Topic: Say. If somebody built another small miner... - page 3. (Read 2409 times)

Been wanting this for years.

Yeah, I guess I didn't think that through from the customer service side. Although you could probably just put in a very low thermal cutoff, it's not worth it. But it would still be cool as an option, I like home mining even if it's not efficient, but hate the way miners look.

-Dave

Intending the end user to supply the case becomes a significant factor for cooling, and I don't trust end users to do this correctly often enough to base a product around the idea. Leaves a lot of room for the occasional idiot to roast his unit and talk without ceasing about service and replacement and to quote the lady from the meme, "ain't nobody got time fo' dat".

I posted something similar the other day.

Makes you wonder if something like the old BFL monarchs would work. They could be hooked up via USB or through a PCIe connector.
They were more or less a massive, at the time before all the insane sized GPUs, PCIe card.

Back then the few people who got them used USB.

Now I have one in a PC case in the corner of the room acting as a space heater. It does not look like miner, it does not sound like a miner.
It's looks like an older mid tower PC.

Yes, it's not hooked up though the PCie at the moment, just though an RPi sitting in the case. But the concept seems interesting.

Would also save on manufacturing costs, since there would be no case, just a board.

-Dave

I'm genuinely surprised that the network hashrate is even viewable, it's gotta be a given that the public pool fee rates are disproportionately applied, i mean i reckon at least half the visible pools are using their own mining rigs, which as you say are manufactured by the same enterprise.

I gather this means Gekko Science has embargos against various corporations... I was really only suggesting from a manufacturing perspective and in terms of managing risk, say have a card compatible with an atx / mini-atx would create several liability, a SSH access for the product - it would mitigate a concern for Gekko Science in the product's distribution (though it is my intention, I am not readily read up on US law, but from a UK comparison). There are benefits to a "third party chassis".

"bigger commercial contractors" can take a hike, they already own the entire hardware market

I'd definitely be considering a few. Been tinkering with an R606 I purchased from bitshopper, its pretty deadly. I tried it on my old slush account, not making much returns there; however, to be expected. So I've put the unit on at http://pool-btc.ga mining solo BVG, its smashes those blocks like an absolute beast, some really clever gadgetry going on, hardware was always my issue back in the day, the S3's I was running were bought second hand, so the r606 coupled with a lower difficulty coin - I've never seen anything like it.

I've been deploying some mining contracts through http://miningrigrentals.com, and the r606 (solo-http://pool-btc.ga) is outmatching an S9 (pplns) http://persianmine.net, though this may be due to the difficulty settings used by the pools on the coins I am targeting (mainly BVG).

I like the r606 form factor, although for a stack something like those FPGA cards in a ATX / mini-ATX case would look pretty good, if cooling wasn't an issue.

I would suggest that keeping the Rpi away from the assembly would be a good idea at least for offering a scalable product, I think issues experienced by R606 users may be associated with RAM, the headless Rpi 4 is running at least one r606 very well, i don't really know enough about what each component in that configuration is doing, but an FPGA style offering to be paired with a proliant system would probably spark alot of interest from bigger commercial contractors.

From the RnD off this device i've bitten the bullet and building a cryptopool at http://moonpool.io

Keep us posted !

Count me in I would take one for sure. 

I use Duplo to place brackets under/around small mining hardware (and my network devices)
Got lots of it from when the kids were young, it's very sturdy even when put together, and it has a very low % contact area on the top.

I intend on making another R606-sized device. Been considering the idea of a ~300W machine with a built-in Pi controller in a simple case. Still just an idea, we're still very early in dev, but it'd be pretty nifty.

I would be interested in three or four of these once you have them ready!  I have been looking for a replacement to the R606's and you do make great hardware!
I would also probably design and 3d print some kind or bracket or case if they came without one.  I've done some carbon fiber style cases that can take a bit of heat.  If I did design something I would release it for everyone to use.

I'm interested, depending on the price, fan noise and whether it will be available in Europe.


A mini itx case is a good idea. I have some metal plates in this form factor with mounting holes to build an adapter plate. I hope that if the new miner should show up, they will have some holes to attach to an adapter plate with spacers.

I'll be doing some testing with soldering heatsinks, but I'm already using a low-temp soldering process for chips so it'll take a lot of experimenting to be safe/reliable. Most likely I'll stick to my screwed-on heatinks. The biggest thing there will be that surface plating is connected to power ground, so heatsinks from adjacent voltage nodes must remain isolated. But that copper plating should be the next best thing to an exposed die for reducing thermal resistance, assuming it doesn't delaminate. But I don't want the kind of heat density and total temperature that should cause that.

What isn't awesome is how they attached heatsinks on these... they missed the mark there big time.

If you can find a way to avoid soldering the heatsink to the top of the chip, do it. I've been repairing 17 series hashboards and it seems like nearly all issues are related in one way or another to those heatsinks. The chips have a film on top to interface between the chip -> solder -> heatsink. Many times the film fails and delaminates from the chip. Then the chip overheats to the point of melting solder and there ya go.

Also see a lot of problems with excess solder from the application of the heatsinks falling down onto the board. Little balls of solder precariously held in place by remaining flux residue... just waiting to short out pins after the right nudge shakes them free.

I just joined this forum literally / specifically because of THIS thread, and the future potential product(s) being discussed here.

@sidehack I just found out about your products, and it is truly awesome and appreciated that you do this type of work.

I wish I had known about the previous products when they were first released, as those latest ones being from Q4 2018 are slightly dated now... which is exactly why I just joined here: to be among the first to find out when the next product(s) get released (especially those based on the 1397)!

(I also looked into the FutureBit gear and as sidehack said it raises the bar... )

Looking forward to your next generation of ASIC USB and pod products... (hopefully this year?)  ;-)

Cheers.

Well I don't know about the protocol. VH is working on that. My job is the hardware, and I gotta say, all the hardware improvements to the chip are AWESOME. Like when I saw a spec sheet I was really confused about a lot of things because it was all so different, but when I saw 'em on a board it started making sense and now I'm genuinely impressed at how well thought-out these chips are. It'll mean a lot of design changes from the old stuff, but it also means the new designs will be a heck of a lot easier.

At least the footprint is the same, so the various iterations of stenciling and process changes we figured out for BM1387 will still apply.

Of course ... why make it easy when you can arbitrarily just switch everything around....

Are you able to get datasheets for these or do you have to reverse-engineer all the details?

Yes I did mean 1387. Had 1397 on the mind lately for, oh, no particular reason.

From 1387 to 1397 they use the same footprint but from what I can tell, entirely different pinouts, power levels, even the protocol has changed somewhat. It's definitely not a drop-in replacement.

Did you mean 1387?

I was wondering why you wouldn't just use the 1397, seems like the pin functions are pretty close to identical. Do they mess with the commands over the serial interface enough to make switching a hassle?

Might double your BOM cost, but also doubles efficiency....

Yea it has got a decent efficiency for us home miners (although awful compared to industrial gear).

Have you got any sort of time frame for the usb sticks or the new chip pod (assuming its a pod/R606 style)?  I know some would hold you to a date but I’m just wondering if its this year or next lol