You all know me and some of my crazier hair brained schemes.
Well, this one is no exception.
After countless attempts at trying to find the ultimate electron reservoir for my itty-bitty Asus EeePC 900HD, I've finally created one that does the job.
Well, sorta... If I could just get the chip in the "original battery" to at least head toward the concept that there is more capacity than what once was...
One problem (aside from the "holy crap, what is THAT?!" factor) is that there is a known issue with Asus batteries and some serious miscalculation that takes place in the management hardware.
All of that aside, this is one hell of a performer, with a whopping 18 hours of runtime, if I decide to take it down to the 'danger' zone. More on that in a minute...
The completed pack is 200Wh by design.
This is what I did:
I took two extended batteries for HP laptops and removed the cells from each pack. These are 100Wh each, with 9 cells per pack, 3 paralleled sets of 3 in series, for 11.1V nominal.
EDIT - I realize that the filename here says "93Wh"... The actual batteries pictured here in fact ARE 93Wh, but this pic was actually taken AFTER I built the new battery, and the 100Wh shells were long gone. There were two flavors of battery available to me at the time - 93Wh and 100Wh. I of course ran with the 100's. Thats an hour plus of extra run time, dangit! My purpose here is served - they illustrate what the original packs LOOKED like.
This little difference in nominal voltage posed a problem for my application however, as the EeePC uses a 7.4V battery. So what to do with one in hand and coming to this realization? I quickly grabbed a second extended pack; to get 9 sets of 2 in series. Here they are after the wiring was complete. I moved the thermistor from the original Asus battery into roughly the center of the pack to get it "as good as it gets". I didn't have the time, resources, or money to play around with trying to convert over to multiple thermistors. This isn't a big deal, because peak charging current for each set of cells is only about 110mA. The thermistor will only ever kick in if the ambient temp has brought them too high to charge. I'm too anal to let that happen, but just in case. I've seen the youtube videos...
This is a shot of the new battery layout, next to my (now seemingly standard) size comparison tool, a good ol' pack 'o smokes, and one half of the case for the original Asus battery. Big difference.
I had the opportunity to leave the original cells in the original battery, but I decided against it for two reasons. One, they only would have added an hour or so, and two, the wiring would not have "neatly" fit up inside like it does. Everything WAS looking like it would be much neater until I ran up against the itty bitty holes that the thermistor originally was soldered into. Dodging the monkey wrench meant a little rat-nesting, but it didn't turn out that bad I suppose. At least it appears "neat" in the finished product...
This is the front view of the completed battery assembly, as if the screen were facing you. The aluminum plate is actually a set of 3 plates, loosely stacked upon one another. They spread the heat out coming from the bottom of the laptop so as to not "thermally age" certain cells before others. Works very well, actually, better than I ever expected. The velcro's purpose is hopefully, well, obvious.
Another shot from the left side...
And another from the back. Now you can see what I mean by "neat"... That's pronounced "contained".
Here it is, laptop perched up on top, held on by the velcro strips (which still need adjustment, but do the job adequately for the moment). And apparently, coupled with the Droid, I can post to here from anywhere (I tried to say that you'd get back all you have given me and more, didn't I?) What will you do now? LOL
And now for the power shots. Here is the voltage discharge curve for a calibration attempt I did. The calibration didn't take, but I have a real good understanding of how much juice is left at any given point by referencing this little gem. The X axis is time in minutes, the Y is mV. "Reaching" the end of this graph to see the knee takes a LOT of patience and careful timing. I didn't bother with all that for this particular discharge however; I disabled all the "oh crap" features of power management and let it go until the battery said "Danger Will Robinson! Danger! Danger!" and cut the cord for itself.
The associated current plot for the above graph. It's worthy to note that the spikes (except at the very end) are when I woke it up to grab a sample here and there. The end spike is as the battery voltage dropped, the current went up drastically to keep internal regulation. It's a snowball effect for sure. Wattage plotting is coming, but I need to write the code. I know however from another plotting program (the built-in plotter in Ubuntu) that the "snoozing" draw is about 10W for this particular laptop. And again, X is time in minutes, and Y is mA.
Here is the voltage plot for the charge cycle that followed the complete discharge above. Same axis attributes apply as the discharge voltage plot above.
And the associated current plot.
Both of the charge graphs are "zoomed" slightly - It actually took 36 hours (ish) to completely recover the charge that was used!!! Some of this is paranoia, as I have the current limiting (presently) set to about an amp - I'm not sure exactly how much juice the MOSFETs in the original battery can handle without popping, and I can't afford to find out! The OEM power brick is rated 12V at 3A, so I am pulling slightly less than an amp from it when charging is at it's highest. It still gets very warm if the laptop is doing any "crunching".
One other thing worthy of mention, is that at one point during the development and trials of many different batteries and configurations, I managed to get the polarity going into the laptop backwards not once, but twice, and got smoke from TWO DIFFERENT LOCATIONS, one for each event! How the hell this thing still powered up and ran after all that was beyond me, honestly...
The ultimate result was that the laptop's internal charging circuit is shot, and therefore won't charge the battery. For this, I modified a buck converter and tinkered with the current limit sense resistors until I got a reasonable balance between charge time and heat. Doesn't work bad, considering.
The other effect is that curiously, the laptop will only turn on if the battery is present. Once powered up, it will continue to run on the power brick if the battery is removed, but plays dead from a cold start. Even reboots are fine... Not sure I'll ever completely understand what is going on there...
My biggest problem with all of this right now is that the original battery management chip doesn't see the full capacity of it's "new and improved you", which isn't to be expected right away, but according to many things I read online, the chips eventually adapt to the larger battery by taking note of the extra energy required to charge it, as well as the additional capacity during discharge. This one is still LOSING capacity, and I don't understand why, other than a known issue with Asus batteries manufactured in a given window.
The most annoying aspect of this however, is that even though I can tell power management to ignore the "I'm dead" signal coming from the chip, I can't tell BIOS the same thing. The result is that once the chip reports "0.0%", BIOS goes into a lock-down, and there is no recovering from a power cycle, or even a reboot. The machine will flat out refuse to power on until it sees a charge source.
Now, funny thing about that... All I have to do is briefly connect the power brick, hit the power button, and then disconnect the brick. Then it's back to binnis-as-uzhul.
There is a wierd hiccup too that happens at about the point where the chip thinks there is 15% or so remaining. It stops counting down (in power management) and hangs there as if it were trying to calibrate or something and "give it a chance to prove itself" kind of deal. I verified that this is in fact the chip getting "stuck" around the "2.7Wh remaining" mark. The problem with that is, this plateau can last several hours, or only 1 or 2, and I can't predict it. I can cycle the machine as long as the chip is reporting that there is usable capacity, but once it's gone, I could have as much as somewhere between 14 and 16 hours remaining, but CANT USE IT!!! At least until I do the little "brick trick"...
Hmmmph... Oh well, I don't regret this one. It sure can run the laptop all day long and then some, and I don't have to worry about the damn thing going dead regardless of whether I'm surfing the web or watching a movie, or if I'm froggy, BOTH!