Update Large review of Finksy/J4bbrwock server psus compared to ATXpsu-photos up

[dmwardjr]

DAAAAAAAAAAMN....  LOL

Those are some big ass caps.   LOL

got to go to a meeting I will post more tonight.

Sure...  Gotta do what you gotta do.

Appreciate all you're doing...

[nhando]

Haha, I want to LIVE.............When you have a chance, please also compare where the 2000W PSU stack up in the efficiency.  I'm assuming it would be somewhere in the middle.   Also can I trust my Wemo Insight?  It's showing that the B8 S7 is pulling 1540W at the wall.  That sounds way too high for such an efficient PSU like my EVGA G2 1300W.

[Finksy]

thanks for spacer I have been told to not double post as people perceive it as advertising my signature. sometimes I want a new post to show info more clearly.

Phil,

Don't worry most of "these people's" barks are worse than their bite.    We appreciate all the testing you are doing here, I am very anxious to see the 2980's numbers, sounds like prelude was right on the money so far.

Also, did you by chance compare voltage by back-probing the loaded PCIe cables to see if there's any noticeable voltage drop vs the loose one?  and if the avalon's gui for voltage is accurate? I've heard varying testimonies about it.

[Finksy]

Haha, I want to LIVE.............When you have a chance, please also compare where the 2000W PSU stack up in the efficiency.  I'm assuming it would be somewhere in the middle.   Also can I trust my Wemo Insight?  It's showing that the B8 S7 is pulling 1540W at the wall.  That sounds way too high for such an efficient PSU like my EVGA G2 1300W.

The B8 S7's pull around 1330W DC (rating of 1422W at the wall with 93% efficient PSU as per bitmain's specs) which is over 100% load on your EVGA 1300. This would make effective efficiency 86-87%%, which would be about right considering how high the load is on it.

[generalt]

Just my luck that I only have enough power to run two 2880 PSU and not three.  I wonder if it's possible to use the 180mm fan with just two units stacked.

[hawkfish007]

Just my luck that I only have enough power to run two 2880 PSU and not three.  I wonder if it's possible to use the 180mm fan with just two units stacked.

Maybe 2 120 or 140 mm fan will work? I have 2 2980W PSUs coming on my way, I will be able to check. Good thing is 2980W has built in 3 pin header for fans so adding any fans would be easy.

[generalt]

Maybe 2 120 or 140 mm fan will work? I have 2 2980W PSUs coming on my way, I will be able to check. Good thing is 2980W has built in 3 pin header for fans so adding any fans would be easy.

Are those standard 3 pin headers?  I hope the 2880s have them too.  I'm not sure that the area is wide enough for two 120mm fans since kilo's 180mm seems to fit pretty well.

[hawkfish007]

Maybe 2 120 or 140 mm fan will work? I have 2 2980W PSUs coming on my way, I will be able to check. Good thing is 2980W has built in 3 pin header for fans so adding any fans would be easy.

Are those standard 3 pin headers?  I hope the 2880s have them too.  I'm not sure that the area is wide enough for two 120mm fans since kilo's 180mm seems to fit pretty well.

They seem like regular 3 pin header, check philip's pictures. I don't think 2880W has 3 pin header, but I may be wrong. All of my 2880W are at a hosting place. May be smaller than 2x 120 mm will work? If somebody can stack 2 PSUs and measure?

[nhando]

Haha, I want to LIVE.............When you have a chance, please also compare where the 2000W PSU stack up in the efficiency.  I'm assuming it would be somewhere in the middle.   Also can I trust my Wemo Insight?  It's showing that the B8 S7 is pulling 1540W at the wall.  That sounds way too high for such an efficient PSU like my EVGA G2 1300W.

The B8 S7's pull around 1330W DC (rating of 1422W at the wall with 93% efficient PSU as per bitmain's specs) which is over 100% load on your EVGA 1300. This would make effective efficiency 86-87%%, which would be about right considering how high the load is on it.

Thanks Finksy, guess no more using the EVGA going forward and just stick to my 2000W PSUs.  Can't wait to get your new 4K boards!!!!!!!!!!!!!!   


