Build Report: Lian Li A76-PC

[Constantin]

Fresh on the heels of my fiasco with the Norco RPC-431, I am happy to add another build report, this time for a Lian Li A76-PC enclosure. The Lian Li is a much larger enclosure and constructed almost entirely of anodized brushed Aluminum. Thus, it's weight is not that different from the Norco RPC-431.

The Lian Li A76-PC features twelve internal 3.5" / 2.5" slots that are nicely separated. Up top, there are two 5.25" drive slots, lots and lots of PCIe expansion slots, and oodles of room to work in.

Up front, three removable 140mm fans provide cooling for the hard drives. Since hard drive and CPU temperature don't necessarily have anything in common with each other in a FreeNAS setup, I opted to add a NZXT Sentry 3 fan speed controller for the three fans up front. While not perfect, it does a decent job monitoring drive temperatures and adjusting fan speed as they heat up.



I replaced the OEM rear 120mm fan with one of the Noctua 120mm PWM models I had from my other build, then added another 120mm model to blow directly onto the CPU / memory section of the motherboard. Another 80mm fan tries to keep the heat sink on the 10GBe NIC happy. Thus, quite a few fans in there. The picture below shows a 80mm fan blowing on the CPU, I'll update it when I get a chance.



I decided to separate the power and data cables as much as possible. Thus, I used the PC card support collumn in the middle of the case as a runner for the power, front panel, etc. cables. Re-usable velcro zip ties keep them in place. Meanwhile, the data cables were physically segregated as much as possible from power lines and either run across the mounting plate in the back or near the mounting plate. The SATADOM was moved to one of the leftover 150MB/s SATA ports as it reduced the need for long cables to reach the back edge of the motherboard.



Through this careful cable management I was able to easily open and close the side panel covering the backside of the motherboard. Two standard SATA power splitters cover all the bases with the help of the optional Lian Li backplanes. I needed to buy 3 additional backplanes, as the case came with just one. But at least they are available. Note how I kept power and data as separated as possible, with the power lines descending and then plugging right into the Seasonic power supply.

Hard drive temperatures are now in the 33-35*C range. The CPU is around 31*C despite the incoming air washing first over the hard drives. I left the two top slots open - one could accommodate another 3.5" drive (SATA DATA connected) but the top most drive cannot be connected to anything with this motherboard due to lack of free SATA ports. The SLOG is on a SATA3 bus, I figured it would be the only drive that really benefits from a fast interface.

I did cover up a number of PCIe slot holes and other case holes to force as much air as possible through the one fan in the back. I did leave the two PCIe slots open adjacent to the 10GBe card, to allow air to wash over the heat sink on the card.

It is surprising how non-adjustable the "Smart Fan" control is in the ASROCK C2750D4I Bios. While you can select stuff, the minimum CPU temperature for the fan to do anything is 35*C. I suppose I could simply keep up the minimum fan speed a bit and / or fit a small air diverter to direct all air flow over the heat sink.

All in all, this looks like a very happy setup. The hard drive temperatures are close to ambient, the fan noise is quieter than the hard drives and the case doesn't exhibit the rattling that others have complained about. That may have to do with me removing the useless door lock. More pictures to follow. But for anyone considering a 11-drive Z3 setup, this is a pretty ideal case as long as you can stomach the rather large case dimensions.

At around $160 over at Newegg, this case is a bargain compared to the Norco RPC-431 as long as you value your time (and hair). Everything fits, Lian Li ships nicely packaged mounting accessories, oodles of fasteners, even a multilingual manual.

The only obvious loss compared to the case that the Mini XL uses is the lack of individual disk activity lights for the hard drives and the lack of a hot-swap system. Since the Lian Li is destined to service nothing but my home, I am OK with that. After all, even the 'hot swap' sleds in the Mini Xl required multiple fasteners per drive. Both the drives and the CPU should remain much cooler inside the Lin Li, achieving my design goal.

 

[Constantin]

Ran a scrub and a SMART test. Note how the CPU temperature remains steady despite running at 100% capacity.


Next, I will have to relocate the thermal probe of the Sentry 3 to be under the circuit board of a mid-stack hard drive since that's apparently the hottest part of a hard drive. Baby steps...

 

[Constantin]

Decided to have a look with a new resource, a FLIR One Pro. Taking off the side of the case, you can immediately spot the usual suspects: the motherboard with its comparatively 'hot' CPU as well as the drives themselves. Of course, one has to be somewhat cautious with what the camera is reporting as the emissivity of all materials is not constant. But, the power supply (image to right) shows that it is remarkably cool.



