GROUP BUY - AVALON6 - $1100 - Closing Jan 12 at 6:00 PM EST

[kipper01]

I will do one avalon 6 at 2.5 BTC

[kilo17]

PPOC - 2 units
Kipper01 - 1 unit

You have been added 

[PPOC]

PPOC - 2 units
Kipper01 - 1 unit

You have been added 

Excellent, I will wait for Philip's PM with total and already saw he created a payment address. Will they ship direct to everyone or someone receiving them all and re-shipping? I saw NJ somewhere, I am in NY so just curious.

Thanks Again,

[soy]

If I'm thinking right, these Avalons have separate 12V rails per board so if necessary you can run them off two PSUs, probably want at least 600-650W rated each. But a decent 650W PSU to run on 120V shouldn't be hard to find.

Yes that would be good.

Yes efficiency would be better off two different PSUs. If you can get your load on each PSU to around 50% that would be most efficient (maybe not on your wallet).
Graph below is for the PSU I plan on using:

If I want to get more efficiency I could add another PSU and hit right at the 50% mark or just run on 1 and still be fine. Thats why I chose the PSU i mentioned earlier. Of course you guys with 240V can just laugh and enjoy you're superior server PSU efficiency.

You guys with good power supplies would know.  What power factor does a Kill-A-Watt give for a good supply?  Personally I've always gone with cheap LED supplies then added a heat sink to the bridge rectifier, changed out the single turn trim pot for a 20 turn, improved the thermal paste on the power transistors then replacing the fans when they get noisy.  Also I add 3½ LEDs measuring the voltage at the PCIE connector.  With a C1 I use four 360 watt 12v LED supplies, one on each blade and a fifth to power the radiator fans and pump.  While I was considering the buy of an Avalon6 I was considering buying four new 5v or 6v 360 watt power supplies for under $30 each, putting a digital readout on each and again at one of the two PCIE connectors on each blade, putting two in series for each blade.  Putting 12v supplies in parallel burns them up.  So, with two of these improved cheap LED 12v supplies on an S3+ right here that I have a Kill-A-Watt monitoring, I get 120.2VAC, 0.70 power factor, 59.9~/s, 554VA, 388 watts, 4.62 amps.  So, I figure the losses using cheap supplies would show up in the power factor figure and wattage.  Can someone tell me how this compares with quality supplies?  Thanks.

What I'm powering with these really doesn't matter.  The 12vdc would appear as a resistive load so power factor doesn't apply.  I really can't get a good reading on the DC current.  I considered buying shunts but then the cost...

[kilo17]

PPOC - 2 units
Kipper01 - 1 unit

You have been added 

Excellent, I will wait for Philip's PM with total and already saw he created a payment address. Will they ship direct to everyone or someone receiving them all and re-shipping? I saw NJ somewhere, I am in NY so just curious.

Thanks Again,

It can be shipped direct but I will have the details of how that will work tonight.  Phil is getting several of the units for to ship out to people as well

[kilo17]

If I'm thinking right, these Avalons have separate 12V rails per board so if necessary you can run them off two PSUs, probably want at least 600-650W rated each. But a decent 650W PSU to run on 120V shouldn't be hard to find.

Yes that would be good.

Yes efficiency would be better off two different PSUs. If you can get your load on each PSU to around 50% that would be most efficient (maybe not on your wallet).
Graph below is for the PSU I plan on using:

If I want to get more efficiency I could add another PSU and hit right at the 50% mark or just run on 1 and still be fine. Thats why I chose the PSU i mentioned earlier. Of course you guys with 240V can just laugh and enjoy you're superior server PSU efficiency.

You guys with good power supplies would know.  What power factor does a Kill-A-Watt give for a good supply?  Personally I've always gone with cheap LED supplies then added a heat sink to the bridge rectifier, changed out the single turn trim pot for a 20 turn, improved the thermal paste on the power transistors then replacing the fans when they get noisy.  Also I add 3½ LEDs measuring the voltage at the PCIE connector.  With a C1 I use four 360 watt 12v LED supplies, one on each blade and a fifth to power the radiator fans and pump.  While I was considering the buy of an Avalon6 I was considering buying four new 5v or 6v 360 watt power supplies for under $30 each, putting a digital readout on each and again at one of the two PCIE connectors on each blade, putting two in series for each blade.  Putting 12v supplies in parallel burns them up.  So, with two of these improved cheap LED 12v supplies on an S3+ right here that I have a Kill-A-Watt monitoring, I get 120.2VAC, 0.70 power factor, 59.9~/s, 554VA, 388 watts, 4.62 amps.  So, I figure the losses using cheap supplies would show up in the power factor figure and wattage.  Can someone tell me how this compares with quality supplies?  Thanks.

