ono3 - Pelican 1430 to S4 Portable VR Workstation

Discussion in 'Build Logs' started by lhl, Jan 27, 2016.

  1. lhl

    lhl Average Stuffer
    Thread Starter

    #1 lhl, Jan 27, 2016
    Last edited: Jun 6, 2016
    (I've been naming/numbering my VR development machines ono's after the Ono-Sendai Cyberspace 7 cyberdecks. ono3 makes me confident that by the time I get to ono7, it should be worth its namesake.)

    About a year ago, I built a portable workstation in a Pelican 1430 top-loader (gallery, parts list) so I could do VR development while traveling. While a bit heavier than I would have liked (fully packed, is usually about 7.7kg, or 17lb), it's comfortably personal-item/carry-on size for air travel (even able to fit under a US domestic economy seat), and I traveled with it across 4 continents, and for tens of thousands of miles without any major issues. Long story short, it did good.

    That being said, with the new Vive and CB/CV1 HMDs, it's not possible to stuff them into the Pelican anymore, and truth be told, I was getting a bit tired of lugging all that weight around (the Pelican case by itself weighed 2.6kg).

    Phase 1 of my new build was a simple transfer job. There were a few minor issues, but overall, it was pretty painless.

    [​IMG]

    The spreadsheet has the specifics, but the highlights of the current build:
    • NFC S4 Mini (#005)
    • Intel Core i7-4790K (Devil's Canyon Haswell)
    • Noctua NH-L9i CPU Cooler
    • Asus ROG Maximum Impact VII
    • 2 x 8GB Crucial Ballistix Sport DDR3 1600
    • Plextor M6e Series 256GB M.2 PCIe SSD
    • GALAX GeForce GTX 970 OC
    • HDPLEX 250W HiFI DC-ATX
    • HP Firebird 350W Power Supply
    I'll have a couple detail posts, but some assembly notes:
    • There are more than a couple things you'll want to take note from the S4 Mini assembly video. Be sure to watch it in full first.
    • I pre-ordered before I saw any mention of HDPLEX mods - the extra predrilled hole wasn't a big deal, attaching the barrel receiver to the case with one screw was fine for me. The studs that needed to removed was a bit harder. Josh posted directions, but I still had some issues - I ended up having to remove all the nuts completely and the fastening screws to remove the aluminum fascia completely, and then rocked the studs w/ some pliers to loosen, and put some gaff on to get some grip to unscrew. It was a slightly PITA.
    • The HDPLEX by default comes with a 6+8pin PCIe power, not a dual 6pin. Josh posted about rewiring them yourself but I ended up dropping HDPLEX a line, and they mailed me a 6+6. Awesome customer service. If you order directly from them, I bet you can request the appropriate PCIe cable in the first place.
    • You'll need to do some major cramming for the HDPLEX wiring, but it wasn't terrible.
    • The Galax GTX 970 OC isn't technically a mini-ITX sized card (it's a tiny bit longer), but it had no problem fitting (pic) - the only slighty weird thing is that even with it flush against w/ the insert slots in, there was a 4mm gap or so where the backplate screws were - not sure if this is normal, but I had to dig out some longer screws to secure the card.
    • I started off with a Zalman CNPS2X and a Scythe Slip Stream 120 for cooling, but it fell a bit short for my CPU, so I switched to the Noctua NH-L9i (I'll post power and heat details next).
    • As a point of reference, this system runs The Division beta at 1080p Ultra at 45-55fps (I'll probably end up doing a bit of GPU OC and turning down some settings to hit a solid 1080p60 if Nvidia driver updates don't fix things). Feel free to add me as a friend if you plan on playing: randomfoo2
    I'll be moving onto step 2, which is using this as part of a Backpack Rig (for completely untethered VR usage) soon...

    [​IMG]

    I finally published a full write-up of the backpack setup on my blog. I was meaning to get some non-janky mounting done for it, but never got around to it, so gaff tape it is, but you get the idea.

    [​IMG]
     
  2. lhl

    lhl Average Stuffer
    Thread Starter

    I posted some of my power/heat measurements in the main S4 thread, but I'll repost, along w/ more numbers w/ my move from the Zalman/Scythe to the Noctua cooler here.

