Late feedback on July Update:SOLVED

@Jack

Quote:
If one adapts rechargable batteries (3xAA=3.6V, they are pretty cheap, 2000+mAh each) with an adapter that plugs into 5 or 12V line of the PSU with proper diode (or whatever it should be, apologies for possible ignorance in electronics)? Or 3.6V will kill something? Anyone? Shouldn't be to complicated for DIY imho.
------------------------------------------------------
One of my pet theories, as most backup circuits use a 3.6V rechargable battery. You might get away with using a higher voltage, but who in there right mind would take that chance? If you could, a better solution would be to get a single 3.6V lithium & make a cell adaptor.

@sundown

Sure, of course!
The old battery that I removed was at 3.06 volts.

I also understand the desire to divide the hardware and software groups. Sometimes it's hard to see the connections between them.

The reason that I took the extra time to install a spare drive and repeat the entire test was to determine the root cause of an apparent problem. I did not find what I expected, but I was able to find that the "problem" can be reliably switched on or off by my choice of which versions of kick modules I loaded.

And no, my machine is still not as stable as it was before the "upgrade". I still get occasional lockups, though not as often as before I changed the battery.

Yes, I'll be going back to the pre-july release.

It is apparent to me that many folks here have trouble with the possibility that a software update can affect the hardware that it runs on. Of course it can! Even something as simple as insufficient bypass capacitors can have a profound effect on the results of digital logic operations. If you don't believe me, try building without them and see how reliable your circuits perform. Clock them slow enough, all is well. Go a little faster, and the random element comes in.

I design electronics, and I write software. Possibly because I know both sides I am less likely to see an impenetrable barrier between them. I can easily design an SRAM device that works properly at lower data rates, but fails miserably when clocked more rapidly. Many would think that since the SRAM is "only" memory, that it's failures would be limited to providing wrong data. Not true. A low-power situation could cause it to latch data onto the bus too slowly, or fail to release the bus at the end of the access cycle. Such a problem could appear as random lockups, even while executing unrelated code.

If this problem was nothing more than me being too cheap to replace a battery, then yes, this would all be a waste of time. But consider this if you would: This is the type of problem that can stay hidden, and cause random crashes for no apparent reason. And it can be an absolute BEAR to track down, since the problem is not in the code that is executing at the time of the failure. Fixing this "little" problem would almost certainly take out some "undefined random" failures along with it. That's why I believe the information I'm providing is useful.

I took the time to dis-assemble my system and revert to the pure, native Dec06 release, and run it long enough to establish that it is stable, without lockups. Then, by ONLY adding the July release, the lockup problem returned immediately. The test was repeated to make sure it wasn't a coincedence, because I understand the need for proof.

I actually did that to figure out which "added software" was causing the conflict. When I realized that it was all happening within the original releases, I thought it worth reporting.

Yes, a low battery is a hardware problem. No argument.

But if installing and removing the July update can turn a lockup problem on and off like a lightswitch, I think it deserves further investigation. It's an opportunity to make the OS more reliable. I'd call that worthwhile.

I saved the battery. Please come on by and I'll demonstrate it for you personally.

LyleHaze

@sundown

Quote:

I know it sounds scary to actually try it, but someone with the knowledge can tell which voltages/tolerances if the cmos memory is designed for. Then proper converter for 5 or 12 V to recharge the battery. The problem to charge the batteries you'll have to feed a bit higher voltage(afaik).
And the issue of using the right batteries is important too, here you right about single battery. But your proposition isn't rechargeable. On the other hand, rechargeable AA battery would drain itself during month or so. So looks like higher capacity lithium one is a better solution. Maybe the ones used in photo-cameras?(don't remember the voltages).

Jack

@all

I'm wondering if somebody with electronic skills could find the battery problem in XEs and make some kind of fix for it.

@LyleHaze

There wasn't many things that was updated in the July update. Can you test them one by one ? (Maybe USB ?) And if I can remember there haven't been any battery problems in MicroA1s (XEs only).

@all

PS. SE doesn't have battery problems.

(Edit:

@LyleHaze

If the battery reads 3.06V, then that wasn't your problem. Save it as it reads the same as a new battery.

You may want to consider changing vcore on your micro. My micro was unstable for the first 5 min or so after bootup. I changed it to 1.49V first, but the unstability was notably worst from the get go. Next I tried 1.39V & the lockups disappeared other then when I made backups (copy locks). I found that if I dropped it to 1.34V, the copy lockup problem was totally gone. My micro has been very stable for the last 12 months now at 1.34V. From talking to others that have done the same, one step up or down has fixed their problem.

Default vcore setting on a micro is 1.44V. Spec is 1.45V +/-.05V. Changing vcore will NOT help with apps that crash as a general rule, but can affect system stability (random lockups).

@sundown

Thanks.
I might change it later, but I want to complete testing first to determine which kick module is affected by this problem.
Funny that my machine has been rock solid since I got it, and suddenly there are new issues..

I am focusing on the kickstart modules because the first lockup, and a fair few since then, have been immediately after loading those up.

I hope to get time this weekend to:
Remove my main hard drives
replace my "test drive"
Put the older battery back in.
swap kick modules individually until the culprit is found.

brand-new open circuit voltage on a CR2032 is 3.25, according to datasheet and my meter.
"working" voltage is 3, so yes, 3.06 should work properly, and indeed does with the December release.

I'll take the time to do the tests. I can only hope that someone is willing to do something with the results.

I'm also hoping that doing a "Mix & Match" with the lickstart modules doesn't cause too many problems, but I guess I'll find out as I go along.

Hmm, "lickstart"?? Must be a typo.

I'll report results here after the weekend, if I get the chance to do it.

LyleHaze

@LyleHaze

Play it safe & create another boot option in kicklayout. I did this to boot using Petunia from ud4. Make a directory in Kickstart, like kickstart2, put the older libs you want to test in there, add a "test" label, copy the "default" loading setup, but change the path for the libs you want to test. That way your system will always boot into "default" if anything goes wrong. Make sure the "default" label is first & "test" second & a blank line between the two options.

@sundown

Sounds interesting.
Are documents on this available?

I vaguely recall the kickmodules always coming from the first available source, and then "selecting" a boot source from there.
Obviously I'm wrong, I'd like to learn more.

Hmm, Maybe "Lickstart" is how to boot an A1 with a weak battery... Might have to give that a try.

(completely off topic)
What I'd really like to know is how to get all the untested, not-working video modes out of my system.
Most or all of the "fakenative" modes are "stretched" so wide that the left half fills my entire screen.
How can I fix or remove these "bad" screenmodes?
/rant off

Thanks for the help,
Lyle

@LyleHaze

Of coarse we have no docs, we learn the hard way.

If you PM me you email address, I could send you a guide file of my Kickstart directory & my kicklayout file as a sample.

Like #6 has said, you could join us in irc. We have a private channel that Sbaitso has set up for us. Very nice talent pool. Let me know & I'll give you the info to join.

@LyleHaze

This is the thread to read on multiple kickstarts

This is not the safest thing in the world to do, so I do not recommend this for anyone but advanced users.

Write to Niels or Detlef, if you have any questions.

#6