markm
Legendary
Offline
Activity: 2954
Merit: 1090
|
|
September 20, 2013, 12:34:18 PM |
|
But these guys are professionals, so no way they'd have a nice reflow station, right? -Markm-
|
|
|
|
Puppet
Legendary
Offline
Activity: 980
Merit: 1040
|
|
September 20, 2013, 12:41:31 PM |
|
You do if you're in a rush to make a headline-breaking deadline, and if it's a significant number of rejects you hire someone to extract what they can off the board to put in the scrap bin.
Makes no sense. You have to test these chips at some point anyway, its generally easier and cheaper to do that before they are mounted on a PCB So regardless of the fact that it saves you the PCB and assembly cost, you would want to do this. The only sensible explanation I can come up with is that packaging and testing is handled by orsoc and KnC just pays for tested functional chips and wont be doing any other testing on them.
|
|
|
|
markm
Legendary
Offline
Activity: 2954
Merit: 1090
|
|
September 20, 2013, 12:48:35 PM |
|
Are the solder spots on the bottom of chips so non-standard that there are not some kind of socket things you can solder on the board instead of a chip and some kind of conductive gel or something so one can just press a chip onto the thing, hash, remove it, press the next one in etc? Doctors have some kind of gel or something they use for electrodes and paddles and such can't electronic engineers form a connection without actually soldering the thing?
Even if it doesn't always form a perfect connection couldn't you just skip chips that don't and keep moving on, worrying about the "poor connection or bad chip" ones later?
Or even use a conductive solder-paste so you place and press, hash, if it hashes convey it into the soldering oven?
-MarkM-
|
|
|
|
vesperwillow
|
|
September 20, 2013, 12:54:53 PM |
|
You do if you're in a rush to make a headline-breaking deadline, and if it's a significant number of rejects you hire someone to extract what they can off the board to put in the scrap bin.
Makes no sense. You have to test these chips at some point anyway, its generally easier and cheaper to do that before they are mounted on a PCB So regardless of the fact that it saves you the PCB and assembly cost, you would want to do this. The only sensible explanation I can come up with is that packaging and testing is handled by orsoc and KnC just pays for tested functional chips and wont be doing any other testing on them. I guess we'll see. I remember them saying they wouldn't be testing the chips themselves--whether that means they as in KNC or themselves as in the chip by itself, I have no idea. They did say they would do tests of the rigs prior to shipment. Again, what may not make sense to us might just be their plan. With a deadline to meet, if they are handling the rejects themselves, they would likely toss the stuff aside to deal with after the deadline. If ORS is handling the testing/flowing then it makes total sense that KNC is simply handling the rig testing/shipment.
|
|
|
|
timmmers
|
|
September 20, 2013, 01:02:41 PM |
|
Sure they can afford to scrap bad boards. Damn, it's the best thing for them to do. Initially. Fuck the cost, think of the cost of their first deliveries not being on time against that, and cancelled orders. Get the initial couple of dozen or so units out of the door ASAP, gain the kudos for that, then relax a bit and test chips once they have that milestone achieved. IF they get 50 perfectly working rigs in customers hands (literally since they are collected), get some good feedback... the lion's share of orders are october delivery so they'll have more time. Then they can test and not waste components.
|
|
|
|
vesperwillow
|
|
September 20, 2013, 01:09:06 PM |
|
Sure they can afford to scrap bad boards. Damn, it's the best thing for them to do. Initially. Fuck the cost, think of the cost of their first deliveries not being on time against that, and cancelled orders. Get the initial couple of dozen or so units out of the door ASAP, gain the kudos for that, then relax a bit and test chips once they have that milestone achieved. IF they get 50 perfectly working rigs in customers hands (literally since they are collected), get some good feedback... the lion's share of orders are october delivery so they'll have more time. Then they can test and not waste components.
That's part of my assessment, especially because the deadline is a bit of a headline-making one. If they meet this deadline and/or simply make folks really happy, you can expect SHTF and folks will be lining up moreso than before, placing orders. Their coffers will be full for awhile. I really want to see KNC and Cointerra delivering and going head to head.
|
|
|
|
Puppet
Legendary
Offline
Activity: 980
Merit: 1040
|
|
September 20, 2013, 01:18:02 PM |
|
Sure they can afford to scrap bad boards. Damn, it's the best thing for them to do. Initially. Fuck the cost, think of the cost of their first deliveries not being on time against that, and cancelled orders. Get the initial couple of dozen or so units out of the door ASAP, gain the kudos for that, then relax a bit and test chips once they have that milestone achieved. IF they get 50 perfectly working rigs in customers hands (literally since they are collected), get some good feedback... the lion's share of orders are october delivery so they'll have more time. Then they can test and not waste components.
