1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Documenting my Bolt Experiences...

Discussion in 'TiVo Bolt DVR/Streamer' started by clay.autery, Feb 3, 2018.

  1. Feb 10, 2018 #81 of 135
    CIR-Engineering

    CIR-Engineering Video Calibration & Electronics Repair Engineer

    176
    81
    Sep 24, 2017
    Chicago USA...
    I have the same concerns as you with cooling on the BOLT however I am even more concerned with the hard drive health than SoC ODT. I think it's getting very hot in that case and that may be what lead to the first hard drive failure. I really want an internal hard drive just to keep everything neat and together. I'm probably going to cut a hole and mount a second 120mm fan right above the heat sink to push a lot more air through. I'll probably make some holes on each side of the case too to allow the air to flow through more freely.

    We'll see, I'm in the early stages of planning. Right now the case cover stays off until I come up with something.

    If you wind up with extra case parts let me know because I'd be interested in buying them.

    Kind regards,
    craigr
     
  2. Feb 10, 2018 #82 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    I will likely do 3.5" Reds in external enclosures temporarily and then incorporate them into a different, better chassis with a proper thermal control design.

    I've read all the discussions about PMR vs SMR and all that, and I am sure that is one factor, but I believe the head parking issue AND thermal cycling from power cycles, and mechanical wear from spin-up/spin-down are equally or in sum, more important in the lifespan calculations.


    Yup, case tops stay off for now except for a few more tests I need to complete. Processor temperature is a concern for me solely from the desire to maximize the time I can keep this particular TSN/MAC functional to maximize the value of the lifetime service agreement.

    Yes, thermal control of the disk drive is important too. Just because the drive has a 60C max doesn't mean you should run it anywhere NEAR that temp.
    I will likely subdivide the disk drive area from the motherboard area in order to isolate the drive for thermal control. Additionally, I intend to mount the drive on isolation dampers to reduce for vibration reasons. Proper dampers convert the vibration to heat which a) stops them from re-transmitting back into the device, and b) lowers acoustic transmissions into the chassis and/or out of the chassis.

    Not sure if I'll have extra chassis parts, but if I do, I'll let you know.

    In fact, I need some Bolt screws right now.... the short and a couple long screws for plastic. The BackBolt was assembled almost entirely with machine screws (from the factory.... no sign at all of it ever having been opened to include inspecting the thread marks in the screw bosses)

    Enjoy the rest of your weekend! :)
     
  3. Feb 10, 2018 #83 of 135
    CIR-Engineering

    CIR-Engineering Video Calibration & Electronics Repair Engineer

    176
    81
    Sep 24, 2017
    Chicago USA...
    Yup, same here. I want my lifetime service to be worth it. I don't care if the SoC is rated to withstand running at 1,000 degrees C. The chips are not designed to last forever and the cooler they are run the longer they will last. We learned that in Electrical Engineering 101; etched chips are designed to last about five years. In practice they typically last longer, but cooler run chips last MUCH longer.

    Who knows, with ATSC 3.0 this all may be a moot point. But, my gut tells me that I will have plenty of use for my ATSC TiVo's for at least another five years or more. I think ATSC 3.0 will take a long time to catch on if it ever does. People don't feel like upgrading all their hardware and network affiliates don't want to spend millions to do it again when they just upgraded less than 15 years ago. IMHO ATSC 3.0 stations will be few and far between.

    craigr
     
    clay.autery likes this.
  4. Feb 11, 2018 #84 of 135
    bootman_head_fi

    bootman_head_fi Member

    35
    0
    Aug 3, 2008
    The bolt really needs to be in a standard case vs the fancy curved case it is in now.
    Marketing vs engineering usually results in increased hardware churn. ;)
     
  5. Feb 11, 2018 #85 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    Yes, it should.

    IMO, the curved case was MORE to make it the top device in a stack of devices in situations where it doesn't have its own perch/shelf, than it was to provide airflow under the case. Taller bumps on a standard box-type chassis would have insured clearance for airflow. In fact, the weird under-chassis space encourages an air-flow short circuit where hot air exiting the fan grill is piped back into the case via the (insufficient open area) intake vents. The ABS case has insufficient material or rigidity to have supported any appreciable weight had it been molded in standard dimensions.

    Additionally, in their zeal to make it tiny (and save materials), the internal volume is woefully insufficient for thermal flow, proper sized heatsink, et al.

