Repair: Apple Macintosh PowerBook 100 “Gotta re-cap ’em all!”

In the last part of this little project, I managed to refurbish the two PowerBook 100 power supplies to a working condition. The next step is to refurbish the main unit itself. Even though I was equipped with the service manual, I was a little hesitant owing to the possibility of breaking something which could be hard to source, or making it worse than it already is. Nothing ventured, nothing gained, I suppose.

I’m aware that others have done the same to their PowerBook 100s and seen positive impacts. It’s hardly surprising when you consider its age. Some were even nice enough to draw out capacitor maps for the boards. Even though I’ve seen them, I decided to “go it alone” and instead base my repair on my own teardown and analysis.

This post will seem a little “jumpy” as both teardown and repair are interspersed, but I feel that’s probably the best way to present it.

Required Material and Methodology

I suppose if you’ve never done any soldering or re-capping of electronics, it’s probably not a good idea to start with a piece of equipment that you value. That being said, from my experience working with the PowerBook 100, it definitely rates a moderate-to-hard difficulty compared with some other work I’ve done.

In order to get the job done, the tools you should have include:

  • A set of Phillips head screwdrivers of various sizes for laptop disassembly.
  • A fine-tipped soldering iron, preferably of the temperature regulated sort.
  • Some solder, 60/40 tin-lead type works well.
  • High quality desoldering braid – I personally use Chemtools branded wick, but formerly used Goot which varies a little in quality. Good wick really makes the job easier, bad wick can leave you very frustrated.
  • Desoldering bulb (or solder sucker) – the bulb type seems to be more useful as you can “blow” the holes open or “suck” them clean, whereas the pen type is really “one shot” with one hand and hardly easy to get into position.
  • Hot air gun (preferable, but not essential) for removing the SMD components with more ease.
  • Foil (I didn’t have this, but I probably should have) to mask sensitive portions of the board during hot air rework.
  • Tweezers to grab a hold of the SMD components as you’re heating them with the hot air gun, and to position replacement components.
  • Cotton tips for use in cleaning the board and getting into the crevices.
  • High purity isopropyl alcohol (e.g. methylated spirits) for cleaning residue from old leaky capacitors.
  • Hobby knife for “picking out” all the disintegrated rubber feet in the laptop.
  • Side cutters for trimming component legs after they’ve been mounted.

Other things you will obviously need include the replacement capacitors themselves, some time and patience to work on the boards, steady hands, decently good eyesight, and preferably a good workspace including an ESD workbench (which I don’t have, but with appropriate precautions, you can get away without it). Some other people like to have a thin stainless steel needle for clearing out holes, although I’ve never had one or used it before. Having storage for screws and small components removed from the machine is also a good idea, and pen-and-paper in case you need to draw a quick sketch or a camera to take photos of how things were originally arranged.

In my case, I decided that I didn’t want to open up the laptop again for a while so I opted to replace the capacitors (where possible) with more durable tantalum capacitors which will not leak. These capacitors are typically a little more expensive. Even though it’s likely the original capacitors were over-rated in terms of voltage, I’ve decided to match the voltage rating like-for-like, and also match the type (SMD vs through-hole radial where possible to avoid straining surface mount pads). This required me actually tearing down the laptop first, assaying all the capacitors and then finding a “compatible” replacement. Unfortunately mechanical fit is hard to verify beforehand, especially when changing capacitor type, so in some cases you might have to be creative.

In hindsight, it may have been more appropriate for someone with a single iron and a basic hot air gun with no SMD experience to have gone with all-through-hole components for a neater look, however, the SMD components function as expected despite the slightly less elegant skewed mounting.

I obtained all components from element14, a major electronic components distributor just down the road. Unfortunately, the range of locally stocked components were, at times, limited, thus I had to choose slightly more expensive alternatives. All in all, the cost of the capacitors came out to be about AU$60 in total, as tantalums are a bit on the pricey side. It is thanks to them that I could even obtain some of the SMD variety, which aren’t easily obtainable elsewhere without a minimum order quantity, longer delays or higher prices.