That is a ton of work for little savings, the server PSU's like the 1000w and 1200w can be had for $40 or less.  Simply jump the "on" with a resistor and solder on some wires and done.  I have a few I did recently and can post it on the PSU thread here if you want.  Also the 1200w has a Pot to adjust voltage and amperage (I have a thread on that as well). Here are the links to the 2 threads:
PSU discussion:
https://bitcointalk.org/index.php?topic=1284563.0
PSU Voltage adjustment (by yours truly  )
https://bitcointalk.org/index.php?topic=1294058.0

[sidehack]

Power factor should be as close to 1 as possible. A good server PSU will have active PFC on the input, which typically yields a PF of 0.97 or better. 0.7 is pretty bad and if you run too much power at that level your utility provider will hate you. My bill goes up 1% for every 1% my PF is below 90% (averaged for the whole shop).

I have some Avalon6 already in hosting; the most recent ones were put with a Dell 750W server PSU per blade. They're running just fine far as I can tell.

I wish I had stock right now, but we won't for another month or I'd probably offer a discount deal for participation within this group buy. We're working on an interface board that's compatible with DPS800 and DPS1200 edge-connector server PSUs, probably the ones kilo just talked about. Maybe next month.

[soy]

If I'm thinking right, these Avalons have separate 12V rails per board so if necessary you can run them off two PSUs, probably want at least 600-650W rated each. But a decent 650W PSU to run on 120V shouldn't be hard to find.

Yes that would be good.

Yes efficiency would be better off two different PSUs. If you can get your load on each PSU to around 50% that would be most efficient (maybe not on your wallet).
Graph below is for the PSU I plan on using:

If I want to get more efficiency I could add another PSU and hit right at the 50% mark or just run on 1 and still be fine. Thats why I chose the PSU i mentioned earlier. Of course you guys with 240V can just laugh and enjoy you're superior server PSU efficiency.

You guys with good power supplies would know.  What power factor does a Kill-A-Watt give for a good supply?  Personally I've always gone with cheap LED supplies then added a heat sink to the bridge rectifier, changed out the single turn trim pot for a 20 turn, improved the thermal paste on the power transistors then replacing the fans when they get noisy.  Also I add 3½ LEDs measuring the voltage at the PCIE connector.  With a C1 I use four 360 watt 12v LED supplies, one on each blade and a fifth to power the radiator fans and pump.  While I was considering the buy of an Avalon6 I was considering buying four new 5v or 6v 360 watt power supplies for under $30 each, putting a digital readout on each and again at one of the two PCIE connectors on each blade, putting two in series for each blade.  Putting 12v supplies in parallel burns them up.  So, with two of these improved cheap LED 12v supplies on an S3+ right here that I have a Kill-A-Watt monitoring, I get 120.2VAC, 0.70 power factor, 59.9~/s, 554VA, 388 watts, 4.62 amps.  So, I figure the losses using cheap supplies would show up in the power factor figure and wattage.  Can someone tell me how this compares with quality supplies?  Thanks.

That is a ton of work for little savings, the server PSU's like the 1000w and 1200w can be had for $40 or less.  Simply jump the "on" with a resistor and solder on some wires and done.  I have a few I did recently and can post it on the PSU thread here if you want.  Also the 1200w has a Pot to adjust voltage and amperage (I have a thread on that as well). Here are the links to the 2 threads:
PSU discussion:
https://bitcointalk.org/index.php?topic=1284563.0
PSU Voltage adjustment (by yours truly   )
https://bitcointalk.org/index.php?topic=1294058.0

Yes I know.  I owned a KNC Mercury back in the day and they were having Jupiter users with burned up supplies.  They were jumping the connector.  Those supplies were designed for motherboards not miners.  The respective voltage loads are way different.  I suspect some supply manufacturers were using a low voltage rail for the maximum current ramp circuit as the low voltage would reach its limit quickest but jumping the connector and never putting a load on the low voltage rail, when a user would turn off the miner, the supply's low voltage rail wasn't brought to ground quickly enough then the user turned the Jupiter back on while the maximum current ramp was still high and bam the voltage spiked blowing the capacitors on the miner board sometimes causing the board to burn.  

By the way the solution there I think was the pre-fab jumpers with a resistor to the low voltage rail allowing it to be drawn down to ground when the miner was shut down.  They sell those on Ebay.

[soy]

Power factor should be as close to 1 as possible. A good server PSU will have active PFC on the input, which typically yields a PF of 0.97 or better. 0.7 is pretty bad and if you run too much power at that level your utility provider will hate you. My bill goes up 1% for every 1% my PF is below 90% (averaged for the whole shop).

