Nifty. I'm actually not fielding customer emails on this round, but the guy in charge of handling shipping stuff sorta pulled an all-nighter on other tasks (he was still here at 8AM) so he might not have gotten to your emails. Tracking numbers are getting entered in the order database, but we haven't automated tracking customer emails yet. That's probably going to be implemented soon.
|
|
|
Wow, that review page really likes animated banner ads.
|
|
|
880W at the wall for an 86% PSU is an output power of about 760W. The same from a 90% is a wall consumption of 840W. That difference, not even considering the potential failure/replacement expense, is pushing $33 a year in extra electric at my rates, which are some of the cheapest in the country.
Unless that was 880W output, which at 86% is 1020W at the wall. 880W output at 90% is about 980W, close enough to the same difference for these calculations, but 880W continuous from a 1000W consumer-grade supply definitely greatly increases your odds of failure.
I wonder how hard it'd be to jack a DPS-800 into it instead? Running off 220V that'd handle the load pretty easily.
|
|
|
I wonder if the internal supplies are 90% efficient? Hopefully the guys making them, especially since they're gonna be charging serious bank, put the best supplies in that they could find.
|
|
|
Kit includes a PSU, interface board and 4 PCIe 6-pin cables. We've shipped units to US, Canada and six other countries; some folks run on 110, some on 220, and between different countries and volt/amp socket ratings there's half a dozen different cords to keep track of so that part's up to you. The PSU's mains socket mates to a standard cable anyone with extra computer parts is probably gonna have lying around, or could be found at a mom-and-pop computer store for a couple bucks if necessary. Worst case, Walmart or the RatShack probably have 'em in stock for a stupid markup.
|
|
|
Now that would be slick - controlling three PSU's together. I'd have to sell off all the boards I've already bought though to convert over to that. False. The three-way board would use the IO headers on existing boards. And all future boards are being designed pin- and signal-compatible, so you could use three DPS-2000BB boards or three DPS-800 boards when those are available, or any past or future revision D750 board. Also, Bawb3 - it's too bad you don't have V0.5 boards, because the V0.4 boards' internal current meter isn't accurate (reads high) or you could get a good measure on how much power the cards are actually eating. For the V0.5 board, 1.8V on the CUR pin should correspond to 60A. I'd say if your V0.4 board's CUR pin reads 1.2V or less, you're at less than 2/3 capacity and it should be safe to add a third card. Should be, no guarantees. I've got one of these boards powering a 7870 Eyefinity 6 on my workbench right now, been running for probably approaching a month with no issues at all. Are you using the EON pin to turn the supply on, or manual toggle switch?
|
|
|
Still got boards and cables available. We're now shipping the V0.5 boards, which have a better supply-side socket (better more reliable fit) and improved current metering. We're working on PCB design for DPS-2000BB boards and should be prototyping soon. We're also working out a package kit for tying three Z750 supplies together with a single control board with better status LEDs and unified control, that'll allow you to load-balance them together into a single ~2200W unit.
|
|
|
One of the benefits of these boards is, if your $110 supply goes out you're out the supply and all the wiring. If the $12 supply goes out you get another $12 supply and don't touch the cabling at all. Plus you can actually get 750W out of them, sustained, which destroys most 750W consumer-grade supplies inside a few weeks, if not days. If you want to have a backup $110 supply to swap out quickly, it costs you $220 up front. If you want a backup for these, it costs an additional $12 up-front and the swap is unsocket-resocket instead of pull-and-rerun-all-cabling.
Better prices on these aren't really gonna happen unless you want to buy them 50 at a time. We're not using junk parts, we're not building them in a Chinese sweatshop. The margins right now aren't that great, enough to stay in business and we're trying to grow a little but it's slow. Manufacturing these boards is two guys' full-time jobs and a couple guys on part-time, and since we don't do preorders and ship from standing inventory, all parts and labor are paid for out-of-pocket before any orders are even placed let alone payment cleared and deposited. It's not easy but we've started the whole thing from zero and are trying to keep expanding with other solid products - not just support resources for bitcoin miners, we're working on some custom desktop hardware and nixie-display stuff one of these days when time and resources allow.
