Hello! As far its my first post here, let me briefly introduce myself. I'm Andrey (maybe you'll prefer Andrew), 29 (almost 30) years old male, living in the middle of Siberia (you know, bears and vodka ). I am network administrator, worked few years @ local ISP. I liked airsoft, my team modelling Bundeswehr (German army) panzer infanterie and I am machinegunner Sorry for my so-called English, feel free to correct my sentences. Task overview As far as I'm IT & ISP man, I organaised my home LAN like semi-industrial solution At corridor I mount 19" enclosure, 12U heigth: DIGIITUS SOHOLine 12U There are mounted patch pannel for my home mini-SCS, ADSL router, managed switch and server (subject of modding). Server man few service like DNS cache, http cache, torrent client, DHCP server and other different services (apache, mysql, ruby on rails for web-dev). OS of my choise - FreeBSD. My current server is roughly cutted mini-ATX case, its shrinked for 15cm to fit the enclosure. Its (enclosure) depth is only 40cm (27.5cm between front rails and its back panel). There is NO manufactured case can fit there! I mean, no cases which can fit this enclosure AND hold significant amount of HDDs, 'coz I'm going to change my Unix server into network attached storage (and feed other computers and notebooks with space to store => no need of large discs there) Updates index (20 nov 2008) First sketches (26 nov 2008) My current setup [mini] (14 dec 2008) HDD cage arrived; some thoughts (26 dec 2008) Case design - general layout and airflow (20 dec 2008) More designs on air filter; two building failures (5 jan 2009) - Frontal filter support (10 jan 2009) Air filer holder - almost complete
I am interested to see where this goes. Do not worry about your English. This is a multi-national forum anyways. As long as you have decent pictures all shall be fine.
Update 1 (20 nov 2008) First sketches First sketches I'm dreaming this case for two years till now and even bought some hardware. But only now I begin to make some real steppes to it. My greediness force me to find out how to hold as many HDDs as possible with minimal amount of space elapsed. Also, HDDs should be easly reached for maintance and replacement. I choosed two layered design. Bottom layer will be filled with drive hot-swap enclosures, top - with mobo, RAID controller(s) and PSU: Due to such design I can pick almost any mobo and use standart (slightly converted/squeezed) PSU. (My prev. designs include 1 or 2 maximum cages + mini-itx mobo and 1U server PSU) As HDD case Supermicro CSE-M35 was chosen: With cost about $170 (in Russia) they hold 5 drives and use 3 standart 5.25" bays. They said Noisy fan can be easly replaced. There is more expencive model CSE-M35TQ with monitoring features onboard; nipping on ahead - I bought that one. Cabinet width allows positioning of 3 cages - 15 HDDs total. At first stage I'm going to place only one cage and put 2 placeholders insteed of two other cages. It's another planned feature - possibility to rise capacity over time. I bought used Highpoint RocketRaid 2220 8 port SATA-2 controller at Ebay year ago for $150 (plus shipping ). This card allows to interconnect up to 4 same cards together and build inter-card RAID (up to 16 devices in one RAID). Non-home device, but price is reasonable. So, with two cards I can put all my 15 drives into single hardware RAID. Sweet! BUT there is one point about it - it uses PCI-X interface! Its hard to find cheap and small form-factor mobo with PCI-X. At time of bidding I didn't pay any attention to it, sad. I'm planning to make drives layer 3U heigth and about 2U for electronics. It'll allow me to use standart PSU. PCI-X is backward compartible with old good PCI 32bit (in exchange of lower speed) and it allows me to use almost any mobo to startup. (There is so-called active raiser cards on the market, they allows to brige one interface into another but they cost bunch of $$$) Power consumption. HDD startup consuption, lets say, 30W. Do we need 15*20=450W to only power up all these discs? No, thanks God! RAID cool controller supports staged spinup. Mobo, controller and other stuff - lets say 200W max. Single HDD eat about 9-13W (source), I'll suppose its very hungry and old one and consume 15W. Total - 425W at maximum load. I'll start with old 350W PSU (remember? only one cage) and will change it to 500-600W one at some point.
Update 2 (26 nov 2008) My current setup [mini] Many thaks to my friend Andrey Veshtebey, he bought and send me Supermicro CSE-M35TQ, it cost about 5200 RUB (about $210 at that time)). This cage can do some SES-2 and SES-3 magic. I'm begining to study SolidWorks. First attempts: Right now my 'server room' looks like this (from up to down - patch-panel, organiser, switch, power distributor, shelf with moded, keyboard, shelf with server): Planned layout: Current server case was made from some old tower case by cutting it for 15cm in depth, then I exchange front and rear pannels by rotating case and mounting power/reset buttons on front pannel. Three evenings of rough work. Funny thing I made - system start up from RAID-1 (mirror) made of ancient 8gb drives I did it just for fun.
