Oh hey, every other time in this thread when someone mentions a clock position for the pot, it's referenced to USB down.

Were you running it hot, without cooling? And then unplugged it, and then the next time you plugged it back in it's detected by Windows (meaning the USB chip is active) but not cgminer (meaning the chips aren't communicating)?

Then yeah, that's what I'm gonna guess.

If you were running it on stock settings and reasonable temperature and it's randomly not detected, consider increasing the voltage a bit and trying again. Could be a balance issue.

I'm still guessing it's that thing I said about heat damage.

Oh also since this doesn't draw any power (seriously- milliwatts) from USB you won't need a powered hub if you want to hook up multiples. Get one of them $8 10-port guys and it'll probably still work.

But if crypto's only 5% of their business, why would that cause record high stocks and dividends? I can see Bitmain having record earnings since the markup on every S9 is roughly 300%, and 5% isn't really a drop in the bucket but it's awful darn close. Certainly not a bank-buster.

Oh also, TSMC.

S9 has 189 chips per miner, so that's about 500 miners.

Funny you should mention the hubs. I picked that one back up over the weekend with a bit of a concept change that should make everything better. I intend to start prototyping sub-circuits as soon as I get more time to run out a board layout.

Because for some reason you like people and don't like big companies?

If it was running without cooling and you unplugged it while very hot, that can cause damage to the ASICs.

Right, but I designed and built them so I have to assume there's a misunderstanding somewhere.

(or he could put it on eBay and get about $1100 from it; even after fees he'd still come out about $100 farther ahead)

If we're talking about the flat, 2 o'clock is just a bit above absolute minimum voltage.

2 o'clock position is stock setting. Looks like it's already turned up a bit.

Bumpitty bump, got about 16 sticks now and another hundred to test. Last chance for a month or so.

Can't imagine where all those CPU coolers came from. Or all those Technobit boards. /sarcasm

Yeah it's pretty common and easy to miss. I still make that error myself on occasion.

Is there a decent web frontend for configuring cgminer available on that OS? I'm a firm believer in Debian's reliability and, as handy as Minera is for configuring, honestly... I really really hate web scripting and that package has dynamic content up the butt.

I'm gonna guess...

What's the OS? Maybe newer versions of Minera suck.

When testing, I try to shoot for fewer than 2 HW per hour at 100MHz and stock voltage. That's just my own standard for what I'm willing to ship. 4 an hour at 150MHz isn't bad. I forget the formula offhand for figuring percent errors, but as long as your hashrate is about what's expected you probably won't have any trouble.

The green light is tied directly to 5V. Basically just tells you the thing's getting power. The white is linked into the comm lines and flashes every time the chips respond with a nonce. So flashing means it's hashing away. Solid white generally indicates one of your chips is hosed up.

Most likely. If the sticks pass my bench testing and move on to assembly and product testing, about the only way they fail later tests in ways unrelated to ASICs is if parts have been broken off the board. Getting to seconds testing is due to mediocre chips, failing seconds testing is due to very mediocre chips.