Back to normal!

Discussion in 'Pyra News' started by EvilDragon, Oct 11, 2018.

  1. Xtreme2k

    Xtreme2k Member

    Joined:
    Apr 10, 2009
    Messages:
    41
    I noticed the same. The whole cases story starts to get ridiculous. After endless back-and-forth discussions, ED decided to have those 500 clear ones produced by FormAction and then move the molds. News from Jul-22: "The molds for the cases are being moved AFTER UniquePlast comes back from holidays, and that is middle to end of August." Now it's middle of October, no clear cases, no moved molds, no explaination why. And now even the clear "final prototypes" seem to be not really useable.
     
    Swordfish II likes this.
  2. TrashyMG

    TrashyMG Sarcasm Dispenser Staff Member

    Joined:
    Jan 18, 2010
    Messages:
    10,474
    Before you get all worked we don't know what Notaz was even given for a casing, it could have come from some of the samples prior to the 500 clear production run, which is what most developers got.
     
    Grench, CommanderB and crazyhorse like this.
  3. Xcl4m4t10n

    Xcl4m4t10n Advanced Member

    Joined:
    Dec 18, 2009
    Messages:
    1,060
    Talking about that, is there something 100% finish? Because ifair, nothing is REALLY 100% finish, only finished in theory but not "doable", due to little issue and/or third party issues.
     
    CommanderB and ThinkPad like this.
  4. Djoga'Ro

    Djoga'Ro moonstruck

    Joined:
    Apr 3, 2016
    Messages:
    831
    @Xcl4m4t10n Talking about hardware - what's not finished? Well, testing isn't done, since software wasn't ready to enable proper testing, but unless testing uncovers design bugs, all HW could be considered done (design-wise). Or am I missing something?
    And then it's just a matter of ordering, lead times, assembly, QM, and shipping, in't it.
     
  5. ThinkPad

    ThinkPad Very Active Member

    Joined:
    Dec 21, 2012
    Messages:
    399

    This is not looking good at all.
     
    Swordfish II likes this.
  6. TrashyMG

    TrashyMG Sarcasm Dispenser Staff Member

    Joined:
    Jan 18, 2010
    Messages:
    10,474
    @ThinkPad one unit isn't indicative of the entire prototype batch.
     
  7. Bosbeetle

    Bosbeetle Terminally lost

    Joined:
    Sep 7, 2008
    Messages:
    3,672
    Location:
    The Netherlands
    Thank god (or somebody or just well lets not go into this) that they're looking at that stuff, will only make the pyra better, worked for the pandora, I expect a lot of 'ironing' on the software side... It's good at this point to see what problems are hardware and which are software...
     
  8. Djoga'Ro

    Djoga'Ro moonstruck

    Joined:
    Apr 3, 2016
    Messages:
    831
    @ThinkPad I tried to hint at the journalistic quality of posting here just the conversation starting email, omitting its response and the response's response ...
    It seems, my hint was too subtle.
     
  9. crazyhorse

    crazyhorse Very Active Member

    Joined:
    Sep 9, 2009
    Messages:
    267
    Yeah, sorry, there is some discussion on this going on at the mailing list. Most of the issues are os related and being looked at.Looks like its a new untested unit.
     
    Djoga'Ro likes this.
  10. ThinkPad

    ThinkPad Very Active Member

    Joined:
    Dec 21, 2012
    Messages:
    399
    Your and the @TrashyMG subtle hints are looking more like damage control.
    Hope i'm wrong. :)

    no hard feelings.
     
  11. ClockworkCoder

    ClockworkCoder Chaotic Neutral

    Joined:
    Jan 21, 2016
    Messages:
    929
    Location:
    Menzoberranzan
    There are a bunch of responses (that I've read) in the newsgroup related to that post. Most issues were cleared up in subsequent replies to so don't worry about the apparent breadth of issues listed in that post.
     