Update 3 (14 dec 2008) HDD cage arrived; some thoughts Cage arrived. Alleluia! Built nicely. Didn't connect it yet, but already made SolidWorks model. Package contains a packet with various bolts, stickers and cable 'hyrda' - 4 SATA and some smaller cable (I suppose - its for activity and fault LEDs). I cant imagine - OH WHY?! - why they supply 5-bays cage with 4 drive cables? Beside this, cables into hydra is quite thin and shielded. Every SATA connector wired with 2-2.5mm cable. I wish I could buy cables like that somewhere. 15 ordinar cables will form huge bundle. Cage are powered by 2 molex power cables, also there are 5 SATA connectors and few jumpers and monitoring & LED connectors. Just founded SolidWorks PhotoView - renders here! Then I build RAID-controller model I found some burned out element on the board, near PCI bracket! So I plugged card into my PC (first time for 1.5 years after I get it!), it start up normally. I hope it was unnecessary part I have troubles with cabling inside of server case. Imagine - you have to connect new drive cage. Server secured into enclosure and you really don't want to remove it and disassemble for it. Some thougts here. Maybe we should path cables near front pannel and make them long enougth? Maybe we should allow all inserts to slide away on the guides (all inernals at once or each layer separatly)? Third possibility - we can make some back-plane with connectors - so cage will clip into it (most industrial-like decision). Now I decide to go easy path - patrially disassemble case to add new cages. Cage will slide into lower bay freely and secured into it with L-shaped bracket (pictures into next update). Back panel will be removable, I'm thinking about some plastic air tunnel behind it around cages.. I have a question (still actual! 30 dec 08) - steel or aluminium?.. Every cage is about 3kg + drives + PSU + mobo + case itself. Is aluminium strong enougth for it? Case will be formed with 3 parts - trough (side pannels, bottom part, mounting brackets made from one sheet), top and back. I'm going to use 3mm alu for now. PS: then I try to post new comment its didn't visible and going to moderation but quick reply works fine... hmm
Looks great! If I didnt have to move around so much I would do something similar. To write a new post use "Go Advanced" then write your post the click "Submit Reply". Moriquendi
Update 4 (26 dec 2008) Case design - general layout and airflow Hi here! As Maks [Sleepstreamer] Erlandsson said in interview for modnews.ru - "If you love, like me, to think up case design in the building time, then building stage is most difficult. If you like to plan desing in tiny details first - then this stage is most difficult…" (rus->eng translation made by me). I am planning all details and dimentions, so this stage is a bit long, sorry guys. Chassis is based on "trough" bended from sheet of 3mm alu - side pannels, bottom and mounting brackets: Top and middle pannels are secured to side pannels with 15x15mm alu angled bars; no other supports planned yet. L-bar in the front of case designed to be support for HDD cages: To the top of drive cage will be mounted mounting plate. Cage freely sliding into case and fastened to angled L-bar with 2 screws Frontal L-bar secured in place this way: At first time I wanted to place angled bar with horizontal part down, but becouse of this disposition there were too much free space above cage. Spacer needs to make some space above shelf(middle pannel) for cage mounting screws and below - for mobo screws and to allow cages slide in freely. Air flow I am going to keep heightened pressure inside of case and feed suction fans thru filter. Two 120mm fans will be placed on the top pannel: Frontal fan will be placed right above (lets say so ) processor on the mobo, so we don't need CPU fan. As you see, I choose microATX form-factor - pretty small size and more than enougth expansion slots. Filtered air will cool contents of top layer. PSU will be located on the left (not shown on pictures), at first I decided to make an air tunnel to it, but maybe I'll just put a fan on it. Then air flow will reach bottom layer thru hole on the back of the shelf and will leave the case by passing thru harddrives. Fans on cages work on exhaust (inside of case); I'm planning to change their direction on clockwise. I planned to make an air tunnel on the bottom part of case to feed air from hole in the shelf to directly to cages, but now I think it's not necessary. All chinks of case will be filled with bondo or some kind of sealant. It will guarantee what our nicely cleaned air will not leave case in unplanned point and will cycle inside as planned. Also i thought about air tunnels on upper layer two-way tunnel from frontal fan to CPU and PSU and from back fan- to lower layer. But there is only 76mm heigth and air will flow inside smoothly without any tunnels. What do you think about it? Air filter take-down case planned to build with 3mm PVC. Extreme (i.e. first and last, not sure it is right word) PVC layers will contain slot for metal mesh. On picture shown only 3 layers, but actually I'll place 4 layers: Filter support will be made of 15x15mm angled bar. It will fastened to top of case. Fans will be fasten to bottom part of support. Filter itself can be removed thru slot on front pannel. Yesterday I bought some raw mats - sheet alu 500x1200mm 1mm and 3mm thick (I wanted to buy 2 and 3 but there were no 2mm alu in small sheets), angled bar 15x15x1.5x3000mm, goldie zink mesh, bolts and nuts. Stay turned! PS: if any forum admins read this - please delete some double posts above. I strip them to bare minimum ('please remove me')
Update 5 (20 dec 2008) More designs on air filter; two building failures Since I'm not ready yet to cut 3m alu sheet, I started with air filter. Yesterday I bought some rubber 1mm thick to make gasket: Do you see a gap(chink)? I'll give up on this chink if it have been located (I feel there is a mess, don't remember this English tense) OVER filter, but it is UNDER and will suck unfiltered air (well, pretty small amount, but lets master as much as we can). I tried to figure how to bend angled bar to fit this gap and make some other bonus usefull application for it. Ended up with some part of sheet metal bended this way: Inside holes are for fasten to filter support frame, top holes - to fasten to top of case, little holes on the front - to secure air filter in place. There will be some frontal pannel for filter, didn't think yet how to connect it and filter. Assembly view: Do you see rectangular pieces of plastic on the sides of filter? It is lock: I made it with 5mm PVC. It should be glued to bottom layers and snap into top I start to cut plastic with hack-saw and slightly screw one side - I cutted it angularly(?). Fretsaw with diamond thread makes things much easy and cleaner. When I cut windows I clearly see what I understand early - PCV is too soft and too flexible material for such application Anyway, I decided to continue and see how it come out. I give up on locking system and fasten it with bolts. Cutting groove for mesh. BTW, greate knife. Very cheap and VERY sharp. What I get after it. I'll rework it with other mats after holidays (I dont have any at home): On the next day I dought Fit miter saw. My prev Matrix was a piece of s**t so I return it. This one not perfect but I can't find better saw. Makint support frame. Start to cut angled bar. Unnecessary material was cut with fretsaw. Diamond thread produce very thin cut - about 0.2-0.3mm. Ready to bend... WTF???! I cut it right but sides doesn't fit. They gives me wrong bar - 20x20 not 15x15! At least I pay for it as for 15x15.
The plans look very nice and being always interested in NAS/Server builds this one is definitely interesting And getting the wrong profile really sucks! Did you buy lots of that stuff? WTF topless modding?
500mph, ModMinded, Moriquendi, Jipa, Mankz Thanks for comments and support! That is what I like in english (i.e. international) forums - kind and active community 3 meters. You get me, haha! Yeah, right, its me topless. Thanks! Nice to hear
Update 6 (5 jan 2009) - Frontal filter support New Year is passed, then my wifes birthday and all flat full of guests in between. Tonight I find some time and desire to continue. After last failures with support frame I decide to start again with front support bracket. I actually begin to work with it 1st january. I cut piece of 1mm alu sheet in size and begin to cut: Unnecessary mats removed, right befor filing (I wanted to glue draw to sheet but printed it not in scale): I skipping drilling, scratching and bending stages, its obvious. Here is result. I can't say I'm very happy with it but it seems to be okey. Tomorrow I'll see how it's fit. Support frame. Process was absolutly same as in prev update, so I skip it. Final result will looks like this: Side note. I often read in worklogs - "my camera was acidently broke, so i take no pictures or take them on my phone" and thik like "Kek, loosers, how it could be?". But! Today my old lovely SLR was broke. At first sensor located behind mirror stop to move and close about 90% of picture. I shoot a picture at bulb expose and flip it by slightly touching. Now seems what it didn't open (unflip) at all and auto-focus not working. So, I rise F-stop number to 10 and shoot on manual focus. Manual focus on Canon cameras is pain in ass but lagre F-number compensate errors. Cheers, there is pictures!
Nice updates, sorry to hear about the hassles along the way. Very nice bending work. I like the little lock idea you had for the filter, sorry it had to go, but bolts are definitely easier. A question for you about the filters: Will you be using some other material between the mesh you have shown? I don't think that mesh will do any filtering, besides keeping things bigger than a pencil out. LOL @ "Side note. I often read in worklogs - "my camera was acidently broke, so i take no pictures or take them on my phone" and thik like "Kek, loosers, how it could be?". But! Today my old lovely SLR was broke." (not loling at the camera breaking... that sucks! Loling at the irony)
Update 7 (10 jan 2009) Air filer holder - almost complete Update time! Lets work on filter frame again. I bought two Thermaltake Thunderblade 120*120*25 fans with green LEDs. Don't get me wrong, I bought them not becouse of green lights! I browse half of night on manufacturer sites in comparision and this model is best I can buy locally. There were better one - with blue lights, but they were out of stock. Max. air flow: 78 CFM Max. air pressure: 2.97 mm-H2O Noise: 21 dBA At first, I filed out parts marked by number 1. Also look on parts marked with 2. It is sticky cloth with flexible electric ribbon under it with soldered LEDs. Becouse of this curved surface fans didn't come close and didn't match drilled holes. I decided to remove lights completly but then find the way to keep it. I remove cloth and cut ribbon: Now fans matches perfectly. Drilling holes in front support and in frame for riverts: Making brackets: BTW, I have a trouble with riverting. My tool is too big to fit my details and after tighting riverts caps were angled. I solve this trouble very easy - by hammering! I cut rubber but then decide to cut frame too - it close too much airflow. Cuts made, rubber cutted more to fit it Assembled view. There is not much to do with this part - only some contact plate with power and RPM connectors. I thinking it up now. Regarding questions ModMinded asks. That filter with bolts is not final result, it is some hmm... test build. Then I find out what things come out bad I decided to not trash made parts but to finish them into some complete shape, just to hold them in hands and keep myself motivated It is just real filter holder. I don't know yet what filter material I'll use.. foam rubber? Waiting for suggestions!
