USB 2.0 and FCC/CE EMC compliance

[galapogos]

Hi guys,

I'm developing a USB 2.0 high speed device that is currently very much
functional. However, I'm having some problems with radiated emissions
test with both FCC and CE standards. Based on the sweep from
30-1000MHz, I'm pretty convinced it's a USB 2.0 problem, because I see
distinct spikes at 240, 480, 720, 960MHz, which are probably harmonics
of the 480MHz USB 2.0 sweep. These spikes occur on 2 different PCs
that I use, and they disappear when I use another laptop that only has
a USB 1.1 port.

I'm wondering if anyone has had experience designing USB 2.0 devices
for FCC/CE compliance? I understand that many factors could cause
these spikes, ranging from the power source, actual test PC used,
cable length/shielding/gauge, cable layout, etc...I'm wondering which
ones are the ones that cause the biggest difference. Booking the test
lab is kinda expensive and I have no equipment to perform debugging so
I hope I can do the most I can before hand to solve this problem
before showing up at the lab.

Thanks in advance!

 

[Kadir Solid Gold Suleyman]

Hi guys,

I'm developing a USB 2.0 high speed device that is currently very much
functional. However, I'm having some problems with radiated emissions
test with both FCC and CE standards. Based on the sweep from
30-1000MHz, I'm pretty convinced it's a USB 2.0 problem, because I see
distinct spikes at 240, 480, 720, 960MHz, which are probably harmonics
of the 480MHz USB 2.0 sweep. These spikes occur on 2 different PCs
that I use, and they disappear when I use another laptop that only has
a USB 1.1 port.

I'm wondering if anyone has had experience designing USB 2.0 devices
for FCC/CE compliance? I understand that many factors could cause
these spikes, ranging from the power source, actual test PC used,
cable length/shielding/gauge, cable layout, etc...I'm wondering which
ones are the ones that cause the biggest difference. Booking the test
lab is kinda expensive and I have no equipment to perform debugging so
I hope I can do the most I can before hand to solve this problem
before showing up at the lab.

Thanks in advance!

Regarding cables etc. ask the range engineer how this is typically
done. I have not done FCC for USB but for other devices. How bad are
you failing? When you quasi-peaked were you still over? Is your
device shielded? In addition to the things you listed above, it could
actually be your device which is failing FCC Class B...when you were
connected to a usb 1.1 port the 480MHz clock on your device was
probably not operating. When you connected to the two different PCs
it was operating...just pointing this out.

If your device is not shielded go to the grocery store and buy some
disposable baking pans cut out appropriate holes for USB, power, etc.
At the range with the baking pan in place, does your emissions go
down? If your emissions goes down it is probably an issue with your
design. You will need to look at PCB layout etc.

regards,
Kadir "Solid Gold" Suleyman

 

[galapogos]

Regarding cables etc. ask the range engineer how this is typically
done. I have not done FCC for USB but for other devices. How bad are
you failing? When you quasi-peaked were you still over? Is your
device shielded? In addition to the things you listed above, it could
actually be your device which is failing FCC Class B...when you were
connected to a usb 1.1 port the 480MHz clock on your device was
probably not operating. When you connected to the two different PCs
it was operating...just pointing this out.

If your device is not shielded go to the grocery store and buy some
disposable baking pans cut out appropriate holes for USB, power, etc.
At the range with the baking pan in place, does your emissions go
down? If your emissions goes down it is probably an issue with your
design. You will need to look at PCB layout etc.

regards,
Kadir "Solid Gold" Suleyman

Thanks. My device is enclosed in a partially metallic(steel)
enclosure. An older design had a fully metallic enclosure, and the
spikes were slightly less but still there. The layout/design for the
older design is different though, so it's not an apples to apples
comparison. I thought about wrapping the device up with aluminum foil
as well, but seeing that the older design was already fully shielded,
would that help?

Unfortunately for us, the testing engineer isn't very interested in
helping us pass. He's just doing his tests and telling us that we're
failing without providing us with much help, and this is the first
time I'm doing FCC/CE testing, so I'm kinda lost as to what I should
be doing as well. I'm hoping it's a simple cable issue, but I'm just
not sure.

