Norco RPC-4020 questions

Speaking of suggestions below is a link to a thread at AVS Norco apparently agreed to monitor re: suggestions for future cases

AVS/Norco: Submit your ideas for the next server case (24 bays, etc)

http://www.avsforum.com/avs-vb/showthread.php?t=1154444

If we are going to post in that thread lets do it as one unRAID community post rather than a hodge podge of replies.

Trayless disks and better ventilation along with better nipple switches would get my vote. Although i dont think i will be buying a Norco again, IMHO smaller arrays are the way of the future.

On a side note i just spent 2 hours trying to get this thing into a cabinet. The rails are good enough to do the job but only barely. They are too generic and need to be front and back mounted for a server of this weight.

bubbaQ, I'll take a look at that fan since I ordered the Xigmatek fans along with the case and had no idea the stock Delta's in the 4220 would be pushing something like ~80cfm (they remind me of my old 80mm Tornado). Right now I'm only going to run 8 drives so using the 140's won't kill me. However, they are no longer my ideal solution given the slight cabling issues I've run into using them with my drop-in replacement bracket.  I'll start work on a revised 120mm plate in the next week or so.  My main goal has been to make the replacement plate a drop-in solution. Using 120mm fans would definitely solve my cabling issues and would also allow me a bit more freedom in designing the entire plate.  Eventually, I'd love to get a high performance 140mm fan plate fully developed but the existing 140mm fans are not really suited for that purpose and maneuvering the plate into position is an art form as well.

I'd like to see both the 4U and 3U boxes:

- Trayless drive bays, or at least trays that don't need screws... such as having pins sticking up from the bottom of the tray that fint into the screw holes in the bottom of the drive.

- 120mm fans or at least a choice of fans, and a temperature-based fan speed controller.

- Thermal diode on each drive bay backplane, with LED/alarm and selectable temp indication (i.e. yellow at 45C and red at 55C, using jumpers on the backplane).  Not complicated... use one setting that applies to all the bays.

- And one FULL SIZED 3.5 bay for a fan controller/multifunction panel OR some fan controls in the front panel.

- a couple of up front USB ports

- side fan mounts (1 each side for 120mm) for extra exhaust.

And while dreaming, along with the standard slim optical bay, a full-height 5.25 bay.

I have 3 XIGMATEK XLF-F1453 fans for the job.  There is not much of a selection in 140mm fans right now and those fans claimed to have good static pressure.  They are rated at 63.5CFM and are very quiet; the stock CPU cooler on the Athlon II makes more noise than 3 of these fans combined.

 

 

The 3 fans side by side leave half an inch or less of gap between the sides and about the same at the bottom. I still have a final test fitting to do with the fiberglass mounting plate and am awaiting the delivery of some 140mm grilles because the fans blades come quite close to the power and data cables plugged into the backplanes. (In regards to clearance for this bracket, I think it would be easier to use 20 angled SATA cables in the 4020 than the SAS cables required for the 4220.) I also have to finish the lower part of the bracket this weekend which will include the wiring openings.  The finished bracket will also have foam rubber stripping to seal any gaps between the sides and the bottom of the bracket.

I had not seen these fans before. The specs look good especially considering I have the older generation Norco with the noisier fans. By my measurements 3 of these fans cable tied together with some foam tape could be a quick hack replcement fan plate. Not the neatest of solution but for those looking for quick and cheap it might be an ideal solution. Since you have all this kit what do you think?

On a side note other ideas for future Norco improvements:

No random vent holes all over the place.

Present the power from the backplane beside the PSU neatly

BETTER FRONT AND BACK MOUNTING RAILS

Blanking plates for empty slots to stop air being drawn through them as path of least resistance

removable/washable dust filter in the caddies

Lid lock that isnt just springy metal clip (try and get this thing open when one of them gets dunted its a 2 man job with hammers)

Option for simple LCD display

 

I had not seen these fans before. The specs look good especially considering I have the older generation Norco with the noisier fans. By my measurements 3 of these fans cable tied together with some foam tape could be a quick hack replcement fan plate. Not the neatest of solution but for those looking for quick and cheap it might be an ideal solution. Since you have all this kit what do you think?

