Probably the PIC samples the voltage to calibrate the adjustment, so if it's off because of a resistor mod it'll just dial in the DPOT setting to where it wants to be instead of where you want it to be.
Damn, forgot about autotune playing with that. Brainfart... Now to figure how else to get the weak card up to speed |
|
|
|
|
Considering that the s7's have been out of production since shortly after the s9's arrived in spring there is no way that Bitmain would cover it as a warranty issue. Even if bought new from them back then and never used, warranty is only 90-days from you or the original buyer getting it.
My guess is that the bad boards have damage to the 3.3v feed to the PIC and GPIO circuits of the BB. Because all 3 boards get the 3.3v from the same point on the controller If 1 board is shorted it will try to take down the other boards as well.
Let me guess, the boards in each s7's were fed from 2 PSU's? That is when something like this is most likely to happen. When using 2 PSU's always be sure to power the controller last so the 3.3v comes alive after the +12v to all hash boards does.
|
|
|
|
3 of my S9 batch 19 are only running at 11Th and some even 10Th.
Very strange problem. My machines are hosted in labrador so I'm waiting to hear back if its a down board or just a firmware issue.
Really hope its just firmware
My batch 18 has a similar problem. main cause is that board 1 is only running at 486MHz while the best board, #3 or chain-8 runs 580MHz. The middle board is around 546 Mhz. Methinks I'll try the pencil mod on #1 to see if it can get up to speed. A side note to this is that the B18 and probably 19 seems to skip most or all of the auto-tune checks when soft booted. Came up within a few minutes and all boards report the exact same settings. |
|
|
|
|
I vote yes for the 2-3 chip stick as well.
As you said it is the least $$ risk for you, good for the community, and at least would prove to BitFury that you can do it right.
|
|
|
|
What do Bitmain put on the Customs Declaration when they send the miners? Do they list the real value of the equipment or do the "adjust" it to help customers out?
Only reason I ask is that the customs duty would kill me having a new miner (which is needed as I would like to make a profit for once).
They damn well should be listing value as what it was bought for. Listing at a lower value is highly illegal and if/when caught by the receivers country will result in hefty fines and good chance the company would be banned from shipping to that country. |
|
|
|
The disturbing part as that the OP tried to contact them in several ways to figure out how they could make it right and they blew him off. You can get away with that if you are a monopoly, which for now they are, but that may not always be the case. I competitor that can't out-compete bitmain on price but can on service, could still make a go of it. A competitor that can compete on cost and also has customer service will bury them.
I agree on all points 100% Ironically, since BitFury and Canaan Creative (Avalon) has had the ah, "luxury", of time to have worked out any kinks to their 14/16nm chips and IF the resulting miners that are built using them can compete on performance/price with Bitmain -- that would be great and a huge wakeup call to Bitmain. Overall reliability will of course be up to the miner makers (hint hint Sidehack PLEASE!!!) along with of course customer service. So far Bitmain is setting a pretty low bar on both counts so YES I would pay more for a competitors miner for equal or better speed/power performance if: 1) Better designed 2) Better customer service. In any large production runs of equipment there will be failures but -- given proper thermal and power design considerations that should be far far less than what is being reported for the s9's. |
|
|
|
Just FYI for anyone who does this, those things are really tiny and you can't tell that from OP's super large images. You are going to want a very sharp pencil. I basically covered half of the resistor with graphite but it wound up being especially shiny along the edge since I was working without a magnifying glass and I am not exactly a surgeon. But I took it apart and did it 3 times total to work up to a decent result. Pretty pleased with how it is performing now. Finally getting the advertised hashrate.
