Linear regulators for high power are gonna cause you some problems. If you're sinking 12V down to 5V, your TO220 are gonna be putting out 40% more heat than the sticks will.

Our original idea for the Gekko hub was to put in per-port power monitoring. I still think it's a cool idea, so nice work there.

As for hub stock, we got hit by parts price increases pretty severely so had to take the thing back to almost zero to redesign for parts we could source affordably. Couple that with a fairly time-inefficient case manufacture process and limited time available for dealing with issues around CompacF manufacture and you see the problem. Good news is we're solving problems and improving our tooling and shipping hubs. And also shipping sticks faster now too. 2022 is shaping up to be a pretty good year for GekkoScience products.

Yeah I never use a config file, just a command line or shell script that's easy to edit because it's basically one command line.

Oh yeah that old guy. Error as old as cgminer running USB devices. From the Compac/2Pac support thread:


You'll want to cd into the correct cgminer folder, but that is probably obvious. Either that, or run cgminer as root.

There's a link in the first post, to a post hanging out on Page 3, with all the instructions you could want.

Those heatsinks look like a great way to destroy a hashboard or power supply.

Bitmain's individual chipsinks are held to the chips using a low-temp-process solder, probably something bismuth that melts around 140C. That's why so many people have problems with them falling off. In order to support solder, the top surface of the ASIC is copper plating pretty much right over top of the silicon die, which gives absolutely phenomenal heat transfer characteristics. However, the copper plating is internally connnected to ground. One of the failure modes of a hashboard is a heatsink gets loosened then drops to short between the chip and a heatsink from an adjacent row, which operates at a different voltage potential because each "domain", that is to say each row of chips, is put in series.

So unless you're 100% certain that the thermal compound you're using is non-conductive and you're 100% certain that your coverage won't thin out such that you're 100% certain there's no contact between heatsink metal and chip plating metal - and you're 100% certain the screws that are mating into a nut soldered to local ground planes also aren't contacting the heatsink metal - I honestly don't see how this heatsink solution won't be a disaster.

There should be a cgminer command to list all devices, and I believe it tells you the bus/port numbers. On my testers the busses enumerate as 0/1 or 0/7 depending on machine, so I run two separate instances of cgminer, one pointed at bus 0, one at bus 1/7, for easier note-taking about which devices are on which hubs during burn-in. We run up to 7 sticks at 400MHz per bus (note the bus usage is equivalent to 5 sticks at 550MHz); any more and we had stability issues like you're seeing. We test on debian linux, which I guarantee will make a more efficient use of the bus than your Windows.

With that many ports, it's likely you have two distinct USB controllers in the machine. If you connect one hub to each controller, they won't be fighting each other for bandwidth. When we test at the factory, we got a bunch of machines with 6 ports in the back, 4 on one controller and 2 on another (shared with the 2 up front) so we connect one hub to each bus and let 'em eat their fill.

Also, performance on Windows is gonna suck more than *nix anyway just because of how Windows handles traffic.

Probably the chip crapped out. Contact the seller for support.

Believe me, when I have something to report it'll be here first.

Adjusting the voltage allows for higher operating speeds, because faster switching speeds require more power to make the switches switch faster without choking. Tuning for the lowest voltage at which the speed is stable is how you optimize efficiency. At super high speeds, even with good cooling, you're not running into hub limitations, you're running into limitations of the power delivery parts. The main inductor is rated for 17 amps and beyond that, you start to saturate the magnetics and regulation suffers. Basically, even without cooling, it wouldn't be safe to operate.

In general, up to that point, better cooling means it'll safely run higher speeds. Might need to pipe in power directly if the USB jack has problems.

You'd probably want something to circulate the oil just in case, but yeah that's mostly what you'd need.

Twenty cent heatsink? I'll have you know, sir, that the heatsinks cost thirty-eight cents!

Also, pod miners aren't complete units with computers included. It's just a multi-chip USB miner with an external power supply. You're right the cooling and VRM are also included, but power supply, USB cable, PC/controller are all external.

Also, the 5V pin on the USB jack can be lifted without breaking anything. I mean mods like you're suggesting probably void the warranty but you can do it without destroying the jack.

Yeah we've intentionally kept the same dimensions for the past three generations of sticks so people can re-use the external hardwares they've conjured up. We'll keep the same formfactor for as long as it's feasible.

It's not easily swap-able for chip upgrades. This assumes an upgraded chip is pin- and signal-compatible, which is almost universally incorrect.

if you've got the skill to modify the USB stick for external power without wrecking anything, it can be done with the stick as-is. This idea has been presented I don't know how many times in the seven years I've been building high-performance stick miners and it gets dismissed every time. It will continue to be dismissed every time. If you want external power, you no longer want a stick miner.

So, the pod miners.


Also regarding soliciting funds from Sledge, that was tongue-in-cheek. We're doing okay, and I couldn't ask for better advertisement than his story anyway.

The Compac regulator we designed for the first stick in 2015 was built to run 0.8 volts 17 amps; the 2Pac used it for 1.6 volts and the NewPac used it for around 0.86V and this stick uses it for about 1.5V

We designed the hub to run 3A per port because we knew our sticks were made to pull 15 watts. The Compac wasn't really capable of it without modification but as a single-chip stick, 10 watts or so was a practical max anyway. But every stick after that could do the juice. We've used the exact same regulator circuit, changing only the parts needed to set the voltage range, from the beginning.

We decided to stock up on hubs when we launched these sticks because we knew people would want them, but we got bit by the chip shortage and some other things that slowed down production. I'm working on a redesign of the hub that'll provide the same quality of power but that's easier to manufacture. We're also investing in some new tooling to help speed things up. Stock of things in 2022 should be more regular than what we've seen this past year.

Also for the record, the Compac F isn't the first stickminer to pull 15 watts. The 2Pac and NewPac could both work on that level as well. The GekkoScience Base Hub was designed during the 2Pac days so it's fun to see that it's still the best for this application.

Also, definitely gotta lay congratulations on Sledge for his block! If you happen to want to, you know, unload some of your winnings, I may know a guy.

The T in MOSFET stands for Transistor

That's an effective summary of why I care so little about USB-C devices. Only two ports at a time, and no PD+data, doesn't quite mean it's useless but certainly doesn't help anything.

Now people do like USB-C cables enough that future pods are likely to have USB-C connectors, but it'll still be a basic USB2 with no power draw. Just using the cable people like.