 Show Posts
|
|
Pages: « 1 2 [3] 4 5 6 »
|
What hash rate are you getting at 400Mhz @ 1.3v for only 10 chips? Right now the most I've OC my Avalon is 358Mhz.
I dont recall exactly. I believe It was a bit over 4ghs |
|
|
|
whoa that wire mod looks sick...
do you think w/o an overvolt that more consistent voltage will help HW% errors drop and thus a bit higher clock even at stock voltage? cuz that mod seems simple enough.
Yup. JFET transistors act like tiny capacitors when they're closed - the current doesn't actually stop flowing until the capacitor is 'charged'. And of course more volts means faster charge time. On the other hand, of course, it means more heat, and more risk of damaging your chips... Just the wire mod for consistent voltage risks damaging the chips?! ... I would think the more consistent voltage has little / no chance of damage to chips w/ stock cooling.. I know its obvious a mod like 1.34v risks damaging / overheating chips if not properly cooled, but wasnt thinking about doing that. Just the wire mode for tad more clockspeed cuz of the more consistent voltages. I think my cooling setup should handle it fine (5k rpm 250cfm fans) Just adding a wire (assuming done correctly) will have no chance of damaging your chips. However if you have never soldered to electronics before I recommend leaving the board as is. where exactly have you soldered the 2 points? i dont see the r10 resistor anywhere on the first pic. If you look closely the resistor in the exact center of the image is R10. It is an 0603 size resistor with a 0402 sitting on top. << look closely, its there. |
|
|
|
|
Stability comes from 2 ways. Voltage and cooling.
Overclockers have been using liquid nitrogen to reach insane speeds for years. To say weather you can hit a specific speed depends on voltage and cooling.
I certainly think that if you can do 370mhz stable(no total crashes) with a lot of errors the wire will probably reduce the error rate by boosting the marginal chips voltage.
Whenever I overclock the first thing I workout is my thermal solution. You have to do something with the extra heat or it will kill your chips. Running cooler also helps with stability and error rates.
|
|
|
|
Wire and voltage mod preliminary results: Successful!  300Mhz 350Mhz 400Mhz Inductor 1.299v 1.299v 1.299v Middle 1.290v 1.290v 1.288v Bottom 1.291v 1.291v 1.289v chip Before the 16awg wire I was seeing 50mv drop from top to bottom. Only one module of 10 chips was powered up. 400Mhz was perfectly stable. I only ran for 10 minutes due to thermal reasons(my watercooling solution is not active yet) From the Burnin thread SebastianJu is getting 450MHz@1340mV I suspect I should get close to that if the VR can provide the current. Note 400Mhz at 1.3v is a LOT hotter than at 300Mhz!!! if you do this mod you MUST upgrade the stock cooling!!!! Thermal paste on the back of the chips at the MINIMUM!!!!! |
|
|
|
whoa that wire mod looks sick...
do you think w/o an overvolt that more consistent voltage will help HW% errors drop and thus a bit higher clock even at stock voltage? cuz that mod seems simple enough.
Yup. JFET transistors act like tiny capacitors when they're closed - the current doesn't actually stop flowing until the capacitor is 'charged'. And of course more volts means faster charge time. On the other hand, of course, it means more heat, and more risk of damaging your chips... Just the wire mod for consistent voltage risks damaging the chips?! ... I would think the more consistent voltage has little / no chance of damage to chips w/ stock cooling.. I know its obvious a mod like 1.34v risks damaging / overheating chips if not properly cooled, but wasnt thinking about doing that. Just the wire mode for tad more clockspeed cuz of the more consistent voltages. I think my cooling setup should handle it fine (5k rpm 250cfm fans) Just adding a wire (assuming done correctly) will have no chance of damaging your chips. However if you have never soldered to electronics before I recommend leaving the board as is. |
|
|
|
|
Then just buy BTC directly and make more profit.
|
|
|
|
whoa that wire mod looks sick...
do you think w/o an overvolt that more consistent voltage will help HW% errors drop and thus a bit higher clock even at stock voltage? cuz that mod seems simple enough.
From what I have read from others on Burnins board thread they are seeing big changes in performance with just a 10mv difference. a 16awg wire should be good for 10a without much voltage drop. it will be a couple of days before I can finish up on the watercooling stuff before I can fire the boards up and take some measurements. Doh! I'm anxious .. oh u already did volt mod huh, that's why gotta wait till WC setup is good...? Where were the voltage readouts listed, tried finding them in this thread - didn't find them, was it in the avalon users thread? Bunch of stuff here and the next few pages after. https://bitcointalk.org/index.php?topic=179769.msg2927914;topicseen#msg2927914They are running at 1.4v and higher it seems!!! These are old chips at 110nm i believe! they will run into signal integrity issues with high clock rates long before we burn one out due to over voltage. Going too high will just waste a bunch of energy. Mine is modded to 1.34v which seems like a good place based on what I have read on Burnin's thread. |
|
|
|
whoa that wire mod looks sick...
do you think w/o an overvolt that more consistent voltage will help HW% errors drop and thus a bit higher clock even at stock voltage? cuz that mod seems simple enough.