I have some Avalon6 already in hosting; the most recent ones were put with a Dell 750W server PSU per blade. They're running just fine far as I can tell.

I wish I had stock right now, but we won't for another month or I'd probably offer a discount deal for participation within this group buy. We're working on an interface board that's compatible with DPS800 and DPS1200 edge-connector server PSUs, probably the ones kilo just talked about. Maybe next month.

Wow, I never thought a Kill-A-Watt could read a PF that high except on an incandescent light bulb!  Is .97 PF an actual Kill-A-Watt reading on a bitcoin miner?

[soy]

I have some Avalon6 already in hosting; the most recent ones were put with a Dell 750W server PSU per blade. They're running just fine far as I can tell.

I have half an S3+ running on a Dell PowerEdge 2950 750W supply right now, the other blade went bad.  I'll move the Kill-A-Watt over and get back with a reading.

[sidehack]

With active PFC, yeah. It actually uses a boost converter off rectified mains and constantly adjusts the current draw to be proportional to voltage draw, scaled against the power draw required to maintain output regulation. You get a bit of HF component and the nonlinearity of your rectifiers but otherwise it's pretty darn good (and that's not hard to filter out). I haven't measured directly but if you can find the 80PLUS sheets on 'em they'll have the expected power factor listed, and all the ones I've seen are well over 0.9. A regular passive PFC is cheaper and less complex but for any kind of high-power load will never get anywhere near what an active PFC can do. That's also why server PSUs (and high-end ATX) don't have a 120/240 switch like a lot of old ATX supplies, and contributes to increased efficiency off higher input voltage; whatever the input voltage, it all goes to the same high-volt DC bus by merit of the boost converter.

[soy]

I have some Avalon6 already in hosting; the most recent ones were put with a Dell 750W server PSU per blade. They're running just fine far as I can tell.

I have half an S3+ running on a Dell PowerEdge 2950 750W supply right now, the other blade went bad.  I'll move the Kill-A-Watt over and get back with a reading.

You are absolutely right.  The Dell PowerEdge is registering (and the miner is on a 12 gauge line I ran for miners closer to the breaker box) 122.1VAC, .98PF, ~59.9/s, 248VA, 244 watts, 2.04 amps.  So, if it had both hashing boards the current would be in the 4.08 ballpark compared to the cheap supplies 4.62 amps.  Cheap supplies are running at 113% of the Dell server supply.  Wow.  Thanks for point it out.

[soy]

Power factor should be as close to 1 as possible. A good server PSU will have active PFC on the input, which typically yields a PF of 0.97 or better. 0.7 is pretty bad and if you run too much power at that level your utility provider will hate you. My bill goes up 1% for every 1% my PF is below 90% (averaged for the whole shop).

I have some Avalon6 already in hosting; the most recent ones were put with a Dell 750W server PSU per blade. They're running just fine far as I can tell.

I wish I had stock right now, but we won't for another month or I'd probably offer a discount deal for participation within this group buy. We're working on an interface board that's compatible with DPS800 and DPS1200 edge-connector server PSUs, probably the ones kilo just talked about. Maybe next month.

When you do please PM me and I'll buy.  For the PowerEdge I hacked a couple of terminal lugs in place but can't really trust them.

[pikachuy]

Power factor should be as close to 1 as possible. A good server PSU will have active PFC on the input, which typically yields a PF of 0.97 or better. 0.7 is pretty bad and if you run too much power at that level your utility provider will hate you. My bill goes up 1% for every 1% my PF is below 90% (averaged for the whole shop).

I have some Avalon6 already in hosting; the most recent ones were put with a Dell 750W server PSU per blade. They're running just fine far as I can tell.

I wish I had stock right now, but we won't for another month or I'd probably offer a discount deal for participation within this group buy. We're working on an interface board that's compatible with DPS800 and DPS1200 edge-connector server PSUs, probably the ones kilo just talked about. Maybe next month.

Hi sidehack, you talking about something like this? www.ebay.com/itm/291653035822 Found it while browsing around for my DPS1200. If you have the same capability, then I rather buy from a fellow member =)

[KeezAZA]

Phil,

Valkir will be in touch for our Avalon. He said he would talk to you tonight.

Thank you again !!

Done I contacted Phil.

Kilo17 I dont know how much Phil wanted himself but you can had 2 to it. 1 for me 1 for herbpean.