 

I finished up my 140mm plate this morning and it works quite well. The server is quieter than my main desktop.  I'll take a few pictures this evening and post them for you.  The main issue I ran into is that I made a fiberglass mold of the original plate and it positioned the fans very close to the SAS connectors and power connectors.  Luckily, my fan grilles came in today.  Once installed, everything worked great.  I'm going to test a 120mm plate, but I'm quite happy with the 140mm plate now that it has been finished.  When I post pics, I'll detail my construction method because I did find some trouble spots when using foam tape to help seal the fans.

Excellent you couldn't have done this at a better time. Due to my new fw being so loud i decided to bite the bullet and look at every bit of kit I have and make them as quiet as possible... the Norco being right at the top of the list followed by my acoustically shielded cabinet (yup thats right my super expensive marketed as a silent server cabinet is probably the loudest thing in my whole system lol).

NAS, here are some pics from my 4220.  The main thing you will notice is the fan plate is mounted very differently than in the 4020.  However, the fit of 3 140mm fans is very good.

In the first shot you can see the foam stripping I've put on all of the edges to help seal the plate. Additionally, I've added grilles to the front and back of each fan and have cut pieces of sheet rubber (available at any Lowe's in the plumbing dept) to help tie all three fans together and to make sure the grilles do not rattle. The one area I did have problems with was the bottom section of foam stripping; the fans don't have enough surface area to prevent the stripping from coming off if it get snagged.  My main hurdles were the EATX motherboard stand-offs as well as getting my power and SAS cables arranged properly.

In the second shot you can see that I've replaced the stock 80mm fans with some quieter fans I had around the house.  While I don't believe they are necessary, I decided that having them help exhaust air from the case fan could only help the performance of the 140mm fan plate.

Finally, in the third shot, you can see how close the SAS and power cables come to the fans.  Also, I direct wired these fans instead of reusing the fan header PCB mainly because I get lazy and wanted to get the project finished.

In all, I'm pleased with the performance of the fans so far.  I really don't have the storage density to really stress these fans at the moment, but they do seem to move quite a bit of air with ease.

Very nice. I like how the 140mm fans essentially fill the entire chassis.

I think I will try this modification without the rear fans as i don't think they will make any difference. I also done some experiments here and cable tieing fans together via the mounting holes seems to work extremely well resulting in a flat strong fan plate.

My aim will be to come up with a way non experienced modders can make this alteration simply using off the shelf products and household tools. Specifically no fiberglass. Think gaffer tape (aka function over form) 

I think I will try this modification without the rear fans as i don't think they will make any difference. I also done some experiments here and cable tieing fans together via the mounting holes seems to work extremely well resulting in a flat strong fan plate.

So far, adding those 2 80mm fans has been good for about 2 degrees C.  With higher storage density the gains might be worthwhile. Even though the 80mm fans are rated at 30db, they are barely audible.  I'm sure you could get some very quiet 80mm fans to do the job but I couldn't justify it given my fan stockpile from previous builds. My line of thinking was that if I can help create a lower pressure area in the motherboard section of the case, then the efficiency of the fan plate moving air into that void should increase.

NAS,

I would say the 4020's mounting screws would be much easier to work with than the 4220.  I decided to use fiberglass because it was a material that I've worked with in the past and could recreate the intricate shape of the stock plate easily.  The main thing I would keep in mind is that you want that fans to be secured to the case because they are very unwieldy and could possibly fall over if not secured properly.

Good tips and thanks for the further info. 2 deg is quite a bit of a drop so i think i will look at fitting some quiet 80mm as well and see what differences it makes here as well.