    After analyzing it just a bit more, I believe I could write up a prioritized bullet list of engineering factors. Near the top would be "Minimize cost of production." Somewhere, some ways down the list would be "Thermal design minimally sufficient for x% of total sold to NOT fail due to high heat until the warranty expires." I'm guessing durability, customer convenience, flexibility, and many more cool things weren't even on the list. ;)
     
  6. Feb 11, 2018 #86 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    BoltBack Thermal Management:

    02/10/2018 - Fan on OE 2-pin power header has dropped idle temperature of the box from 54° C to 49° C fully assembled. Will post back once the materials arrive to build the inline tap for the 4-pin, HDD power header.

    02/11/2018 - This will be updated in a piecemeal fashion as I accomplish other tasks.

    At Idle, OE Fan (NCC): 54° C
    At Idle, OE Fan (CC): 60° C
    At Idle, OE Fan, LS elevated (CC): 59° C
    At Idle, Noiseblocker BlackSilentFan XS-2 (NCC): 49° C
    At Idle, Noiseblocker BlackSilentFan XS-2 (CC): 61° C
    At Idle, Noiseblocker BlackSilentFan XS-2, LS elevated (CC): 56° C
    Sustained transfer, OE Fan (NCC): 58° C
    Sustained transfer, OE Fan (CC): 64° C
    Sustained transfer, OE Fan, LS elevated (CC): 64° C
    Sustained transfer, Noiseblocker BlackSilentFan XS-2 (NCC): 59° C

    * CC = with CableCard; NCC = No CableCard
    * LS elevated = Left side of chassis elevated until internal fan is parallel to ground. Idea is that this will assist in breaking the air circulation short circuit and allow a little more efficient internal convection.

    ====================

    Variable Fan Speed on 2-pin Header..... redux.

    Added 02/11/2018 - Answer? Still don't know, for sure. Yes, I ran the test below, but I MISSED the much higher speed during cold boot-up. The fan CLEARLY runs at a higher speed when the machine is first plugged in. I am awaiting arrival of some pre-made cables from China for another purpose. I will use some of the spares to make an inline tap for the 2-pin header.

    This will make it easier to run a boot-up test while monitoring the fan voltage. This voltage measurement and logic should better answer the fan speed control method question.

    Then, I will run ANOTHER in operation test where I monitor the voltage and fan speed while I artificially raise it to temps higher than 82° C (as high as I can stand it, or until the high speed switch kicks in.... whichever comes first.)
     
    Last edited: Feb 12, 2018
  7. Feb 11, 2018 #87 of 135
    CIR-Engineering

    CIR-Engineering Video Calibration & Electronics Repair Engineer

    176
    81
    Sep 24, 2017
    Chicago USA...
    Hey clay,

    Have you taken the heat sink off any of your BOLT's yet? If so, was it a thermal pad or thermal past you found? If you haven't yet I would like it if you'd take the heat sink off your junk BOLT because I really want to know. If it's a pad I'm probaly not going to change anything, but if it's past I will definitely level the heat sink and replace the paste.

    I just took the heat sink off my junk Roamio and was rather surprised to find paste. However, the Roamio also has a metal inner case that doubles as a second heat sink with a heat transfer pad attached to the PCB directly under the Broadcom. So it's cooled from above and below.

    I'd like to know your findings.

    craigr
     
  8. Feb 12, 2018 #88 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    Nope, haven't taken it off yet, but from what I can tell with my mag glasses and a really bright light, it is some kind of phase change pad like these from Bergquist:

    Phase Change Pads

    They can be machine applied, and they are temperature activated, actually flowing at a set temp... The idea is that you thermal cycle the process/pad/sink sandwich at the right temp, typically ABOVE the normal op temp. Keep it there for a certain time, the pad flows and bonds the sink to the processor (thermally, not really mechanically). The springs in the pin fasteners put a minimal clamping pressure on the bond. Then you operate it below the flow temp and there ya go.

    I don't have a "junk" Bolt yet. I have BoltBack running as a backup with a 4TB drive in it and am also using it for testing purposes.... That's where all those temps in Post #86/#2 came from.

    I AM going to remove the heatsinks though. I just dropped an order for custom hardware for mounting the heatsinks with higher spring tension, and another setup that will be semi-rigid. I won't be replacing the pad with another pad, even the top-shelf Bergquist stuff I have here UNLESS the processor top is not true or has some massive run-out. I will be using Arctic Silver in the thinnest layer possible after I lap the heatsink mounting pedestal flat and smooth. Hence the custom hardware. Those spring-loaded nylon push pins are not intended to be removed and replaced.... you can get away with it maybe 6-12 times if you are careful... maybe more.

