And if you're running the 2Pac off the Pi's USB, there's current limiting to USBs built into the Pi and even if that's not the problem you're pulling 2A through a weak-sauce USB Micro connection which is not a good idea long-term.

Better to find a good hub that can power the Pi and the stick in parallel.

I have exactly zero idea what that statement means.

Yeah, 6-pin will definitely be useful. That was a must-have feature on the larger pod idea, but since that project got scrapped this one will have to do its best to fill the gap. That's why right now I'm working on a few added features to spiff it up a bit. Probably over the weekend I'll punch out the new board design.

I'm hoping to take some time and work up an improved Bitfury test also, since the whole signals side of things has changed a lot from the original idea and the old test board doesn't work right, possibly because of how a bunch of new stuff is scabbed into it.

Um, yeah. That's been the plan the whole time. The 2Pac and everything were just to build resources for that project. I've had Bitfury 16nm samples since last year.

I would much rather go with "sturdy" than "fancy" which is why I tend to like USB B. I've never liked Micro, that won't even be an option.

Anyone saying toss the barrel jack, y'all aren't thinking clearly. This is an entry-level device designed to be quiet and entertaining. The guy who wants one sitting on a shelf somewhere lottery-ticketing away isn't gonna want to strap an old ATX to it to get 50 watts of 12V when a cheap and silent brick will do.

I'm not building this for the pros. It's for the n00bs. I like that there's a learning curve for operating it, because I really like when people actually learn things. Folks these days really seem to hate thinking, and I despise that attitude. But I'm not interested in forcing newbies to deal with something ugly, jankety or loud just because people who can already hack it like it that way.

In any case, thanks for the input. USB-B is out, 6-pin is in.

I wanted input on this decision because the Terminus isn't planned to be a one-off product. I want to build a Bitfury in the same formfactor, so I'd like to keep things physically compatible wherever possible.

Well heck. I guess I'm the only one still uses them nice sturdy B jacks.

6-pin it is.

I'm working right now on prototyping a simple controller that'll handle fan throttling, overheat shutdown and string lockup recovery. If the string is running unstable, like the buck is at capacity and briefly trips for overcurrent or overheat, or the chips are cold-started at a threshold speed, the string risks a lockup so the micro will detect that condition and force a reset. Problem with a reset is, the chips default to 200MHz and there's a decent chance that's not the speed you want. VH has some recovery code I need to test out which will detect this condition and restore the user-set frequency. Should be pretty handy for long-term stability especially at threshold voltage/speed combinations, for those of you concerned about efficiency or pushing the peak overclock.

So if anyone's wondering, dev work for that is what's holding up the Terminus schedule, and dev work for that is delayed by 2Pac manufacture. But we're getting there. I've already got a test board set up.

(I also have a degree in EE, plus background in construction and residential electric, and the guys at the supply house answered a few questions)

But yeah, high-leg delta looks to be pretty easy to screw up especially as far as balance goes, and then there's the risk of accidentally 208ing your 120V circuits if you grab the wrong leg.

Okay, show of hands. What would people rather see - a 6-pin jack for power (in addition to 2.5mm barrel), or a USB-B for signal (in addition to USB mini)?

Recall the max power on this guy is about 75 watts.

Jstefanop - okay. You said it was different than previous boards which is what got me confused, didn't know if they had changed something. What you describe is present on almost every miner after the S5. S7 (145-chip batches), S9, T9, R4 all have a string regulator like that. Novak and I were thinking about bucked strings back in 2014. To me, what makes it either the best or almost the worst topology is whether or not the user is able to adjust the voltage. If core voltage regulation is in play, the user absolutely should be allowed to adjust it.

Okay well, now I feel like I got a discount.

Dang, that's a tall bill. I'm in a rural co-op; for my 400A upgrade they swapped the transformers at no charge and I did the rest. Material cost for the upgrade, including adding 200A pulls and panels to the hosting shelves, was under $5000. Mostly the 3-phase distro panel and 1300 feet of 250kcmil aluminum. Fortunately the pipe from the pole was way overspec for the 200A pull or retrenching would have doubled that budget.

$75K, that's a lot of dollars. Have fun.

Before I have upgrade boards I'll have a ~70W pod miner with BF16 chips.

Pretty sure those adapters are going the wrong way.

I'm not sure how Corsairs are pinned out at the PSU (I don't have one to reference), but I've made cables from the SATA/PERF ports on an EVGA to 4-pin Molex for risers. Not too difficult, and rated for the power you'd be drawing.

My pick-and-place is pretty mediocre, but I've been working with it for almost two years so I've figured out how to make it behave. I do actually use a kitchen convection oven with one of those controller kits for all my reflowing. Started out with an IR oven that came with the robot; it's handy because it's big but it also realy sucks. Hot and cold spots everywhere, really bad regulation. I was looking for a basic convection reflow oven and found one I kinda liked, but it cost about four times what I thought it was worth. The one I got now cost $250 in parts plus one day of work and the boards come out beautiful, way better than that big IR oven on its best day.

Thinking about building another one. Pretty handy. Probably won't get another robot anytime soon. Got my eye on one I like, US-made, but the base model with some necessary extra feeders costs something like six times what I paid for my car. We're not there yet.

I do have my own pick-and-place. I farm out nothing. If I sell it, it was designed and manufactured in-house. Were this not the case, I wouldn't be talking about using up all my time with manufacturing.

No, that would take years. But there's still a lot of manual processes and a buttload of assembly steps, being as it's a double-sided board with salvage chips, USB jack, heatsink and stock voltage setting with three testing steps.

All y'all wanting to be on the waiting list will have to wait a bit longer. I won't start a waiting list until it's proven ready. Once I announce that, make sure to come back and let me know.

Even though I proved base functionality of the new version last week, I don't have a new PCB down yet. Last week I also got orders for the entire rest of this batch of 2Pacs at the same time my assembly minion decided to bail so I'm a bit backed up. Gonna be making time this week to test out a few new things I want to integrate into the Terminus final design, but I also got about 500 2Pacs to assemble post-haste without my trusty ass*cough*istant to help so it's definitely juggling time.

Please explain this point.

My shop was wired for 200A 3phase wye, which was one of the deciding factors on whether to lease it or not. I set up all my miners on three shelves, each with a sub-panel pulling from two legs so I can pull 120V or 208V with a single- or double-pole breaker. At the junction box where the legs drop to the subpanels I have current meters installed so I can check loads at a glance, handy for knowing where to install new gear to keep things balanced. All the miners are run on 208V because, well, it's better.

The pole has three transformers wired on a common neutral, almost trivial to handle when I upgraded to 400A service. The new feed from the pole and the new distribution panel weren't cheap but the whole thing took only one day to make the change.

Delta would have just made everything way more difficult to wrangle. If you're already doing the work of swapping up to 800A, I'd definitely see how difficult it'd be to rewire the existing setup for wye.

Well, looks like I'll be doing another 2Pac batch. These things sold out pretty fast and I'm still getting requests.