    Wattage is measured w/ a regular Kill-a-watt, so good mainly for ballparks. CPU monitoring/stress testing is w/ OCCT. GPU/CPU+GPU is with Furmark.

    With the Zalman/Scythe (Scythe blowing air in, so the guard side towards the Zalman):

    Using HDPLEX for power, GTX 970 unplugged
    • Idle (800MHz) - 40-45W, 42C
    • w/ Turbo (4.4GHz) - 173W, hits 100C (thermal throttle) in <25s
      • throttles to 4-4.2GHz - 150-160W, 100C
    • w/ Turbo disabled (4GHz) - 126W, 88-92C
    Using SX600-G for power w/ GTX 970:
    • Idle (800MHz) - 95W, 42C
    • w/ Turbo disabled (4GHz) - 180W, 88-92C
    • Furmark (GPU 1088Mhz, 80-85C, 105% TDP, 85% fan, power limited) - 255W (CPU is about 65C, runs at 4GHz about 30% utilization)
    • Furmark + CPU Burner maxes out between 275-290W, both CPU and GPU around 80C.
    Today I got a 6+6pin PCIe power cable for the HDPLEX. Here's what it looks like w/ the HDPLEX+Firebird PS and the Noctua cooler:
    • Idle (800MHz) - 100W, 42C
    • w/ Turbo disabled (4GHz) - 179W, 75C
    • Furmark (GPU 1088Mhz, 79C, 104% TDP, 77% fan, power limited) - 269W (CPU 60C, about 30% utilization)
    • Furmark + CPU Burner - 290-310W, both CPU and GPU around 77C.
    Honestly I probably could have left the Zalman in - it's quite impressive considering the entire cooler is 85g and would handle a 35W/65W CPU fine, I imagine. The Noctua is much heavier (411g), although removing the Scythe and fan mount saves 100g, so it's not the worst tradeoff for a 15C cooler CPU under load.
     
    Josh | NFC likes this.
  3. Josh | NFC

    Josh | NFC Not From Concentrate
    NFC Systems Silver Supporter

    Thanks for the detailed write-up, especially the cooler comparison for people. I'm super interested in the battery as we discussed.

    The PCIE key not inserting all the way (4mm!!) sounds really weird. I'd love to know what caused that if you ever figure it out. Looking at the picture it seems like the ribbon cable might be the offender.
     
  4. lhl

    lhl Average Stuffer
    Thread Starter

    Yeah, for the battery, I had to order some XT60/XT90 connectors/cabling and the a 7.4*5mm barrel connector (the cheapest and easiest way which was to Amazon Prime an $8 90W PS that I'll chop off). I'll post some pics of all the various connectors and parts for the battery stuff since that I think will be the of the most interest. I believe my setup will end up being more compact/lighter than the other backpack rigs I've seen. The full S4 unit weighs in at about 3.3kg. The 12Ah 6S battery is 1.4kg.

    I'll post some detail shots on the PCIe key tmw (just leaving the office now) - it looks fully slotted and I double checked to make sure there weren't any wires snagged. The gap is weird.
     
  5. iFreilicht

    iFreilicht FlexATX Authority
    Gold Supporter

    Oh you were the guy with the VR Pelican build, I remember seeing that on [H]! Very interested in the whole untethered VR business, what sort of battery life are you looking to aim for? My expectation right now would be about an hour, probably less.
     
  6. lhl

    lhl Average Stuffer
    Thread Starter

    I need to run some in-VR wattage testing, but I expect a bit over an hour w/ the 12Ah battery. I actually have a spreadsheet with a bunch of different battery options (tradeoffs primarily around weight, Wh/kg and $/Wh). If I were getting paid to build this I'd add in a hot swap battery controller, but since I'm doing this just for personal use, I'll just be directly connecting the battery to the HDPLEX.
     
  7. lhl

    lhl Average Stuffer
    Thread Starter

    Quick update: it took a few orders of connectors more than I would have. I might take some more photos later...