Also working on a DPS-800GBA board. I don't believe those have a current-share interface, at least not one that I've been able to find, so I have to build that circuit separately. DPS-2000BB dev is higher priority right now though.
|
|
|
It's come down from $125 plus shipping.
|
|
|
I'm gonna be checking over the initial PCB design today or tomorrow, hopefully we can get a quick-turn small batch in and tested by the end of next week. It'd be probably four weeks after that before we had any of a full batch to start shipping, so six weeks is probably a minimum wait. The DPS-2000BB will incorporate the same 10-pin IO header and functionality (external-on, current meter, 3V3SB, 5V, SHR etc), and a built-in fan controller with manual/external speed adjustments and headers for two 4-pin fans. We're designing it for a 200A rating, which is more than the supplies are rated for but not more than they can handle. Once the test PCBs come back we'll stress-test them for stupidly high loads and make changes where necessary for the full batch, which will probably take a couple days. Starting in about six weeks we'll have at least one of the minions working full time, and hope to have some better shop space, so once we get to that point stuff should start rolling out quicker than it has been. We've done the whole thing so far on basically two people doing three people's worth of man-hours, and that's for the entirety of design, manufacture, website maintenance, order handling, inventory management and packing. It's been busy and I'm really looking forward to still being busy for a long time - but also having help. Busy means revenue, revenue plus help means expansion, expansion means y'all get better stuff, more stuff and all of it gets done faster.
|
|
|
If it's pin- and signal-compatible with DPS-2000BB, then technically yes. But it takes a lot of time and resources to design, test and manufacture so if I do make a board specifically for that supply you shouldn't expect it for three or four months.
|
|
|
As far as I know, two people are. One is us (eBay seller midwestrefurb), and the other is TracerX who bought a few in the first batch and has been trying to unload them at substantial markup for a little over a month. Course now I'm not seeing tracerx' listing so who knows.
|
|
|
Good to hear it's working properly. I don't like building things that ever fail.
Any orders placed now will get V0.5 boards, whose fans work on a wider variety of supplies. Still busting up stock though. One of my minions I think is quitting on me, the other two are halfway through training and the other full-time guy has been working on PCB design and adminstrative all week so we don't have as many boards on hand as I would like. I just spent about 11 hours manufacturing and will be at it again tomorrow afternoon to get some (large) current orders met Monday and then stock'll be back up on Tuesday.
|
|
|
Has the board's fan speed indeed failed, or is it just accidentally set to external control? I believe the question has been asked before but not yet answered.
Also, no the V0.6 board doesn't exist yet. We just started shipping V0.5 boards, and are making the list of things to change on the next batch. V0.6 boards won't be avaibale for probably at least a month. The V0.5 boards have better connectors, more accurate current measurement, better labeling and a more solidly mounted toggle switch. Any other changes were made to help expedite manufacture.
|
|
|
Snazzy. We'll have DPS-800 boards soon, I just need to find time to prototype external current-share and finish PCB design. DPS-2000 boards took priority.
|
|
|
I really want to make a "your mom" joke out of that response, but this isn't really the place.
|
|
|
The Z750 supplies also really like running off 220. Go ahead. Go for it. I'm gonna, as soon as I have free time.
|
|
|
I'm really looking forward to getting some DPS-2000BB boards done and ready for testing. We should also have some DPS-800GBA boards in the works, which'll be handy for folks that are retiring Blade backplanes in exchange for other hardware. The DPS-2000BB boards will have integrated fan controllers (4-wire) with adjustable speed and two headers.
And within a few days we should have some test three-way-controller boards for hooking supplies in parallel and controlling them all as one. That'll take a lot of headache out of people trying to figure it out for themselves. The 10-pin headers on all our boards will be pin- and signal-compatible so the control board should work with all of the different supplies.
|
|
|
Is that DC power, or power at the wall? If DC power, your numbers disagree with what I've found. If at the wall, what conversion efficiency is your supply operating at? What level of overclock? https://bitcointalk.org/index.php?topic=476645.0 might be more appropriate for discussion on that topic.
|
|
|
|