Reference to posts #1232 and #1235 on page 83, regarding the non-linear battery reading below 20%, now I discovered something about the upper range... I'm not sure if this is actually the first time I observed on fw3.0... my impression is so. On fw 2.0, I'm pretty sure it wasn't like that because a large part of my N5 burn-in (~150hrs so far) was with 2.0, and I have been checking quite regularly and have established a general trend of the reading dropping about 10% per hour. My impression is that by the time 3.0 came, I have slowed down the burn-in frequency and ramped up DX80's, hence I may not have observed the battery level drop trend of 3.0.
Today I happened to check the battery level after running for 6hrs and I was shocked to see 71% remaining. This couldn't be. Unfortunately I didn't manage to catch the battery flat as I was out.
I have since recharged and started another run. As of 5hrs, I'm looking at 76%. If it still lasts 9hrs (means the reading really screwed up big time), it should end around 3am later (1130pm now, GMT+8). I will be staying up till around 2am and will check again. If it is still running, I won't be too worried about the battery's physical condition. But however it would mean a probable issue in fw3.0 compared to 2.0.
Update + consolidate:
- At 5hrs, battery reads 76%.
- 6hr 00min --- 71%.
- 8hr 30min --- 61%.
- Once it falls below 60%, it's speeding up. At ~8hr 40min onwards, it dropped 5% in 3mins.
- 8hr 45min --- 48%.
- 8hr 50min --- 42%.
- 8hr 55min --- 32%, now it's dropping 2% by 2%.
- 9hr 00min --- 17%.
- 9hr 05min --- 5%.
Exercise cut. Stop before player poof lol...
So battery is still ok. Could it be fw3.0? I'm tempted to flash back to 2.0 to try...
Sidetrack... btw how do I export the playlist and later import back? I've been doing firmware flashing for both N5 and DX80 recently and really dreading each time gotta setup my playlist again
Update:
- Flashed to 2.0 and running now. After 2hrs, reading is 91%. Ohno... looks like this behavior unlikely to be 3.0 related
- 6.5hrs --- 70%. I stopped the exercise and flashed back to 3.0.
I have promised to follow up on the non-linear battery reading issue raised by samuelboy, it takes longer then expected, but we kind of having a preliminary understanding of the issue by now.
FIrst of all, we have conducted a battery consumption test on a large sample set of N5. We setup the testing condition as follows:
(1) New N5 from production line, never been used, firmware v3.0
(2) Backlight set to maximum, and screen always on
(3) 36ohm Transistor at Balanced phone out to stimulate Headphone loading
(4) Volume at 80%
We take the battery reading of each machine at 10 minutes intervals, and the result are as follows:
Based on the result of the test, we consider the reading as fairly linear from 90% to 20% range, so the cause of nonlinear battery reading might not be related to firmware v3.0
We take this result and samuelboy's data and consult various related suppliers including the MCU Solution provider and we finally worked out a hypothesis. Within the Power Management module of N5, we have a table to store the Battery parameter and we work out the battery reading base on the battery parameter and a certain algorithm that we constantly refined to maintain the linearity of battery reading. We now assume that the battery parameter table is damaged for unknown reason, it can be caused by software malfunction or hardware crashed, we just can't be sure at this moment, we need to get hold of a problematic N5 sample before we can investigate the cost of the event. WIth a damaged parameter table, all the battery level computation become unreliable.
This actually explained samuelboy's obseration.
On fw 2.0, I'm pretty sure it wasn't like that because a large part of my N5 burn-in (~150hrs so far) was with 2.0, and I have been checking quite regularly and have established a general trend of the reading dropping about 10% per hour.
But when Samuelboy refreshed his N5 to v2.0 and crosscheck again, he notice the battery level become nonlinear now. That's because the previous observation was gathering before the battery parameter was damaged, and the recent battery test on V2.0 was after the battery parameter was damaged.
We shall further our investigation to proof our assumption, and develop solution to rectify the situation. One of the solution under discussion is to include "rewritten of battery parameter" as a standard step in a firmware update, so if the issue happen again, user can simply refreshed the firmware of the player and the battery level computation should be reset to normal after that.
I hope we can establish the solution promptly, but for the meantime, please don't panic because we are quite certain that if your N5 can run for around 9 hours between charge, the battery and the battery management function of your N5 is intact, it is the reading that screwed up, please establish a regular charging patter instead of trusting the battery reading.
That's all for now, we'll flying to LA early next week for the CanJam SoCal. This is our first appearance in CanJam so I am completed jammed up right now and there are loads of thing I need to complete before the Show. Please pardon I can't join the discussion or respond to request/PM promptly as before, I'll try to catch up.
Stay updated on Cayin at their sponsor profile on Head-Fi.
|