callmeedin Posted October 31, 2011 Share Posted October 31, 2011 I am building the 15 HD UCD tower and I had a couple of stupid/simple questions: 1. Why are there two molex power adapters on each SS-500? Do I need to get splitters to plug in both of them or will the unit run with just one molex connected? 2. Am I correct in assuming that it is OK to have all three SS-500 units plugged into the same line coming out of the power unit, since the power supplies listed only have molex adapters on one line? That you for the detailed list of components for the various unRAID configurations. Can't wait to have my first unRAID server up and running later this week. Quote Link to comment
Rajahal Posted October 31, 2011 Share Posted October 31, 2011 1. Why are there two molex power adapters on each SS-500? Do I need to get splitters to plug in both of them or will the unit run with just one molex connected? The molex plugs are for two redundant power supplies. Since you will only be using one power supply, you only need to plug in one molex connector to power all 5 drives. 2. Am I correct in assuming that it is OK to have all three SS-500 units plugged into the same line coming out of the power unit, since the power supplies listed only have molex adapters on one line? Yes, that is fine. Quote Link to comment
ohlwiler Posted October 31, 2011 Share Posted October 31, 2011 2. Am I correct in assuming that it is OK to have all three SS-500 units plugged into the same line coming out of the power unit, since the power supplies listed only have molex adapters on one line? I'm going to have to disagree with Rajahal here. Having 15 disks on a single line most likely made up of 18 gauge wire is well beyond the current capacity of the wire. Beyond the safety aspects, there is a voltage drop consideration. It depends upon the drives and the length of the leads, but I would estimate about a .5 to .75 volt drop on the 12 volt line during disk spin up. I posted some more detailed information here: http://lime-technology.com/forum/index.php?topic=15773.msg146800#msg146800 I would recommend finding a power supply with two strands of Molex connectors. The Corsair TX650 had two strands of Molex connectors. Quote Link to comment
Rajahal Posted October 31, 2011 Share Posted October 31, 2011 2. Am I correct in assuming that it is OK to have all three SS-500 units plugged into the same line coming out of the power unit, since the power supplies listed only have molex adapters on one line? I'm going to have to disagree with Rajahal here. Having 15 disks on a single line most likely made up of 18 gauge wire is well beyond the current capacity of the wire. Beyond the safety aspects, there is a voltage drop consideration. It depends upon the drives and the length of the leads, but I would estimate about a .5 to .75 volt drop on the 12 volt line during disk spin up. I posted some more detailed information here: http://lime-technology.com/forum/index.php?topic=15773.msg146800#msg146800 I would recommend finding a power supply with two strands of Molex connectors. The Corsair TX650 had two strands of Molex connectors. I've built many 15 drive servers based on the Corsair CX500 and CX500 V2 500W power supplies, both of which have only one line of molex connectors. Some have been running for years with no issues. I will look more into the safety concerns you've brought up, but at the moment I don't see any reason why this is a problem. As I understand it, you are claiming that the wires used have lower voltage ratings than the circuitry inside the PSU? I've been working under the impression that you only need to consider the PSU's max amperage on its +12V rail(s). I'm not an EE, I'm just working with what I've read and my personal experience in building countless servers. Quote Link to comment
ohlwiler Posted October 31, 2011 Share Posted October 31, 2011 Voltage drop on yellow wire Strand length = 2 feet Current = 15 drives * 2 Amps/drive = 30 Amps Resistance = 6.385 ohms/1000' / 500 = .01277 Ohms Voltage = Current * Resistance = 30 Amps * .01277 Ohms = .383 Volts Voltage drop on ground wire Strand length = 2 feet Current = 15 drives * 2 Amps/drive (12 Volt) + 15 drives * .8 Amps/drive (5 Volt) = 42 Amps Resistance = .5 (two wires) * 6.385 ohms/1000' / 500 = .00639 Ohms Voltage = Current * Resistance = 42 Amps * .00639 Ohms = .268 Volts Total Voltage drop on 12 Volt line = .383 Volts + .268 Volts = .651 Volts We are seeing 11.35 Volts at the Molex connector if the power supply is really putting out 12 Volts. This neglects the resistance of connectors and the voltage drop from the circuit boards in the drive cages. Quote Link to comment
Rajahal Posted October 31, 2011 Share Posted October 31, 2011 In the resistance calculation, why are the ohms/feet divided by 500? I'm sending off an email to Corsair about this, I'm curious to hear their input on it. Quote Link to comment
jumperalex Posted October 31, 2011 Share Posted October 31, 2011 and why are you using a common ground for 5v and 12v?? Don't the 12v and 5v each have their own individual grounds? Quote Link to comment
ohlwiler Posted November 1, 2011 Share Posted November 1, 2011 and why are you using a common ground for 5v and 12v?? Don't the 12v and 5v each have their own individual grounds? Good question. I wondered about that too, I know at the power supply all the grounds are grouped together. I don't know about at the drive, so I made an assumption. If indeed they are separate then the voltage drop on the 12 volt line is even greater. Quote Link to comment
Rajahal Posted November 1, 2011 Share Posted November 1, 2011 I emailed Corsair tech support and send them ohlwiler's calculations above, and explained that I have been using their CX430 V2 PSU for up to 12 green drives and their CX500 V2 PSU for up to 15 green drives. I asked them if ohlwiler's calculations were correct, and if it was unsafe to power 12-15 drives off a single cable. Here's the response from Corsair: 11.35 is very low for the 12v rail. I recommend you have the PSU replaced. 10/31/2011 6:18:49 PM I think the support rep misinterpreted my original question, so I clarified: Those were just calculations' date=' they were not actual measurements. Are the cables insulated to compensate for the drop in voltage over a length of 2 feet? If the PSU is rated for 12V, then I believe it should be able to provide 12V at the molex connectors, not 11.35V. 10/31/2011 6:25:59 PM[/quote'] The followup response: there shouldn't be any drop in voltages the 12v should supply 12v with a +/- 5% allowance. 10/31/2011 7:00:23 PM Well, 12V x .05 = .6V, so the molex connectors should be delivering between 11.4V and 12.6V. I'm still not quite sure what to make of all this. Quote Link to comment
ohlwiler Posted November 1, 2011 Share Posted November 1, 2011 In the resistance calculation, why are the ohms/feet divided by 500? In the chart I linked to in my other post the resistance was given in ohms/1000 feet. http://www.powerstream.com/Wire_Size.htm The ATX specification states that the power supply voltage must be within 5% of nominal, thus 11.4 to 12.6 volts. Quote Link to comment
ohlwiler Posted November 1, 2011 Share Posted November 1, 2011 Quote from: Corsair Technical Support there shouldn't be any drop in voltages the 12v should supply 12v with a +/- 5% allowance. 10/31/2011 7:00:23 PM I'm sure the Corsair engineers would just cringe at that statement. There was a discussion at Hard OCP about this issue: http://hardforum.com/showthread.php?t=1616682 Quote Link to comment
jumperalex Posted November 1, 2011 Share Posted November 1, 2011 and why are you using a common ground for 5v and 12v?? Don't the 12v and 5v each have their own individual grounds? Good question. I wondered about that too, I know at the power supply all the grounds are grouped together. I don't know about at the drive, so I made an assumption. If indeed they are separate then the voltage drop on the 12 volt line is even greater. greater? I do know there are two ground wires on a molex connector, so if the 12v current has its own ground, then voltage drop is less since there is less current flow. [shrug] either way, the real answer will come from corsair. Quote Link to comment
jumperalex Posted November 1, 2011 Share Posted November 1, 2011 I think the support rep misinterpreted my original question, so I clarified: Are the cables insulated to compensate for the drop in voltage over a length of 2 feet? FYI, the insulation has nothing to do with it. Insulation protects against shorts/arcing between the conductor and something "else" like the chassis. It doesn't protect against voltage drop. The question is if the conductor gauge (thickness) is enough to handle the current without 1) excessive voltage drop, 2) excessive heat. Now the heat ... that is where insulation matters because it needs to handle the heat created. But if we are at the point of worrying about the heat rating of the insulation, then we are already beyond the current capacity of the wire and the associated voltage drop. Easy answer: use a multimeter to test the voltage at the far most molex during spin up of all 15 drives. it won't be perfect since most multimeters aren't usually very fast acting, but it will give you a good idea. Quote Link to comment
ohlwiler Posted November 1, 2011 Share Posted November 1, 2011 then voltage drop is less since there is less current flow Less current flow but double the resistance. Since the 12 volt circuit current is greater than the 5 volt circuit current the 12 volt drop would be greater if the grounds are not shared. either way, the real answer will come from corsair. Actually, the real answer comes from Georg Ohm. http://en.wikipedia.org/wiki/Ohm%27s_law Quote Link to comment
Johnm Posted November 1, 2011 Share Posted November 1, 2011 The phrase for today is "staggered spin-up".... A high quality PSU is designed for start up spikes for short durations. Quote Link to comment
jumperalex Posted November 1, 2011 Share Posted November 1, 2011 IIRC staggarded spinnup only helps during boot, not during array spin-up after spin-down. The issue isn't the PS supplying the spike in power, it is the drives at the end of the line tolerating the voltage drop ... whatever it might actually be. Though Rajahal's experience so far would say anecdotally that it is safe. Quote Link to comment
jumperalex Posted November 1, 2011 Share Posted November 1, 2011 then voltage drop is less since there is less current flow Less current flow but double the resistance. Since the 12 volt circuit current is greater than the 5 volt circuit current the 12 volt drop would be greater if the grounds are not shared. either way, the real answer will come from corsair. Actually, the real answer comes from Georg Ohm. http://en.wikipedia.org/wiki/Ohm%27s_law Ahhh yes, I missed that you did already assume two, shared, ground lines. Should have read closer. Quote Link to comment
jumperalex Posted November 1, 2011 Share Posted November 1, 2011 So isn't the solution here just to get a couple of sata-to-molex adapters. I have one sitting right here so I know it exists. and i just found it at newegg http://www.newegg.com/Product/Product.aspx?Item=N82E16812530007 and http://www.monoprice.com/products/product.asp?c_id=102&cp_id=10226&cs_id=1022604&p_id=7642&seq=1&format=2 or http://www.newegg.com/Product/Product.aspx?Item=N82E16812119262 with molex splitters. makes for less cable i guess. or another http://www.newegg.com/Product/Product.aspx?Item=N82E16812198026&Tpk=CABLE-S15M42 Quote Link to comment
Johnm Posted November 1, 2011 Share Posted November 1, 2011 IIRC staggarded spinnup only helps during boot, not during array spin-up after spin-down. The issue isn't the PS supplying the spike in power, it is the drives at the end of the line tolerating the voltage drop ... whatever it might actually be. Though Rajahal's experience so far would say anecdotally that it is safe. ok, spin-up groups. Most retail servers are seriously under-wired then.. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.