Topic: GekkoScience BM1384 Project Development Discussion - page 95. (Read 146665 times)

I almost didn't recognize you with a SpongeBob pic. Had to double check it was you. I'd figure you more of Walter Mathau guy, being curmudgeonly and all.

Oh, hey there bick. Didn't recognize you with the new haircut. Looks like avatars are working again? As for the coin price, I highly suspect it'll drop several percent by this time tomorrow because it's invoice time and every time we get invoices paid in coin the price bottoms just before I have time to cash out.

Undoubtedly, let us hope the price of BTC cooperates at the time the boards arrive in the buyers hands.

https://www.youtube.com/watch?v=R8d3dr7-fdU

Seems likely, which is exactly why


An S1 board alone would make a good quiet desk miner, and four a sexy S1 upgrade. But the Prisma refit would also be a good blade in a larger unit, so there's also a dual merit to its design. But it's not priority. Speaking of priority, I might be back to Compac design tomorrow and will probably be able to get a finished first draft of the whole board.

I think there are much more S1's around on total

I'm ashamed of myself for only now seeing this thread.

Keep up the good work mate 

Keen to see this go all the way.

Oh. Well the problem with that is TypeZero boards in that size neighborhood would probably stock at 250W or so, and the S2 heatsinks are built for 100W boards.

The controller is a BeagleBoner so all it'd require is a different OS image to make it do different things, and the backplane is already wired up for blade comms and such, so an upgrade kit is really not that intensive to pull off. I was thinking about it for a bit but Bitmain announced initial info for their probably-already-completed design when they had prototype hashboards on display at the conference three weeks ago and some. Once all our other projects are done, if they get as far as manufacturing, I might look into building S2 boards with the next chip  - by the time we get to it someone will probably have released a chip better than the BM1384, whether it's Bitmain themselves, Avalon, a resurrected ASICMiner, who knows.

Well, I wasn't thinking of using their controller/backplane and what not.  Just the heatsinks and PSU and maybe the case, however we can jam the Type0 in there.

Oh. Righto then. Our primary goal in this project is to make a set of boards which can be run independently but also mounted on an S1 chassis. A secondary goal is to make a board which would mount on a Prisma.
The stickminers are basically making a product out of stages we'd be at during prototype anyway, and also since our stick miners would be pretty nifty (which is to say, possibly the best stickminers in existence at present given the efficiency and feature set) and also not terribly expensive. But the primary goal is TypeZero boards.

Quit killing my dreams guys.  


Edit:  I think I just realized... when I said either I meant either as in "one or the other" not either as in "both."

Yeah the dimensions are close but off in ways that you can't really reconcile without totally re-drilling at the least- and even then it probably would not fit as well as you would hope. 

Most of the development so far has been sidehack's show but today I etched a breakout board for prototyping the Compac/Amita buck circuit.  Parts for it should be here some time this week as should the test boards for the BM1384 breakout.

--
novak

A formfactor that would fit either is approximately impossible, or at best it would be heavily mediocre on both platforms. I'm hoping we have the resources to make Prisma refit boards, but then I'm also hoping we have the resources to make a batch of stick miners so who knows.

Not really THAT many, but it's taking up space.  I have more S1s though.  A form factor that would fit either would be awesome.  Recycling the parts that are too heavy to ship efficiently would also add resale value to the old hardware and cutdown on production costs.

Prisma is on our list of things to work on, if we can get the stickminers and S1-upgrade boards going. How many Prisma parts you got?

If only I could purchase boards and put the tons of prisma parts I have in my warehouse to use.

Or even S1 hardware.

Pretty unlikely since they already have them designed and prototyped. It'd be a FIFTH thing for me to work on.

But if they bail on the idea, I might look into it.

Maybe you can beat Bitmain to the punch on the S2 upgrade kits and we can just use our S2 heatsinks and what not for your TypeZero boards!

I'm not sure what you're talking about. Please describe what you're thinking with a bit more detail (or a schematic sketch would be pretty great). Because if you're talking about forward voltage on the diode, then I'm pretty sure you're connecting a forward-biased diode across a high-current rail and expecting it to not immediately explode? If the IfVf curve I just looked at for the ES2A is accurate, there's an expected Vf of 0.8V at If of 2A, which means when the chips are running at top-end voltage on the TypeZero board we'd have  about 9W being cooked off in diodes. That's 5% of the machine's total power dissipation at that point, and at Tjc of 25C. I don't have a chart for 60C but I believe PN junction gain tends to increase with temperature, so just assuming at operating conditions it's 3A instead we now have 14W being burned off, so 8%.
It may be possible to find a part with a much flatter I/V curve, and shifted up the axis for generally lower currents in the operating regions we'll be using, but I'm not sure I'll ever trust a string of forward-biased diodes stuck directly between 30A power rails to not cause more problems than they solve. It might be worth testing, but probably won't go on at least the initial board design because I'll need to test the crap out of it to make sure we're not creating more fire hazards or wasting a lot of power.

What I'd rather do is grab a little SOT23-5 comparator and take a direct measure of the node voltage, and if it gets over a safety threshold I can use that to kick out the regulator. Ideally the chips and their associated capacitors will be fairly close to balanced, but if there's no provision to prevent runaway something could pop.

What about a passive way? Just connect either two parallel schottky or single universal diode. ES2(A,F or J suffix) has Ifsm=50A and can be choosed for right Vmax for each chip. ES2A has 40mA at 0.6V = 0.024W per chip additional loss, while at 1V (still safe) it reaches 5A. It is not 100% protection, but it should perfectly balance each chip's voltage. Only disadvantage is that treshold voltage depends on temperature, so cold and hot Vmax will be different a bit. I think if you look around you can find diode with better V/I and V/Temp characteristic.

Yeah I probably won't be quite as active with sharing fine details on the TypeZero (there's a few aspects of it we're gonna keep in our pockets until the official release) but we'll definitely keep folks posted on the project development as it develops. Without the community there's no reason to do any of this anyway. One of my goals, in general, is to run a business without being a greedy bastard and prove to greedy bastards that it can be done successfully. One primary aspect of not being a greedy bastard is deliberately not ripping people off, and deliberately avoiding scenarios where even accidentally ripping people off is possible or likely, so being simultaneously self-sufficient and fairly transparent is conducive to all of that.