DL088 – Dell TL-2000 Tape Library & LTO-3 Tape Drive Teardown

DL088 – Dell TL-2000 Tape Library & LTO-3 Tape Drive Teardown


Hey everyone, so we’ve got another teardown to do today, what we’re going to look at is a LTO-3 tape drive. An LTO drive is used as tape storage, backup for computers. Developed in the late 1990s So we’re going to take a look at that, but first i’m going to show you the device it came out of, because i just picked this up off ebay broken i’ve just fixed it up. I am going to take the drive out and i am going to teardown the drive but the actual unit here is going to go off to a friend of mine who wants to put an LTO-5 tape drive in it and convert it. So this is a dell TL-2000 tape library, it’s a 19″ rack unit which can take one or two LTO tape drives and can also store 24 LTO tapes and there is a robot mechanism which takes the tapes and pops them into the drive so you can access them over the network. As i said this takes upto 24 of these LTO tapes and it’s completely automated system so you don’t have to keep going down to the server room and swapping tapes in and out you can just possibly have one or two of the cartridge and you just swap those out weekly or something. It just makes it a little bit easier. So i think this unit dates from about 2008 i am not entirely sure, the LTO-3 drive can store up to 400Gb on a single tape, they do have compression as well so they can theoretically go to 800Gb, but don’t support you ever really get that. So you can have 400Gb per tape, and there is 24 tapes so that gives this unit a total capacity of 9.6Tb now obviously when this is upgraded it will be a whole lot more if you upgraded this to an LTO-7 drive which is the largest available at the moment you would get 24 tapes each storing 6Tb each so that would be 144Tb which is quite impressive really. So anyway, back to the case in point i’m going to just quickly run through this because it’s kind of interesting on the inside and then we’ll get cracking on the tape drive, so i just take the lid off this… so you can see here we’ve got two tape cassettes, one on each side, we have a robot mechanism for loading and unloading the tapes in the center. Just at the back here a small controller board, the actual tape drive in the middle and a power supply. There are a few user controls on the front, i won’t bother showing them because they are not particularly interesting The tape magazines, these would release and slide out of the rack unit that way and these store the… This is a 12 tape unit and there is another one on the other side, so all these tapes would be stored in there like that they all have barcodes on and there is barcodes in the back as well, the robot mechanism actually has a bar code reader so it can identify all the tapes and it knows where each of the tapes are as it does get a bit complicated when it comes to managing backups and things like that, so everything is bar coded. Now one thing i did find interesting with this is these little clear plastic sections in each of the tape cartridges these are light pipes as far as i can tell, there is no LEDs or anything in these part, the LEDs are in the drive head and as it moves i think an IR LED shines though these ports and you get a reflection back so it uses these to find the exact center of the actual tape slot, i think they have done this for size reasons, i wonder why they just didn’t have a bar code or something it could read but maybe they wanted to keep everything as slim as possible so they have gone with this optical solution which is kind of neat. So when i picked this up it was described as having a robot fault now when i took the lid off it was a bit of a mess in here, there was tapes all over the place, it had obviously been turned on with all the tapes just strewn so this tried to move, it’s jammed up. This drive belt here was out of place. It also looks like it’s been dropped as well there is some damaged plastics, some of the plastics around here were slightly damaged. Now i’ve fixed those up and i have fixed up the robot head as well which had some very very minor damage on one of the gears so that i’ve fixed up so it does all appear to be working fine so this should be ready to be re-used Right i can actually show you this running which is kind of interesting because it’s nice seeing electro-mechanical devices whirring away doing their thing so it’s quite cool so i’m going to power this one, it’s a bit noisy because as anybody who’s been in a server room knows it’s always noisy as hell! So i’m just going to turn this on you probably won’t be able to hear too much but the fan noise does quieten down after it’s finished booting up. Now this does take a bit of time to boot up so, i’ll see you in a moment… Now what it’s doing here is just going through and cataloging all the tapes, so it’s reading the bar codes that would be on the tape. I believe the clicking seems to be a focus mechanism so it’s adjusting the focus. looks like it’s picked up that black tape is actually a cleaning tape, so i think what it’s going to do is load that into the drive and do a cleaning cycle. Right well that’s enough of that, let take the drive out and take it to pieces! So i need to just take the drive mechanism off this drive sled. Right we’ve got the drive extracted out of the sled which plugs into the tape library and obviously it’s just a standard form factor, 5.25″ and half height. The tape goes in the front here, there is no bezel on this because they often omit that on drives that are intended to be used in embedded applications because you don’t need a