They are continuing to make BM1387B ASICs, I know that. I haven't seen the new S9 models but they're probably pretty much identical to the original except with more reliable binning or small tweaks to improve efficiency of the chips. 16nm has come a long way in the last three years.

Chips are available to registered repair guys through Bitmain's internal sales channels. They've also just about doubled in price in the last few weeks.

Are you using one of the 2019 builds, or an older version that only recognizes NewPacs? Remember this device didn't exist before about April.

My last job was electronics refurbishing, a skill I already had and brought to the company. The PCs used for testers were either pulled off a pallet at my scrap guy's or fetched in a bulk lot for like $12 apiece.

The biggest miner I have any interest in building would draw around 500W. A decent ATX PSU will always be enough for GekkoScience gear.

Your "ancient" machine is newer than anything in my shop, including the computers used to design, program and test these miners. The only processor of comparable age, I think, is in the hosting router which handles 22 firewalled VLANs with independent VPNs through a rebuilt gigabit managed switch.

As a third-generation Broke Person it's always hilarious to me what people consider obsolete. Your machine should be sufficient, with appropriate hubs, to run a dozen or two R606 no problem.

Handy that the PSU is enough for both.

I test up to six R606 at 550MHz (AB) to a hub on an old Dell 755 running Debian 7 with no speed bottlenecks.

R606 are out of stock everywhere, even for the manufacturer (me). We got slammed with NewPac orders from almost literally every reseller in the world in a 48-hour period so basically everything is sidelined until we're caught up on sticks.

I may have revived one of our dead robots which I can use to make R606 without affecting the NewPac assembly line, but I'll have to juggle that inbetween work trips to help build the new shop (which ate the last three days). So, we'll have more but they'll be slow coming because the main line is all NewPacs for at least the next couple weeks.

If I'm reading things right, the S15 uses a 72-chip single string per board. That's why the PSU runs around 19V but would only need to source ~25A per phase (assuming it uses separate rails per board, which is best for per-board tuning but also unlikely given Bitmain's cheap-out design philosophy).

You can still do a board-level regulator with narrow strings and get decent efficiency (95-97% conversion) depending on the Vin/Vout ratio and most importantly the current draw.
S9 with a 3-wide string was asking close to 50A from a single phase, and resistive losses in FETs and copper are proportional to current squared, so by shifting to a two-phase design one could halve those losses. But Bitmain likes to build cheap and sell high and cares less about longevity so that never happened.

Dedicated PSU can give you a 5-10% efficiency boost depending on what implementation you're replacing, but it also comes at a cost - you're now locked in to their hardware and their replacement mechanism.

Really? I'm genuinely impressed the three-port-shared regulator is able to handle that kind of juice.

Controller is an old Dell 755 running Debian 7. Hub is GekkoScience, and it and all six pods were powered off the same DPS1200. Everything stock, no additional fans, volt setting 7.

If you're only wanting to run at 100MHz, you can turn the voltage down (tiny knob in the bottom corner, rotate anticlockwise maybe 20 degrees) which will help it run cooler and draw less USB power.

Is it just downclocked or also undervolted? Because you can get an S7 at 3.7TH to around 800W with an undervolt hack, and then replace the firmware with S7-F1 version you can run a single fan. Not dead silent but pretty quiet.

Obviously an S9 board is better, as you can get that hashrate from less than half the power draw. But it's still an option.

Also Phil's right, the housings aren't the same length. S9 is somewhere 1-2 inches longer.

With the troubleshooting steps in the first post.

And a fan? If you're running sticks that high you need a fan.

Salvage S9 chips? New generation?

The NewPacs and R606 are built with BM1387 same as used in the S9, but not raggedy old salvage chips. Pfft.

(we tried; harvesting sucks)

ASICBoost is probably the biggest factor in that. It's basically impossible to push past 800GH on a single pod without it due to USB packet limitations.

NERRRRRRD

This really makes me want to see if I can fold up a simple single-board case with board-facing fans. Not sure if I'll have the energy to do it this weekend.

It's not a chip resistor, it's a 100uF ceramic capacitor. Anything over 1V rated will be fine (I believe the part on there is 6.3V rated).