It’s become clear from comments on my previous article, that Lenovo have released a new Windows 7 driver for the Ericsson F3507g Mobile Broadband card.

You would never know from glancing at the downloads page because it doesn’t list versions numbers or publish dates…

The driver is now listed under Vodafone and AT&T Wireless WAN (HSPA) Driver. Just to make things extra confusing (and harder to find) they removed the actual card model number from the description. The new filename  is 7uw706ww.zip (direct download link), and comes in at 40MB (10 MB smaller than the last driver). It comes as a Setup.exe, including both 32-bit and 64-bit drivers.

After installing the new driver (no need to uninstall the previous one), you will have the following new devices in device manager:

• F3507g Mobile Broadband Driver (Network Adapters)
• F3507g Mobile Broadband Device Management (COM3)
• F3507g Mobile Broadband GPS Port (COM6)
• F3507g Mobile Broadband Device

The crucial thing about the new driver is the change in version from 4.4.50.6.0 (dated 16/12/2008) to 4.54.7103.1 (dated 26/06/2009). The drivers are also HCL certified by Microsoft. Some devices are dated 29/07/2009 and version 1.0.0.52. Either way, all the drivers are much newer

This driver is radically different from the previous one. The previous driver installs your broadband card as an old school modem (with COM ports et al). This is what made it particularly tricky to setup.

The new driver actually gets the broadband card to show up as… (drum roll)… a Mobile Broadband Connection!

The upshot of this is that, after installing these drivers you don’t need any third party software to get online!

I am not sure if this device type was available in Vista, but I have certainly never seen it before. It integrates with Windows differently to a dialup modem. One obvious change is that it shows up in the networks list accessed via the system tray like so:

You can set it to automatically connect when there is no other connection available, which is quite handy and works quite well.

There is only one trick you may need for setting the APN. To do this you need to go to the networks list and right click, show properties. Note: This is completely different than if you select properties in the Network Connections windows. You should see the following:

Here you can enter the APN and username/password if required by your ISP.

There are only a couple of negatives I have discovered so far:

• Using Wireless Manager it was possible to force the connection to only use HSDPA etc. I can’t find a way to do this with the new setup
• There is not as much status information available as when using Wireless Manager. For instance, when I am connected, it ALWAYS seems to display UMTS as the connection type, even though I know I am on HSDPA. This may or may not be a bug.

One very exciting feature that I just stumbled upon when researching this article is the Windows 7 Sensor compatible GPS Location driver that gets installed.

This new feature for Windows 7 allows applications to access your GPS data through a standard API, for example Google Chrome can use this to position you on Google maps! I have not got this to work so far but I am hoping that’s because I have no GPS signal indoors.

Overall I am happy with the new drivers – it’s certainly a better way to integrate with Windows, and the connection seems to be much more reliable.

For the last few months I’ve had Windows 7 RC installed on two home computers, and both of them have exhibited a strange problem: They would randomly wake from sleep .

The computers would often wake up just seconds after being shut down. At other times they would wake up in the middle of the night. To make matters worse one PC, which had recently had its graphics card upgraded, would go into an endless loop of restarting without ever getting to the Windows login screen. I suspected a power problem relating to the new card.

I spent quite some time investigating these wake problems (which I believed were unrelated) but could not diagnose the issue. Eventually I gave up and had to disable sleep on both machines (which pained me greatly!).

Last week I had a second wind and was determined to resolve this problem once and for all. I figured there must be a tool that could tell me what caused the wake up events, and after some research I discovered the Vista/Windows 7 powercfg tool.

To use this tool you simply type powercfg –lastwake at the command prompt

In the above screenshot you can see the computer was woken by something on the USB hub (my mouse in this case). But much to my annoyance, when I ran this tool after a phantom wake it reported the type was “Unknown”!

Because I had a suspicion this could be a wake-on-lan (WOL) behaviour I used the WireShark network analysis tool to monitor network traffic for WOL packets. Perhaps a third computer had been infected with malware and was trying to wake other computers to have its wicked way? Unfortunately nothing showed up on the wire.

No further progress was made until I randomly stumbled across a forum posting about a network adaptor setting called “Wake on pattern match“. I checked both my PCs and this setting was enabled. It appears this obscure setting is enabled by default in Windows 7 for Realtek RTL8168B/8111B and RTL8168C(P)/8111C(P) Family Gigabit Ethernet NICs.

What does “Wake on pattern match” actually do?

Most people are familiar with Wake-on-LAN which uses a “magic packet” (hard-coded) to trigger a wake event. Well it seems that feature has been superseded by the more flexible “Wake on pattern match” capability. According to this Microsoft Article on Power Management for Network Devices:

The packet patterns that define the wake-up frames [for "wake on pattern match"] are provided to the NDIS 5.0 miniport driver by the operating system. At runtime, the protocol sets the wake-up policy using OIDs. These are, for example, Enable WakeupSet Packet PatternRemove Packet Pattern. Currently, the Microsoft TCP/IP is the only Microsoft protocol stack that supports network power management. It will register the following packet patterns at miniport initialization:

• Directed Layer Two packet
• Address resolution protocol (ARP) broadcast for station IP address (frames with DIX header)
• NetBIOS over TCP/IP broadcast for station’s assigned computername (frames with DIX header)

I am no networking expert but the last two sound like they could be pretty common activities on your typical home network. If so it’s no wonder my computers were exhibiting such crazy wakeup symptoms. (Either that or I have got a renegade PC on my networking sending out these packets).

The solution? Simply disable the “Wake on pattern match” setting under the network adaptors advanced properties tab, as shown below:

My biggest mistake? I could have short circuited the resolution of this problem by testing my computers without the Ethernet cable plugged in right at the start. The reason I didn’t was I couldn’t understand how a WOL packet could possibly be being sent so I ignored the possibility.

Now I know about “Wake on pattern match”, next time I will be prepared.