Re-Capping Technique

Even though I’ve replaced capacitors in many devices thanks to the “capacitor plague”, I’ve never really ever written about the technique, which I think, deserves some mention. I’ve developed the technique out of experience and “feel” for it, and it helps to minimise any risk of damaging the board. It might seem like a trivial thing to “desolder” something and replace it with something else, but it’s actually more difficult than it seems. Part of the reason is that the boards are often produced for automated manufacturing and have pads and component placement which are not optimized for hand soldering or rework. These boards can have multiple layers with plated-through-hole vias which can be damaged if not desoldered with care, and traces can easily overheat and delaminate, or be damaged by applying too much force.

The most basic sort are the through-hole capacitors, which are mounted in holes that go through the board. This sort is quite common, especially for larger electrolytic capacitors. A naive approach may be to wick away the solder using braid or a solder sucker and then just yank the component out. However, this approach rarely works for a number of reasons – often the solder pads and holes are just barely larger than the component, so effective removal of all solder is unlikely. Yanking the component out while solder remains partially in place risks pulling off pads and plated-through holes and damaging traces making it impossible to repair. Further to this, if the board is manufactured with lead-free solder, or the solder has oxidised a bit, it will be very difficult to wick the solder away at all.

My approach is to use a moderate iron temperature about 325-360 degrees and first add solder to the joints, making sure to alloy the whole mass of solder. In the case of reworking lead-free with leaded, this will contaminate the solder a bit but also slowly bring down the melting point due to incorporation of more lead. In the case of lead-free, I might choose to wick away the excess solder and add more fresh 60/40 to make the joint more “workable”. My goal is to make the joints somewhat “bulbous” in shape, as the excess solder also helps with thermal mass, in keeping the joint fluid when heated.

Next, I heat up one of the legs and wait until it becomes fluid. With my other hand, I then gently rock the capacitor slightly away from the melted leg so that it levers out slightly. There’s no need to go overboard, as you will stress the other joint if you do so.

I then alternate and heat the other joint, and rock the capacitor in the opposite direction. This may gain enough leverage that this leg is now freed from the board. I then return to the original leg, and repeat until the capacitor is entirely free from the board. The heat isn’t removed at all during the rocking process, otherwise the solder may set and “grab” onto a trace and pull it through.

Then, I clean the holes by using good quality desoldering braid which should wick up the solder and leave the holes somewhat clean. If this is not the case, because there’s some “tough” solder left somewhere, I heat up the whole joint until the solder is fluid and use a desoldering bulb to blow the hole clean. Others have had success using stainless steel needles, and other implements.

This works well in my experience where the naive approach never really works except for single-sided boards such as power supplies with large holes and simple layout. SMD capacitors are a little trickier, but with a single iron, you can do the same thing just noting not to rock the capacitor too hard otherwise you might tear off the pad. The safer way is to use a hot air gun and heat up the whole vicinity until the solder begins to show signs of change and then gently pick the components off with a tweezer while the board is still hot. It’s probably a good idea to mask sensitive areas of the board which you’re not reworking using foil to prevent them from being thermally stressed.

For soldering the through hole capacitors, there’s nothing to it really – insert the components in the right orientation, check for clearances, bend legs, solder as normal. For the SMD capacitors, it’s tricky to do it properly at home – I’ve decided just to “bodge” it by first reflowing a small amount of solder onto each pad on the board, tacking down one end of the capacitor with the soldering iron while being held with tweezers, then soldering the other end, and finally returning to the first side and touching up the joint there. It’s not a good look – but I suppose you could go over it with a hot air gun to straighten it out. To do it properly probably entails brushing with solder paste and hitting it with the hot air gun – but the pads were probably not optimized for the replacement SMD capacitors so the results may be a little underwhelming.

Teardown and Repair


For the PowerBook 100, I followed the service manual almost completely when it came to disassembly. To remove the rubber feet, I used the knife to “stab” and “lever” out the gooey mess as much as possible. The top screen assembly came apart from the laptop after undoing three screws, and the keyboard was next to be removed. The keyboard has two flex connectors – probably one for rows and another for columns or something like that, which contrasts with modern keyboards that have a single connector.

The next thing to come off was the palm-rest which revealed the hard drive and trackball assemblies.


One of the biggest surprises to me was that the trackball was a Logitech product. I didn’t expect that. Because I wasn’t paying attention, I didn’t even notice C1 or C2 and so I forgot to order replacement capacitors for this entirely. Whoops. I noted down the value of C1 (0.22uF/50V), but not C2 … sometimes these things are a little tricky.