From what I have read from others on Burnins board thread they are seeing big changes in performance with just a 10mv difference. a 16awg wire should be good for 10a without much voltage drop. it will be a couple of days before I can finish up on the watercooling stuff before I can fire the boards up and take some measurements. |
|
|
|
Preview of what im working on. Im going for 1.34v, watercooling, and a 16awg wire to distribute current to the bottom of the board. 39.2k R10 piggyback resistor  watercooling channels. Extruded alu is not fun to weld.  16awg wire.  |
|
|
|
|
Has anyone been able to meet these guys face to face yet?
|
|
|
|
How about B3 machines that have changed design ? I see on one photos hash unit are probably grouped in 4 on one PCB and added thermal sensor on it.
My unit is a batch 3. It looks like we are seeing the exact same voltage drop. |
|
|
|
I measure my board with my 100MHz oscilloscope with 1.3V mod and did not see noise on power lines. I check it now after over 10 days mining.
About drop:
Standby: 1.298V
395MHz:
1.278 on coil VR and nearest capacitor
1.268 first two chip
1.228 last chip.
I solder silver-plated wire 0,8mm across one board and check how its help.
Thanks for the info. It looks like it's more work to do. Loosing 3% hashrate for voltage drops would be sad. 50mv drop is horrible! adding a couple thick wires from the output of the VR to the last chips should help a lot. although the problem could be a return path issue. I dont have my unit in front of me at the moment. I assume the bottom layer is Ground and the top is VCC? There may be an issue with the ground vias/pins not being able to handle the current. |
|
|
|
|
Thanks lastbit and GandalfG for the info. I have access to a nice 12ghz oscilloscope and will take noise and ripple measurements of the output of the VR and see if it gets unstable past 20a.
I am running a 1200w supply so I should be fine on the supply side. Going to be switching to 220v as well to gain a bit more efficiency.
I will be running coolant to a radiator outside for watercooling... Unfortunately it has been 105f or 40c for the last month outside. Which is even more reason I want to get the 800+watt this thing produces out of my house!
The main issue I see with cooling is getting the heat from the chips into the heatsink. lastbit's photo earlier show the exact bowing away Im seeing.
As the board heats up you can watch it bow away. I have a clamping solution im working on as well to hold the chips flat against the heatsink.
Just took some measurements. At the top of the board closest to the VR im seeing 1.2v at the bottom im seeing 1.15v! Im surprised to see so much of a drop! They must have cheeped out on the amount of copper on the board.
|
|
|
|
|
Thanks lastbit!
Ill just pull the spec sheet for the controller.
Ill probably just drop to a 6k resistor and a 10k pot in series till I can figure out what voltage it takes to get to 400mhz.
Avalon chip spec shows 1.2v typical 1.32 max!
I wonder if I can shut off all of the individual modules except for the one Im working on.
|
|
|
|
@lastbit.
Have you reworked a VR with the 1.5uH? How much current have you been able to push?
I haven't. I have to solve cooling first. I hope next week. Im in the process of watercooling mine. Should be done next week as well. I plan on switching out the sense resistors(R10 i believe) on the VR to get to 1.2v. Where can I find the schematics for this thing? |
|
|
|
|
@lastbit.
Have you reworked a VR with the 1.5uH? How much current have you been able to push?
|
|
|
|
I'm interested in power consumption so I'd appreciate if someone could report that.
Some power measurements for Avalon.Unit is batch#3 w/4 modules, fw 20130723. Temperature target 65C. Keep in mind this is equiv with 45 for batch#1&2. Temp1 (input temperature) is 21C . Temp3 (output temperature) is roughly 15C lower then Temp2. Wall is wall power in watts. Errors is percentage of HW errors versus diff1 shares. | freq | wall | fan1 | fan3 | temp2 | errors | | 256 | 685 | 2280 | 2040 | 60 | | | 270 | 722 | 2280 | 2160 | 60 | | | 282 | 755 | 2400 | 2160 | 61 | | | 300 | 803 | 2400 | 2160 | 63 | 0.94% | | 325 | 873 | 2520 | 2280 | 66 | 1.08% | | 350 | 943 | 3480 | 3240 | 63 | 1.35% |
Produces arround 109-110Gh/s @350MHz, confirmed by btcguild (more or less). At 375MHz unit cannot maintain 65C temperature target. Hash is less, around 105Gh/s, due to large percentage of HW errors. |
|
|
|
|
If I recall my 4 module batch 3 with Corsair AX1200 gold supply pulls ~800 watts at the wall at 325mhz.
i am in the process of water cooling my unit. Will try and get past 350mhz soon.
The thermal/mechanicals on this thing are horrific! You can actively watch the PCB bow up from the heatsink as the ASICs start hashing!!
I will post pictures of my solution once it is all finished.
|
|
|
|
|
Your minercount should be 32.
|
|
|
|
|
I would like to see Batch 3 orders get free upgrade modules with 2nd gen chips!
|
|
|
|
|
