I appreciate the shielding Apple added, especially with two resonators in the HF band. The shielding is, BTW, probably to prevent the device from causing interference, not to prevent other devices from interfering with it. (It's not a radio, after all.)<p>Note that if you do choose to buy the cheaper one without any shielding, it's your responsibility (in the US) to prevent it from interfering with licensed users of the RF spectrum. (In this case, 25m shortwave broadcasters on 12MHz and government-run time-transfer services, like WWV and WWVH, on 25MHz.)<p>If your device is particularly annoying to some shortwave listener, expect to get a letter from the FCC telling you to shut the thing off.<p>(I looked up a few of the letters, and I really appreciate the politeness from the government: <a href="http://www.arrl.org/files/file/Part%2015%20Letters/04-06-24-Battery%20Charger.pdf" rel="nofollow">http://www.arrl.org/files/file/Part%2015%20Letters/04-06-24-...</a> "There are
several simple and inexpensive steps that can be taken to eliminate interference from battery chargers, and we would be glad to send them to you if needed.")
According to Bunnie's investigation into microSD cards[1], that low serial number probably indicates that the components were produced on a "ghost shift", when a rogue worker comes in at night and runs the plant off the books<p>[1] <a href="http://www.bunniestudios.com/blog/?page_id=1022" rel="nofollow">http://www.bunniestudios.com/blog/?page_id=1022</a>
I'm equally impressed by the Apple adapter. They make expensive stuff, but they legitimately don't cut corners. It's nice to see there's still one consumer electronics company that still does that.
The lack of shielding is rarely any problem. Two oscillators is not better than one, since PLLs can generate the required clocks (480MHz = 25 / 5 x 96, 100MHz = 25 x 4, 125MHz = 25 x 5); SMSC has some USB/Ethernet solutions that use the same 25MHz frequency.<p>> The Windows drivers are the exact same digitally signed ones that Microsoft distributes through Windows Update<p>AFAIK there is a standard device class for "USB ethernet controllers", so any chipset that conforms to it will work fine with the standard drivers.<p>As for the unmarked chip - many IC companies are not averse to creating custom designs for a specific customer (and marking it however you want) if you're willing to buy enough. I don't think it's ASIX since they don't have the single 25MHz clock source; more willing to bet on Microchip or SMSC.
I recently purchased a Gigabit ethernet adapter for my rMBP. If you're fine with something a bit larger and are willing to go above "dirt cheap" I'd strongly recommend the Inateck HBU3VL3-4. It's made by a German company so I'm not sure how available it is over in the USA.<p>It has an ASIX AX88179 which is a USB 3.0 to Gigabit chip and it my tests I was easily able to get around 800-900Mbps. It also has 3x USB 3.0 ports, so not only do you not lose your USB 3.0 port you gain 2 more. I've not dismantled it but it feels solidly built.<p>Here in the UK I paid the same price as the Apple USB to Ethernet Adapter, which is only 100BASE-T and offers no hub.
We have used a similar dongle at our company for a 2nd ethernet adapter on our Foxconn-built boxes and found out that they usually don't have unique MAC addresses. This made our lives very difficult because Cobranet audio network uses level-2 addressing. We eventually found a supplier with real MAC addresses.
An old analog hacker rule of thumb is that if your product is going to emit RF, you'll find spurious RF on input and output terminals (including ground terminals), which you can check with a scope. Such a pre-test is cheaper than getting a field test done and failing.<p>The scope test might be do-able on a comparison basis between the two designs. It's possible that Apple over-engineered their shielding.<p>It's not necessarily a cloned design. The cheap part may simply implement standard protocols, and use the name of the Apple part to ensure driver compatibility without too much testing. Still, that seems rather under-handed if it's what it seems.
OP isn't alone - I find these parts dissections fascinating as well. Posts like these show you how and why hardware is becoming a comodity. It's going to be a very interesting century.
>I’d love to learn more about these secretive industries and the engineers who work in them.<p>Isn't it pretty well known that 1) they aren't secretive and 2) they don't actually design their own hardware. They start off by producing the legitimate item, in the factory that's contracted by the company who designed the product. If the product is successful or easily fits in their existing knockoff fabs, they plans are basically copied over
We use a ton of USB adapters where I work and are using them all day long with several different systems. A few months ago we got a batch of cheap adapters and one noticeable difference was the cheap ones get really hot compared to the more expensive name brand adapters. Appear to work OK, though.
The taobao link in the original post leads to a listing which does not specify the chipset or support for Mac. This suggests it's a cheaper chipset. The 4-5 best-selling cheap USB-to-Ethernet adapters on taobao use the RD9700 chipset, and can be had for 22RMB including shipping.<p>The cheapest Asix 88772-based one I can find from a reliable seller is 25RMB including shipping.<p>It seems like there is another 50 cents that can be cut from the design, although the RD9700 may not have an OSX driver:
<a href="http://forums.macrumors.com/showthread.php?t=1450026" rel="nofollow">http://forums.macrumors.com/showthread.php?t=1450026</a>
This is interesting - and more importantly, is there any part of the ethernet standard they don't support? I only ask, as at my university there were a number of networks that students simply couldn't connect to if they used <i>any</i> ethernet adaptors. Oddly enough, thunderbolt adaptors, and ethernet on the motherboard worked fine, just USB adaptors didn't work.
Great writeup.<p>So I guess the question is, is the one with the fruit on it worth $25 more? It seems like the cheap one is more "Wozniak" in approach.