The circuit board has a bit of a hood built over it, directing air flow from the 120mm blower over the memory sticks and the CPU heat sink. Neat how the MSX mode melds stuff even if there are some pretty obvious parallax issues with the dual lenses of the camera module. When I discovered that the main chip on the 10GBe card was allegedly over 50*C (at bottom of picture), I re-positioned the smaller 80mm fan to blow more directly at that heat sink. It then allegedly cooled down to the 40's.



The picture below is a close up of the motherboard, showing the section not covered by the fan hood. Again parallax issues but the hot spots roughly correspond with nearby chips.



And now for the hard drive stack. Same parallax issues at close range but a story that is consistent with the SMART data that the drives put out. Interestingly, the 7K4000-series drive with a P8Hxxxx S/N puts out far more heat than the other P8Gxxxx and P9xxxx drives I have in that array. It's the hot spot near the bottom. I wonder if this is a one-off issue or part and parcel of that batch of drives? Or a sign of impending drive failure?



Per the camera and the SMART data, the drive with the P8Hxxxx serial number runs 5*C hotter than the other drives in the array.

 

[Chris Moore]

Per the camera and the SMART data, the drive with the P8Hxxxx serial number runs 5*C hotter than the other drives in the array.

I found about that much variance between a hot drive and cool drive even in drives from the same batch. It could be some variation in the exact nature of the airflow around that drive or it could be some minor variation in the production process. I have not seen a direct correlation between a drive that runs a couple degrees hotter and early drive failure.

PS. This is both in my experience at home (28 drives between two servers) and my experience at work where I maintain many racks full of servers filled with hundreds of drives.

 

[Constantin]

Thanks so much for the info. I won't worry about it too much other than re-positioning the drive to a "cold" spot (i.e. the spot right above the SLOG SSD).

Speaking of failures, had an interesting one today. A simple, unmanaged network switch (GS116 - v1) gave up the ghost. When I took it apart, it appears that at least one Al electrolytic cap has decided to bulge itself into non-existence. Have two more such switches at that site and will have to inspect them also...

Funny to think that way back when such bulging caps were the inception for my personal web site; teaching anyone interested how to repair the blown Lelon caps in their first gen Apple Airport base stations. Funny where life can take you and how these issues keep cropping up.

 

[Chris Moore]

Are you going to fix it?

Sent from my SAMSUNG-SGH-I537 using Tapatalk

 

[Constantin]

Sure! Should be easy enough since it's a through-hole capacitor. Found the dimensions of the KZG series, it's 8x11.5mm. Ebay sells them for $0.40 ea, delivered. So an exact replacement it is...



The sick patient is on the right, a similar, non-bulging cap is on the left. Since the cost-cutters at Netgear left several designated cap spots open on the PCB, I'll likely fill one or two of them with the other caps I'll be getting (minimum order = 10). That should reduce the stress somewhat.

 

[joeinaz]

Thanks for the review. Newegg currently has the Lian Li PC-A76 on sale for about $117 after promo code. One review of this case suggested there may be an issue with case vibration once disk drives are added. Did you have and issues with vibrations with your case?

 

[Constantin]

There is the occasional hum but it's not hard drive related - they are silent since they "float" on silicone rubber mounts. Most likely a fan is the source of the noise that the reviewer was objecting to.

Based on my limited experience, the silly lock on the front of the case is one likely culprit. Not sure what purpose it's supposed to serve, I removed mine. My case is virtually silent, the loudest thing is the head moving sounds the drives make when they all move to add data. The fans are virtually silent (Noctua PWMs, plus the Sentry controller for the front fans) as is the power supply. Properly routing and securing wires likely helps a lot.

Your price is 50% off my NewEgg price, so fortune smiles upon you. If you need storage for 12 drives, that's a great deal. I recently got my hands on a Q26 so I will compare and contrast the Q26 with the PC-A76 and the mini XL case. I wonder if the Q26 would have been a better case than the Mini XL case iXysystems chose?

 

[joeinaz]

My Lian PC-A76 should arrive tomorrow. I originally purchased a Rosewill Blackhawk Ultra. It holds ten 3.5" disks internally and has four 5.25" disk bays. When the PC-A76 became available at a good price, I bought it with the intention of returning the Blackhawk. When I look at both cases the each have the following advantages:

Lian PC-A76: 12 internal 3.5" disks, SATA backplanes, numerous fan filters and a weight of 20 pounds. (versus 36 lbs for the Blackhawk)
Blackhawk : 8 internal fans, four 5.25" bays, and wheels.

My final build will be a dual socket Xeon system with 2 E5640 CPUs, a minimum of 32GB of RAM, 4 PCIe cards and an internal LTO tape drive. I assume the Liam would be able to cool my system with the number of available fans. FreeNAS will be the base O/S with Windows Server and Linux also running as virtual machines.

I may attempt to do some testing this weekend but I am torn as to which case to keep...

