Underneath??
He just has the modules raised up so the fan on the right of the pic is blowing on the bottom of each unit. That will drastically reduce your VRM temps.
Topic: Hacking The KNC Firmware: Overclocking - page 12. (Read 144378 times)
He just has the modules raised up so the fan on the right of the pic is blowing on the bottom of each unit. That will drastically reduce your VRM temps.
Underneath??
See DC/DC numbers 0 and 1 are hot:
I was afraid of that, they can most likely be hacked to run faster using the spi just like the Nov but those VRMs simply don't have much headroom, it looks like they scrimped on this design to maximize profits unlike the margins they built into the November boards.. The VRMs used have a thermal shutdown so they shouldn't cook themselves, within 10C of thermal shutdown (starts in the high 90s if I remember right) it will start throwing a FAULT 4 indicating it's hot..
They put heatsinks on the VRMs, which was a great move but with stock clocks have them at almost 50A they simply may not have a lot left.. With that said the GE VRMs run about 40% over spec with no issues so perhaps the Ericsson VRMs have a lot of headroom designed in, there might be hope..
I need a system for a few days to figure something out but I cancelled my order so best I could do is customize something and send it for testing but without being able to read the bus it will be tough unless by chance they didn't change things at all..
Hi Padrino,
I found a way to solve the 105+ C readings. I have modified one box (will do 2 more) and now the DC/DC 1 is at the same levels as the others (around 90). I'm sure that with some better and additional Fans we can lower the temps of all VRM's.
oh elenelen..i wish i had that neptune
it should be easy to lower the temp on vrm...buy little square copper heatsinks for the vrms...scrap the crappy thick thermal pad that doesn't do crap and then just take the yeller sticker off vrm n put copper on...also get some heatsinks for the bottom...take it out the case...raise board off table an inch or so and have fan blow on it...i have nov. mods oc'd really high and highest vrm temp is around 80c all with heatsinks top n bottom raised n 140mm fan blowing on itCan you give us a picture please??
See DC/DC numbers 0 and 1 are hot:
I was afraid of that, they can most likely be hacked to run faster using the spi just like the Nov but those VRMs simply don't have much headroom, it looks like they scrimped on this design to maximize profits unlike the margins they built into the November boards.. The VRMs used have a thermal shutdown so they shouldn't cook themselves, within 10C of thermal shutdown (starts in the high 90s if I remember right) it will start throwing a FAULT 4 indicating it's hot..
They put heatsinks on the VRMs, which was a great move but with stock clocks have them at almost 50A they simply may not have a lot left.. With that said the GE VRMs run about 40% over spec with no issues so perhaps the Ericsson VRMs have a lot of headroom designed in, there might be hope..
I need a system for a few days to figure something out but I cancelled my order so best I could do is customize something and send it for testing but without being able to read the bus it will be tough unless by chance they didn't change things at all..
Hi Padrino,
I found a way to solve the 105+ C readings. I have modified one box (will do 2 more) and now the DC/DC 1 is at the same levels as the others (around 90). I'm sure that with some better and additional Fans we can lower the temps of all VRM's.
oh elenelen..i wish i had that neptune
it should be easy to lower the temp on vrm...buy little square copper heatsinks for the vrms...scrap the crappy thick thermal pad that doesn't do crap and then just take the yeller sticker off vrm n put copper on...also get some heatsinks for the bottom...take it out the case...raise board off table an inch or so and have fan blow on it...i have nov. mods oc'd really high and highest vrm temp is around 80c all with heatsinks top n bottom raised n 140mm fan blowing on itCan you give us a picture please??
Looking at the picture of the ASIC board
http://archive.ericsson.net/service/internet/picov/get?DocNo=28701-EN/LZT146435&Lang=EN&HighestFree=Y
And:
http://www.ericsson.com/ourportfolio/products/bmr464-series
Curious that the manufacturer states that only up to 7 of these can be used in parallel.
Looking forward to hearing more details.
http://archive.ericsson.net/service/internet/picov/get?DocNo=28701-EN/LZT146435&Lang=EN&HighestFree=Y
And:
http://www.ericsson.com/ourportfolio/products/bmr464-series
Curious that the manufacturer states that only up to 7 of these can be used in parallel.
Looking forward to hearing more details.
See DC/DC numbers 0 and 1 are hot:
I was afraid of that, they can most likely be hacked to run faster using the spi just like the Nov but those VRMs simply don't have much headroom, it looks like they scrimped on this design to maximize profits unlike the margins they built into the November boards.. The VRMs used have a thermal shutdown so they shouldn't cook themselves, within 10C of thermal shutdown (starts in the high 90s if I remember right) it will start throwing a FAULT 4 indicating it's hot..