 

[Kadir Solid Gold Suleyman]

Thanks. My device is enclosed in a partially metallic(steel)
enclosure. An older design had a fully metallic enclosure, and the
spikes were slightly less but still there. The layout/design for the
older design is different though, so it's not an apples to apples
comparison. I thought about wrapping the device up with aluminum foil
as well, but seeing that the older design was already fully shielded,
would that help?

Unfortunately for us, the testing engineer isn't very interested in
helping us pass. He's just doing his tests and telling us that we're
failing without providing us with much help, and this is the first
time I'm doing FCC/CE testing, so I'm kinda lost as to what I should
be doing as well. I'm hoping it's a simple cable issue, but I'm just
not sure.- Hide quoted text -

- Show quoted text -

Hmm.. Partially metalized is pretty good for a consumer device. You
never mentioned how bad you are failing with the spikes. It's
impossible to say without looking at your layout, and having someone
who is knowledgeable of USB testing advise you of the correct
procedure. But...

Do you have an EMC analyzer? If you can at least rent one, and a
close-field probe (or make one using coax and an improvised antenna),
you should be able to probe around and take relative readings and
isolate out where the radiation is coming out. Check the data sheet
and make sure you don't have any unnecessary clocks turned on. Try
both shielded and unshielded cables.

Eliott labs (Silicon Valley area) is an excellent EMC lab and you can
ask their EMC engineers questions. First thing I would do, tho is
check your layout.

 

[Eeyore]

Hi guys,

I'm developing a USB 2.0 high speed device that is currently very much
functional. However, I'm having some problems with radiated emissions
test with both FCC and CE standards. Based on the sweep from
30-1000MHz, I'm pretty convinced it's a USB 2.0 problem, because I see
distinct spikes at 240, 480, 720, 960MHz, which are probably harmonics
of the 480MHz USB 2.0 sweep. These spikes occur on 2 different PCs
that I use, and they disappear when I use another laptop that only has
a USB 1.1 port.

I'm wondering if anyone has had experience designing USB 2.0 devices
for FCC/CE compliance? I understand that many factors could cause
these spikes, ranging from the power source, actual test PC used,
cable length/shielding/gauge, cable layout, etc...I'm wondering which
ones are the ones that cause the biggest difference. Booking the test
lab is kinda expensive and I have no equipment to perform debugging so
I hope I can do the most I can before hand to solve this problem
before showing up at the lab.

Did you do any pre-compliance testing ?

What design principles did you adhere to in order to minimise your emissions ?

Graham

 

[galapogos]

Hmm.. Partially metalized is pretty good for a consumer device. You
never mentioned how bad you are failing with the spikes. It's
impossible to say without looking at your layout, and having someone
who is knowledgeable of USB testing advise you of the correct
procedure. But...

Do you have an EMC analyzer? If you can at least rent one, and a
close-field probe (or make one using coax and an improvised antenna),
you should be able to probe around and take relative readings and
isolate out where the radiation is coming out. Check the data sheet
and make sure you don't have any unnecessary clocks turned on. Try
both shielded and unshielded cables.

Eliott labs (Silicon Valley area) is an excellent EMC lab and you can
ask their EMC engineers questions. First thing I would do, tho is
check your layout.

The spikes are pretty bad. For FCC, here's how much it's failing by
240MHz - 20.2dB
480MHz - 17.9dB
720MHz - 22dB
960MHz - 16.8dB

For CE, here's how much it's failing by
240MHz - 14.2-25dB
480MHz - 8.7-11.5dB
720MHz - 6.2-18.3dB
960MHz - 2.4-5.4dB

Theses are over a few different PCs.

I might have access to a spectrum analyzer, but I'm not sure if I have
antennas or near field probes available. I'll check though. I'm not in
the USA so I don't think I'll be able to use their services.

 

[galapogos]

Did you do any pre-compliance testing ?

What design principles did you adhere to in order to minimise your emissions ?

Graham

No pre-compliance testing. We followed mainly reference designs, and
for the USB/IDE portions(this is a USB storage device), we adhered to
USB and IDE design guidelines for trace length/PCB stackup/etc. This
is mainly for SI issues. Emissions was not specifically designed for.