  12. rv6502-2

    rv6502-2 Still Fresh

    Joined:
    Jan 17, 2016
    Messages:
    20
    Location:
    Volcano Base
    Normally the DLL (Delay-Locked Loop) should only be disabled to run RAM at very low clock speeds (for power saving).
    Disabling the dynamic DDR RAM clock speed might also fix the issue.

    That would explain why the RAM crashes when the device is idle and the RAM controller drops the memory clock.
    It might be possible to turn DLL off before power save and re-enabling it when going back to full clock rate but it could be risky if that causes bus glitches during retraining/resync.

    DDR RAM is always a pain to get stable.
     
    Djhg2000, error and klapse like this.
  13. Djoga'Ro

    Djoga'Ro moonstruck

    Joined:
    Apr 3, 2016
    Messages:
    831
    @ThinkPad It is damage control. Just damage, that hasn't manifested yet and ahs no physical cause - i.e. purely informational.
     
  14. Swordfish II

    Swordfish II Very Active Member

    Joined:
    May 20, 2015
    Messages:
    748
    Ok this is super troubling...
     
  15. directive0

    directive0 Very Active Member

    Joined:
    Apr 8, 2015
    Messages:
    762
    Location:
    Toronto, Canada
    Yep thats a big scary list, isn't it?

    Now I see why big companies don't share their production prototype troubleshooting with consumers. Still looks pretty good to me.
     
    xnopasaranx and error like this.
  16. EvilDragon

    EvilDragon Administrator Staff Member

    Joined:
    Mar 4, 2003
    Messages:
    19,652
    Location:
    Ingolstadt
    Well, that's not really scary, and here's why:

    That's simply because the PWM is currently programmed to run at 82Hz, which flickers.
    This is not a hardware issue but software, as the PWM can run up to 1MHz.

    No one has yet worked on finding the perfect results.

    That's something that needs to be checked and confirmed. As we were busy getting the units to work without freezing, all the boards I have here (over 20) have not been tested yet.
    Doing tests when the units freezes regularly is no fun ;)

    The audio driver doesn't work with the current kernel and needs to be fixed.
    This is quite an annoying issue, and most probably the biggest thing we need to work on softwarewise right now.
    Hardware is fine.

    As mentioned multiple times if you followed the newsposts. They are still working on the shoulder buttons, but I don't expect them to be perfect until I actively work with the new company on them.
    Pandoras shoulder buttons were the same, and we got them to work eventually in the Rebirth / Ghz units.

    No one has seen that yet, so it could be an issue with his board.

    OTG is working on Nikolaus unit, but it has been fixed recently, so it's very likely notaz doesn't have those fixes in his setup yet.

    Probably magnet not strong enough. This is the first unit that has a magnet in the lid, but with the freezing units, I couldn't test them.
    It can very well be that I simply need to source stronger magnets.

    The switch itself works, as Nikolaus confirmed with his unit.

    The unit I've built for notaz has been built pretty fast, without any testing done (as he was leaving on a trip and wanted to take it with him).
    The way it sounds like is that I accidentally grabbed one of the LCDs that I haven't approved for the production - as that was exactly the issue with them.
    The ones I approved are working fine.

    That's certainly broken, as these are tested and working components.

    Nikolaus has tested all the hardware and the design itself is fine.

    Just as a reminder:

    These prototype boards have MULTIPLE purposes.

    1. They are needed to we can TEST if everything works well and then report back to GC, letting them know what's wrong. A high failure rate is normal for such production runs, as the reason you do that is to find out what can possibly go wrong when populating and then find fixes for them before you go into mass production.
    Just a quck information (as most of you don't know this): The first production run of the 1GHz Pandora had a failure rate of around 29%! And basically the only thing that has changed compared to the previous version was the SoC! Later production runs went down to 2% failure rate.