I suppose element14 is probably not going to be happy with me ordering one or two capacitors to fix this one up. Oh well. Anyhow, it turns out that the hard drive was a Quantum Godrive 120Mb SCSI drive, with a relatively thick form factor.


Sadly, these are known to be failure prone nowadays, because of their internal rubber parts which have also probably disintegrated into goo.


The underside has a few chips which take up the majority of the area of the board. No electrolytic caps that I can see, which is good.


Of course, it has a 2.5″ SCSI interface which looks visually similar to what you might find on the back of a 2.5″ IDE drive. I’ve never handled a 2.5″ SCSI drive, so this was a first for me.


With these in hand, the next step was to take off the expansion RAM board. As it turns out, the expansion board was a third party board from a company called Sunland Microsystems. It had four unpopulated footprints on the top, which made me think whether it would be feasible to mount a few more pSRAM chips on the top and make it twice the capacity (4Mb rather than 2Mb expansion).


The pSRAM in question is a Hitachi 658512LFP-10. I managed to find some for sale online, but in a larger batch than I needed and at a price I’m not sure I could justify – after all, to do the modification could risk damaging the board as it stands, and I’m not entirely sure if it would work. The capacitors on the board look like they’re not the right size for the footprint, hence their “crooked” orientation. No caps need replacing here, which is good.

The system daughterboard was next to come off, and this is basically the processor, ROM and RAM on a card.

2016092119518580 2016092119518579

This would turn out to be our first patient, as this has an SMD electrolytic capacitor near the small connector. The tantalum next to it doesn’t necessarily need replacement.


The capacitor (C310) was replaced with an equivalent value SMD tantalum capacitor – noting that the band on tantalum capacitors faces the positive. The orientation is a bit crooked, but considering I’m doing it by hand with a single iron, I think it’s okay.


The next step is the main-board, which has a whole lot of capacitors that look like they need attention. The surface mount electrolytics show signs of distress, namely the solder joints around them look a little corroded, indicating that there is probably some leaking of electrolyte over time.

The best way to deal with the leaky electrolytics from my experience is simply to first scrub the area near the suspected leaking capacitors with a cotton tip soaked in high purity isopropyl alcohol. If the tip turns brown, you know you’ve got an issue. It’s good as it allows you to clean up some of the chemical residue which reduces the awful smell that comes out of heating up the residue during desoldering. A lot of the mess won’t be accessible as it’s underneath the capacitor, but once removed, you can then scrub the area again using a fresh cotton tip with isopropyl and that should bring the area back to cleanliness.

There’s a good number of SMD and through-hole radial capacitors on this board, and the best way to get it done would be to go slow, and exercise caution and patience. The through-hole ones really caused trouble as the solder had somehow formed a “hardened” alloy which resisted melting and required me to up the iron temperature and suck the solder out as it wouldn’t alloy nor wick out. At least I caused no damage there.

For the SMD chips, removing them one by one with an iron got tedious and risky in terms of tearing off a pad, so I used a hot air gun. The smell was rancid and ultimately, I didn’t have any foil on hand and didn’t mask anything so I melted the keyboard flex connectors slightly. That’s a lesson learnt – connector plastic and hot air don’t really mix. One of my tantalums was physically bigger than the footprint, so I applied solder, soldered one end and then “waved” the hot air over to get the other contact to solder. In the process, I “cooked” the outside of the capacitor … oh well.


In the end, it’s probably not my best work, but it’s my first attempt at SMD work with a single iron. It’s not ideal, but functionally, it should be just fine. The larger capacitors weren’t available in tantalum, so I settled to go with one solid-electrolytic and a regular 105 degree rated electrolytic. It’s a compromise I had to make.


The underside of the board after it was repaired, and it looks mostly the same. The original factory seems to have left some sort of flux or coating on the rear which isn’t really a nice conformal coating or anything, but it gives it an uneven glossy appearance.


With that major piece out of the road, the next stop was the display – a known troublesome part in my case. To get to it involved removing the Sony speaker from the body, clutch plate, clutch covers, brightness and contrast knobs, more screws and screw covers etc.


Inside, I found a Sharp LM64P791 LCD screen. There is a connector board and LCD backlight inverter underneath behind the shielding.


The backlight inverter had two electrolytic capacitors, which might need replacement. I decided to be safe and replace it anyway – although the replacement tantalums were a bit tall even when bent over, resulting in some stress on the board when reassembled.