    • Battery uses an XT90 connector. You might be tempted to buy an XT60 connector since it looks exactly the same - they are however, hilariously, not the same size (wah wah)
    • Battery charger comes w/ a Dean's connector to banana plugs, boo.
    • It was cheaper/faster for me to buy a power supply w/ a 7.4*5.0mm barrel connector to cut off than the connector by itself. That being said, it has terrible 16 gauge wire, so I may end up replacing it anyway.
    • I think I can use this parallel connector to do a battery hotswap...


    [​IMG]

    Oh, and the moment of truth after some somewhat shoddy soldering, heat shrink and electrical taping, typing this post from DC power. :)

    [​IMG]

    Sometime over the next few days I'll connect this to the charger which has a monitoring board and do a rundown test. (If anyone knows about a balance board that has a USB or serial interface to report into the system that'd be nice)
     
    Phuncz, iFreilicht and Josh | NFC like this.
  8. lhl

    lhl Average Stuffer
    Thread Starter

    Ran some Furmark without problems, wiring got a bit warm, but nothing serious (I really need to find my IR gun).

    Also, out of curiousity, ran Unigine Valley just to compare numbers.

    My 4790K/GTX 970 ends up performing about 50% better than the 6500/GTX 960 (even running 1200p vs 1080p):

    Unigine Valley Benchmark 1.0
    FPS:
    108.7
    Score:
    4548
    Min FPS:
    37.1
    Max FPS:
    203.4
    System
    Platform:
    Windows 8 (build 9200) 64bit
    CPU model:
    Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz (3997MHz) x4
    GPU model:
    NVIDIA GeForce GTX 970 10.18.13.6175 (4095MB) x1
    Settings
    Render:
    Direct3D11
    Mode:
    1920x1200 fullscreen
    Preset
    Custom
    Quality
    High

    One nice thing is that when running Unigine, my GTX 970 clocks much higher - rock solid at 1430MHz (vs 1088-1101MHz in Furmark) - memory speed remains the same at 3505MHz.

    The battery started at 24.6V (4.1V * 6S) and it's currently at 22.76V (3.79V * 6S). I'll run it until it gets down to 3.6V, but will need to rerun under a production load to get a feel for how long it'll last in real world usage...
     
  9. PlayfulPhoenix

    PlayfulPhoenix Just another human
    Site Staff Kimera Industries Gold Supporter

    This is insane. I only wish I had the ability to cobble something like this together... I'm thoroughly impressed :)

    What's the overall weight of all of this equipment? Seems backpack-able to me, but I'm also a fairly tall fellow...
     
  10. lhl

    lhl Average Stuffer
    Thread Starter

    Oh yeah, totally backpackable, is actually the point, but that'll be in a future installment. I'll do a full post on the DC system after I run some numbers (just killed my battery playing the Division Beta :) Dipped below 3V/cell! Oops (I have an alarm coming in which will make this safer/better), right now I'm just plugged into the battery charger monitor which doesn't have an undervolt alarm.

    (Once I finish putting it together I'll actually take all these posts together and do an all in one writeup so anyone should be able to put this together. Having some basic soldering skills would probably make life easier - there's just lots of bits and bobs to deal with. Also I suspect for the backpack mount, a 3D printer will make life easier, but you could also probably just use a lot of gaff :))

    Have some work deadlines and slowly trickling in parts (although, hooray for Amazon Prime). Once I get a full BOM would definitely be cheaper to assemble another.
     
  11. iFreilicht

    iFreilicht FlexATX Authority
    Gold Supporter

    Very cool! Good to see that it's stable under unrealistic loads, that's the first step!

    From what I know, a simple parallel connector could maybe work for a hot swap, ideally the output voltage of the battery on high and low charge should be very close to each other. Otherwise the nearly-dead battery will just be charged by the new one during the swapping process and could potentially blow up right in your face if that exchange happens too quickly (which it probably won't).
     
  12. Josh | NFC

    Josh | NFC Not From Concentrate
    NFC Systems Silver Supporter

    I love how this forum does all my R&D for me...

    Before you know it Westinghouse and Ratheon will be checking in. :)
     
    LooseNeutral and jØrd like this.
  13. lhl

    lhl Average Stuffer
    Thread Starter

    #13 lhl, Feb 2, 2016
    Last edited: Feb 2, 2016
    A 3.2V to 4.2V/cell swap is probably fine - I'm going to guess amperage will be well below 1C - I'll probably do some multimeter testing with some smaller voltage difference before doing a full/empty test, but honestly, not all that worried for an instantaneous swap - the main question will be will the HDPLEX get enough juice w/ the mismatch, but I'd guess there'd be a <1C charge, and these batteries are good for 10-20C discharge so should have lots of headroom.