They put heatsinks on the VRMs, which was a great move but with stock clocks have them at almost 50A they simply may not have a lot left.. With that said the GE VRMs run about 40% over spec with no issues so perhaps the Ericsson VRMs have a lot of headroom designed in, there might be hope..
I need a system for a few days to figure something out but I cancelled my order so best I could do is customize something and send it for testing but without being able to read the bus it will be tough unless by chance they didn't change things at all..
Hi Padrino,
I found a way to solve the 105+ C readings. I have modified one box (will do 2 more) and now the DC/DC 1 is at the same levels as the others (around 90). I'm sure that with some better and additional Fans we can lower the temps of all VRM's.
oh elenelen..i wish i had that neptune
it should be easy to lower the temp on vrm...buy little square copper heatsinks for the vrms...scrap the crappy thick thermal pad that doesn't do crap and then just take the yeller sticker off vrm n put copper on...also get some heatsinks for the bottom...take it out the case...raise board off table an inch or so and have fan blow on it...i have nov. mods oc'd really high and highest vrm temp is around 80c all with heatsinks top n bottom raised n 140mm fan blowing on it See DC/DC numbers 0 and 1 are hot:
I was afraid of that, they can most likely be hacked to run faster using the spi just like the Nov but those VRMs simply don't have much headroom, it looks like they scrimped on this design to maximize profits unlike the margins they built into the November boards.. The VRMs used have a thermal shutdown so they shouldn't cook themselves, within 10C of thermal shutdown (starts in the high 90s if I remember right) it will start throwing a FAULT 4 indicating it's hot..
They put heatsinks on the VRMs, which was a great move but with stock clocks have them at almost 50A they simply may not have a lot left.. With that said the GE VRMs run about 40% over spec with no issues so perhaps the Ericsson VRMs have a lot of headroom designed in, there might be hope..
I need a system for a few days to figure something out but I cancelled my order so best I could do is customize something and send it for testing but without being able to read the bus it will be tough unless by chance they didn't change things at all..
Hi Padrino,
I found a way to solve the 105+ C readings. I have modified one box (will do 2 more) and now the DC/DC 1 is at the same levels as the others (around 90). I'm sure that with some better and additional Fans we can lower the temps of all VRM's.
See DC/DC numbers 0 and 1 are hot:
I was afraid of that, they can most likely be hacked to run faster using the spi just like the Nov but those VRMs simply don't have much headroom, it looks like they scrimped on this design to maximize profits unlike the margins they built into the November boards.. The VRMs used have a thermal shutdown so they shouldn't cook themselves, within 10C of thermal shutdown (starts in the high 90s if I remember right) it will start throwing a FAULT 4 indicating it's hot..
They put heatsinks on the VRMs, which was a great move but with stock clocks have them at almost 50A they simply may not have a lot left.. With that said the GE VRMs run about 40% over spec with no issues so perhaps the Ericsson VRMs have a lot of headroom designed in, there might be hope..
I need a system for a few days to figure something out but I cancelled my order so best I could do is customize something and send it for testing but without being able to read the bus it will be tough unless by chance they didn't change things at all..
It is exactly the same VRMs as Ericsson BMR464 series 50 Ampere.
I wonder, since this setup needs more power how could they use the same number of Vrm modules ( 8 of them) as the Jup boards?
Jup were designed for 170W , with 8x vrms, handling each approximately 22W.
Now with a stock Nep card, it uses around 36-37W each vrm ?
Is it only the metal plate+thermal pad for saving the extra vrms?
No space for overclocking.
I wonder, since this setup needs more power how could they use the same number of Vrm modules ( 8 of them) as the Jup boards?
Jup were designed for 170W , with 8x vrms, handling each approximately 22W.
Now with a stock Nep card, it uses around 36-37W each vrm ?
Is it only the metal plate+thermal pad for saving the extra vrms?
No space for overclocking.
Welcome Elenelen,
what we need is the part number and manufacturer of the vrm chips, or does any one here has them already?
If we know that, we can take a look to the original specs for this part. And the we can point you / every Neptun customer to an
point where you can´t damage your miner. Without this knowledge it is stupid to say tweak anything.
Loosing warranty!!
what we need is the part number and manufacturer of the vrm chips, or does any one here has them already?
If we know that, we can take a look to the original specs for this part. And the we can point you / every Neptun customer to an
point where you can´t damage your miner. Without this knowledge it is stupid to say tweak anything.
Loosing warranty!!