I am still pondering the simplest way to attach the new fan plate. Some stickt back foam tape should deal with any vibraytion and gaps but that wont stop them falling over. My current though is to stick a couple of rubber stop on so the plate would slide into a simple slot. Until i try it some of these ideas might just be useless obviously.

One thing i should point out for future generations is my aim is not ultra cool HDDs.... its mid 30 deg with a much lower dB level and as easy and cheap to install as possible.

NAS, here are some pics from my 4220.  The main thing you will notice is the fan plate is mounted very differently than in the 4020.  However, the fit of 3 140mm fans is very good.

This came out great, You should cross post in the pimp your rig thread.

http://lime-technology.com/forum/index.php?topic=2031.msg14752#msg14752

One thing i should point out for future generations is my aim is not ultra cool HDDs.... its mid 30 deg with a much lower dB level and as easy and cheap to install as possible.

NAS, I've added 4 more disks to my array since I got my pro key this afternoon.  I am still in the process of formatting them and running a parity check.  After about 2 hours the drive temps went from about 34-36 to 40-42C.  As long as they do not get much warmer then I think everything is fine, but I will definitely be experimenting with some other plates to see if I can improve the cooling.  FYI I am currently using 3 1.5TB Seagate and 3 750GB Western Digital drives.  All are 7200rpm drives which would definitely contribute to heat soak in the case.

*Update*

I've been swapping out some of my various 80mm fans and have noticed that fan plate performance is heavily dependent on the performance of the 80mm fans.  To test the extreme case I put two of the Delta 80mm EHE fans that I got stock with the case into the rear fan mounts, sealed up the rear vents, and ran a parity check.  Instead of hanging out in the the 40-42C range with spikes to 43C, the whole array plummeted to 39C.  I've been watching it the past 10 minutes in the web management screen and the temps just keep dropping. Of course, that is because I'm pulling 160 CFM and the fans are so strong that it is sucking air THROUGH the PSU.  Bottom line, the air pressure in the motherboard area is very important to the performance of the fan plate.

I am really surprised that the rear fans make that much difference. What kind of ambient temperature are you running.?

What we have is really two states of operation:

1. A couple of disks are delivering content. Low temperature for NAS as a whole. this is the majority state of the device.

2. Parity or high load. Lots of heat being generated. Comparatively the device is in this state a low percentage of the time.

This suggests that we want two states of fan operation as well. We know 140mm fan plate is quiet at approximately 22dB overall so this lends itself to the default state. What we need is a mechanism to jump to state 2 automatically where we kick in the two rear 80mm fans. In this scenario i would even consider using two Vantec Tornados.

Ideas?

Bottom line, the air pressure in the motherboard area is very important to the performance of the fan plate.

This is why I want an exhaust fan mount on each SIDE of the case, to increase exhaust.  It may not help much in a rack room with multiple racks packed in tight side-to-side, but with a single rack, it will make a world of difference.

This suggests that we want two states of fan operation as well. We know 140mm fan plate is quiet at approximately 22dB overall so this lends itself to the default state. What we need is a mechanism to jump to state 2 automatically where we kick in the two rear 80mm fans. In this scenario i would even consider using two Vantec Tornados.

 

Ideas?

I thought BubbaQ invented some idea of modulating fans based on temperatures from lm_sensors.

Perhaps those fans could be kicked into high gear if a parity check is detected.

Guys,

I'm going to start a general Norco rackmount thread since the 4020 and 4220 both share similar problems but do have unique design differences that make discussing this in the 4020 thread somewhat misleading to those who may not know the differences between the two cases.  Additionally, I'm sure the issues we've been working on with the 20bay case extend to the smaller rackmounts and might extend to the unannounced 24bay Norco and NAS rackmounts theorized in the AVS thread.

I thought BubbaQ invented some idea of modulating fans based on temperatures from lm_sensors.

Perhaps those fans could be kicked into high gear if a parity check is detected.

I didn't invent it... I just exploited it.

