4-in-1 Sata Power with Capacitor - Thoughts?

[landS]

Silverstone Tek Four-in-One SATA Power Connectors with Power Stabilizing Capacitors (CP06)

$11

http://www.amazon.com/Silverstone-Tek-Connectors-Stabilizing-CP06/dp/B005DD28XG/ref=pd_ys_sf_s_172282_a1_2_p

 

Includes two 2200?F capacitors for stabilizing +12V and +5V voltages

Compatible with most cases with 43mm gap between each SATA connectors

 

My Understanding is that one should only use a single power cable for up to 4 Hard Drives (I believe I read this on a Norco Site at one point).

 

However, most PSU cables that I have used have far too much slack between cables.  Some of the cords have 4 connectors, others have a lesser amount.

 

This looks like a cheap way to manage the power connectors for those of us not using backplates.

 

Anyone have any experience using these?

Are the line stabilizing caps just a gimmick?

 

Cheers

[garycase]

A bit of transient-suppression certainly won't hurt anything, so whether they're needed or not, the capacitors aren't a bad idea (especially with modern single-rail PSUs).

 

[electron286]

  Cool!  I think I will order some of those!  I often need additional drive power connectors, and this looks like the best and possibly the easiest and least cluttered way to do it!  :-)

 

  The capacitors should also help a bit with a small/marginally sized power supply when all drives are spun up at the same time, like when shutting down an array before a server power down.

 

  Although not intended to do so... I would think...  The capacitors also could have the effect of masking a marginal power supply... making it very usable... or as it is failing make it seem like it is ok...

 

 

[garycase]

The capacitors aren't going to make a power supply "better" ... if it doesn't have ample power, it still won't.    What they WILL do is suppress transients that may be a result of inadequate filtering; or caused by multiple drives "pulling" peak current at the same time (thus causing spikes in demand, which can cause corresponding blips in the PSU voltage).

 

[jumperalex]

The capacitors aren't going to make a power supply "better" ... if it doesn't have ample power, it still won't.    What they WILL do is suppress transients that may be a result of inadequate filtering; or caused by multiple drives "pulling" peak current at the same time (thus causing spikes in demand, which can cause corresponding blips in the PSU voltage).

 

In fact I would  argue they are too small to do even that.  They will cleanup transient NOISE  but not dips caused by spin up of other drives and a too small psu.  That would require a much larger cap than can fit in that connector.  They are filter caps not power caps.

 

Sent from my Nexus 7 using Tapatalk HD

[electron286]

  ok, I will admit I probably got too excited there...  The manufactures web site does say that there are 2 2200µF capacitors though... if correct, it should help a bit...

 

  Yes there are caps rated that high that would fit in the seemingly too small connector.  There are two types of specialized caps, that are often used for memory back-up purposes in low power memory applications.  I have a LOT of 1 FARAD caps, that are only rated at 5.5 volts.  And they are very small indeed!  They are also lots of fun to play with!  :-)

[landS]

Ordered 2

[jumperalex]

Alibi: My current assignment is what would call, at best, not intellectually challenging.  So this just seemed like a fun little thought and google experiment.  Been a few years since i had to even think about basic EE equations.

 

So some googling to refresh my memory and finally to find someone who did all the math for me ... http://forum.allaboutcircuits.com/showthread.php?t=71676

 

C = I * (delta T) / (delta V) which rearranged is  (delta T) = C * (delta V) / I

 

C = capacitance

I = current

T = time

V = volts

 

If you already knew all that, don't take it personally please

 

Right so these two caps, on the 5v and 12v lines are 2.2X10^-6F and they are charged to 5v and 12v respectively max because that is what is on the lines.  Based on a max 10% voltage drop we get delta V of 0.5v and 1.2v respectively.

 

Now lets assume, reasonably, that a "green" HDD will pull 2.25W on the 5v line and 4.5W on the 12v line.  Those equate 0.45A and 0.375A respectively.

 

So how long can a capacitor provide that power?