bezel on it you can see the little clip marks, you could put one on if you wanted to. Round the back we have a small little fan and we’ve got the SAS interface, that is Serial Attached SCSI, a couple of odd little connectors, there was one of these connected which which is known as an LDI interface, which is Library Device Interface which allows the tape library to talk to the drive without going through the SAS interface. Right so under here we have clearly two motors there, i will explain why we have two of those in a second So lets whip the top off. Bit dusty in there, according to the menus on the library it had actually been turn on for about 41,000 hours so it’s not surprising there is a bit of dust in here. So here we can see the main mechanism, we’ve got what looks like a load and unload mechanics here, and this is the tape transport mechanism. Now if i start putting a tape in, i don’t have this powered up yet we can have a look to see how it detects the tape. We can see just down in here we’ve got a small optical sensor there so when you push a tape in it moves this piece of plastic and breaks the beam there so then it will detect there is a tape being inserted and start drawing the tape inside. I’m just going to put a tape in but unfortunately at the moment it doesn’t want to load this in. The display on the front is showing an error so i wonder if it’s expecting to be plugged into the library so it won’t seem to do anything but i think if i put this into service mode it should be able to get it to load the tape I’m just pushing the button on the front here. That’s in service mode now so lets try popping this in again Ok, what that has done is loaded the tape and extracted the tape out of the cassette itself and it looked like it pulled it all the way around here and attached it to the take up spool. By the sounds of it wound off a fair amount of tape, obviously there is going to be quite a bit of lead-in tape before you get to the good bit of tape, because the beginning of the tape is always going to get slightly damaged and distorted. I’m not quite sure what it’s doing, but i have noticed that the tape head which is just here, you can see that moving slightly You can see it tracking up and down very slightly just here and you can see the bottom of the head just down in here as well it should bring the tape now around the head… …and you can just see the head moving up and down slightly. Ok if we have a look at this in closer detail We’ve got two motor drives here, the take up spool and this one is for the tape spool, so as the tape is on here the tape needs to be wound off and onto this. There is going to be accurate speed control on these two motors to make sure they both run at the same speed otherwise you’ll end up with slack tape or stretching the tape so that’s kind of interesting. You can see the take up spool inside there as i said the head is just under there we;ve got some nice machined rollers and a whole load of cogs and mechanisms there to take the tape in and out can’t really see too much yet. There is a motor over here in this corner, which looks like it drives all of these, quite a number of stages! One, two… …eight, so a fair bit of reduction on that. There is another small motor just on the corner here and another down in there as well. There are a few motors on this! Plastic protector, well there is a few devices on this side but you can clearly see there is a big footprint of something underneath there some here, here so these will be on the other side. Some fairly big high density connectors it looks like. Some pass through connectors onto the other side of the board. I’ve released all the flat-flex cables, there is quite a few actually, one, two… …nine, ten, eleven …thirteen of them so quite a lot, it also looks like the board is not screwed down it’s just literally held in with the flat flex and the metal plate that was on it. Right here’s the controller board, i’m not going to do into a huge amount of detail because it’s a bit beyond me, yea there are going to be some fairly large devices motor controllers, stepper motor drivers and all sorts of other stuff so on this side we have some large BGA devices all big name brands, we’ve got IBM LSI, NXP, IBM again quite a lot of these inductors around here so there going to be a whole load of local voltage regulation going on these are possibly power supply circuits here looks like we’ve got an optical sensor there mounted on the board itself. Yea i’m not going to go into a lot of detail on this. It is just a controller board after all. There is some nice large BGA devices there IBM one with a big heat spreader. Right lets have a look at the mechanics, we’ve got all these flat flexes running to various sensors and stuff, the two main motor drives. With quite a number of connections… well actually they are not… there is not a huge amount of connectors on there, it looks like it but you can see there is actually quite a number of connections on there but most of them are paralleled onto these three main tracks so that’s for power so you can more current down them because i suspect these are going to draw a fair bit of current. So they have just taken the individual pins and then just and turned them into large traces. On the front here we’ve got a flat flex with 4 LEDs, those are status indicators if you had a front panel installed. The small tactile switch is just the eject button. This flex here looks to run upto this PCB which is has a circular trace on here so i think that is the RFID tag reader that reads the tag inside the tapes Motor connection, we’ve