    2. Once the boards are fully tested and fixed, they go out to developers and the prototype-preorders so they can work on the remaining software stuff (like audio driver, making the desktop more usable on standard setup, etc.)
    Except for some few people which do low-level stuff (like fixing the memory freezes!), thes boards won't be shipped before they have been fully tested. Thanks to the freezes, I couldn't test the mainboard notaz received, so some defects are nothing special but to be expected.

    3. Now that the freezes are basically fixed, I can work on the tests, give feedback to GC, fix any boards that are broken and send the working units to devs / prototype preorders. This will take some time (I need to write the scripts to do the tests and then test all the boards), but it can be done now and couldn't be done as long as the unit I was working with froze every few minutes.

    It's nothing new that there's still a lot of work to be done to the software, this has never been hidden. It can be seen on mailing lists and on the developer-website.
    It's also nothing special that there are various priorities here. Fixing the freeze had a higher priority than testing the backlight flicker, especially as it was clear that this is not a hardware issue but needs to be setup properly using software.

    The biggest issue were the freezes - as these held up all the testing and getting the units in the hands of developers. This was kind of a vicious circle.
    And, of course, the case company, which is REEEALLY slow (which is also a reason we switch).

    So, everything clear now?
    --- Double Post Merged, Oct 15, 2018, Original Post Date: Oct 15, 2018 ---
    Yes, it's testing and tweaking the machine setup for mass production (but that's something GC does themselves based on our feedback on the mainboards, which I couldn't give them because of the missing tests).
    Basically, there should be no known hardware issue left right now.
    --- Double Post Merged, Oct 15, 2018 ---
    Yep, exactly that. The one I still wasn't satisfied with.
    --- Double Post Merged, Oct 15, 2018 ---
    Explanation? FormAction was still tweaking the shoulder buttons (I even mentioned that) and they're still incredibly slow.
    Both are reasons I want to switch to UniquePlast, I guess you can now understand why.

    The clear cases ARE usable. Even if the shoulder buttons aren't working perfectly from the beginning, they can be made to work with some knife.
    Even thr final Pandora cases needed manual fixes on the shoulder buttons to make them work properly.
     
    Djhg2000, rSl, javaJake and 14 others like this.
  17. levi

    levi Still fresh, damnit!

    Joined:
    Oct 6, 2008
    Messages:
    10,910
    Location:
    Somewhere off the coast of the EU
    Edit: Seems I was a little slow in posting this and most of the points I made have already been made by more informed people.

    Please see the reponses to notaz's post which I think had been posted by the time your post went up.

    Needs configuring. Currently it's a known issue, because ED and now Notaz can see it, but hns can't.
    AtC's unit doesn't have sound so he never noticed this. Can now be fixed.
    Clear plastics are shoddy as I already reported.
    All under investigation.

    It's a little disappointing that we're still at this point, but at least we now have people testing that can actually test this stuff. The rate of software fixes does seem to have picked up in the last couple of months too.
     
  18. Grench

    Grench Forum Addict!

    Joined:
    Oct 3, 2008
    Messages:
    6,031
    Curiosity only - how many boards, how many developers and how many remaining prototype-preorders?
     
  19. fusion_power

    fusion_power Advanced Member

    Joined:
    Dec 11, 2006
    Messages:
    7,020
    Location:
    germany
    Let's hope it does not take till an "Rebirth Pyra" this time until the shoulder buttons are working perfeclty. ;) Besides, it sounded like this part of the Pyra was actualy finalized a while ago. They tinkered ages on these shoulder buttons, longer at least then I did when I made the first 3D concept renderings for. XD
    Try some silcone oil by the way, should solve most issues with stuck parts like this.
     
  20. EvilDragon

    EvilDragon Administrator Staff Member

    Joined:
    Mar 4, 2003
    Messages:
    19,652
    Location:
    Ingolstadt
    Should be around 24 boards, 8 prototype-preorders... well, and the rest for devs, not sure yet whom though.
     
    rSl, spud42, Silent-Hunter and 6 others like this.

Share This Page

Loading...