Tweaking the S7 boosting performance with small hardware tweaks

Alright, let's get into it. This discussion is really about the 135-chip S7 model. The older 154-chip versions, with their taller string and fixed voltage, have basically hit their peak performance unless you add a regulator to your power supply. But the 135-chip has a bucked string, which means it has that regulator built right onto the board. The stock voltage for an S7 is around 10.3V, or roughly 690mV per node. This seems to be spot on for 700MHz according to the BM1385 datasheet. If you’ve got a slower S7, the voltage might be lower. I just got my hands on an S7LN, designed for 600MHz across two blades; I found the setpoint to be 10.16V or 677mV per node. The voltage setting here is fixed. There’s no dial or anything to tweak this. BUT... The voltage from the regulator is actually controlled by a digital pot more specifically, a digital rheostat (MCP4017). It’s like a variable resistor whose resistance changes based on a number sent through a serial line. The default setting is right in the middle, which is why some 'dead' S7s might start at 9.3V; the value doesn’t get updated, so it stays at that middle setting. Ideally, a small microcontroller (PIC12F1572) updates the value from its memory as soon as it powers on. That’s probably why Bitmain suggests powering the controller last (or at least one reason), since the DPOT and PIC draw power from the 3.3V wire on the 18-cable to the controller. The DPOT’s 'off' setting is at max, which is why it starts under 9V; once the circuit...

Interesting project for sure. Looking forward to the updates. Cheers, - zed

Interesting... On the batch ones, is the string voltage set by what the PSU provides (at the PCIe plugs) or a fixed board-level buck? I ask because I have a b1 at work that has oc'd beautifully - set to 700MHz both ck and M's Miner Monitor pegs its average hashrate @ 5.43THs with peaks of 6THs and higher   I have a b5 that does almost as well. Wonder what they would do with a tad more Vcore done by raising the PSU voltage a tad?

If it's a 154-chip, it's straight off the PSU with no regulation. The later 135-chip batches are regulated. As far as I know, the setpoint is not affected by the PSU voltage.

So if I dial up the output from a IBM 2kw supply another 1/10v or more, things might get interesting... Sidenote: After the bonded T1 service at work that supplies the VOIP phones and i'net being down since ~ 7:30 this morning, as of 1 hour ago it looks like it's back up. I left 2 s5's powered up there to watch on ck pool. That outage dropped me 86.4THs for (over) 14hrs.   More of a bugger is that now I just don't feeling like going 20 miles to bring it all back up. That can wait til the morning I...

And by "154-chip" obviously I mean "162-chip", which is to say 54 chips per board. Whoops.

So since each blade is about 200 watts, can these operate at close to the S3 noise levels? Your temps are 47C can you run these S7 safely at like 60C with a slower fan speed?

I bet you could. The 47C it's currently running, it's only got the pusher fan installed. I bet with both, even with both blades powered up, it'd be a lot cooler. I'm not going to hardware-hack the other blade, gonna use that as a guinea pig for software changes, so I won't be able to report on that for a while. But there's probably talk about it in the S7LN thread.

Good to see someone hacking this... I have some PIC hacking experience as was part of a team that reverse engineered and reprogrammed the PIC in the Scalextric Digital Powerbase. Have Pickit 2 and Logic Analyser etc, however no 135 Chip S7. I will look into solving that if the couple of S5 I have on ebay sell. Rich

If I'd had the budget I'd have bought one ages ago just for this reason, but it wasn't until I sold the rest of my Compac batch and Bitmain had one for $300 that I was able to bite. Think you can write a bit-bang I2C capable of sending a single two-byte instruction, waiting for a minute and sending another? The pick-and-place is keeping me pretty busy today but in spare moments I'm looking up details on I2C timings and such, next step is learning the PIC instruction set and breadboarding a bit.

Yes probably but like you have a lot of other stuff on my plate at the moment. I had however when I first saw the 135 chip S7 bought some PIC12F1572 and some MCP4017 so time permitting can knock something up without an S7. Will PM you with and email as might be interesting to look at the hex code you have downloaded. Rich

Alright. I'll keep working on my end and see what I come up with. The only S7 blade I had access to before yesterday was a roasted 54-chip from a Batch 1 miner, or I'd have been working on this around February.

wow !!! thx for the experiment & find, keeping an eye on this for sure.

a 2 board  7 lite down clocked to 2000gh doing .18 watts a gh would be a 360 watt machine. it would be a decent item

If it'll hit 0.18 - that's a stretch. You'll probably lose 5% in the buck, then add 30W for fans and controller, and 10% to the PSU, gives you 450W wall. Still almost twice the hashrate of an S5 for 3/4 the power draw, and probably a lot quieter given a sealed chassis and push-pull fans. It looks like I'll be working on code this evening and probably tomorrow, so who knows but I might not actually have it working by burger time.

Cant wait to see what you come up with sidehack.

so 450 to 2000gh  is .225 watts` per gh with a titanium maybe 430 watts  or .215 watts a gh. good numbers don't know how many I would get , but at least one.

If everything works how I think it should, anyone with a PIC programmer and a screwdriver could rework any S7 to run undervolted. At the same numbers as estimated above, a full 135-chip S7 would run 3TH at under 700W. I don't actually know how good it'll be until I do some more testing, and once I have functional writeable boards I can start measuring actual power draws. I'm really gonna try and get it working tomorrow, pretty excited anyways. I have a working flowchart drawn up for the I2C comms, but I'll need to learn some of the instruction set and the specifics of the chip to be able to get the timings right.

Me likey. Me likey a LOT! Considering with the 3 s7's I had planned taking off line at home (being replaced with ma s9 whenever it gets here) and the 2x batch-18 um, spares, I have. This is looking very good: After undervolting the farm to stably run ~ 4THs per miner pulling under 700w (edit, 1,100w) each would let me put most if not all the s'7s I have back online with still less than my current power budget. 'Taint ran the #'s but  methinks more THs ta boot Tossing a coin into the tip jar for this!

Actually, 4TH would be more like 1100W.