    If you remove your heatsink and it is a pad, and you want to continue with a pad, it will need to be replaced. And if you do that, it'd be better to just go with AS5 or similar.

    Yep, I've never seen a Roamio up close, but from your description, it is clear that it was much better built at least from a thermal design perspective.

    I will keep you updated! :)
     
    Last edited: Feb 12, 2018
  9. Feb 12, 2018 #89 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    BoltPrime Hard Disk Drive Replacement

    02/11/2018 - Installed the 3TB Toshiba drive in BoltPrime today using essentially the same procedure used on BoltBack on Saturday.

    The only issue that arose was that the CableCard issues returned. Same 8 channels as the first time. Spent an hour on phone with Suddenlink. Verified that we agree on the Cable Card numbers. They sent the re-pair signal several times with various combinations of TiVo reconnects, cold reboots, et al. No joy. They are rolling a truck for the 3rd time... supposedly with a replacement CableCard.

    Hardware Order Submitted

    Researched and submitted an order today for a complete complement of Bolt fasteners to include the plastic and metal self-tappers, the machine screws for the WiFi card retention, and the machine screws for the CableCard bracket/slot.

    Additionally, I calculated the specifications and submitted an order for heatsink mounting hardware in anticipation of replacing the assumed phase change pad with proper silver-based thermal interface compound.
     
  10. Feb 12, 2018 #90 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    BoltPrime Thermal Observations: Had a truck roll from Suddenlink again. The CableCard mysteriously lost the same 8 channels it had a problem with originally. So, I decided to put the case tops on and make some thermal observations with the replacement fan installed.

    DATA - Here are the temperatures gathered for BoltPrime to date:

    At Idle, OE Fan (CC): 73 ° C
    At Idle, Noiseblocker BlackSilentFan XS-2 (CC): 64° C
    At Idle, Noiseblocker BlackSilentFan XS-2, LS elevated (CC): 61° C
    At Idle, Noiseblocker BlackSilentFan XS-2, LSFI elevated (CC): 59° C
    At Idle, Noiseblocker BlackSilentFan XS-2, Vertical (CC): 54° C
    Sustained transfer, Noiseblocker BlackSilentFan XS-2 (CC): 67° C
    Sustained transfer, Noiseblocker BlackSilentFan XS-2, LS elevated (CC): 62° C
    Sustained transfer, Noiseblocker BlackSilentFan XS-2, LSFI elevated (CC): 60° C
    Sustained transfer, Noiseblocker BlackSilentFan XS-2, Vertical (CC): 56° C
    • CC = with CableCard; NCC = No CableCard
    • LS elevated = Left side of chassis elevated until internal fan is parallel to ground. Idea is that this will assist in breaking the air circulation short circuit and allow a little more efficient internal convection.
    • LSFI = LS as above with addition of fan exhaust isolator. (aka: . See images/dimensions.
    • "Sustained transfer" = Recordings being written TO the device under test (DUT) from another device.
    Note: All temperatures after "At Idle, OE Fan" are with 3TB Toshiba drive in place of the stock WD 500GB drive.

    The following images illustrate "Left Side elevated" (LS), fan isolator prototype, and "Left Side elevated with Fan Isolator (LSFI), and Vertical.

    [​IMG][​IMG]
    [​IMG][​IMG]

    ASSERTIONS/ASSUMPTIONS - As expected, the data here, combined with the BoltBack temperature observations, support the following:

    a) The chassis "legs" are too short. On the right end of the case, half the vents leading to the case interior and CableCard bay are heavily obstructed. All of the limited inlet venting could benefit from being elevated to some degree.

    b) The stock 15mm tall fan combined with the short interior height combines to reduce the headroom over the fan to such a point as to drastically reduce it ability to exhaust hot air from the chassis. The increased headroom over the 10mm tall fan clearly improves the exhaust flow.

    c) An air-flow short circuit exists under the chassis. The curve under-chassis space inhibits rather than assists airflow. Air exhausting from the fan appears to be constrained under the chassis and actually feeds back into the intake vents.

    d) Elevating the left side until the internal fan is parallel to the ground reduces observed ODT. This reduction is most likely due to a partial correction of the airflow short-circuit combined with improved interior conductive airflow. Increasing angles appear to further improve the disconnection between exhaust and intake (see e below) and encourage further increases in convective airflow interior to the case. Placing the chassis vertical resulted in an additional 5° C ODT temperature reduction, the lowest idle and loaded temps for any position thus far. NOTE: Simply elevating the left side helps, but insuring the right end inlets have a clear "view" of the exterior air supply also helps. (See right end in images below)

    e) Further remedies for the airflow short-circuit lower the ODT temperature. In horizontally oriented positions, adding a barrier that walls off the higher temperature exhaust from the inlet vents that also serves as the elevating method assisted in reducing the ODT. Not a bunch, but it serves to prove that keeping the inlet air supply as cool as possible is important.

    f) The configuration above is probably 90%+ of the maximum performance that can be achieved in stock trim.