Ja. The resistor is about the size of a kernel of short-grained rice. When done, also be sure to carefully blow/brush off any graphite dust that might be around the area. |
|
|
|
Wozers!  1st I gotta say that I would think that a large bulk buyer like you *should* have a direct sales rep at Bitmain and through them better contact with support... Ya know, like most large direct sales companies would do for commercial buying customers. I'd certainly get in writing/email from Bitmain specifics on how many boards or other defective parts can be put on 1-ticket and sent in 1 package under that ticket#. Given Bitmain's Service history best to keep their tracking work load to a minimum. Wanna bet that they have folks manually entering info into their systems that communicate between Customs/shippers/and of course the repair company?  Just another spoiled brat customer........ sacasm lolEr? We are talking about what is expected in normal large biz-to-biz relations even in Asia. How us lowly peon general Consumers are treated and communicated with should be a different matter. The fact that large volume accounts are treated no differently rather speaks volumes on Bitmain service overall. Not good and seems like their business processes need a LOT of work in all departments from initial support ticketing to sending out the repaired/replaced parts. I'm just glad that so far I've had only 2 s9 board failures and BitmainWarrenty in Colorado took care of them pronto for around $140 each. |
|
|
|
<snip>
as far as hosting I'll be back in January or February and would love to get together
++++. Really tho my miners get unplugged because they're loud if they could be spun down easily by worker types that would be cool
Getting OT here but, if you are running s7's do apply Sidehacks mod. Power drops from around 1.2kw each to between 930 to 980w, speeds around 3.7ish THs per miner. Some folks have gotten them down to ~800w but stability could be an issue there. In a warm area, fans can be throttled down a fair bit but still not exactly quiet. They certainly don't scream anymore though  I've ran those settings most of summer with room ambient around 85-90F. Since plants don't like temps that high then putting miners somewhere in the area around them should work well. Sample of mod'ed s7's in my farm  |
|
|
|
To OP: Have you taken any resistance measurements of R10 before and after pencil tweak along with resulting output voltage to give us target values for the R10 mod? edit: nvm... just looked back at OP and saw measure resistor R10 resistance (mark C). Mine shows 4.86kOhm. Lowering this resistance increases voltage of whole chain, if you use pencil mod and draw on the top of R10, you will decrease its resistance. Be careful, small adjust will change voltage relatively high. Resistance of 4.5KOhm rise voltage to about 9.6V across chain.
If you want to use soldering tools, you can unsolder R10 and solder here serially connected 3.7kOhm resistor with 2.2kOhm multiturn potentiometer. This will allow you both over/underclock. |
|
|
|
As a Blast form the Past, here is the link to the original How to undervolt an s1 https://bitcointalk.org/index.php?topic=526060.msg5838713#msg5838713 Dropped them from 1.19w/GHs to an amazing 0.89w/GHs Compare that to today's >7-8x lower efficiency. Woof! Thought about doing it to the dozen s1's I had back then but -- each board had 4 regulators to tweak by hand and x2 per miner x12 miners.... just not worth my time to bother with. Now with the s9 and 1 tweak per board and then only as-needed -- thass a while 'nother story. Thx for the info. |
|
|
|
Wozers! 1st I gotta say that I would think that a large bulk buyer like you *should* have a direct sales rep at Bitmain and through them better contact with support... Ya know, like most large direct sales companies would do for commercial buying customers. I'd certainly get in writing/email from Bitmain specifics on how many boards or other defective parts can be put on 1-ticket and sent in 1 package under that ticket#. Given Bitmain's Service history best to keep their tracking work load to a minimum. Wanna bet that they have folks manually entering info into their systems that communicate between Customs/shippers/and of course the repair company? |
|
|
|
As I see all miners are sold out at bitmaintech? https://shop.bitmain.com/main.htm?lang=enAlso I see the increasing price of the Bitcoin that's why I want to buy this miner and have found it on Aliexpress here http://ali.pub/a7b40It costs $2,150.00/piece with Free Shipping to United States via DHL. Estimated Delivery Time:6-13days Only three pieces available. Another seller's Total Price: US $1,748.27 - http://ali.pub/lti1nWith Estimated Delivery Time:8-16days Also five orders and only two pieces available. Why these prices are so different? Maybe I don't understand something? It is called charging whatever price the market will bear. I would also check to see what batch the miners are and if they are 'new' or have been used. |
|
|
|
3-cents eh? Hmm. I have 7 semi-retired s7's using Sidehack under-volt/clock mod that pull 930w each and I'm Detroit area. Care to host them? Calc says @ 3 cents they would be good for $1.65 each after deduct for power cost. For that matter, if they can go to a good home...
are these miners not being utilized? Correct. They've been replaced with s9's and another should join them tomorrow if the latest s9 I ordered comes today. Still have 11 more of them (s7's) online. I've limited myself to ~22kw of free power at work so have been slowly replacing the s7's with 9's but am starting to get unsightly retired miner buildup... Well, the miners aren't so much the problem, it's their damn burros... |
|
|
|
|
Got my stick in the mail today. Cute lil' thing.
|
|
|
|
Hi all,
Call me crazy but I want a KNOB, is that too much to ask?
