Hi, I just wanted to share some communications I am having with the owner of passmark.com which whom I purcased the USB 2.0 loopback plug from. Maybe some of you can shed some light or offer some direction to my problem. Or maybe just feel releaved that your setup doesnt behave like mine, who knows.


You might want to browse the bottom portion first.


---------------------------------------

Hello Ian,

This was really helpful but I need some more insight to a problem I am having. I will break it down to you as best as I could.

Problem: At cold boot (when computer is not on for a long while) my wifi USB device does not start. Tried Netgear, and D-Link (108mps). Both same results.

Setup: Sumicom S620 which is those supersmall pc's, P4 2.8, Intel 865 (I think) chipset, intel USB chipset.

Lots of USB devices:

2 Belkin self power hubs
1 16" Range Extender, powered.
Audigy 2 NX sound card
GPS device
USB to Serial device
Wireless firefly remote
Diamond PVR600 TV device
Touch screen
4-1 card reader.

Everything is self powered either by direct power or powered hubs except the wifi. Basically this problem only happens when the computer has been off for a while. BTW, this setup is in my vehicle and I have a really good PSU powering the computer and quite a few POL's powering the USB devices. When the system starts up for the first time and dosent enable the wifi I can unplug it and here the driver unloading even though I donít see the driver. I then can plug it back in and the wifi will activate. If I shut the computer off breifly, including removing the power cable from the computer so the wireless is not getting the 5V and then turn it back on the wireless is load correctly. It is only when the system has been off for quite a while when the wireless will not start on the first attempt. Like I said above I tried this with two different wireless devices and went through the whole driver uninstalling, reinstalling bit. At first I thought it was a power issue so I made a USB cable that used power from a separate power source for the wireless instead of using the computers power, still no go at first bootup.

I can get it to work right on the first startup when a lot of the devices are disconnected. So to find out why this was happening I hooked up the USB loopback and starting benchmarking the bus. I knew this from day one that when no devices are connected I get a 350-370 USB throughput. With all devices connected I get a 68mps reading. Some devices aid to pulling the USB benchmark down more then others like my wireless keyboard and mouse, I think I noticed it drop like 30mps alone. I rememberd what you said about noise and shielding in the usb cable and the cable to keyboard and mouse receiver was long and not shielded so I replaced it with a shorter higher quality shielded cable. I didnít benchmark the difference with this new cable but it had no effect on the wifi booting up from a long cold start. Though, with the netgear wifi, which by the way was the more stubborn one, it didnít work when this kb/mouse receiver was in from a cold start and even subsequent cold starts if the computer was off for as little as 5 minutes.

The strange thing is the wifi is the only device doing this...I even investaged on how I can get the wifi drivers to load before other USB drivers do. I am close but I havent fully made the necessary changes to the registry to change the ServiceGroupOrder and Start reg keys for the wifi device.

My question is, why does it do this on the first cold boot only?, and when it is powered down and started up immediately it works? Btw, the system has to not have power on the computer completely for it to load correctly. Meaning, when the driver fails to load a reboot will not work, the 5V on the USB port needs to be interrupted before the computer starts back up. This is why I thought it was a power issue, but appling 5V to the USB from a separate source and before the computer even started up didnít work.

The computer has a pcmcia slot and I am ready to give up on trying to get this to work and just use a pcmcia wifi, but its killing me that I cannot get this working right, nor I cannot seem to put my finger on the problem. Maybe I am at my limits with how many USB devices I can have but I thought USB can support 255 (or close to it) devices??

Other things I am doing are, using the shortest possible USB cables or of high quality. I just purchased a Monster Cable USB cable 7ft. To replace the one going from the PC to one of the Belkin 4 ports. The other Belkin 4 port is connected to computer on a 12" cable connected to a high quality 16' USB range extender. I am not sure if I can improve the USB line alltogether or how to bring up the avarage bandwidth throughput when all devices are connected. I am wondering if theres a way to log the failure of the driver load and to know why but that might require debugging the system and I am not completely familiar with that.


-----Original Message-----
From: IanR (Passmark) [mailto:[email protected]]
Sent: Wednesday, August 24, 2005 3:16 PM
To: Joseph Puma
Subject: Re: USB 2.0 Tester

Joseph,

Receiving a recovered bus error does not have any impact on the correctness of the data transmitted or received (data verification errors indicate this). The recovered bus error count will increment on any USB error...any noise on the USB cable or any kind of glitch seen by the USB transceiver. It is possible that connecting other USB hardware to the bus could introduce suffiicent electrical noise on the bus to trigger such an event. This should not be seen as a problem with the hardware, but rather, a guide that further investigation may be required. The typical cause of recovered bus error is poor quality cabling.

Different chipsets have different maximum throughputs, this could be the case with what you are seeing. It is important to note that as you add USB devices to the (Universal Serial) Bus you will observe a reduction in maximum performance of each USB device. This is to be expected and is due to increasing overheads on the bus as more devices are added. The overheads introduced from each USB device on the bus reduces the number of bytes that can be utilized by other USB devices during a frame (i.e.. less data blocks can be transferred per frame ). Overheads include packet organization, framing information, polling, clock adjustment and reserved time.

Best regards,
Ian Robinson
PassMark Software

> ----- Original Message -----
> From: "Joseph Puma" <[email protected]>
> To: "David Wren" <[email protected]>
> Sent: Wednesday, August 24, 2005 6:46 AM
> Subject: RE: USB 2.0 Tester
>
>
> Just curious. Is it normal to get recoverable errors when doing a test
> and having other USB hardware hooked up. I don't get any errors when
> only just the USB2Test is hooked up. Though on a HP computer I ran the
> test on I had the same wireless USB keyboard and mouse receiver
> (Gyration) and I didn't get any errors, but it could have been that
> the wireless was on a different controller since I was plugged into
> the front with the usb2test and the wireless was in the rear. Hmmm....
>
> Also, I found that that the intel chipset performs much better then
> lets say a SIS chipset. I was getting a 379high on the intel and like
> a 290 on with a sis chipset. Of course when you add USB hardware that
> number drops down. Why is that?
>
> Joseph Puma