    Oh, also a 10C 12Ah battery = good for a 120A discharge, so stability was never an issue - it's able to provide much more amperage than any of the DC power supplies available - you'd have no problems running a Fury Nano w/ the battery. As a reference, 350W @ 24V = 14.6A.

    The only currently fishy bit of my setup is the 16AWG wire is only rated for 13A, but like I mentioned, it didn't get to hot even under full load, so that was fine. Also at 6S*4.2V, it's 25.2V which is technically out of spec for the HDPLEX, but it seems pretty well built, it levels off below 24V after a few minutes (once it goes below 4V/cell) and I have a fire extinguisher handy if it's not ok. :)
     
  14. lhl

    lhl Average Stuffer
    Thread Starter

    OK, today I ran my first rundown test. Basically, took the fully/freshly charged battery (25.12V, 4.17-4.2V/cell) and was planning to run it until it ran flat, but one of the cells dropped out right before the 5hr mark, so I figure I wouldn't push it (as soon as load was removed, it popped back up to about 3.0V very quickly.

    So, for this first test, I booted up into my non-VR/game system - a bespoke Arch Linux install (aren't they all - mine is a pretty minimal Openbox/compton/tint2 setup). I just worked as usual today- about 30 Chrome tabs, a mess of terminals, Spotify, some YouTube videos and started a timer (also, had htop, nvidia-smi dmon, and powertop running on a desktop that I would check into).

    Hardware connected was the mobo GigE ethernet (I218-V), a USB3 hub/GigE, a Pure Pro KB (love this thing), mouse, and (powered) USB DAC/amp. About 4 hours in I plugged two phones to charge. As far as usage goes, CPU ended up about 25% idle, 35% 4GHz, and 30% 800MHz. GPU basically never broke 696 MHz and about 30W when playing videos and was normally around 20-25W and at 324MHz or even 135MHz according to nvidia-smi.

    Voltages stayed above 4V/cell well into the first hour. Here are the numbers I ended up tracking:

    0:00 - 25.12V - 4.17-4.2V
    0:10 - N/A - 4.12-4.15V
    0:20 - N/A - 4.08-4.11V
    0:50 - N/A - 4.00-4.06V
    2:13 - 23.08V - 3.82-3.85V
    3:07 - 22.68V - 3.77-3.79V
    4:00 - 22.40V - 3.72-3.74V
    4:52 - 21.40V - 3.19V-3.66V

    A couple minutes later, one low cell dropped off and I didn't feel like pushing things. When load was removed it came up at 2.75V, but as soon as it cooled (over the course of maybe a minute or two) things weren't looking so bad:

    [​IMG]

    All in all, not bad for what should be a 266.4Wh nominal battery. (Wall power on the Killawatt shows 50-55W in browser/Spotify use, and 90W watching YouTube vids.)

    More aggressive workload testing will have to wait until I finish some work deadlines, so probably not until end of the week or next week!
     
    Phuncz and iFreilicht like this.
  15. iFreilicht

    iFreilicht FlexATX Authority
    Gold Supporter

    You are better read in the matter of batteries than me, that's obvious. Good to know you're not just fooling around like an amateur. Now that the normal workload test has concluded so promising, I'm very curious to see you run the VR-Tests.
     
  16. Phuncz

    Phuncz Spatial Philosopher
    Moderator Gold Supporter

    Hmm I always thought Lithium-Ion and Lithium-Polymer batteries had a 3.5V min and 4.2V max. 3.10V sounds like a defective cell.
     
  17. lhl

    lhl Average Stuffer
    Thread Starter

    That 3.10V was at a full discharge. It seems like it's definitely the weakest of the bunch holding charge (or it may be heating up the most for some reason) since it started out balanced w/ the others. LiPos can generally safely discharge to 3.0V, although most people set alarms for 3.4-3.6V. I'll probably set my alarm for 3.2V since I don't have to worry about landing a drone or anything, and as soon as you lighten the load, the voltage should pop up. 3.7V is nominal rating, 4.2V is max.
     
