Bobsag has offered to put up some BTC, but I'll have to decide how to split that up between the cables he wants and an order for these connectors. And then decide how many connectors we want to order.

I have a few overclocked Blades for sale again, looking to get 0.3BTC apiece plus a bit for shipping. Total of five available, all running about 14.5-14.6GH 99+% in my current setup.

Typically what happens is a VRM resistor isn't installed properly, and overvolts a VRM. This causes the ASICs to fail, which since they also draw from the 3.3V line, shorts the 3.3V to ground (or, well, 0.7 to 0.8V) and sometimes explodes the 3.3V buck regulator (IC102). Probably if you can identify which VRM isn't working properly (one somewhat reliable way is to see which one's outputs are shorted, or fire it up and measure for improper output voltages) and pull those ASICs. That might get the rest of the board working. I have a few here that I got from others (some overclocked, some not) where VRMs had failed and roasted a bank of ASICs, and some of them are working again on 7 banks.

That's how I did my first test, wired straight to the supply terminals with pin header plugs and hot-snotted the whole thing solid.

The Z750P connector is not the same as on the 2000. I think, actually, that the Z750P is the only effin' supply that uses that connector; there's a bunch with the same pinout but ever-so-slightly different spacing on the power blades which is why it's hard to find mating connectors.

Hope you don't mind if I ask you any questions while designing the board for this supply.

DPS800 is the next board we'll have available. Just gotta get some time to make one - in addition to Blade/Cube repair service and Z750P designs, I've got a "real job" and a few other side projects to juggle.


Also looking at cables; Cheshyr got some good numbers, need to find out a bit more information but we'll likely have cables available for purchase alongside the boards for around $4-$5 depending on wire diameter and length; this'll have a PCIe 6-pin and spade connectors already assembled.

Had about a week's holdup with initial funding for a first batch, but we should be ordering the parts and PCBs before end of business today. Depending what happens, we should have orders opened in about ten days for the Z750P boards, and a solid design for DPS800 boards and for a several-supply control panel not too much later.

Surprised it lasted that long.

Got any kind of measure for power consumption off that? A 25GH Blade would be something like 400W, run at 28MHz oscillator and require probably 2V core (which should explode the ASIC). Pretty sure a 1203 resistor would lower the core voltage.

Thanks for linking my website, by the way. I'm still selling the kits to overclock these Blades to 13/14.4/14.9GH stable (instead of fake) if anyone's interested.

The Z750P connectors aren't, at least not in stock. I have a jerry rig solution but for the next round I'll have to order from the factory.

In the very near future I'll have boards for the DPS800 and probably that 2000W fanless supply y'all seem to like; that one will integrate a PWM fan speed controller with internal/external drive options and at least one fan header.

Manufacture on the Z750P boards has been delayed by finances, but we should start rolling them off about the end of next week or early the week after. Hopefully the boards for other supplies won't be too far behind.

Which is why the board gives you fan speed control.

Pretty sure they're compatible, as far as I know the differences are internal.

Which is what I currently have a bid for, and should have in-hand by the time boards are ready to ship if everything works out.

Also, since these are 750W supplies instead of 2000W supplies, running a whole lot of stuff off a single supply isn't really that much of an issue. Two AntMiner S1, two AM Cubes, two or three decent GPUs per supply. I've got a DPS800 to start prototyping for, and will look into those 2000W supplies as well, so it'll need to be a consideration in the future for sure.

The distro block is built into these boards. The simplest solution is to run a cable from there straight to the device; that saves connections and adapters and stuff. If you need to power a few hundred devices, wrangling exactly one wire per device is a whole lot easier and that's currently what we're looking at.

I'll probably focus on making 16AWG available alongside the boards once everything's ironed out.

That's what I'm figuring on doing. It'll take a goodly while to get them from the factory, but probably worth it as long as FCI product engineers didn't tell me wrong about proper mating connectors, and the mechanical diagram data isn't wrong. What I don't get is this is apparently the only power supply with that pinout type, that uses 0.2" spacing instead of 0.25" but why are there so many of them running around?

I'd be fine with 16AWG; all my stuff is currently running on 12 inches 18AWG which is too dern small but it's what I had. 14AWG was requested, so that's what I'll deliver to the requestor. He's been smoking 18AWG wires.

I had a cables contact already, getting bids from him for production which is, unfortunately, overseas. I'd prefer domestic manufacturers, and Cheshyr is following up on some leads for me in that neighborhood. The initial batch's primary customer has requested 24" 14AWG, and I figure having spades crimped on the other end will be really handy.

I would think those voltages would be good enough. The next consideration is, perhaps the high-clock oscillator output isn't getting to the chips. The problem with that though, is I believe the two oscillators are wired in parallel. This works because each has an enable line, which when disabled gives a high impedance on the output so it's more-or-less isolated from the signals generated by the other when enabled. They share the logic buffer and signal paths to the chips. If it were not a clock-dependent thing, I'd consider that either the address decoder was missing a bit, or the oscillator's logic buffer was missing an output or two.

As far as I know, the only things that change when switching to high clock are the VRM output and the select lines to the oscillators. Weird.


I do have on hand a few Cube cards in need of repair, the symptom of which is half the chips are X'd on high clock. I'll raise the priority for diagnosing and repairing them, then report back my results.

Yes, Cubes should run with any number of cards in any arbitrary arrangement.

Also, thank you.

Well I didn't actually get them up to 43. I'd expect to see about 400W though, based on my Blade numbers.