    Bottom Line: To achieve idle temps much lower than 59° C, we must a) get more heat out of the chassis, put more cool air into the chassis, increase airflow over the heat generators, especially the processor heatsink.

    Potential TODO List:

    1) Build a second fan isolation/elevation prototype.
    2) Add ducted/forced air inlet to the prototype in #1
    3) Test #2 with internal fan removed
    4) Modify prototype to add a bigger fan to "vacuum" air OUT of the exhaust section.
    5) Other stuff as it occurs to me... Some things will be tested on BoltBack for proof of concept prior to use on BoltPrime.

    I'm hungry! :D
     
    Last edited: Feb 13, 2018
  11. Feb 12, 2018 #91 of 135
    justen_m

    justen_m Cheesehead

    8,824
    221
    Jan 14, 2004
    Boise, ID
    Clay, you're a bigger geek than me! LOL,
     
    clay.autery likes this.
  12. Feb 12, 2018 #92 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    LOL! Could be, could be... Love this stuff! Haven't done much of it in decades, but it's fun.

    Ever lapped a heatsink or a processor flat? :laughing:
     
  13. Feb 12, 2018 #93 of 135
    CIR-Engineering

    CIR-Engineering Video Calibration & Electronics Repair Engineer

    176
    81
    Sep 24, 2017
    Chicago USA...
    Please do keep us updated.

    That said, I suspect that the heat transfer between the Broadcom SoC and heat sink is adequate based on two simple findings:

    1) Just removing the cover lowers the ODT temp dramatically.
    2) Putting a fan on the heat sink proved to lower the ODT temp even more and also dramatically.

    If heat transfer was an issue between the Broadcom and heat sink I don't think opening the case, or especially adding a fan over the heat sink, would have such a large decrease in ODT temperature.

    I am eager to see what you find when you replace the thermal pad with Arctic Silver 5 and level the heat sink. However, my hypothesis is that there will not be much change.

    Also, I'm not making this guess in ignorance. Though I do not practice it, I have a 25 year old degree in mechanical engineering degree on heat and mass transfer. If you were able to put a thermal probe right under the heat sink next to the Broadcom and compare that with the ODT that would tell us something. A large temperature differential between the ODT and the hottest part of the heat sink would indicate a problem. A smaller temperature difference would obviously indicate the heat transfer is adequate.

    Happy hunting,
    craigr
     
  14. Feb 12, 2018 #94 of 135
    justen_m

    justen_m Cheesehead

    8,824
    221
    Jan 14, 2004
    Boise, ID
    Sorry, no not that big of a geek.
     
  15. Feb 12, 2018 #95 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    Some 18 years ago when I was developing the original AS, we went through great expense and effort to accurately determine the best conditions under which the product would perform, application procedures, etc, etc ad nauseum. We used thermal probes, data acquisition hardware/software, all kinds of fixtures with sensors, and the like. We did conclusively prove that it was the best formulation on the market at the time if preparation and application were done properly.

    The physics are not trivial of course. And you are correct in assuming that in general, optimizing the thermal junction between a heatsink and a processor (or SoC) is well down the list of things to do for improving thermal performance. But when you've done everything else and those last few degrees are needed, it can be a game maker/breaker.

    In an ideal world, the thermal interface would be infinitesimally thin because the mating surfaces were so flat and so smooth that they met each other at 100% contact. Real world never gets there. but properly done, you can with little investment get to 0.001". I've actually made it to sub 1/1000th of an inch before. At these super thin layers is where the original AS stood out, because it could conform. Most thermal compounds have particle size variations such that the interface thickness is a function of compound particles holding the gap open in highly prepared interfaces. AS used really consistently sized and specially manufactured silver particles that would naturally WANT to lay out in a nice layer, fill in microscopic valleys, and minimize carrier "pooling". The carrier has poor thermal transfer qualities compared to the silver.

