Right, details then.
Quick teardown of the iCointech iMinerProduct link:
http://www.icointech.com/product1_en.phpA friend of mine ordered some (one of those BTC 'rich kids' who mined way back when) and threw one my way to dissect before he was going to plug them in, himself. I'm all set up for USB peripheral shenanigans, he is not.
I have to admit that I was skeptical as well well. Who wouldn't be? It's a super tiny miner, supposedly, in a package not unlike that of a USB stick. Like these:
http://www.dhgate.com/store/product/flash-drive-10pcs-lot-usb-2-0-cap-u-disk/139494557.htmlhttp://www.legitreviews.com/images/reviews/923/ocz_diesel.jpghttp://www.usbbook.com/productinfo.asp?id=325I did remark earlier in the thread that these are pretty standard husks that you could fill with whatever if you really wanted to:
http://www.wholesale-usb-flash-drive.com/index.php?main_page=product_info&cPath=6&products_id=133Note that there are some slight variations, mostly in the cap and endcap, but also in whether there's any grooves and, if so, where (thin sides vs broad sides) I believe this is closer to what the iMiner's husk actually is:
http://www.promo-wholesale.com/china/Hard-Drives/8/Sprite-1-Gb-USB-Flash-Drive-47848.htmI actually have a USB stick quite like it, and I took a comparison picture.
PrecautionsMachine completely disconnected from the internet, running a sandboxed VM, all ready to go with a USB sniffer, network sniffer and procmon in case it turned out to be a USB stick with some executable on it.
Turns out, all of that was unnecessary. The cgminer version pointed to is stock version 3.1.1, the batch file simply adds the parameters:
--icarus-options 115200:1:1 --icarus-timing 3.3=120So we can already tell a little bit about what we could expect to find later, that along with the CP2102 drivers (note: there's newer available that add a cert for Windows 8.1 - the version number has not been updated by SI, though) at the very least suggests something other than a regular USB stick, which tend to have a controller+usb interface chip fully integrated together.
Plugging it into the machine, nothing untoward appears to be happening. The usual banter back and forth getting USB details. It just registers as a virtual COM port - there's no custom Vendor ID or anything in here ( USB\VID_10C4&PID_EA60\0001 ). No flash drive, no HID trying to quickly take over the machine through simulating keypresses - nothing. All that happened externally was 4 blinks and then the LED remaining on.
Benign, let's give 'rConsidering the device itself appeared to be benign, and cgminer was benign, I figured what the heck and powered it up on an easy-going pool (50BTC - yeah, I read all the reports - I don't usually mine at all and they were less hassle than eligius, this is not an endorsement).
Lo and behold, the LED turned off, and shortly after, cgminer happily reported almost 300Mhash/s from the device. Along with, eventually, shares (short blink of the LED) and hardware errors (longer blink of the LED).
Edit: After 10,000 Satoshi at 50BTC, stats:
cgminer version 3.1.1 - Started: [2014-03-02 05:32:21]
--------------------------------------------------------------------------------
(5s):374.4M (avg):298.5Mh/s | A:6833 R:1 HW:147 U:2.0/m WU:4.1/m
ST: 2 SS: 0 NB: 331 LW: 26777 GF: 283 RF: 1
/Edit
So, it certainly appears to be legitimately a Bitcoin miner, albeit obviously one 'for fun' and not for ROI (like 95% of the devices available, right? right.)
But what is inside it?
What is inside?There are two types of these husks. One where a thin sheet metal is clamped around plastic, and one where the main body is a thick metal. Unfortunately, the iMiner uses the latter. I say unfortunately, but that's really a good thing as the metal acts as not just the main body, but also the heat sink. Still, it meant that getting into it was a lot more difficult. Removing the end cap just showed that the board was practically epoxied down with thermal goop (and possibly some thermal tape), so it certainly wasn't going to be sliding out the way it had slid in during assembly.
Thankfully, it's just aluminum or some other soft metal like it, and a small hacksaw made short work of not being able to get at the innards.
Once the innards were exposed, what became visible as the main components were:
Silicon Labs CP2102 USB-UART Bridge used for communication between the USB host and whatever is inside the stick.
A crystal at 8.380MHz.
Could be 8MHz round, but some goop that won't budge (didn't want to touch it with rubbing alcohol just yet) is covering up the last digits.Alpha & Omega AO Z1038PI buck voltage regulator. Probably because the other chips don't want none of that USB 5V goodness.
Silicon Labs C8051F330 microcontroller.. Probably to offload all the communication tasks.
A big fat chip that gets really hot with the markings: QT20 / CA3755. That 'QT20' sounded familiar, but a friend of mine asked an FPGA guy and he wasn't sure (because it's probably an ASIC, duh). The other code on there - X1349 - is just a batch and date code (49th week of 2013), similar codes are present on the SI components while the AOZ1038 may have been produced in 2014. Nonetheless, given the parameters, this is probably an ASIC. Somebody will probably pop up and remind me why the 'QT20' sounds familiar.
Mining awayAfter I took it apart, did it still work? Absolutely. With a little extra cooling from an otherwise completely useless 'laptop helper fan' (no, seriously, don't buy them for that - the physics say it doesn't work well, and reality agrees), it's mining away even as I write this.
Full imagesFull images? Of course, full images! This post is long enough without a bunch of full size (even if proxied) inline images, so please enjoy them, along with descriptions, over at imgur:
http://imgur.com/a/5wmv2#0Edit March 6th: And since I figured I'd better put it back together again, the tearup:
Album:
http://imgur.com/a/9Jc0X
Scam or no scam?So now the question.. is it a scam or is it not a scam?
If the definition of the scam were "it's just a USB flash drive", then no - it is not a scam. It certainly behaves like a miner, and the pool is accepting shares so as far as I can tell, it's not faking it either.
If the definition of scam is "getting people to part with money they'll never see positive return on", then I'd say that's still a no. Mining hasn't been particularly profitable for most devices, let alone low power ones like these.
While communication could be handled better (all these 'scam!' posts could have been prevented pretty easily by sending one to a trusted member, or at least posting more details), I think the 'scam!' battle cry should be reserved for those products/sales/companies that have been verified as scams, rather than just 'gut feeling'.