 

0.0000022 * 0.5 / 0.45  = 0.00000244 = 2.44 x 10^-6 = 2.44 uSec

0.0000022 * 1.2 / 0.375 = 0.00000704 = 7.04 x 10^-6 = 7.04 uSec

 

Now I'm not gonna take any solid guesses at how long a voltage sag might last during the spin up, but I bet it is longer than 2.44uSec.  I also know from more googling that there is a bit more to it since the capacitors aren't having to carry the entire load BUT the sensitivity is not multiple orders of magnitude. 

 

So I'm still standing by this is only for line smoothing and not making up for voltage sag from a marginal PSU.  Which is GREAT mind you ... but i'd also rather buy a good power supply in the first place.  Finally, I ,ight get these myself just for the better wire management anyway

[S80_UK]

 

Right so these two caps, on the 5v and 12v lines are 2.2X10^-6F and they are charged to 5v and 12v respectively max because that is what is on the lines.  Based on a max 10% voltage drop we get delta V of 0.5v and 1.2v respectively.

 

Now lets assume, reasonably, that a "green" HDD will pull 2.25W on the 5v line and 4.5W on the 12v line.  Those equate 0.45A and 0.375A respectively.

 

So how long can a capacitor provide that power?

 

0.0000022 * 0.5 / 0.45  = 0.00000244 = 2.44 x 10^-6 = 2.44 uSec

0.0000022 * 1.2 / 0.375 = 0.00000704 = 7.04 x 10^-6 = 7.04 uSec

 

Now I'm not gonna take any solid guesses at how long a voltage sag might last during the spin up, but I bet it is longer than 2.44uSec.  I also know from more googling that there is a bit more to it since the capacitors aren't having to carry the entire load BUT the sensitivity is not multiple orders of magnitude. 

 

So I'm still standing by this is only for line smoothing and not making up for voltage sag from a marginal PSU.  Which is GREAT mind you ... but i'd also rather buy a good power supply in the first place.  Finally, I ,ight get these myself just for the better wire management anyway

 

But the caps are not 0.0000022F.  They are quoted as being 2200uF (micro-farads), or 0.0022F.  Your result is therefore 1000 times too small, so the hold-up times should be milli-seconds rather than micro-seconds.  If the values are as the manufacturer claims then I would say that the additional filtering could be helpful. 

 

What I do not like about this product would be the way that the input power connector appears not to have any means of retention, thus introducing a significant measure of unreliability into the system.  I would also question whether the SATA power connector (as used for the power input) was ever designed to carry the current for four drives.  That's a further reason for possible problems arising after installation.

[garycase]

Ah gee, what's a mere factor of 1000 among friends 

 

Kinda like our Congress (in the US) discussing budgets -- million, billion, what's the difference 

[jumperalex]

Yup you are 1000x right   I entered 2.2x10^-6 in my sheet and not 2200.

 

That said, even talking milliseconds and not microseconds that still makes these noise/ripple filters and not sag support.  Remember all I'm disagreeing with is their utility to help with a marginal PSU.

[electron286]

... sorry about the marginal power supply thread invasion I seem to have created... :-)

 

  Please skip over reading the rest of this post... if not interested in my thoughts on marginal component testing...  :-)

 

  I have played around a bit with old and what I would definitely call marginal components with unRAID for fault failure mode testing and worst case data loss and recovery scenarios...  One of my fonder test machines has a power supply that works great with an array of 6 older 400GB SATA data drives and 1 500GB parity drive.  It never has any problems with monthly parity checks, or anything OTHER than if you want to shut down the server... if you attempt to stop the array when all the drives are spun down... the server reboots because of the loading on the +12V line when all the drives spin up together...  This was NOT a one time fluke, but was very repeatable, and I never was able to manually spin up all drives at the same time without forcing a re-boot.  (I always wondered why the monthly parity check process NEVER seemed to cause the same condition however...)

 

  Adding a 1000uf capacitor to the +12 volt rail externally on one of the 4-pin molex power cables coming from the power supply, was enough to make it work properly.  I have not had any conditions of the system re-booting for any reason since.