    Somehow, I managed to NOT get out of California with any of that lab equipment... long story.
     
  16. Feb 13, 2018 #96 of 135
    CIR-Engineering

    CIR-Engineering Video Calibration & Electronics Repair Engineer

    176
    81
    Sep 24, 2017
    Chicago USA...
    No doubt that AS and proper leveling will allow easier heat transfer between the Broadcom and heat sink. I’m just saying that your experiments have proven simply removing the case and plopping a fan on the heat sink lowered the ODT 32 degrees C to a comfortable 41 degree C operating temperature. IMHO running the chip that cool is a plenty fine improvement without the trouble of chasing a few more degrees. I think the BOLT will outlast its usefulness with the SoC at 41 degrees C before the SoC fails ;-)

    craigr
     
  17. Feb 13, 2018 #97 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    Quite likely, but stopping now wouldn't be any fun at all! And I already have the replacement heatsink hardware on the way. ;)
    And as much as I dig the open chassis look, it's not a viable long-term solution. Although, I might consider clear sides and top for the new chassis. :cool:

    Check this out.... Propped BoltPrime up vertically a couple of hours ago. Right this minute, the ODT is down to 52° C. It's 0049 hrs, and the thermostat has been set at 65° F since 2200 hrs, so I suspect the temp is a bit lower than it was a few hours ago. In fact, the temp on the front of the Rosewill eSATA box with the 1 TB drive in it is exactly 2 degrees cooler than average. So I will normalize the vertical temp to 54° C. That is still a FULL 5° C reduction over the "Left Side elevated, Fan Exhaust Isolated" (LSFI) pictured in Post #90 and #1 above. I'm a little surprised.

    The vertical position does concern me a little, primarily for the disk drive. I'm wondering what the drive temp is right now. Reckon the temp is higher than when the case sits flat or in the LS/LSFI positions? Additionally, you think having the disk drive motor almost on its side has any longevity implications regarding lubrication retention?

    NOTE: BoltPrime is my ALL-IN device that runs 73° C in stock trim with no mods and sitting horizontally.

    [​IMG]
     
    Last edited: Feb 13, 2018
  18. Feb 13, 2018 #98 of 135
    Worf

    Worf Well-Known Member

    2,281
    64
    Sep 15, 2000
    Wait... you put your TiVo on top of your (high) heat producing AVR? That already seems a recipe for disaster - no amount of built in cooling can help if the device is soaked in hot ambient air from your receiver.
     
  19. Feb 13, 2018 #99 of 135
    clay.autery

    clay.autery Member

    151
    11
    Feb 2, 2018
    Northwest...
    I thought VERY same exact thing when I saw the initial ODT measurement of 73° C. Here's what I said in Post #1 of this thread:

    POST #1: "This temperature reading shocked me. Isolating BoltPrime completely from external heat sources reduced the temp perhaps by a degree..."

    I temporarily set the machine up in a completely different place with no heat producers anywhere close. It MIGHT have made a one degree difference. So I moved it back out of convenience. ALL temperature readings in this thread are from that position atop my Samsumg Blu-Ray Player, which sits between BoltPrime and the Onkyo A/V Receiver. Apparently, it isn't having as much impact as one might expect, the ambient airflow is sufficient to reduce its impact to </= 1° C. OR, maybe the BlueRay Player (OFF 90%+) with its aluminum chassis is deflecting/absorbing most of the heat coming up from the AVR. Note that in the pictures the AVR vents beneath BoltPrime's right side vents are covered with CD/DVD sleeves to deflect the AVR exhaust. Additionally, the temps recorded in this thread are based on revised positions that keep BoltPrime wholly on top of the Blu-Ray Player.

    While the AVR IS certainly a consideration, it is not at the top of the list.... yet. At some point, I will likely add a boxed riser between the AVR and components above with the "bottom panel" lifted on the diagonal as a baffle to direct the exhaust air to the rear away from other components.

    Good thinking though!!! :)
     
    Last edited: Feb 13, 2018
  20. aaronwt

    aaronwt UHD Addict

    21,343
    476
    Jan 31, 2002
    Northern...
    I still have a 15mm, 4TB drive running in my Bolt. It has been running since October 2015. Although I did have another 4TB drive die last fall. But the drives did go through multiple Bolts. But since I got lifetime Bolts last year, I don't plan on getting any more. My previous two Bolts were on yearly. And the first two were from the eBay scam. Which I still have sitting in a closet for spare parts.
     

Share This Page