Glad to hear it!

I since tried to connect the board using a USB2.0 cable of 5m in length and
can’t establish communication with it. The cable length is within the USB2.0
specifications. It may just be a poor quality or bad cable, but would like to
hear if anyone else has been successful in using a long USB cable.

While 5m is indeed the spec, my experience with many USRPs and USB
cables has been that only high quality cables that have been USB 2.0
certified can be depended upon. I’ve had brand new cables that would
otherwise be in spec, but lack the 2.0 certification logo, completely
fail. On other occasions, they’ll work fine, but then fail mysteriously
at odd times. I don’t think I’ve ever seen a 3m cable fail, or a
certified 5m cable fail, so it just looks like there isn’t much margin.

-Johnathan