Part of the lm_sensors package is sensors-detect and pwmconfig.  You need perl to run the initial setup, but after that, you don't, so you can just install a perl package once, run the setups, and then have no need to run the perl package install at every boot (it is big).

In any event, to control the fan headers on your mobo is simple IF they are controllable..... all are not.

Twop pseuofiles are used:

   /sys/class/hwmon/hwmon0/device/pwmX_enable

   /sys/class/hwmon/hwmon0/device/pwmX

where "X" is a digit from 1 to n, denoting which fan header you want to control.  For example:

This enables speed control on port 2:

   echo 1 > /sys/class/hwmon/hwmon0/device/pwm2_enable 

This sets the speed on port 2:

   echo 95 > /sys/class/hwmon/hwmon0/device/pwm2

Values are from 0 to 255... you have to use trial and error to learn which vlues result in which fan speeds.  95 on my system sets the fans to a speed that is inaudible.  110 will be barely audible.  40 or below will stop them, and about 180 or higher will put them at max.

There is no way to know WHICH fan header is controlled by which psuedofile except by trial and error.  Some may not be controllable, some may.  Hook a fan up to each header and then run the setup files from lm_sensors to find out which fan headers are controllable.

Here is a quick-n-dirty shell script that is run every 2 minutes by cron:

#!/bin/sh
#
SILENT=95
QUIET=110
COOLING=120
FAST=125
HIGH=130
MAX=220

maxTemp=`smartctl -d ata -A /dev/sdd|awk '{if($2~"Temperature_Celsius"){print $10}}' 2>/dev/null`
parityTemp=`smartctl -d ata -A /dev/sdb|awk '{if($2~"Temperature_Celsius"){print $10}}' 2>/dev/null`

SPEED=$SILENT

  if [ $parityTemp -gt $maxTemp ]
        then
    maxTemp=$parityTemp
        fi
  if [ $maxTemp -gt 29 ]
        then
    SPEED=$QUIET
        fi
  if [ $maxTemp -gt 34 ]
        then
    SPEED=$COOLING
        fi
  if [ $maxTemp -gt 39 ]
        then
    SPEED=$FAST
        fi
  if [ $maxTemp -gt 41 ]
        then
    SPEED=$HIGH
        fi
  if [ $maxTemp -gt 43 ]
        then
    SPEED=$MAX
        fi

echo $maxTemp
echo 1 > /sys/class/hwmon/hwmon0/device/pwm2_enable
echo $SPEED > /sys/class/hwmon/hwmon0/device/pwm2

I assume there is no such thing as a PCI card that can be used to control fans?

NAS, I have several ideas I'm kicking around at the moment.  Once I get off work I'll do some more research on them to evaluate their effectiveness.  I know from using programs such as SpeedFan and unRAID web GUI that all of the pieces are there; it's just putting them all together in an effective package.

In any event, to control the fan headers on your mobo is simple IF they are controllable..... all are not.

BubbaQ, I assume you would need a 4pin header on the motherboard and for the fan to support PWM.  Is this a correct assumption?

NAS, I have several ideas I'm kicking around at the moment.  Once I get off work I'll do some more research on them to evaluate their effectiveness.  I know from using programs such as SpeedFan and unRAID web GUI that all of the pieces are there; it's just putting them all together in an effective package.

Keep me posted I am researching as well

BubbaQ, I assume you would need a 4pin header on the motherboard and for the fan to support PWM.  Is this a correct assumption?

No, it works with both 3-pin or 4-pin fan headers

Does anyone just replace the stock 80mm fans with quieter ones?  My drives are spun down most of time and this cooling system although not great is much better than the rigged 15 drive case I previously had where drives would burn you if you touched them (and they were raid5 so they were spun 24/7).  Really my issue is just with the noise, I just want them quieter, the performance is fine for me - has anyone had good experience with a very very quiet 80mm fan in the 4020 case?

Nate