 

  I also was not meaning that these cool little power cables, which I am mostly attracted to for cable management purposes also, should be considered a fix for a marginal power supply, or to make a bad power supply better.  I was actually more thinking that if a person were to use these, that it is quite conceivable that as a power supply degrades normally from aging, that the added capacitors in the cables could mask, for a time, the failing power supply.  Making in at least some situations, the power supply usable for a longer life span.

 

  Years ago... back in the older AT (pre ATX) days.  It was not all that uncommon to add external capacitors to power supplies to increase stability.  Of course switching power supply designs themselves have made major leaps forward in how they are designed, and the components that are now used in such power supplies have seen similar improvements.

 

  On another more recent test machine, one with very bloated CPU and support chip capacitors on the motherboard, (old P4 machine), I had to test 3 different power supplies before finding one that had a stable enough output to allow the machine to post... it would then operate with no problems under various endurance test benchmarks.  A 4th tested power supply would allow the machine to post, but would not allow full operations, without an eventual blue screen in Windows.  Adding an external 2200 uf Capacitor to the 12 volt line again made this test configuration stable.  Note that all 4 of these power supplies work fine on all other GOOD systems I have used them on.  This has allowed me to rank these test power supplies for testing for marginal motherboards in the future...  Just because a new power supply works on an old computer, it does not mean that there may not also be a problem with aged components on a motherboard...

 

 

[garycase]

I'm not surprised a 1000?F capacitor stabilized your start-up surge.    While a few milliseconds doesn't seem like a long time, it IS a long time in the world of surges caused by reactive components.    And a 2200?F capacitor will do an even better job.

 

Certainly it's not a fix for marginal supplies ... but a supply that's not really "marginal", but perhaps struggles with the combined startup load of drives that are spinning up should be much more stable with these.    I suppose any supply that has that issue is "marginal" to some  extent => but with modern relatively low-power systems, you SHOULD use as small a PSU as makes sense, so you're not running too far below the 80+ % efficiency range, which starts at 20%.

 

So even a 350w unit is running at less than optimal efficiency when the system's drawing less than 70 watts -- not unusual for many boards today [My system idles at 20 watts].

 

One feature I think would be a BIG improvement in UnRAID is staggered spinup under ALL conditions ... i.e. a 1/4 or 1/2 second timer after a startup that prohibits further startups during that time.

 

[electron286]

Staggered spin-up... I have wished for that for years...  :-)  It has also been discussed MANY times in the forums.  I would also like to see the timer delay you mentioned, be a USER configurable parameter!  :-)  It would also probably make sense to use it in conjunction with the spin-up groups in some manner... so a user could selectively spin-up groups in priority, or by drive/directory access priorities...  while still having staggered spin-ups.  possibly a choice of 1, 2, or 3 drives at a time spin-up options...  (I would think 4 drive MAX simultaneous spin up would be a good limit, if it was user configurable.)

[garycase]

Agree -- if the user could set (a) the timer value (in perhaps 1/4 sec increments); and (b) the max number of simultaneous spin-ups;  then we could use MUCH lower capacity PSU's that would let us run them at least closer to their 80+ efficiency range.

 

Folks buying 650w and 750w PSUs for UnRAID boxes think they're getting great efficiency, since they tend to be 80+ Gold or Platinum units ... but if they're drawing less than 130W (650w unit) or 150w (750w unit) they're running far below 80% efficiency => the efficiency generally drops off rapidly below the 20% threshold.    A staggered spinup would let most folks use 350w or 450w units, which would be much more efficient at typical operating draws.

 

[landS]

I am running a Preclear on 4 disks right now (4 TB Seagate  STCA4000100) on a X7SPA-HF-D525 mobo.  This is pulling 45w on my APC SMC1500.  This places my  SeaSonic X Series X650 Gold in a very inefficient mode. 

 

While a little bit of noise filtration would be nice, I am more interested in the wire management these offer.  The SeaSonic comes with 2 Sata cable sets -0 1 with 2 ports and 1 with 4 ports.  However, I JUST discovered taht SeaSonic DOES sell the 4 port cables separately and this PSU has room for a bunch of em.