 

[Constantin]

Are you going to fix it?

The parts for the GS116 v1 finally came in. Last night, it was time to replace the blown capacitor and fill the two empty spots on the PCB with like capacitors. Replacing the original capacitor was relatively easy, getting the RoHS solder out of the filled vias in the empty spots was relatively difficult by comparison. I ended up using quite a bit of chipquick to coax the solder out of the filled vias.



The image above shows the switch with the cover off (you only have to remove two small screws, no clips, etc!). The heat sink covers the two ASICs or whatever Netgear uses to deal with network traffic. The sick capacitor was in the lower left section of the PCB. The following two images show the replacement capacitor (left image, rightmost capacitor) as well as the two capacitors I added to the empty spots on the PCB (two through-hole capacitors, brown).


Eventually, I prevailed and better yet, the switch didn't release any blue smoke when I powered it back on. It even works as a network switch! FWIW, the sick capacitor had a capacitance below 350uF, so that would help explain why the power supply reached its limits. The new trio of capacitors should help, particularly if the two new additions are on the same power bus as the old sick one.

 

[wblock]

Sure! Should be easy enough since it's a through-hole capacitor. Found the dimensions of the KZG series, it's 8x11.5mm. Ebay sells them for $0.40 ea, delivered. So an exact replacement it is...

Sorry, I realize this is probably too late to help, but I would not suggest buying replacement capacitors on ebay, many are fakes. Digikey has several lines of Panasonic low-ESR that are really good.

 

[Constantin]

Hi wblock and thanks for the reply.

I don't doubt that Digikey has some good caps, but they didn't seem to have the whole United Chemi-con KZG series in stock. I couldn't find anything equivalent with a 820uf, 6.3V rating, a 8x11mm form factor, and 0.036 Ohm impedance or better at 100kHz that was available for purchase. The only two available products that met these specs were obsolete and discontinued. Maybe I missed something?

FWIW, I did check the capacitance of all the caps I got via eBay and they were all within 5% of spec. Couldn't check the impedance at 100kHz, though. I doubt anyone would play with the voltage rating (can't go lower than 6.3V with a standard through-hole Aluminum cap) leaving the quality of the electrolyte and other like construction details as the most obvious avenue for cheating.

If these capacitors last a while (IIRC, this switch was purchased around 2005) then I'm more than happy. This switch will likely spend much of its time in a box awaiting an opportunity to serve. It's companion is slated for removal as well, also to be replaced with a Mikrotik that also features two SFP+ ports.

 

[Constantin]

Did you have and issues with vibrations with your case?

I spoke too soon. I had my first vibration / hum from the chassis last night. Still have to hunt down the source, expect it's fan related.

These panels could benefit from a very thin layer of foam at mating points.

 

[wblock]

I don't doubt that Digikey has some good caps, but they didn't seem to have the whole United Chemi-con KZG series in stock.

Possibly not. I stuck with only Panasonics unless a close match was not available, for the simple reason that I have never found a bad Panasonic capacitor in any equipment I worked on. Incidentally, capacitance is not nearly as important as ESR for these caps, and that's difficult to test without special equipment.

 

[p1etr0]

The Lian Li A76-PC...

Hi there! I am Pietro from Italy. Sorry if I go a litte OT bothering you about the Lian-Li PC-A76X. I was planning to build a new pc with a PC-A75X, but it went out of stock in every shop, so I turn my choice to 76X. I have to thank you really much. Nowhere I searched if was confirmed that front door is removable (I spot some screw on pivot door), but you did. Luckyly I found this thread on google.

In this review: https://m.hardocp.com/article/2013/01/31/lian_li_pca76_fulltower_case_review/7 is pointed out that vibrations come from panels, that bend and flex too easily.

 

[giovannife]

hello
i'm Giovanni from Italy
i want to buy and build my freenas server
my decision is to buy a lian-li pc a76x and 3 x satabackplane to create a raid zfs1x2 with 8 wd red 3 tb
my question is regards sata backplane
when a hdd fail can i remove the hdd failed without shutdown the server and replace with a new one
someone have tested this swap ?
thanks in advance for your reply

 

[giovannife]

sorry
the sta backplane is this:

Lian Li BP3SATA
X10SLM+-LN4F
LSI MegaRAID SATA / SAS 9260-8i 6Gb/s flash to it mode
intel xen e3-1270
4 x 8gb ecc dimm

 

[p1etr0]

Hello Giovanni,

I do not know if this or another backplane is compatible with pc-a76x. Lian li website and its support contacts are very useless for these kind of informations. If you need, I can send you a scan of the manual (hard to be found on internet) or take some picture of hd cage.

 

[giovannife]

Thank you for your answer
i write an email to the customer support team