Pages:
Author

Topic: Server Power Supply Interface Board - for standalone miners and GPU rigs - page 23. (Read 120715 times)

hero member
Activity: 854
Merit: 500
I was going to buy a bunch of these, but at $42.50 each, even with power supply for $12 that's more than $52 then I have to buy cables and such and put it all together.


I can get a power supply with cables ready to go for $50

http://www.outletpc.com/zk2737-sdgr-750e-solid-gear-750w-power-supply.html

So it does not make any sense unless you have these PSUs laying around and you want to make good use of them.

At $25 each I'd take a bunch, at $42.50 I'd buy one.

legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
I haven't lit up any loads on a DPS-800 yet, but I have a Z750P powering a Cube on full clock and a AntMiner running 400MHz with zero problems for a month now. I did some load testing and the Ant pulled about 30A, but the cables I tested with were kinda crappy so the terminal voltage at the Ant was only about 11.5V
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
I have a problem - a second DPS-800GB A that was powering a single antminer just died. wont start up - just clicks and seems to spin the fan but no DC power output. In comparison i have an identical HP supply ( ATSN-7001044-Y000) running 2 antminers without any trouble. Is this an issue with the older supply?

I swapped in a DPS-670CB A and after a few hours that failed and seems like it intermittently switched on and off for about 10minutes at a time until I found it with the warning LED lit and not providing power. Unplugging and replugging has the miner running again.  At 375MHz I dont beleive that the miner can be using anything over 500W (<75% of the rated power). Is this something to be concerned about in terms of either the particular antminer or the 670W server PSU?

edit: the psu is pushing only 11.82V - probably that is the issue, considering I have a second identical suply that only provides 11.24V and makes a strange 'clickering' noise (as in a randomized clicking similar to a bulb that is flickering) somewhere near the 120V side of the board
full member
Activity: 238
Merit: 100
power draw, energy crisis? i heard these are up to 95% efficient

They are, however servers now days use 1/4 the load, everything is low power blades.
The machines these were originally powering, were full of gold, built of stainless steel and massive, 100+ lbs each.
One ALR server with 6 PPRO cpus was $226,000 retail.
Some brands used three of the 2000watt jobbers.
It all comes from a time long ago, lost to us.
full member
Activity: 168
Merit: 100
Once we get some standing stock and a webstore put together, we'll be selling direct. Should be next week. It's taken longer than I'd have liked, to the tune of about two weeks, but my goal is to open general sales by next Wednesday.
Nice work dude! Glad to see this coming together.

I've still got that open quote with FCI if you need it.
legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
Once we get some standing stock and a webstore put together, we'll be selling direct. Should be next week. It's taken longer than I'd have liked, to the tune of about two weeks, but my goal is to open general sales by next Wednesday.
sr. member
Activity: 546
Merit: 269

Looking good, if you can add me to a list if you have one when these come available, Cheers Smiley 
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
Yep, the DPS-800GBA same as all the Blade backplanes use (if I'm remembering right). The board will also integrate current monitoring, and hopefully load balancing, same as the Z750P boards.

As for load-testing high power, I've used multiple Cubes so far but am working on a water-cooled adjustable dummy load for up to 2000W basically using current feedback to drive a parallel set of bigass FETs. Otherwise... dunno? Bunch of truck headlights?

 have just modified a DPS-800GB A power supply following the instructions;
pins 31 + 34
pins 30 + 12V

and testing without load i see a voltage of 12.8V - without the 30+12V pins shorted, voltage seemed much closer to 12.0V

I do NOT want >12.2V since this is for computer components rather than RC battery charging - Is the pin 30+12V connection designed to achieve a higher (12.8V) output, or will it drop to 12.0V once a load is applied? I followed instruction on the rcgroup forums

with some testing, found my solution:
shorting pins 31+34 turns on the system, with about 12.3V idle and 12.09V under a 400W load
shorting  pin 30 + 12V results in it jumping to about 12.8V (good for rcgroups, but not computer equipment)

no issues after about 15 minutes now

EDIT: MASSIVE FML - I assumed that because one PSU worked great the other would, and didnt fully test it before installing it at the very bottom of my crate build under 2 antminers and the working DPS-800GB. Had to cut all the power lines to the antminer (the antminer was also buried at the bottom of the build) and route them to a working supply

Are you using 1xDPS-800GB to power 2 Ants or just one? Also...what's the efficiency of the DPS-800GB? I mean is it 80+ 90+ or worse than that?!