 

At this point, I would be most workied about S80_UK's points:

"What I do not like about this product would be the way that the input power connector appears not to have any means of retention, thus introducing a significant measure of unreliability into the system.  I would also question whether the SATA power connector (as used for the power input) was ever designed to carry the current for four drives.  That's a further reason for possible problems arising after installation." 

 

 

 

 

[jumperalex]

well my math, suspect as it was, didn't deter me.  i just got my cable from Amazon Prime.  I like it.  Haven't install it yet but a test fitting tells me both the head end and the drive ends will hold very tight. also for anyone really worried about the head end coming lose, there is plenty of room to use a small zip tie to secure it to the PSU connector.  i'll add a picture of my effort in a bit.

[jumperalex]

Oh and as for current carriying ... well what i wold NOT do is add several of these connectors to a single PSU line.  so say a PSU line has 3, or even 4, SATA connectors, don't go adding to each one of those for upwards of 12 drives on a single line.  That said, consider that this PSU http://www.newegg.com/Product/Product.aspx?Item=N82E16817139010 offers 4 sata connectors on a single line.  so at least as that is designed, with the gauge wire used, they seem to think it is ok.  don't forget internally they all connect to the same spot (assuming single rail).

[garycase]

This is pulling 45w on my APC SMC1500.  This places my  SeaSonic X Series X650 Gold in a very inefficient mode. 

 

Why did you use a 650 on an Atom-based system ??

 

Even my 400w unit is running way below its efficiency range with my D525 setup ... but there's really nothing available in a high-quality 80+ PSU that's in a range appropriate for these Atom-based servers ==> AND the fact that UnRAID doesn't do staggered spinup means you need a PSU with enough power to handle spinup loads, so there's a real Catch-22 with regard to efficiency vs. suitable load.    I've toyed with using a PicoPSU unit;  but don't think it would handle the 6-drive spinup load, so I'm sure I'll just stay with what I have.

 

Can't be all-that inefficient anyway ... my system idles at 20w and never draws over 45w (per my Kill-a-Watt) ... so even if it's only operating at 60-65% efficiency, it's only wasting 3-5 watts.

 

[jumperalex]

Before

[garycase]

Is there an "After" picture coming ??

 

[jumperalex]

After.  Note that my drives are closer together than planned for the connector.  so it took some work to make the short wires bend to fit.  i had to do the same thing with the PSU lead as you can see, but that was easier because they were longer.  of course that also made it less clean [shrug]. 

 

in the end, i think when it comes time for me to add another (with 4 more drives), i might not because the large head connector will be right in the middle of all the drives, and since most PSU leads have multiple connectors, a few unused ones that are connected to the second 4-in-1 will be sitting around making it a mess.  though some creative cable management around the "back" of the case might make it neater.

[garycase]

Still looks a good bit neater than in the "Before".    I suppose they need to allow a bit of slack to ensure the connectors work with various cases with slightly different drive spacing (although I'd think that was pretty standard).

 

[jumperalex]

Yeah it does, but notice i don't have the sata cables in place in the after.  My main worry is they put stress on the drive's power connector.  again, i had the same concern with the PSU line, but it was easier to force them to conform.  But so far so good and you're right, it does still look a little better.  For whatever that is really worth

[landS]

Why did you use a 650 on an Atom-based system ??

[...]

Can't be all-that inefficient anyway ... my system idles at 20w and never draws over 45w (per my Kill-a-Watt) ... so even if it's only operating at 60-65% efficiency, it's only wasting 3-5 watts.

 

The 650 was destined for a build that did not happen - so I had a brand new PSU sitting at home unused.  It was either this PSU or an older Enermax Revolution!  Originally I had planned on building the server out all recycled equipment.  During the first preclear the thrice recyled biostar video died and hence the low power draw build was born... I wanted it to be cool running (or able to take heat), quite, and highly reliable. 

 

3-5 watts waste is something I can certainly live with