got these sensors over here, these are obviously the connections that run to the head, we’ve got 80 pins on that one and 80 pins on that one so we’ve got 160 pins running to the head. If that the head then i would imagine that is the head positioning servo there is obviously some sensor coming up on the same flat flex little plastic bits and bobs, that is a different plastic to these bits and had the feel of being a different material looks like… this wheel here runs on this so this is some kind of plastic designed to have things rubbing on it so it doesn’t wear. This little mechanism here which basically allows this… wow what a lovely articulated little thing! So I think what this is there is some small brushes on there and they are black i wonder if they might be little carbon brushes maybe some static reduction to stop static build up on the tape? I couldn’t see that being used for any kind of cleaning or anything like that, i think that is static control. This looks like it’s the claw that grabs the end of the tape from the tape spool and drags it… round and passes it to something in here guessing there is going to be some kind of latch mechanism on here yea, there is probably something down in there it shoves it into now we’ve got a couple of optical sensors mounted on the top of this plate. Interestingly none of these cogs and gears have any ballraces, they just run against these machined posts, so these plastics will be designed to be effectively self lubricating if you want to put it like that. So they don’t need any lubrication. Here we’ve got the main pickup spool. The end of the tape is just in there so think this would move around, probably capture that, not quite sure how it would do that. It would grab it and then pull it… …not quite sure how that would have got round there, maybe i missed something, so it would grab this and feed the end of the tape around here continuing to work around, this would line up and that would then drag the tape inside here, lock into place and that would stay locked in the center of the spool allowing the claw to stay in place. This can continue to spin… bit difficult really, you get the idea! We’re starting to get some plastic identifiers on here, that’s PC+GF50 so that’s polycarbonate with 50% fiberglass, yea it’s definitely got quite a colour to it. That is going to be tough stuff! Tape guides, three of those on spring loaded mounts. Lovely machined steel there, some quite high quality bearings in there i would imagine. On this spool which is almost identical with only change there is some small ridges machined into the surface of the spool, which is there to reduce the air-bearing effect between the tape and this spool, increases the friction and that allows this spool to reduce high frequency fluttering of the tape before it runs over the read/write head. Wow, listen to the sound of that, sounds like metal but it’s not, it’s plastic! We’ve got PPS+GFM65 not quite sure what that would be the GF will be Glass Fiber PPS? Let’s see what that is… that is quite an advanced plastic actually, Polyphenylene Sulfide + fiberglass as you can hear, it sounds metallic. I bet that would not have been cheap to make. There is the tape head itself, i bet a lot of the cost of buying one of these LTO drives goes into that little bit just there! It actually looks like this has another degree of movement i had not noticed before you can move it that way and this stepper motor drives it in the vertical position, so the stepper motor does it that way, but is also able to move this way as well. I didn’t notice what thing actually adjusted that. Really smooth operation on that, oh yea, there’s why. This is why you pay your money, there is two small ball bearings there to allow this to move backwards and forwards smoothly Although these would have been built to a price but they certainly didn’t skimp on it they certainly expected people to pay the money. Optical sensors there tracking the position of the tape loading mechanism. You can see how when the tape is drawn in that when it gets to the end position it drops the tape onto the drive wheel. Yea, more bearings here… Nice piece of mechanical engineering. This doesn’t sound very healthy but i wonder if something i’ve taken off has disappeared up inside, looks like it’s a permanent magnet motor. Yea there is something got stuck up inside there. The rating on there 12 Volts 2.5 Amps so yea i thought they would draw a fair bit of current, they are 30 Watt motors. Nothing more to see on this now there is just the odd optical sensor, let take out the RFID reader, not much to see on there just a trace on a PCB. Let’s have a quick look inside one of the tapes. Little door with a nice fine spring. Big spool of tape and the end of the tape with it’s pin which that claw grabbed. The locking mechanism to stop the tape moving when your not using it, spring thingamabob… We have the RFID tag quite cool you can see the Die in it, they have not used black goop. On an LTO-3 tape we have 680 Meters of tape, i’m not going to pull that off! It would make far too much of a mess! Right i think that’s the end of that, you can guess this isn’t going back together! So nice big pile of bits there! Hope you found that one interesting, it was kind of interesting i enjoyed that one! It’s been a long time since i looked inside a tape drive especially one of these new modern or fairly modern as this one is. Right thanks for watching everybody, if you liked it give me a thumbs up if your not subscribed hit the subscribe button and i will see you on the next video. Catch you later!