2016092316138601 2016092316138602

Then, thanks to an online tip-off by someone else with LCD trouble, I realized there were some sneaky electrolytic capacitors on the Tcon board itself. These were surface mount type with a very thin profile and connections on the same side.


The units had frosty joints, a sure sign of bad capacitors. To get to it involved undoing four screws, bending out the retainer tabs, taking out the side metal strips, then removing the backlight assembly entirely, exposing the LCD screen itself.


We can see the flat-panel chip-on-flex drivers top and bottom. The flexible flat cable was very carefully threaded out from the frame – it’s not on a connector on the PCB side which makes it both fragile and annoying.

To work with the board and screen in such close proximity is dangerous, so it’s advisable to have some paper to cover the LCD during work, so that it doesn’t become the victim of solder/flux splashes and scratches when repositioning. Due to the close spacing, desoldering was a bit of a challenge. Desoldering the capacitors shows you just how bad residue can be.


It’s smelly, it’s affecting the solder by causing some oxidation, it’s sticky and it’s a mess.


A good scrub with isopropyl alcohol and a cotton tip later and it’s practically good as new. Because the capacitors were not available, I substituted tantalum radial lead type capacitors, and with some creative lead cutting and bending, the best fit looked like this:


Unfortunately, when reassembling, I found a slight interference between C5 and the plastic frame, so a little bit of a snip was needed.


It was no big deal, because the main criterion is that they have the same vertical profile so they can fit into the casing. That was easily satisfied.


Now that all the internal bits have been attended to, it’s also worth attending to the floppy drive, just in case there’s something that needs to be serviced there. I didn’t have any teardown instructions, but I seemed to have worked it out. If you unclip the front swivel cover, you can remove two plastic rings which allows you to slide the two parts of the case apart. That’s literally it.


In this case, I opened the drive upside down, and the Sony model number is in clear view.


Right side up, the drive is covered by a shield, so nothing to see there.

2016092321508626 2016092321508627

The rear board is nothing but a passive wiring adapter that goes from the FFC to the external cable with a plug, so allowing the cable to be replaced if damaged.


Taking a peek under the shell shows a clean drive, with a PCB on the top and no obvious signs of electrolytic capacitors. There’s one adjustment potentiometer in the bottom left which might be for drive speed? Who knows. I’m not touching it … until I have to.

Now that basically the whole ensemble has been torn apart and everything has been re-capped with the exception of the trackball PCB, I have to reassemble everything and see if it still works.


Before I get to the result, I decided to save all the extracted capacitors and do a quick measurement of their status to see just how bad they were.


Note that because the trackball board C1 wasn’t replaced, we don’t have any readings for it. Also, I didn’t see C2 on that board when I made the table so it’s missing. On the whole, the mainboard really only had two good capacitors of the lot – most of them were outright failures in both capacitance and dissipation factor (and thus, ESR). There was a bunch of Nichicon capacitors which had acceptable capacitance values but elevated ESRs which wouldn’t have performed well. The logical surprise was that the LCD inverter board had both capacitors turn out good – the Sanyo OS CON solid electrolytic is probably no surprise as they’re more durable than regular electrolytics, but the “SE18” was also good. The question is, for how long. All of the LCD capacitors were very very bad …

The leap of faith was to plug the unit into the refurbished power supplies and power it up. To my delight, the unit still works, even after the operation that took a whole afternoon and resulted in the loss of several rubber screw caps.


I also got a close look at the edge of the screen where the matrix lines were visible. It’s a pretty nice pattern.


I decided to leave the unit to “soak” test and see whether it would be okay in the long run. The unit remained stable, but the problematic screen returned.


At this point, I was a bit annoyed and disappointed that my time investment didn’t cure the machine entirely. The screen still exhibited column interference and occasional blanking of the bottom-third. I suspect this may be a problem with one of the chips on the flex cables to the array, as when the unit is cold, it is reliably clean and only plays up as it warms up.

Despite my disappointment, I was still elated that it was my first SMD re-cap, and despite my fear that my work would potentially damage the unit, it seems that I was successful this time around.


It took a lot of love, patience, time and energy to get this far. But now, at least, the power supply and the unit have been entirely re-capped (with the exception of the trackball PCB). At this stage, it’s disappointing that the LCD wasn’t completely happy, but the system did run reliably without a battery which was good and it managed a few hours sitting with a game on loop.