Just curious, do you think they should test assembled miners before shipping? Like at least turning them on and running them for a few minutes? Or ship them asap and let customers find out if they work? The latter also saves them time and money. This isnt only about wasting components, automated testing of those chips will if anything, speed up production of functional miners. It takes so much longer to assemble (and test!) a complete miner based on a bad chip than it does to do chip level testing.
|
|
|
|
itod
Legendary
Offline
Activity: 1974
Merit: 1077
^ Will code for Bitcoins
|
|
September 20, 2013, 01:21:58 PM |
|
Sure they can afford to scrap bad boards. Damn, it's the best thing for them to do. Initially. Fuck the cost, think of the cost of their first deliveries not being on time against that, and cancelled orders. Get the initial couple of dozen or so units out of the door ASAP, gain the kudos for that, then relax a bit and test chips once they have that milestone achieved. IF they get 50 perfectly working rigs in customers hands (literally since they are collected), get some good feedback... the lion's share of orders are october delivery so they'll have more time. Then they can test and not waste components. Exactly, it would be unreasonable for them do this any other way. People seem to forget how seriously overpriced these machines are, their production cost is a fraction of a retail price. The time is the only thing they can't afford to waste. When the time comes to reduce price 50% each month to adjust to expected ROI they will start to pay attention to optimized manufacturing.
|
|
|
|
timmmers
|
|
September 20, 2013, 01:34:16 PM |
|
Sure they can afford to scrap bad boards. Damn, it's the best thing for them to do. Initially. Fuck the cost, think of the cost of their first deliveries not being on time against that, and cancelled orders. Get the initial couple of dozen or so units out of the door ASAP, gain the kudos for that, then relax a bit and test chips once they have that milestone achieved. IF they get 50 perfectly working rigs in customers hands (literally since they are collected), get some good feedback... the lion's share of orders are october delivery so they'll have more time. Then they can test and not waste components.
Just curious, do you think they should test assembled miners before shipping? Like at least turning them on and running them for a few minutes? Or ship them asap and let customers find out if they work? The latter also saves them time and money. This isnt only about wasting components, automated testing of those chips will if anything, speed up production of functional miners. It takes so much longer to assemble (and test!) a complete miner based on a bad chip than it does to do chip level testing. This has been covered. They are burning in the rigs. They said so. Not for long though. I'll spell it out for you a lot simpler since you seem unable or unwilling to understand. 1) First few orders only...assemble as fast as possible, test them to check they work as spec ..get them to customers as is Wylie Coyote is chasing them. Any of that 1st batch that don't measure up...put to one side/scrap. These rigs are modular, and they are going to supply DIY upgrades when all orders are done. So any problems can probably be fixed that way for speed. 2) They fuck about, test everything...end up with the same number of working miners, less scrap...and their reputation alongside the scrap in the bin. Late. Missed the deadline. That's a big thing. To scrap the bad units from a batch of 100 rigs...what's that cost compared to their rep? Not a lot. Boils down to wasting a grand or two of components initially, against a shitload of future sales. It's nothing. They have given more than that away in free rigs and loads more on advertising.
|
|
|
|
Puppet
Legendary
Offline
Activity: 980
Merit: 1040
|
|
September 20, 2013, 01:39:46 PM |
|
I'll spell it out for you a lot simpler since you seem unable or unwilling to understand. 1) First few orders only...assemble as fast as possible,
I get it that time to market is crucial, but if that is your only argument, testing a die or chip (that has been processed for about a month or more in the fab) takes about 5 seconds in a machine. How long do you think it takes to assemble a complete miner based on a faulty chip? Seriously you guys are misinterpreting whats been said, or they misunderstood the question. There is no way anyone is going to assemble a complete pcb around a chip thats had neither wafer level nor chip level testing.
|
|
|
|
vesperwillow
|
|
September 20, 2013, 01:45:17 PM |
|
I've always maintained they will ensure some of the chips are tested initially. What's up for interpretation is who's doing the testing, and how it's being done (in circuit or out).
Once they have a green light on some random tests, they won't be testing each chip prior to PCB assembly unless it makes sense to do so (ie, takes longer to test out of circuit). Their time to market is far too critical. They'll do what's best to achieve time to market, whether its loss of whole PCBs or not. Keep in mind they're allowing refunds.
To do otherwise is actually a shot in the foot.
|
|
|
|
cognoscente
Member
Offline
Activity: 111
Merit: 10
|
|
September 20, 2013, 01:48:32 PM |
|
The shortcut that they took is that they are trusting the DESIGN. They didn't waste a couple of months doing a 'test batch' to verify the design. They never hinted that they were taking any other shortcuts in assembly. NONE. Some of us need to check our medication.
|
|
|
|
itod
Legendary
Offline
Activity: 1974
Merit: 1077
^ Will code for Bitcoins
|
|
September 20, 2013, 01:50:58 PM |
|
Seriously you guys are misinterpreting whats been said, or they misunderstood the question. There is no way anyone is going to assemble a complete pcb around a chip thats had neither wafer level nor chip level testing.