  Phil will confirm.

yes  Valkir and HerbPean  will have their 2 shipped to my house in NJ  and I will get them up to Montreal

So that is 2

I have decided on 2   so 2+2 = 4 to my home in NJ

I have some confusion on shipping  are all USA orders going to be dropped shipped to kilo17?

 at 1112usd   then you ship them out?  Would  a secondary shipping be needed?


I get that Valkir and HerbPean  are in Montreal  and I am going to charge them as little as possible to get the two from NJ to Can.

@ the guy in South Africa  KeezAza    if I get a fifth I could ship it to you  but I need to see how much extra it will be. 1112 + ?

Fast check with us post office if I send it with 300 insurance priority express it is 95 usd to ship


so you would be 1112 + 95 = 1207 usd

Thank you for the offer.
I am going to have to give this offer a miss due to RSA weak currancy right now. I was hopeing it would strengthing to R15.80/ $, but instead it chooses to stay @ R16.60/$.

[soy]

I have some Avalon6 already in hosting; the most recent ones were put with a Dell 750W server PSU per blade. They're running just fine far as I can tell.

I have half an S3+ running on a Dell PowerEdge 2950 750W supply right now, the other blade went bad.  I'll move the Kill-A-Watt over and get back with a reading.

You are absolutely right.  The Dell PowerEdge is registering (and the miner is on a 12 gauge line I ran for miners closer to the breaker box) 122.1VAC, .98PF, ~59.9/s, 248VA, 244 watts, 2.04 amps.  So, if it had both hashing boards the current would be in the 4.08 ballpark compared to the cheap supplies 4.62 amps.  Cheap supplies are running at 113% of the Dell server supply.  Wow.  Thanks for point it out.

Wait, I'm rethinking this.  I made an error looking at the current of course as it's out of phase with the voltage.  I've gone to the Kill-A-Watt dollars per day running cost and am waiting for numbers....

I've doubled the S3+ with the Dell PowerEdge cost per day as it's only running one blade - not accurate I know as the other doesn't use two controller cards but I'm only counting one fan on the S3+ half.  Now, cost per day running the full S3+ getting 457GH/s avg times 100 divided by the cost per day of the ½ S3+ getting 225GH/s avg with the Dell at .98pf is 79%.  So, the cheap supplies at pf .7 according to the Kill-A-Watt is more efficient by 11%.  A true test would be to swap out the Dell PowerEdge for a cheap supply and see what it costs.

[philipma1957]

Okay At kilo17   lets say I collect btc for 10 miners  or 25btc

I get all mailing info  etc.

And I am ready to send the 25 btc   .. 

I do not have a btc address to send to.


So if you pm me with details  of the btc address I need to vet it to some degree.

I want to be sure I am sending 25 btc or whatever we collect to the correct btc addy.

[sabercatt]

If I buy 4 or 5, can they be sent straight to sidehack for hosting?

[pikachuy]

Should post all the ID's and BTC address here, so we all can see, for transparency reasons. That way, other's can confirm and view the record of the transaction if they wanted too so there's nothing to hide.

[soy]

I have some Avalon6 already in hosting; the most recent ones were put with a Dell 750W server PSU per blade. They're running just fine far as I can tell.

I have half an S3+ running on a Dell PowerEdge 2950 750W supply right now, the other blade went bad.  I'll move the Kill-A-Watt over and get back with a reading.

You are absolutely right.  The Dell PowerEdge is registering (and the miner is on a 12 gauge line I ran for miners closer to the breaker box) 122.1VAC, .98PF, ~59.9/s, 248VA, 244 watts, 2.04 amps.  So, if it had both hashing boards the current would be in the 4.08 ballpark compared to the cheap supplies 4.62 amps.  Cheap supplies are running at 113% of the Dell server supply.  Wow.  Thanks for point it out.

Wait, I'm rethinking this.  I made an error looking at the current of course as it's out of phase with the voltage.  I've gone to the Kill-A-Watt dollars per day running cost and am waiting for numbers....

I've doubled the S3+ with the Dell PowerEdge cost per day as it's only running one blade - not accurate I know as the other doesn't use two controller cards but I'm only counting one fan on the S3+ half.  Now, cost per day running the full S3+ getting 457GH/s avg times 100 divided by the cost per day of the ½ S3+ getting 225GH/s avg with the Dell at .98pf is 79%.  So, the cheap supplies at pf .7 according to the Kill-A-Watt is more efficient by 11%.  A true test would be to swap out the Dell PowerEdge for a cheap supply and see what it costs.

I remember now.  When I got the Dell I did a Kill-A-Watt test and it didn't power as cheaply as my modified cheap supplies.  The difficulty is the question if the power meter correctly calculates the power factor or not.  If not then the cheap supplies would be worse.