HCT testing for USB device

Hi All,

Could anybody comment on this?

Scenario:

We want to do stress tesing of a target USB device using HCT (Hardware
Compatibility Test). This device works as a network device over its USB
connection (Remote NDIS over USB).The device is working fine in normal
conditions. We connected the device to WINXP PC with HCT 12.1 running
in it. Driver category selected is Universal and the driver selected
for testing is USB8023.sys.


Problem:

In HCT testing "USB Address Description", "USB Device Control
Requests", and "USB Enumeration Stress tests" are failed. Only the "USB
Descriptor" test is passed.

In the failed tests, it is said "Data received from request does not
match previous data" for a Device Descriptor query. But from USB
analyzer, it is seen that all descriptors are trasnffered correctly.



Errata 1519 on HCT:
(http://www.microsoft.com/whdc/hwtest/search/details.aspx?Type=err&ID=1519)
Microsoft says "Universal serial bus (USB) devices require a hub
upstream from the device in order to pass these tests. USB devices that
are embedded in mobile or desktop systems, including USB ExpressCard
devices, that do not have a hub upstream of the device cannot be tested
in the test environment and configuration stated in the HCT
documentation. These tests are not required for embedded USB devices"


Question:

(1) Is our failure related to the above Errata ?

(2) Are the above tests valid for an USB RNDIS device?


Thanks in Advance,
Mathew.

Tim
Thu Nov 16 18:01:08 CST 2006

mathew.sebastian@gmail.com wrote:
> Errata 1519 on HCT:
> (http://www.microsoft.com/whdc/hwtest/search/details.aspx?Type=err&ID=1519)
> Microsoft says "Universal serial bus (USB) devices require a hub
> upstream from the device in order to pass these tests. USB devices that
> are embedded in mobile or desktop systems, including USB ExpressCard
> devices, that do not have a hub upstream of the device cannot be tested
> in the test environment and configuration stated in the HCT
> documentation. These tests are not required for embedded USB devices"
>
> Question:
>
> (1) Is our failure related to the above Errata ?
>
> (2) Are the above tests valid for an USB RNDIS device?

The bit about embedded USB devices applies to USB devices embedded in
the PC. There is no way to add the required USB hub if isn't one
already embedded too. You are testing the device with a hub, aren't
you?

But I couldn't say if this is relevant to your problem. Sorry.

Regards,
Tim.

mathew
Sun Nov 19 18:37:30 CST 2006

Hello Tim,

Thanks for your response.

You are correct; we use a target device that has a hub. The target
device is a WINCE based one and over its USB, it exposes a network
connection with PC (Remote NDIS over USB).

Could you please give an example of USB device embedded in PC that has
no HUB to connect to USB HOST?

Without connecting it to USB HOST how is the communication possible?

Waiting for your reply,

Thanks and regards,
Mathew

Tim wrote:
> mathew.sebastian@gmail.com wrote:
> > Errata 1519 on HCT:
> > (http://www.microsoft.com/whdc/hwtest/search/details.aspx?Type=err&ID=1519)
> > Microsoft says "Universal serial bus (USB) devices require a hub
> > upstream from the device in order to pass these tests. USB devices that
> > are embedded in mobile or desktop systems, including USB ExpressCard
> > devices, that do not have a hub upstream of the device cannot be tested
> > in the test environment and configuration stated in the HCT
> > documentation. These tests are not required for embedded USB devices"
> >
> > Question:
> >
> > (1) Is our failure related to the above Errata ?
> >
> > (2) Are the above tests valid for an USB RNDIS device?
>
> The bit about embedded USB devices applies to USB devices embedded in
> the PC. There is no way to add the required USB hub if isn't one
> already embedded too. You are testing the device with a hub, aren't
> you?
>
> But I couldn't say if this is relevant to your problem. Sorry.
>
> Regards,
> Tim.

Tim
Mon Nov 20 22:57:42 CST 2006

mathew.sebastian@gmail.com wrote:
>
>You are correct; we use a target device that has a hub. The target
>device is a WINCE based one and over its USB, it exposes a network
>connection with PC (Remote NDIS over USB).
>
>Could you please give an example of USB device embedded in PC that has
>no HUB to connect to USB HOST?
>
>Without connecting it to USB HOST how is the communication possible?

As an example, embedded systems that use USB-To-Go often use a direct
connection to the host controller with no hub.

A hub is not necessarily part of a host controller.
--
Tim Roberts, timr@probo.com
Providenza & Boekelheide, Inc.

mathew
Wed Nov 22 01:02:54 CST 2006

Hello Tim,

Thanks for your response.

So, can we conclude it like, the Microsoft errata on HCT
(http://www.microsoft.com/whdc/hwtest/search/details.aspx?Type=err&ID=1519)
is applicable only to USB On-The-Go devices?

I am confused because Microsoft has listed some device categories which
are affected by the above erratum.

Thanks,
Mathew.

Tim Roberts $B$N%a%C%;!<%8(B:

> mathew.sebastian@gmail.com wrote:
> >
> >You are correct; we use a target device that has a hub. The target
> >device is a WINCE based one and over its USB, it exposes a network
> >connection with PC (Remote NDIS over USB).
> >
> >Could you please give an example of USB device embedded in PC that has
> >no HUB to connect to USB HOST?
> >
> >Without connecting it to USB HOST how is the communication possible?
>
> As an example, embedded systems that use USB-To-Go often use a direct
> connection to the host controller with no hub.
>
> A hub is not necessarily part of a host controller.
> --
> Tim Roberts, timr@probo.com
> Providenza & Boekelheide, Inc.

Tim
Thu Nov 23 12:27:13 CST 2006

mathew.sebastian@gmail.com wrote:
>
>So, can we conclude it like, the Microsoft errata on HCT
>(http://www.microsoft.com/whdc/hwtest/search/details.aspx?Type=err&ID=1519)
>is applicable only to USB On-The-Go devices?

No, On-The-Go is just an example. It is possible to design an ordinary USB
host controller without a hub. It's also possible to create an embedded
device with a USB peripheral hard-wired to the host controller, without
using a USB cable at all. The criterion listed in the errata seems to be
pretty clear: if you ship an embedded device without a hub that includes
one of the USB peripherals shown, you do not need to run the four tests in
the title.

>I am confused because Microsoft has listed some device categories which
>are affected by the above erratum.

I'm not sure why you are confused. Are you creating an embedded device?
--
Tim Roberts, timr@probo.com
Providenza & Boekelheide, Inc.

mathew
Tue Dec 05 02:31:08 CST 2006

Dear Tim,

Sorry for a late response.

I would like to conclude my posting as follows

My development is on USB target side uses Philips Isp1581 USB
controller. Developed an RNDIS driver to make it as a netwrok device.
It connects to USB Host PC using USB Hub.

So, according to your explanation my device should be able to pass all
HCT tests. Now the device passes only USB Device Descriptor tests. So,
it is expected something wrong at my device side.

Thank you very much for your time and response.

Regards,
Mathew.


Tim Roberts wrote:

> mathew.sebastian@gmail.com wrote:
> >
> >So, can we conclude it like, the Microsoft errata on HCT
> >(http://www.microsoft.com/whdc/hwtest/search/details.aspx?Type=err&ID=1519)
> >is applicable only to USB On-The-Go devices?
>
> No, On-The-Go is just an example. It is possible to design an ordinary USB
> host controller without a hub. It's also possible to create an embedded
> device with a USB peripheral hard-wired to the host controller, without
> using a USB cable at all. The criterion listed in the errata seems to be
> pretty clear: if you ship an embedded device without a hub that includes
> one of the USB peripherals shown, you do not need to run the four tests in
> the title.
>
> >I am confused because Microsoft has listed some device categories which
> >are affected by the above erratum.
>
> I'm not sure why you are confused. Are you creating an embedded device?
> --
> Tim Roberts, timr@probo.com
> Providenza & Boekelheide, Inc.