Control buck converter, core clock, and fan speeds on 6+ miners with a thermal override so as to avaoid hard shutdowns like the s7
Rackmountable, I want to start colocating these things so a clean rack apearance is a must (6U or W.E. works)
- string of miners off a PI would be a must
- wifi is cool
- 5 or 10 pcie sockets, 1200 watts maximum so we can use the little HP guys you sell, maybe 750 so that 2 could hook up to an ant miner psu.
My machines live in liscence grow ops in Detroit and The Bay
Detroit is 3 cents For primary hook ups and we have TONS of space
Trueco2oil.com that's me and we don't steal anything
3-cents eh? Hmm. I have 7 semi-retired s7's using Sidehack under-volt/clock mod that pull 930w each and I'm Detroit area. Care to host them? Calc says @ 3 cents they would be good for $1.65 per-day each after deduct for power cost. For that matter, if they can go to a good home... Trueco2.com - Wonder if Chronic Releaf here carries you? Last few oils I got are from Illuminati Extracts (also CO2 process). |
|
|
|
once bitten twice shy, i will never trust you again.
I'm in this camp. asiabtc talked me into paying a ridiculous price to purchase their soon to be released blade miner, and after many weeks of waiting a bunch of gridseed miners eventually showed up. This was after he began selling the blades for nearly half what I was charged. Never got any sort of compensation or even an apology. I assumed it was communication issues involving a language barrier and eventually gave up. I wonder how many other people have been similarly screwed by this company with no recourse... Anyway, the writing is here on the wall for you. Trust them if you want to be parted with your coins. yeah if I recall 3000 usd to start and drop drop drop drop drop to about 350 usd in under six months. I would also say they killed ltc and it never really recovered. but that could be quite an argument as other factors stand in for that. To op I am still willing to test and review a model plus pay a fair price for it. But Most people here would say I am trustworthy and you are not trustworthy. I am like I will give a guy a second chance. So when you are ready to send me a demo I will run it test it do a long thread on it. And then pay you for it. or ship it back on my dime. your risk is I run away with one unit which since it is shipping to my home in NJ, USA means I would be running away from my home. Which I am not looking to do at the age of 59. Gotta say, ditto to what Phil says. I jumped into the hardware end of BTC because of an A1 miner I ordered from AMT/Josh Zipkin (@ notlisted - same miner for you?) back in Feb 2014. A whole saga in itself there... Point is, I did forensics on the A1 boards as-designed by the Swiss company Bitmine.ch and replicated by IMET for AMT. I said it before in the A1 and Innosilicon threads: Bitmine.ch/AMT were clueless as to even the most basic power network and thermal design rules. As for IMET's role - as a PCB assembly house they should have known better and after a proper design review given a polite"ahem" followed by their concerns about what should have been very obvious design flaws - not just taken the no-review approach of 'here is BOM, replicate the board'. Anywho, if you want a technical write up of it -- possibly with warts and weak points and all -- from an Industrial electronics designers viewpoint <raises hand>. Bear in mind that systems I've designed have proven lifetimes of decades... *if* anything stands out, of course remedies to weak points will be supplied as well |
|
|
|
Has anyone figured out the HEX files for the Antminer S9? I wonder if this same procedure could be done to the S9. I'd assume it can/could be done with pre-batch-16 miners that are using the old firmware. Batch-16 and higher with the auto-tune pretty sure this will not work. |
|
|
|
UMC 14nm FinFET ready for production in 2Q17 October 27, 2016 http://digitimes.com/news/a20161027PD200.htmlUMC will ship 14nm chips to BitFury in small volume in the second quarter of 2017, said the sources, adding that shipments will expand in the following quarter. BitFury going for a 14nm FinFET chip next year, wow! Hmm. Wonder *who* UMC will use for the fab? Per-their website, they do not have either a 16 much less 14nm fab. http://www.umc.com/English/class_300/c.asp They only list in-house @ 28nm and (much) bigger nodes. I highly doubt that Samsung would be interested and that leaves Global Foundries with their fab in upstate NY. More specifically, what was the old IBM fab in Fishkill. GloFo licensed parts of Samsung's process so now offer 14nm to all comers. |
|
|
|
|
Show Posts
