The 14V only sources power to local LDOs on the top couple nodes; the rest are fed by Vcore from nodes higher up the chain. Those LDOs provide 1.8V for control and IO and 0.9V for the PLL. Total current requirement per chip is almost nothing, on the order of 10mA. On the 2Pac I use a resistor divider for the 0.9V and it works fine. As long as your boost rail is at least about 1.5V higher than the Vcore of the top node (since ground reference there is about 0.7V lower and a good 1.8V LDO will work at 2.1-2.2V) and it can source ~100mA it should work. You're right, that is easy.

It's USB-B. Legacy? Surely I'm not the only one who still uses B. I prefer it to all newer versions. But there's also a mini on there for folks who'd rather use that. I like giving people options.
12V power from a barrel connector, so any kind of 5V supply will be basically useless. This draws zero power from the USB bus. USB for power on a 30-60W device makes no sense. (Please don't bring up USB-C, that's clearly not what's going on here.)

Yeah, I bought several tubes all at once back in my PC hardware refurb days and kept using it, then just bought more when it ran out. Which I think is about to happen again, we've assembled about a thousand 2Pacs already and it goes faster with two chips.

http://solo.ckpool.org/workers/1BURGERAXHH6Yi6LRybRJK7ybEm5m5HwTr_TERMINUS

Whoops I must have forgot to mention I have a new pod miner in development, the Terminus R808. It still uses BM1384 (8 of 'em, hence the name) that should run 100GH off a 12V3A brick and 140GH from a 5A brick. Haven't fully tested yet but it should run that 100GH without a fan (turn it on its side for improved convection), and it has adjustable core voltage and built-in 80C thermal shutoff. At four inches square, it's the smallest not-stick Bitcoin miner made in what, two years?

Everything works except the main Vcore power is misbehaving so I'm redoing that part from scratch. The new layout is already done and I'll be sending off for prototype PCBs tomorrow. It also already works with VH's current cgminer build.

My test unit (running off a hacked external buck) is hashing away at 250MHz (110GH) as seen in the link above.

If you wait just a little bit though I'll have some news you'll probably like.

NFW - no, I didn't ever consider that. Why I didn't ever consider that is because it becomes no longer a USB stick miner. The 2Pac is already kinda large compared to most stickminers, but once you make it bigger by adding jumpers and a barrel jack and then redo the entire buck to for 5V or 12V input you might as well just add a few more chips and call it a pod miner. I've already made a stick with outrageous overclockability, but there's no point trying to make a stick no longer a stick while still being a stick. If you want to power it with external 5V, just wire it up to external 5V. If you want to power it with external 12V it'll explode because of all the 6.3V caps and also the buck controller is specifically made for USB applications so it only works between 4 and 5.5 volts.

Also PCBs are made from fiberglass, so if there's silicon around the chips something is terribly wrong because only the chip die itself, nestled inside the IC package, is silicon.

The thermal paste is Arctic Alumina, nonconductive ceramic thermal compound.

If it's a USB3 hub, that's a well-documented problem. Look it up.

You mean like having a second frequency setting that you can trigger by flipping a switch or something? Because if you mean "the ability to set the clock lower", it's not so much an introduction as a re-introduction since until recently all their gear had that option. I'd like to see the ability to adjust voltage too, but since circumventing planned obsolescence marginally injures their substantial bottom line it's not terribly likely.

Because it would automatically crank back up to full speed when nobody was around to be annoyed. Earnings would be higher than if you left it on "I don't loathe this" full-time.

What's the network difficulty?

The closer to "cold" you run it, the better. The chip will actually use less current for a given hashrate if it stays cooler, and extreme heat is also hard on the parts. That said, 55C is pretty good for 250MHz.

Ideally all miners could be run at the lowest Vcore that doesn't result in errors, but since nobody these days gives us volt-adjustment ability, and since it seems for some manufacturers even frequency-adjusting ability is going away...

The way I see it, both frequency and voltage adjustment are mandatory features for any good miner. If you want to run a certain hashrate, just calculate what clock rate you need and then adjust the voltage to the minimum stable, that way you know it's the most efficient it can be for that hashrate. Optimizing for a specific power use is a bit trickier because you might have to work with both frequency and voltage together to find the best hashrate for a given wattage, but the tables given (expected power draw for a given frequency and voltage) are a good place to start.

USB power meters are super handy. Every stick gets hooked up to one on my bench and most of the time I can tell not just if there's something wrong with it, but what the problem is, just by looking at the power draw.

Dude. You already sent me an email, and sent a forum message saying you sent an email, and the forum sent me an email telling me you sent me a message saying you sent an email. Remember what happened last time you wouldn't stop bothering me? Just send it back and I'll fix it and please stop bothering me about everything forever before I decide never to do business with you again.

No seems about it, it's definitely smarter to run the lowest stable power for any given frequency. Every recommendation for tuning overclocking says to do exactly that. Extra power means extra cost and extra heat; extra heat means potentially destroying the miner or reducing its functional lifetime.

I've got a hosting customer with some solid used units he'd let go at quite a bit less than $400. I can provide PSUs as well.

Is it possible to restore that WU scaling in the code?

Look at your WU number for a practical reference of how much work is being done. I believe the hashrate is calculated statistically off the rate of share returns, but the WU (work units) is a reflection of the actual data being processed. For 100MHz you should see about 150, so 250MHz is 2.5 times that at around 380. If everything's working as expected, your HW will be zero and WU will be a lot higher.

Hashrate (GH) is calculated as 0.11 times frequency (MHz), so 100MHz gets 11GH and 250 should see 27.5GH. At that frequency, you need a core voltage about 700mV (1.4V device-level). The stock voltage is about 620mV, good for 150MHz or so. The stock voltage setting, your wiper (the point opposite the flat) is in about 7o'clock position; try turning it to 12o'clock. That should get you close. But as stated, you will definitely need a fan for cooling because the sticks will be drawing about 10W by the time you get them leveled out.

Oh yeah, I should have warned you guys that price drop was coming. It always tanks right after I get a big invoice paid in bitcoins but before I'm able to cash out and use it. That pattern has been repeating since the big drop in early 2014 so it's a pretty good predictor.

Mining bitcoins is hard with power over six or eight cents. Over ten it's difficult even with the current best gear. It also doesn't help that the current best gear is priced so ridiculously high it'd take a most of a year to pay off even with free power.

I did it, never received it, got a refund, bought a 10GH AM Blade with that money, and turned it into 2TH within 8 months.

If the Bitfury stick works as expected, it'll take a stout hub but it'd run 2/3 of an S1 by itself. I like that the 2Pac will push as much hash as an AM Cube.