Phillip - What should I expect at the walls w the 2K PSU for Batch 8 @standard Freq?

[philipma1957]

those  headers for the fans are small four pin not three pin.

i am not ready to test the 2980 but the bottomline is it appears to be 100 watts better on a 2000 watt load.

say 2000 vs 2100.
 
i am going to check watts tonight and tommorrow on the 2880 then do full

watts on the 2980 the fans are due on fri so i will post fans on sat or sun.


i found on my batch 2 s-7 that it liked the 12.2 volt setting on the 4000 watt psu

also it did well on the 2880 watter which about  12.2 volts.

i got  freq 637 5150gh on a freq 575 unit.

i only have the two avalon sixes for testing.

the s-7 is at my buddies office.

[VirosaGITS]

Haha, I want to LIVE.............When you have a chance, please also compare where the 2000W PSU stack up in the efficiency.  I'm assuming it would be somewhere in the middle.   Also can I trust my Wemo Insight?  It's showing that the B8 S7 is pulling 1540W at the wall.  That sounds way too high for such an efficient PSU like my EVGA G2 1300W.

The B8 S7's pull around 1330W DC (rating of 1422W at the wall with 93% efficient PSU as per bitmain's specs) which is over 100% load on your EVGA 1300. This would make effective efficiency 86-87%%, which would be about right considering how high the load is on it.

Thanks Finksy, guess no more using the EVGA going forward and just stick to my 2000W PSUs.  Can't wait to get your new 4K boards!!!!!!!!!!!!!!   


Phillip - What should I expect at the walls w the 2K PSU for Batch 8 @standard Freq?

It should be 99.46% load on a 1300w G2. The math is 1293W/0.88eff(@120v input)=1470W at the wall. Just a bit over the 12A/15A 80% limit. Not excellent but not bad either. Should be getting mine tomorrow to get some real world tests.

[nhando]

Thanks......WEMO is showing 1540W..............  Not loving that number too much.  Don't even like 1400W number.  Definitely want better efficiency.

[philipma1957]

2880 watt psu tests

30 min

amps 9.1
volts 241
dc  12.20
dc  12.10
dc   12.00
hash 6879
watts 2193
watts per gh 0.3188

1 hour

amps  = 9.0
volts   = 241
dc      = 12.20 at the one un - used pcie cable
dc      = 12.10 at avalon 1 software reading
dc      = 12.00 at avalon 2 software reading
hash  = 6872
watts = 2169
watts per gh = 0.3156

2 hour

amps  = 9.0
volts   = 241
dc      = 12.20 at the one un - used pcie cable
dc      = 12.10 at avalon 1 software reading
dc      = 12.00 at avalon 2 software reading
hash  = 6849
watts = 2169
watts per gh = 0.3166


7 hour

amps            = 8.9 9.0
volts             = 240
dc                = 12.20
dc                = 12.2    at avalon 1
dc                = 11.9    at avalon 2
dc                = 11.98   backprobe of both avalons were 11.97 and 11.98
watts           =  2160
hash            =  6830
watts per gh =  0.3162


I am done with   2880 watter

Next is 2980

30 Min

amps          =   8.5
volts           =  240
dc               =  12.15--------empty pcie
dc               =  12.10 -------avalon 1
dc               =  12. 00 ------ avalon 2
dc               =    ?         backprobe is hard to do
watts           =  2040     this is a 120 watt improvement   but dc is 12.15  vs 12.20
hash            =   6780        this is slightly less due to lower volts
watts per gh =  0.3008      this is  better then  any of the 2880, but by a little more like 4.9%  not quite 6%

early test  of the 2980  will do more in the morning.

A note on the 2880w  pulling the fan pack is easy to do
a note on the 2980w  pulling the fan pack is hard to do.

second note the 2880w  will allow you to do the kilo17 mod.

the 2880w is a special 4 pin with a jumper setup for each of the 3  fans the psu's will run with fans not attached.

the 2980w is  a tiny four pin pwm  you may not be able to run the 2980w with out the fan pack attached.



so all I am sure of is the 2980 stock runs 5% better then the 2880 stock. in the 30 minute test.

[Finksy]

It should be 99.46% load on a 1300w G2. The math is 1293W/0.88eff(@120v input)=1470W at the wall. Just a bit over the 12A/15A 80% limit. Not excellent but not bad either. Should be getting mine tomorrow to get some real world tests.

1293W isn't their DC power consumption though.

As per bitmain's specs:

Power Consumption: 1293W + 10% (at the wall, with APW3, 93% efficiency, 25C ambient temp) 1293 x 1.1 = 1,422.3 W at the wall with 93% efficient PSU. . Back-tracking you get 1422.3W x 0.93 (taking out their rated PSU) =1322.74W DC, 101.75% load on a 1300W G2.

They started adding the +10% for power consumption when they started offering the 135 chip S7's, that isn't the PSU efficiency being added.  At least that's how I interpret it, I don't have anything post B6.

[marvykkio]

 You guys are too awesome, your detail reviews definitely helps others decide what is best for them (size, efficiency,  noise level, etc).  I just hooked up my next pair of IBM 2K PSU and of course I found out they are the newer Rev so I can't use them with the 4K board voltage controls (Only work with Rev 8 or lower per Finksy).  Going to order a pair of older rev PSU from FINKSY to go with the new 4K board bundle.
  
  For this new pair of IBM 2K PSU, I simply just use the Scotch mounting tape Phillip recommended.
http://www.amazon.com/gp/product/B00004Z498?psc=1&redirect=true&ref_=oh_aui_detailpage_o04_s00

 Although they feel weak to the touch, these things are strong as heck when you tape them to something.  They're so strong that I didn't even bother to split out the other pair of PSU to find out the Rev #.  I only use a small piece to attach the FAN together.  Since they're at the bottom of the PSU, I didn't even need to use a mount, just place the PSU (taped together) on top and it's nice and steady on top.  I love this SCOTCH TAPE!  Make this solution clean, sexy and no obstruction to flow.  Guess I wont be needing to buy any fan mounting kit with this Tape.  Op, please quote my message so picture will show.

http://imgur.com/a/MaTjR

3m double side tape works if you have the older single breakout boards board and grilled shelf system

this is nhando's twist on my mount



these feeders who sells them?
and how much they cost?
here you go :
he double side tape the two fans



 he uses the shelf as a grill and base


lastly he gets good wire management  this is 2 single 2000 watt breakout finksy j4bberwock boards

[nhando]

Amazon sells them.  Less than $5, can be use to tape together the PSU and also Fans

http://www.amazon.com/gp/product/B00004Z498?psc=1&redirect=true&ref_=oh_aui_search_detailpage

[philipma1957]

2880 watt psu tests

30 min

amps 9.1
volts 241
dc  12.20
dc  12.10
dc   12.00
hash 6879
watts 2193
watts per gh 0.3188

1 hour

amps  = 9.0
volts   = 241
dc      = 12.20 at the one un - used pcie cable
dc      = 12.10 at avalon 1 software reading
dc      = 12.00 at avalon 2 software reading
hash  = 6872
watts = 2169
watts per gh = 0.3156

2 hour

amps  = 9.0
volts   = 241
dc      = 12.20 at the one un - used pcie cable
dc      = 12.10 at avalon 1 software reading
dc      = 12.00 at avalon 2 software reading
hash  = 6849
watts = 2169
watts per gh = 0.3166


7 hour

amps            = 8.9 9.0
volts             = 240
dc                = 12.20
dc                = 12.2    at avalon 1
dc                = 11.9    at avalon 2
dc                = 11.98   backprobe of both avalons were 11.97 and 11.98
watts           =  2160
hash            =  6830
watts per gh =  0.3162


I am done with   2880 watter

Next is 2980

30 Min

amps          =   8.5
volts           =  240
dc               =  12.15--------empty pcie
dc               =  12.10 -------avalon 1
dc               =  12. 00 ------ avalon 2
dc               =    ?         backprobe is hard to do
watts           =   2040     this is a 120 watt improvement   but dc is 12.15  vs 12.20
hash            =   6780        this is slightly less due to lower volts
watts per gh =  0.3008      this is  better then  any of the 2880, but by a little more like 4.9%  not quite 6%

14 hour results

amps      = 8.5
volts       = 240
dc           = 12.15 -----empty pcie
dc           = 12.10 ----- avalon 1
dc           = 11.90 ----- avalon 2
dc           = 11.94  ------ backprobe of a pcie cable under load
watts      =  2040 ------  
hash       =  6798 -----
watts per gh =  0.30008    



A note on the 2880w  pulling the fan pack is easy to do
a note on the 2980w  pulling the fan pack is hard to do.

second note the 2880w  will allow you to do the kilo17 mod.

the 2880w is a special 4 pin with a jumper setup for each of the 3  fans the psu's will run with fans not attached.

the 2980w is  a tiny four pin pwm  you may not be able to run the 2980w with out the fan pack attached.



 I am sure o the 2980 stock runs 5% better then the 2880 stock. in the 30 minute test. and a longer overnight test.



I am going to do back probe at the avalon on dc  done


 then do the dual 2000/4000 tests    next to do.

_______________________________________________________________________________ ___________
the dual 2000/net 4000 setup has 3 volts setting and some more tweaking with a small pot.

top volts setting 90 minutes

amps =      12  could be more my pdu does .0 until I reach 10   so if I go above 10 readings are really +- 240 watts
volts  =    240
dc     = 12.84
dc     = 12.7
dc     = 12.5
dc     = 12.2 backprobe numbers
watts              =2880    
hash               = 8000
watts per gh    = 0.3597   I know the avalon chip burns more power when overclocked so this is not a helpful judge of the psu's eff



middle volts setting 90 minutes

amps       =      9.6  
volts        =    240
dc            = 12.22
dc            = 12.3 avalon 1
dc            = 12.0 avalon 2
dc            = 12.01 backprobe numbers
watts        = 2304
hash         = 7167
watt per gh = 0.3214   this is disappointing  I need to check again as this puts the dual 2000/4000 net in last place at the close to 12.15-12.22 volt out

also this is 4000 watt capacity so 2300/4000 is closer to half load I would have thought I would do better.

 now results at  12.15-12.22 volt settings

2880watter  at 12.20 did   .3188 watts a gh

2980watter  at  12.15 did  .30008 watts a gh

4000watter at 12.22 did     .3214 watts a gh


the 4000 watt dual is vey quiet and is using only 15-18 watts for cooling

both the 2980 and the 2880   use the stock 3 fan pack of tiny loud deltas at 60 watts.

So I need to keep testing the 4000 watter to see if I have an error.
I also nee to setup the 180mm cooling fan with the 2980 psu.

that will be for the weekend.

[nhando]

the dual 2000/net 4000 setup has 3 volts setting and some more tweaking with a small pot.

volts  =    240
dc     = 12.84
dc     = 12.7
dc     = 12.5
dc     = 12.2 backprobe numbers
watts              =2880   
hash               = 8000
watts per gh    = 0.3597   I know the avalon chip burns more power when overclocked so this is not a helpful judge of the psu's eff


= Me LIKEY LIKEY......  I can't wait until my new 4K board on Monday to Just START my A6 to get it to 3.9-4TH.   It's gimping along at 3.2 now.  That's almost like getting 2 S3 for free with less power used.   

[philipma1957]

the dual 2000/net 4000 setup has 3 volts setting and some more tweaking with a small pot.

volts  =    240
dc     = 12.84
dc     = 12.7
dc     = 12.5
dc     = 12.2 backprobe numbers
watts              =2880   
hash               = 8000
watts per gh    = 0.3597   I know the avalon chip burns more power when overclocked so this is not a helpful judge of the psu's eff


= Me LIKEY LIKEY......  I can't wait until my new 4K board on Monday to Just START my A6 to get it to 3.9-4TH.   It's gimping along at 3.2 now.  That's almost like getting 2 S3 for free with less power used.   

the big kilo17 fans just got delivered I will be testing more.

One thing more on the low under volt clock of the 4000 watt setup

you can run the gear really quiet.  and I am using 7.6 amps for 2 avalon6s  so    you can run 4 avalon6's  easy with the dual 2000  at low speed

[marvykkio]

where you buy these psu IBM?
they are cheap?