I run 2 ants per dps-800gb with no issues this is on 240v. The fans in the psu do spin up pretty loud compared to running just one ant per psu though.
The spec for ATX 12v power allows for up to 12.6v so I think 12.8 will be fine. I run mine at this.
  

I have been running 1 per ant so far, but am tempted to try 2 ants on a single DPS-800 since it has closer to 840W at 120V. I would assume fan noise may get crazy though, so I have been avoiding it so far!

EDIT: running two antminers on a single ATSN-7001044-Y000 (identical to a DPS-800GB) at stock speeds with no issues for 30min. If it can keep stable I will try overclocking to 375 or 387.5
hero member
Activity: 826
Merit: 1000
°^°
power draw, energy crisis? i heard these are up to 95% efficient
legendary
Activity: 1596
Merit: 1000
I think some info is needed, for the people who do not quite get it yet, plain jane below:

These server Power Supplies:

Originally costs 2000+ dollars were bought as spares in case of failure, they were all Hotswappable (redundant)
Protected the most valuable data at the time.

Close to 100% efficiency with a Mean Time Between Failure (MTBF) of years (400000hrs plus) All 100% UL approved.

Make all but the very best PC supplies look like Junk, toys.... take one apart and look at it, beautiful expensive components inside.

You can arc weld with these supplies, all day.

Scrapped out for pennies on the dollar, during the energy crisis years ago, Bygone era. Never again to be seen.

I had 30-40 big old servers (PPRO PIIIxeon 4 6 8 way) on 300amp commercial main, and the breaker was warm. Two to four of these type power supplys in each box.

Why would you even compare a PC supply to these ? There is none.

The Negative on em is only the powerdraw, if you can live with that, these are perfect for what we use them for.

And now sidehack has made it easy to use for hobbiests who did not come from a blue collar (electrical/mechanical/engineering) background.

Be happy, I am. Cost of PS for your big rigs, has just evaporated Smiley Get ten of these adapters. I must have 100's of these powersupplies, all different kinds.


Thanks for the interesting post. Smiley I disagree with the bolded text, though. I'm holding a 2000w Delta monster right now that's about 91% efficient at full load, so that's a bit better than the 2 Seasonic 1000w units I'll be replacing and a hell of a lot cheaper too! Smiley Had a bit of fun taking it apart earlier. Much higher quality than any consumer power supply, for sure.
full member
Activity: 238
Merit: 100
I think some info is needed, for the people who do not quite get it yet, plain jane below:

These server Power Supplies:

Originally costs 2000+ dollars were bought as spares in case of failure, they were all Hotswappable (redundant)
Protected the most valuable data at the time.

Close to 100% efficiency with a Mean Time Between Failure (MTBF) of years (400000hrs plus) All 100% UL approved.

Make all but the very best PC supplies look like Junk, toys.... take one apart and look at it, beautiful expensive components inside.

You can arc weld with these supplies, all day.

Scrapped out for pennies on the dollar, during the energy crisis years ago, Bygone era. Never again to be seen.

I had 30-40 big old servers (PPRO PIIIxeon 4 6 8 way) on 300amp commercial main, and the breaker was warm. Two to four of these type power supplys in each box.

Why would you even compare a PC supply to these ? There is none.

The Negative on em is only the powerdraw, if you can live with that, these are perfect for what we use them for.

And now sidehack has made it easy to use for hobbiests who did not come from a blue collar (electrical/mechanical/engineering) background.

Be happy, I am. Cost of PS for your big rigs, has just evaporated Smiley Get ten of these adapters. I must have 100's of these powersupplies, all different kinds.
full member
Activity: 158
Merit: 100
Yep, the DPS-800GBA same as all the Blade backplanes use (if I'm remembering right). The board will also integrate current monitoring, and hopefully load balancing, same as the Z750P boards.