Only registered users can comment.

  1. Like the new intro Marc! Great video as always 🙂 always look forward to new uploads, keep up the top work.

    Jason

  2. Hi, thank you for the video. And the great subtitles. It helps me to learn new words. I was hoping to see more details of the controller card, and the power supply. Anyways, a big thumbs up. Happy to see you again.

  3. You should get a cheapie RC ESC 3-phase motor drive and see if you can run these motors!
    Should be pretty straightforward.
    Oh, and regarding the noise – all enterprise & server gear manufacturers have a secret minimum noise requirement. If it's not noisy, it isn't worthy of being in a rack.

  4. Wow, SIX motors in one drive! Interesting and impressive! Even my DLP TV has 3 motors and one more electromechanical part (the DMD, but which isn't motorized).

  5. At 13:30 you say that the two motors should "run at the same speed". If you think about it, they will only run at the same speed when you are reading the mid-point of the tape. The rest of the time the two motors will run at different speeds to account for the fact that each revolution of the "full" spool pulls more tape than a full revolution of the "empty" spool.

    **vp

  6. I wonder why the reels in the cartridge where not stacked like in the Philips NV1500 VCR and using a rotary video R/W head to enable the reading of data from tape and writing updated data back to the tape.

  7. I want one. I dont need it, and I really dont have anywhere to put it or anything useful to do with it, but I want it because it looks cool. That is another interesting thing. Here we are with super advanced SSDs that can hold terabytes and access that data extremely fast, and I want a tape drive library that uses technology invented 20-30 years ago. I know you said that this particular tape library was made in 2008, but that is still 8 years ago.

  8. Wtf is this thing taking so long? My MSL6000 is instant on when fully loaded, it takes like 15 seconds to do the whole inventory.

  9. Now that was neato! Thank you for this video. Always wanted to see automated back up systems in action.

  10. lord…i remember being the guy that had to "do tapes". The wait for that thing to do its thing and unlock the caddies was brutal at the end of the day:)

  11. These are still used to backup servers, right? Different generation I'm assuming. SSD's I've heard do not retain their data forever whereas a tape can be run after a couple decades and still retrieve good data.

  12. These tape libraries and drives are the the standard on the systems I work on. The ones I install are the IBM TS3100. The systems they connect to are called Power systems i5 O/S.
    I just get them hooked up and operational. Pretty interesting to see inside. Thanks!

  13. Poor Tapedrive "tape streamer" he had so much to offer.

    I think there were still enough people who wanted to take over for a small amount.

    I would like to buy a LTO 6 or 7 Tapestreamer. Unfortunately, for the consumer, it is unpayable. $ 5000,- to $ 7000,-
    than $150,- to $200,- voor a tape.

    It remains beautiful technique.

  14. Ahhhhhhhhhhh, that pile of parts at the end just brought a tear to my eye. What a well put together machine. Just wish our old VCR units were built as well.

  15. That's an absolutely insane amount of silicon on that control board. I understand now why these drives cost as much as they do. Yet I can't help but wonder why they'd need such huge amounts of logic. Have they just implemented everything on huge FPGAs due to volumes being too low, or what? It would have been interesting if you had gotten close-ups of the chip markings.

  16. At 12:08 you say you don't know what it is doing. It is quickly calibrating the tape head after loading the tape.

  17. Can you just change the LTO drive in a library for increased capacity?? Or does some other controllers need to be replaced as well.
    I habe an older library system that uses LTO-1 tapes and I want to use LTO-3 tapes. Its a Dell unit with dual drives

  18. Hm, I cringed inside as he showed the pile of broken parts at 28:21 … I can't even afford a LTO2 drive and here one was destroyed 😛

  19. Watching this in 2019. Very interesting look inside the operation of these drives. I kind of figured there was a grab mechanism to pull the tape out but I thought there was a handoff to the take-up spool where it locked the pin – I would never have imagined that the grab hook goes into the spool and actually spins with the tape.

    Those tape drive mechanisms are also far better built and more durable than any 2.5 inch hard drive including RDX. You pay a pretty sum for all those machined bearings, exotic plastics, built-in error correction and that massive head.

    I just purchased an external LTO drive for my media server backup. My enormous 512 disc binder of DVD movies all fit on a single LTO6 tape – that's' a massive space savings right there.

  20. if you didn´t put it back together while making it work again then you should die VERY SLOWLY AND IN EXCRUCIATING PAIN !! cuz the computer gods do not allow such atrocities to go unpunished.. if you had assembled it and made it work then I feel a deep respect, admiration and gratitude that you´ve shown us such a nice piece of hardware 😉

  21. I though the two motors doesn't work at the same time? One to unreel the tape to the spool, and the one under the tape drive is to reel back the tape.

  22. No capstan/pinch roller. Speed is regulated by tracking data encoded on the tape. The motors are 3 phase sensored brushless motors rather than stepper motors. 1 motor runs in braking mode to manetane tension while the other motor pulls. Imagine if this kind of software defined transport existed when audio tape was just invented.

Leave a Reply

Your email address will not be published. Required fields are marked *