Why is that so difficult for you to accept this concept? If you: - manufacture something for 300-400 EUR and sell it for 3.000-4.000 EUR, - and already sold thousands of them, - and you are in mortal fear of mass refund requests, how much are you gonna care for a few percent of bad units?
|
|
|
|
markm
Legendary
Offline
Activity: 2954
Merit: 1090
|
|
September 20, 2013, 01:55:14 PM |
|
No no, you don't understand, when you are in a formula one race, every penny counts, so damn right you want to slow way the fuck down to keep your tires from wearing out because tires cost money plus changing them takes time... -MarkM-
|
|
|
|
vesperwillow
|
|
September 20, 2013, 01:57:13 PM |
|
No no, you don't understand, when you are in a formula one race, every penny counts, so damn right you want to slow way the fuck down to keep your tires from wearing out because tires cost money plus changing them takes time... -MarkM- Stop being logical! lol
|
|
|
|
Anduck
Legendary
Offline
Activity: 1511
Merit: 1072
quack
|
|
September 20, 2013, 02:03:51 PM |
|
No no, you don't understand, when you are in a formula one race, every penny counts, so damn right you want to slow way the fuck down to keep your tires from wearing out because tires cost money plus changing them takes time... -MarkM- You might end up having to stop the whole race because you didn't change the tires and one of them boomed away? Some stuff - that also takes some of the precious time - are still needed parts of the production of these things. They must do proper testing or they might end up with 100% of the units failing.. for some silly thing that is easily fixable but it will eat a lot more time than if they tested it in the first hand. Who really thinks it's good if they ship everything they get without any testing? Would you like to receive unit that hasn't been tested and not only isn't working but also breaks your PSU? And it's not about the design failures.. There's always differences in the actual chip and the designed chip. Even if you're super pro (and we somehow assume OrSoC is in that category?).. it still happens.
|
|
|
|
Puppet
Legendary
Offline
Activity: 980
Merit: 1040
|
|
September 20, 2013, 02:04:54 PM |
|
how much are you gonna care for a few percent of bad units?
You dont. You know they will happen and it will be a non trivial percentage. Could be anywhere from 5 to 50% on a first run. But thats not the question. The question is if you want to discover all those bad chips before you spent time and money putting the defective chips in a fully populated PCB and assemble it to a functional testable miner, possibly with components that are harder to score than the asics (ask BFL) , or if you are going to do what is done with probably every other asic ever produced, test the dies and/or packaged chips before you waste those components, time and money. Im nt sure you realize those chips might be among the cheapest components and with the highest defect rate by far. Thats the one you want throw away, not the rest.
|
|
|
|
vesperwillow
|
|
September 20, 2013, 02:14:55 PM |
|
It sounds like we're just philosphizing over semantics. KNC said they wouldn't ship bum units, and that they would ship ontime.
Presuming this to be fact, how they achieve their goal is moot. They have to meet their time to market goal. I don't think any of us cares if they toss 1 or 15,000 $250 PCB's as long as they meet that goal.
I also don't believe they care either--meeting that goal is like hitting a goldmine to them.
|
|
|
|
timmmers
|
|
September 20, 2013, 02:16:18 PM |
|
how much are you gonna care for a few percent of bad units?
You dont. You know they will happen and it will be a non trivial percentage. Could be anywhere from 5 to 50% on a first run. But thats not the question. The question is if you want to discover all those bad chips before you spent time and money putting the defective chips in a fully populated PCB and assemble it to a functional testable miner, possibly with components that are harder to score than the asics (ask BFL) , or if you are going to do what is done with probably every other asic ever produced, test the dies and/or packaged chips before you waste those components, time and money. Im nt sure you realize those chips might be among the cheapest components and with the highest defect rate by far. Thats the one you want throw away, not the rest. Every other ASIC has been late and the reputation of the company destroyed. We're talking about normal QA procedures here, international standards, AFTER the initial few rigs are out of the door. After they have met that deadline or come close..they can afford to be more intense with their QC if they wish. Formula one is a pretty good analogy. Their initial production costing and unit retail prices will have taken into account waste..if not they shouldn't be even allowed to build with Lego.
|
|
|
|
itod
Legendary
Offline
Activity: 1974
Merit: 1077
^ Will code for Bitcoins
|
|
September 20, 2013, 02:19:01 PM |
|
how much are you gonna care for a few percent of bad units?
You dont. You know they will happen and it will be a non trivial percentage. Could be anywhere from 5 to 50% on a first run. But thats not the question. The question is if you want to discover all those bad chips before you spent time and money putting the defective chips in a fully populated PCB and assemble it to a functional testable miner, possibly with components that are harder to score than the asics (ask BFL) , or if you are going to do what is done with probably every other asic ever produced, test the dies and/or packaged chips before you waste those components, time and money. You keep forgetting some facts: - KNC are in different situation than any other ASIC ever produced. We've never seen 100% a month difficulty growth, and we may see much higher figures very soon. - They don't have a facilities to do the chip-level resting in-house, only the board level testing. - If they try to do chip-level testing for a whole batch #1 of their chips in China, it could take a week (or two). - They don't have tested design yet. They may have a single problem in the design which can slow them down a week (or two) to solve. Can they afford to loose some headroom for design changes gaining less faulty complete units? Can they afford to reduce the worth of their chips 50% (two weeks late) to avoid some percentage of wasted complete units? Can they afford the risk of mass refunds if they don't deliver before mid-october?
|
|
|
|
|