As for load-testing high power, I've used multiple Cubes so far but am working on a water-cooled adjustable dummy load for up to 2000W basically using current feedback to drive a parallel set of bigass FETs. Otherwise... dunno? Bunch of truck headlights?

 have just modified a DPS-800GB A power supply following the instructions;
pins 31 + 34
pins 30 + 12V

and testing without load i see a voltage of 12.8V - without the 30+12V pins shorted, voltage seemed much closer to 12.0V

I do NOT want >12.2V since this is for computer components rather than RC battery charging - Is the pin 30+12V connection designed to achieve a higher (12.8V) output, or will it drop to 12.0V once a load is applied? I followed instruction on the rcgroup forums

with some testing, found my solution:
shorting pins 31+34 turns on the system, with about 12.3V idle and 12.09V under a 400W load
shorting  pin 30 + 12V results in it jumping to about 12.8V (good for rcgroups, but not computer equipment)

no issues after about 15 minutes now

EDIT: MASSIVE FML - I assumed that because one PSU worked great the other would, and didnt fully test it before installing it at the very bottom of my crate build under 2 antminers and the working DPS-800GB. Had to cut all the power lines to the antminer (the antminer was also buried at the bottom of the build) and route them to a working supply

Are you using 1xDPS-800GB to power 2 Ants or just one? Also...what's the efficiency of the DPS-800GB? I mean is it 80+ 90+ or worse than that?!

I run 2 ants per dps-800gb with no issues this is on 240v. The fans in the psu do spin up pretty loud compared to running just one ant per psu though.
The spec for ATX 12v power allows for up to 12.6v so I think 12.8 will be fine. I run mine at this.
  
legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
I don't have specific numbers, but most server supplies I've seen have a typical-load efficiency of 89-93%
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
Yep, the DPS-800GBA same as all the Blade backplanes use (if I'm remembering right). The board will also integrate current monitoring, and hopefully load balancing, same as the Z750P boards.

As for load-testing high power, I've used multiple Cubes so far but am working on a water-cooled adjustable dummy load for up to 2000W basically using current feedback to drive a parallel set of bigass FETs. Otherwise... dunno? Bunch of truck headlights?

 have just modified a DPS-800GB A power supply following the instructions;
pins 31 + 34
pins 30 + 12V

and testing without load i see a voltage of 12.8V - without the 30+12V pins shorted, voltage seemed much closer to 12.0V

I do NOT want >12.2V since this is for computer components rather than RC battery charging - Is the pin 30+12V connection designed to achieve a higher (12.8V) output, or will it drop to 12.0V once a load is applied? I followed instruction on the rcgroup forums

with some testing, found my solution:
shorting pins 31+34 turns on the system, with about 12.3V idle and 12.09V under a 400W load
shorting  pin 30 + 12V results in it jumping to about 12.8V (good for rcgroups, but not computer equipment)

no issues after about 15 minutes now

EDIT: MASSIVE FML - I assumed that because one PSU worked great the other would, and didnt fully test it before installing it at the very bottom of my crate build under 2 antminers and the working DPS-800GB. Had to cut all the power lines to the antminer (the antminer was also buried at the bottom of the build) and route them to a working supply
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
Yep, the DPS-800GBA same as all the Blade backplanes use (if I'm remembering right). The board will also integrate current monitoring, and hopefully load balancing, same as the Z750P boards.

As for load-testing high power, I've used multiple Cubes so far but am working on a water-cooled adjustable dummy load for up to 2000W basically using current feedback to drive a parallel set of bigass FETs. Otherwise... dunno? Bunch of truck headlights?

 have just modified a DPS-800GB A power supply following the instructions;
pins 31 + 34
pins 30 + 12V

and testing without load i see a voltage of 12.8V - without the 30+12V pins shorted, voltage seemed much closer to 12.0V

I do NOT want >12.2V since this is for computer components rather than RC battery charging - Is the pin 30+12V connection designed to achieve a higher (12.8V) output, or will it drop to 12.0V once a load is applied? I followed instruction on the rcgroup forums
sr. member
Activity: 252
Merit: 250


DPS-800GB A?

I just hand-soldered two of these to operate and maintain 12V (short pin 31->34 and 30->12V) but would love to see a simple attachment made for them to avoid some of the annoying close-quarters soldering on the side-by-side pins 30&31

I have a box of about 20 of these things that I plan to convert if the 2 test units work.

ps: any good method for load-testing 12V at 200W, 500W, or 700W?

A bunch of coffee pots or hot plates from the good will.
legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
Yep, the DPS-800GBA same as all the Blade backplanes use (if I'm remembering right). The board will also integrate current monitoring, and hopefully load balancing, same as the Z750P boards.

As for load-testing high power, I've used multiple Cubes so far but am working on a water-cooled adjustable dummy load for up to 2000W basically using current feedback to drive a parallel set of bigass FETs. Otherwise... dunno? Bunch of truck headlights?
Pages:
Jump to: