What do you do when you have some device that only supports wired-Ethernet, but it’s going to be deployed in a place where it isn’t practical to install cabling? Well, the easy solution is to use Wi-Fi, through a device known as a wireless bridge. In the early days of Wi-Fi, where integrated Wi-Fi functionality was less common, wireless bridges were easy to come by. They were a standalone device, similar to a router in size, with an Ethernet port and a Wi-Fi card. Nowadays, wireless bridges are rarer to find, with some fairly expensive gaming-oriented devices for hooking up consoles, but can still be made from a router running DD-WRT or OpenWRT. Of course, that’s still fairly complicated, requiring a separate power supply and sizeable box … so I was on the look-out for something smaller that could work for one device.
Enter the Vonets VAP11G wireless bridge. This device is produced by Shenzhen HouTian Network Communication Technology and set me back just AU$20 from eBay. This is a relatively old-school device, supporting 802.11g (54Mbit/s) standard, requiring 802.11n access points to operate in backwards compatible mode.
This device was manufactured in March 2014, and is version 5.0.0. It accepts DC power from 5-15v, and is specially designed for use with a variety of devices.
The size of a large dongle, the adapter is designed with flexible power configuration. It uses a 2.1mm DC jack, so that you can use it with devices like satellite set-top boxes and VoIP ATA’s, by “daisy chaining” the adapter in the power path. The adapter’s DC barrel plug goes into the device, and the power supply is plugged into the bridge itself. This will work for any device with a 2.1mm positive centre tip with a DC voltage of 5-15v. If this isn’t met, the other option is to use the adapter which allows you to power it from a USB port. This is very convenient for use with a computer, especially for configuration. The Ethernet plug has only two pairs connected, for 100Mbit/s operation, but that’s more than sufficient. Two very bright LEDs (blue, amber) indicate the status of connection and activity.
The bridge itself comes with an 8cm CD-ROM with a piece of configuration software that will configure the bridge to connect to your network of choice. Unfortunately, this software is pretty primitive, and functions only 32-bit versions of Windows. Luckily I had one such machine, and I could configure it that way. Others could try using the open source method, but I had no luck with it unfortunately. There is no web administration or IP for the device that I could find.
I was a happy user of this device, as it always came in handy to allow me to connect “guest” desktops under diagnosis to my network without needing to install any hardware on their machine or divulge my network parameters to them.
Sadly, when I came to use it a few days ago, it simply stopped working. The device had no LED activity, but continued to consume power at about 130mA. No Ethernet link could be established. I checked the power to the board, and I checked the continuity of the connections, and all were satisfactory. Hmm, it certainly didn’t last long :(.
I decided to take it apart to see what can be done to salvage it, and to find out what it is made out of. The unit normally runs rather warm, and we can definitely see why, because it’s a pretty substantial device. Lets get closer to the board.
The main board is marked with VONETS.COM, Model: VAP11G, Ver: 5.1 and dated 13.10.11. The top of the board shows that it is built around a Marvell 88W8515 Libertas SoC, a device used in many repeaters and extenders, dated Week 16 of 2006 and Made in Singapore. This SoC is teamed with a Marvell 88W8010 RF baseband transciever dated Week 40 of 2005 and Made in Taiwan. It seems as if the PCB has an allowance to shield these components in a metal can, but this was not actually provided. An on-board antenna is used, although with movement of a few surface mount capacitors and a jumper, the board seems to have provision for an external antenna connection.
To the left, there is a Samsung 8Mb SDRAM, marked K4S643232E, Made in Korea. There is a Pulse H1102NL magnetic transformer for the Ethernet connection, dated Week 49 of 2013, Made in China. As a result, it seems these components have a very mixed age and origin. Other chips include what appears to be a switching voltage regulator marked GT5152GGI, for which no information is available, and another linear regulator (possibly).
The PCB is dated Week 50 of 2013, and the underside contains another IC which seems to be a switching regulator, identical to the one above. The inductors for the two switching converters can be seen as well. The underside hosts the MX 29LV800 1MiB flash chip, Made in Taiwan, which houses the firmware. A provision for a polyfuse/fuse is provided but not used.
A repair was attempted by going over the board with a hot air gun, focusing on reflowing the main SoC. Unfortunately, no change in behaviour was seen. I tried the same with the SDRAM and Flash, but I could not recover the device. I suspect a failure of the Flash memory may have led to corruption of the firmware resulting in a bricked device, but since no firmware is available and recovery would be rather complicated to do, it wasn’t worth the hassle of further investigating.
I think this will lead to an issue which will become more and more common, as Flash is very common for field-upgradable devices. Flash itself has a pretty poor data retention lifetime of about 10 years (changing depending on temperature, and type of flash), where charge leakage can erode the data to levels which cause read errors. Bare flash chips do not have much intelligence in them, they are generally not read and re-written after a certain number of reads or when “weak” and just slowly lose their data. Eventually, when enough charge is lost, the data is lost, and devices suddenly stop functioning. Maybe it’s worth re-flashing old motherboard BIOSes even if not necessary, just to ensure the data is refreshed.