  18. Phuncz

    Phuncz Spatial Philosopher
    Moderator Gold Supporter

    My memory must have screwed me on that one, that makes much more sense :D
     
  19. confusis

    confusis Writer, Editor, Awesome Person
    Site Staff

    Utter craziness. I'd love to see you use a 2nd S4 as the battery pack XD
     
    iFreilicht likes this.
  20. lhl

    lhl Average Stuffer
    Thread Starter

    #20 lhl, Feb 5, 2016
    Last edited: Feb 8, 2016
    Well, since I missed my Friday deadline and will be working all weekend (grumble grumble Android bitrot), I figure I'll start my DC power writeup (happy to take feedback. eventually this will probably end up somewhere more permanent)

    Power 201: What you should know about DC power

    So, this isn't "101" since I'm not going to cover *all* the basics, but instead will start off w/ some stuff:
    • The electrical grid runs on AC power. All your computer components use DC power. Typical computer power supplies (ATX, SFX, etc) convert AC to DC. If you use something like a PicoPSU or HDPLEX, these are DC to DC converters that take a certain input voltage and convert them to an output your computer components will use (typically 12V and 5V). The big wall-warts are the power supplies that convert AC grid power to DC power.
    • Batteries supply DC power. If you plan on battery powering your computer, you should prefer to use a DC-DC converter board, otherwise you will suffer power-loss conversion going from DC to AC to DC, which is just silly.
    • Here's a decent summary of AC vs DC power, and here's a summary of what a dick Edison was (or in Oatmeal form).
    • IMPORTANT: Before you start dealing with power, you should know that electricity can kill you. Currents as low as 200mA (0.2A) of DC can kill you. Here's a good video that has some info and here's a page that has some electrical safety notes. Please don't do any electrical work if you don't know what you're doing!
    Basic Safety
    • Always unplug before doing any work, including batteries. Note, AC power supplies can remain dangerous even unplugged, but you shouldn't need to touch those at all within the context of building a DC/battery-powered system.
    • Never leave any exposed wires - always use heatshrink/electrical tape around anything carrying live current.
    • Treat electricity like it can kill you - err, because it can
    • Don't cross the streams
    Note, if that last thing doesn't make sense, you should 1) download and watch Ghostbusters, you philistine, and 2) make sure that you read up enough on electrical safety that you do get it before proceeding. Or not, I'm not your mother. ;)

    Basic Theory

    When you hear PC folk talk about power, you mostly hear about wattage. A Watt is measure of power (the rate of doing work), 1 Joule / second.

    1W = 1J / 1s

    For those of you that remember your first physics classes, a way to visualize this is 1 cal = 4.18J, so 1W is basically the power needed to heat about 240g of water by 1 degree C in a second. This is useful to keep in mind as far as cooling goes, just for ballparking.

    More relevant to us though is the relationship Watts have with Amps and Volts.

    1W = 1A * 1V

    You can easily calculate wattage by multiplying amperage (current) by voltage. For more about this relationship, and its relationship with resistance, this primer is pretty good.

    When you are on AC grid power, your voltage is relatively constant (110V in the US, 220V in EU for example), and your amperage is limited by the breakers in your house. :)

    Doing simple math says that if you have a PC system that averages 300W from the wall, that at 110V, it'd be pulling about 2.7A.

    Battery/DC consumption is of course calculated the same way. 300W at say 19V means you need 15.8A.

    If you didn't know about this relationship, hopefully those numbers on your wallwarts make more sense now. You can also get a better sense what happens when too many amps are drawn.