Naahh.. that's a bit too risky for me: to see the part number you need to take apart the whole block and aluminum plate plus foam...
Welcome Elenelen,
what we need is the part number and manufacturer of the vrm chips, or does any one here has them already?
If we know that, we can take a look to the original specs for this part. And the we can point you / every Neptun customer to an
point where you can´t damage your miner. Without this knowledge it is stupid to say tweak anything.
Loosing warranty!!
what we need is the part number and manufacturer of the vrm chips, or does any one here has them already?
If we know that, we can take a look to the original specs for this part. And the we can point you / every Neptun customer to an
point where you can´t damage your miner. Without this knowledge it is stupid to say tweak anything.
Loosing warranty!!
Yes but some of your vrms are running too hot.
You need to tweak it carefully.
You need to tweak it carefully.
What is the level, you guys have experience of, giving trouble ? Some are already at 104 to 106 C
I would not let it run that high.
I know it spits an error at around 95C.
Ericsson says the limit is 120C but I would not dare to let it run more than 95C.
Anyone else to share experiences?
See DC/DC numbers 0 and 1 are hot:
Yes but some of your vrms are running too hot.
You need to tweak it carefully.
You need to tweak it carefully.
What is the level, you guys have experience of, giving trouble ? Some are already at 104 to 106 C
Yes but some of your vrms are running too hot.
You need to tweak it carefully.
You need to tweak it carefully.
Hi Guys, can we try to overclock the Neptune to 575 Mhz ?
I have posted some pictures over here:
https://bitcointalk.org/index.php?topic=170332.35816
Advanced Page gives a list with clocking speed from 50 to 500 Mhz (with steps of 25). It's defaulted on 475 already (so not much built-in upside).
I have posted some pictures over here:
https://bitcointalk.org/index.php?topic=170332.35816
Advanced Page gives a list with clocking speed from 50 to 500 Mhz (with steps of 25). It's defaulted on 475 already (so not much built-in upside).
just don't get it...have a 14in. 1300cfm fan on top of a wooden case i built...it'll blow >> w/o the box to put on top of it for the vent...but as soon as i put the box on it, it blows <<< how is that possible? i guess is it because i'm trying to go from a 14in fan to 7in vent tube? also seems it doesn't matter which way i have the fan..cause i flipped it around and it does the same thing...its a fan to remove heat from an attic...thought it would work great here 
EDIT: ok so i put the fan in the window turned it on and it was blowing out...put the box over it...WTH it still blows out...so i guess i'll try that...just reverse the way i intended to use it...must be the position cause i was laying it flat and now its on the window...wierd I just hope a 7in sucking vent is enuff to suck away 3 jups worth of heat instead of 14in.
And now i think i messed up my controller board for my nov. unit for 1 asic...tryed different board n different psu cable and it does the same thing...ramps up to really high volt then shuts off the vrms they come back on eventually but only at like 10a and throws a fault 81 code on bertmode...oh happy days
EDIT: ok so i put the fan in the window turned it on and it was blowing out...put the box over it...WTH it still blows out...so i guess i'll try that...just reverse the way i intended to use it...must be the position cause i was laying it flat and now its on the window...wierd I just hope a 7in sucking vent is enuff to suck away 3 jups worth of heat instead of 14in.
And now i think i messed up my controller board for my nov. unit for 1 asic...tryed different board n different psu cable and it does the same thing...ramps up to really high volt then shuts off the vrms
they come back on eventually but only at like 10a and throws a fault 81 code on bertmode...oh happy days 
another download link to the exact same custom firmware as page 50 http://www.megafileupload.com/en/file/543339/kncminer-1-01-custom-bin.html
Nove Jupiter here. Just applied the custom fw and my chips are running at 900MHz. Can I overclock it from the Advanced tab without logging to ssh?
Yes that's the beauty of the custom fw, switching between advanced and the bertmod output allows you to quickly get it running nicely.
another download link to the exact same custom firmware as page 50 http://www.megafileupload.com/en/file/543339/kncminer-1-01-custom-bin.html
Nove Jupiter here. Just applied the custom fw and my chips are running at 900MHz. Can I overclock it from the Advanced tab without logging to ssh?
i follow the guide October: -0.0879 and 0x231 - 680Gh/s
i wait the "On initial startup CGMiner/BFGMiner will restart after 60 seconds..."
and when restart hashing 820 arround 2 min and after down to 600
all cores Enabled
i wait the "On initial startup CGMiner/BFGMiner will restart after 60 seconds..."
and when restart hashing 820 arround 2 min and after down to 600
all cores Enabled
Jump to: