[ANN] Selling Server PSU's with PCI-e cable

[Burks]

3x ~900watt?

If they deliver 93+% peak efficiency I'll take them.

 

When do you need them?
Early next week I can test the actual efficiency for you.

[Grim]

Within 10 to 15 working days would be ok. (the sooner the better)

I appreciate it that you do an efficiency test beforehand.

[Grim]

any news?

[Burks]

any news?

Just got them in.
I will test it with 2, 2.5 and 3x S1

When running 3 PSU's on 2 miner
2x 103A = 206A
206/3 = 69A per PSU
69x12 = 828W per PSU
A S1 is about 350w --> 2.4 x S1 should be the same load

results: Tonight!

[Burks]

any news?

Just got them in.
I will test it with 2, 2.5 and 3x S1

When running 3 PSU's on 2 miner
2x 103A = 206A
206/3 = 69A per PSU
69x12 = 828W per PSU
A S1 is about 350w --> 2.4 x S1 should be the same load

results: Tonight!

Unfortunately my power meter broke a couple of days ago.
When I test it with the Amp meter i get somewhere between 90 and 92% with 2,5 S1's attached....

I bought a new power meter, hopefully it arrives before the weekend...

BTW: I just finished my webshop: http://www.bitcoinpsu.eu/ I hope you gys like it

regards

[Grim]

Wow, these 2000 watt units are appealing 

Running 2 units at 75% to 80% (1500 to 1600 watt) from one breaker seems ideal.
(at least for me, I have exactly one double cable = 2 plugs to each 240v/15a breaker, max 3600 watt)

At 75% they should normally run at PEAK efficiency as well.



Very nice page you have.

Efficiency numbers for each unit would be great of course.
If you can eg. load test 60 % or up to 80% on each unit, that should be enough to provide useable numbers.




Moreover I never checked what protection features these units offer:

Some very crucial ones are:

1. SCP (short circuit protection = MOST important, can save your life    )

2. OVP (over voltage protection)

3. OCP or OPP (over current/power protection)
This one isn't that important with HUGE single rail PSUs which I assume they all are.

4. OTP (over temp protection)



btw on your page I didn't find these 900 watt units you talked about ...

[Burks]

Yesterday the new meter came in and I deliverd the first 25 PSU's.

The customer has some S7's (batch and bitmaintech PSU's

2x Bitmaintech: 2672
3x 1300w+: 2792w

So the Bitmaintech is still 4% more efficient :s

Id did some extra measurements with the new powermeter


What I noticed about the 2880w power supply: As soon as the fans spin up, the usage increases about 30w (and that is already with 2x S1, not even full fan speed).

Wow, these 2000 watt units are appealing  

Running 2 units at 75% to 80% (1500 to 1600 watt) from one breaker seems ideal.
(at least for me, I have exactly one double cable = 2 plugs to each 240v/15a breaker, max 3600 watt)

At 75% they should normally run at PEAK efficiency as well.

Very nice page you have.
Thanks 

Efficiency numbers for each unit would be great of course.
If you can eg. load test 60 % or up to 80% on each unit, that should be enough to provide useable numbers.

Moreover I never checked what protection features these units offer:

Some very crucial ones are:

1. SCP (short circuit protection = MOST important, can save your life    )
I'll check that with a cheaper PSU if you don't mind.

2. OVP (over voltage protection)
14V (Almost on every PSU it is 14v)

3. OCP or OPP (over current/power protection)
This one isn't that important with HUGE single rail PSUs which I assume they all are.
They're all singel rail. I'll test this again with a cheap PSU

4. OTP (over temp protection)
I know a 1400w PSU just quits at high temps


btw on your page I didn't find these 900 watt units you talked about ...
I've quit with those ones, since they're quite bulky and noisy

[Grim]

You can't compare the efficiency if you don't take the total wattage of the PSU into account.

If you test each unit at approximately 70% load then you can compare them.

1300 watt test @ ~  900 load
2000 watt test @ ~1400 load
2880 watt test @ ~2000 load


I'm sorry to bother you that much but I do believe many customers would like to know anyway.


Moreover I thought MOST server PSUs delivery platine or even titanium efficiency yet the numbers you provide are hardly even gold efficiency.

I heard of 94-96% efficiency server units, which are they?

[Burks]

You can't compare the efficiency if you don't take the total wattage of the PSU into account.

If you test each unit at approximately 70% load then you can compare them.

1300 watt test @ ~  900 load
2000 watt test @ ~1400 load
2880 watt test @ ~2000 load


I'm sorry to bother you that much but I do believe many customers would like to know anyway.


Moreover I thought MOST server PSUs delivery platinum or even titanium efficiency yet the numbers you provide are hardly even gold efficiency.

I heard of 94-96% efficiency server units, which are they?

Well, Efficiency is basically OUT/IN.
In the end, you have a load which needs to be powered. It doesn't matter if you do that with a 2880w PSU or a bunch of 900w PSU's.

This week I'll extend the measurements up to 6x S1.
I will add a 1100w ATX PSU to the test.
To be honest, I expected higher efficiency as well If you have a suggestion, let me know so I can test them the same way...

[Grim]

In the end, you have a load which needs to be powered. It doesn't matter if you do that with a 2880w PSU or a bunch of 900w PSU's.

uhm that's quite wrong sorry ... it matters VERY much if you care about efficiency ... otherwise it's true yes

The efficiency is a curve which normally peaks around 60 to 75 % load of the PSU.

Outside that load the psu efficiency is WAY worse!!!

You can actually see it in your chart the efficiency for the large units climbed and it would climb further if you applied more load up to ~75% where it would slightly go down again.
So without a 2000watt+ load you wont find out the true efficiency of your 2880 watt unit.

[Burks]

In the end, you have a load which needs to be powered. It doesn't matter if you do that with a 2880w PSU or a bunch of 900w PSU's.

uhm that's quite wrong sorry ... it matters VERY much if you care about efficiency ... otherwise it's true yes

The efficiency is a curve which normally peaks around 60 to 75 % load of the PSU.

Outside that load the psu efficiency is WAY worse!!!

You can actually see it in your chart the efficiency for the large units climbed and it would climb further if you applied more load up to ~75% where it would slightly go down again.
So without a 2000watt+ load you wont find out the true efficiency of your 2880 watt unit.

I think you misunderstood what I said.

I completely agree with you on the fact that efficiency is a function of the drawn wattage.

However: you can approach a situation from 2 vieuws:
- the PSU is most efficient at 70%, so I will seek a load that is 70%
- I have a load of X, how efficient is each PSU on that load.
I prefer the latter, that is why I will test all PSU's at the whole spectrum of loads.

with 6x S1 we will draw 2160w, enough to test all PSU's

[philipma1957]

The 2880 is a silver psu.

The 2980 is a platinum psu.

They both use the same fans and the fans pull close to 60 watts.

In hardware section I did a fan mod to the 2980's and they give true platinum numbers.

I am on an iPad I will post link to the fan mod later.

As I type I am running five Avalon 6    miners using 4800 watts on a pair of modded 2980 psu's

[mavericklm]

We need a psu ~1500watt for last s7 batches.....

[philipma1957]

the 2980 or 2880 with the fan mod saves 40 watts

lot of photos on the fan mod.

https://bitcointalk.org/index.php?topic=1284563.msg13708402#msg13708402

[Burks]

We need a psu ~1500watt for last s7 batches.....

The power rating on a PSU is based on the output, not the input...
So if 1500w goes in, only 1350 comes out (at 90%)

the 2980 or 2880 with the fan mod saves 40 watts

lot of photos on the fan mod.

https://bitcointalk.org/index.php?topic=1284563.msg13708402#msg13708402

It looks like a nice build, but I think it's going to be difficult to give a warranty on that :S
Unfortunately, the 2980 PSU's a quite expensive 

[Burks]

Today I tested with up to 5x S1, here the results

[Grim]

These server PSU's deliver 12.2 volt afaik.

If you calculate with 12.2 volt instead of 12.0 the efficiency is slightly higher (~1%).

[Burks]

These server PSU's deliver 12.2 volt afaik.

If you calculate with 12.2 volt instead of 12.0 the efficiency is slightly higher (~1%).

I calculated it with the actual voltage, but since voltages differ across PSU's and load, I didn't post all those values

[Grim]

This list includes the efficiency standards for supermicro PSUs:

http://www.supermicro.com/support/resources/pws/

[Burks]

A little update

The website is live: www.bitcoinpsu.eu

with the following discount code:

Code:

iloveBCT

you get 20% discount

Some of the products

1000w+ (85A) - 8 PCI-e - 90.5% efficient
Suitable for 2/3 of a S7, one Avalon or 2 S5's
In theory it can deliver 1325 watt's however, due to heating it is sold as 1000w

With discount: €60,-

2000w+ (175A) - 10-16x PCI-e
Suitable for up to 1,5 S7, two avalons or four S5's
*It must me noted that this PSU is very loud, about the same of a S7*

With discount: €92,-

2800w+ (240A) - 12x PCI-e
Perfect for a S7 of 1,5 Avalon

With discount: €144,-

3000w+ (255A) - 20-24x PCI-e - 90.5% efficient
This unit consists of 3 1000w PSU's. good for 2 S7's.
All units can be powered on at once.

With discount: €160,-