    Batteries

    There are lots of different kinds of batteries. Lets start with basic battery chemistries:
    • Alkaline - these are your old-fashioned coppertops. They're not rechargeable and completely inapplicable to our discussion
    • Lead Acid - car/marine batteries, also used for UPS's - they retain charge well and are cheap, but their energy density sucks (you should be getting at least 3-5X w/ Lithium chemistries). They have lots of subtypes (AGM, Gel, etc), but we don't really care
    • NiMH - better energy density than lead acid, but still, don't care. These guys have a memory effect and heavy metals (definitely need to recycle/ewaste). Again, don't care.
    • Li-ion/LiPo - These are the lithium ion batteries that are in pretty much all electronics these days (typically LiCoO2). They have the highest energy density available (until Li-S batteries start popping up), and are standardized at 3.7V nominal, 4.2V max charge per cell. Note, the only difference with "polymers" is referencing the form factor (a polymer pouch vs a cell casing). As you should know from the news or traveling, these can get flamey/explodey if undercharged, overcharged, or otherwise mistreated (temperature, pressure)
    • LiFe - These are worth noting as they have lower energy densities than other Li-ion chemistries (there's also NMC but you won't really find those around) but are more stable. These are commonly used in electric bikes and for utility storage and are 3.3V nominal, 3.6V max charge per cell.
    OK, that about covers it. LiFe are an okay (safer, longer cycle life) option, but lets be honest, for energy density (Wh/kg), cost ($/Wh), capacity (Ah), and continuous discharge (A or C) it turns out LiPo's used for RCs and drones are by far the best suited for powering a computer.

    Lets break down those terms really quickly:
    • Watt-hours (Wh) - simply how many watts/hr a battery can supply. In theory, a 500Wh battery for example would be able to power a 100W system for 5 hours or 250W system for 2 hours. In practice though, the higher your draw (amperage) the less power capacity you have (heat, resistance, etc increase)
    • Amp-hours (Ah) - Wh is typically a computed number from the *actual* this capacity (how batteries are properly rated) - well, technically it's all just spitballing, but you get the idea. Wh is Ah * nominal voltage. (BTW, for those wondering what "nominal means" - that's because the batteries have a range of voltages that is a curve based on their discharged state (and rate!))
    • Voltage (V) - your DC-to-DC board takes a certain input voltage. Note, some DC boards like the picoPSU expects 12V and does not do any voltage conversion. While a bit more efficient, this is the best since a 12V nominal battery will probably pass in something like 15V/16V at max charge and less than 12V when it gets low. Also, a lower voltage requires higher amperage. Consuming 240W @ 12V means you need 20A of current, vs a 24V batter only needing 10A.
    • Amperage (A) - this is the current that's being drawn (or charged). Higher current drawn will negatively effect battery life as temperature and internal resistance increase. If none of what I'm saying makes sense, this guide to LiPo batteries does a good job of giving more background.
    • Capacity/C Rating (C) - this is a standard used by RC/drone batteries in particular for measuring discharge rate. C is a multiplier number against the Ah capacity of a battery. There is typically a continuous and instantaneous/peak discharge number. A 10C 12Ah battery is rated at 120A continuous discharge.
    • Energy Density (Wh/kg) - Watt-hour per kilogram is the standard metric for energy density. Consumer grade RC/drone cells are about 180-200 Wh/kg. LiFe batteries are around 100-120 Wh/kg. Perhaps one day soon, Li-S batteries w/ 400-500 Wh/kg energy densities will show up, but I wouldn't hold your breath, next-gen battery tech is always coming next year (or better yet, 5-10yrs). Calculating this number is just what's on the tin: nominal voltage * capacity / weight.
    • Cost Ratio ($/Wh) - Even more obvious. Typically the price decreases as capacity goes up (since there are fixed costs associated w/ a battery). This is most useful for comparing different battery types and seeing if some other characteristic is worthwhile.
    Let's pick a battery!

    OK, so I have a spreadsheet of options, but I'll make this easy and just tell you which one to get. You want to buy a MultiStar High Capacity 6S battery from Hobby King. The 6S wiring gives you a 22.2V nominal voltage (25.2V max, 18.0V min) which puts it squarely in the HDPLEX HiFi 250W DC board's 16-24V input range.

    The Multistar "only" have a 10C continuous discharge rating (120A for a 12000mAh, 160A for a 16000mAh), which is 100-140A more current than you should ever need for your system. You wires will melt long before you touch those amperages. (Oh yeah, definitely take a look a the American Wire Gauge (AWG) current capacity tables). For our purposes 10 gauge is overkill, 12-14 gauge is good, 16 gauge is a bit weak sauce.)

    Post too long, continued...
     
    iFreilicht, Phuncz and confusis like this.