It was my first attempt at reworking SMD, just “going blind” as a hobbyist with no training and a cheap hot air gun. Of course, I melted parts of the keyboard flex sockets, and then I realized I should have shielded heat sensitive components. Oh well. Mistakes were made, but lessons were learnt. Hopefully, this information (along with tips and techniques) will be useful for others doing similar work.

I suspect I won’t do anything to the machine for a while, as I wait for a CF Powermonster-II to arrive to replace the Quantum GoDrive. Then, I may redo the trackball PCB capacitors at the same time. Wish me luck!

Posted in Computing, Tech Flashback | Tagged , , , | Leave a comment

Achievement Unlocked: PhD in Civil and Environmental Engineering

It was a few months ago when I breathed a sigh of relief as I submitted my thesis and understood that the journey would soon come to an end. I wasn’t a “free man” by any means, as it turns out.

The Assessment

The interim period is one that began with quietness and enjoyment. At the beginning, with no role to fulfill, I could indulge in the things I enjoy most and catch up on all the laziness I’ve been missing out on while preparing the thesis. The joy was real, and I think it was a nice “decompress” to not have to think about it all for a while. The first month was by far the most enjoyable.

But then, after that, it dawns on you that you’re not done yet. Maybe it’s because I’m a bit of a pessimist, and I like to worry about things even though I shouldn’t. But no matter what, I decided to login to myUNSW and check Thesis Examination Management (TEM) for any updates. Assessor due dates were marked on the calendar. Fingers crossed that reports would come in on time. No big deal initially.

Then, as the date draws nearer, you remember to keep checking on a daily basis with hope that examiners would return their reports early, even though you know in your heart that it would be exceptional for that to happen. Then you resign yourself to be happy if they were submitted on time. But before long, you get a little kick in the gut to see that the examiners requested an extension and had it granted, so you’re pretty much saddled with an increasingly nervous wait hoping to hear the result.

I suppose the thing that takes you out of the nervous wait-loop is if you’re successful in getting a write-up scholarship to finish up a few papers and get them submitted to journals. I dutifully applied, but got knocked back owing to the lack of funds this year, but luckily for me, my supervisors arranged an equivalent offer from funds sourced elsewhere. As a result, I was back to what was quite similar to the normal routine of coming into uni and supervisor meetings. But as we all know, paper writing is something that’s hard to get motivated for, and in fact, I’m still trying to finish it all even today!

Regardless, once you see the reports have (at last) rolled in, then begins the administrative game of passing the result to you. My results were sent to the school for vetting, and somehow, managed to stay in the school for 27 days. I didn’t expect that at all. The result was unexpected as well – a C from one examiner and an A from the other. This meant that revisions were required and an approvals process would be required.

Chasing a Deadline

I got right onto the changes, and the supervisors were supportive in advising me as to how to respond to the comments. I managed to get them all finished in 14 days and submitted it on 15th August for consideration. I expected it would take two weeks or so, as the graduation window for the year would close firmly on the 16th September. One day shy of a month for them to review the changes should be enough, right?

I waited patiently, waiting to hear something positive. The friends around me were all saying “yeah, you’re done.” But I knew I wasn’t done until I saw the result.

Two weeks before the deadline, and it’s 5th September. I’m starting to get a little nervous, but it seems my supervisors haven’t heard anything. By the Friday of that week, I implore the supervisors to follow up and see what’s going on.

The week of the deadline, and nothing is heard. It gets to Wednesday 12th. I’m almost ready to accept my fate. I might not be part of this years’ graduation ceremony. Thursday morning, and still no news. I decide to try chasing the matter with the GRS and my supervisor, so we can approach it from both sides “administratively speaking”. It took three calls to get to a person … so I guess it must have been a busy time for all.

I was asked to e-mail everything to them, so they could use it for reference. It was my luck that at that exact moment, UNSW’s MX servers were having a meltdown and e-mails were getting caught on the inbound … for over an hour. I was desperate, so I decided to send myself test e-mails just to check – from four different mail providers, all were much the same:


At least, they had the messages by the close of business, and they could chase it up on Friday – the deadline day.


Rather luckily for me, just after lunch-time, I got an e-mail saying that the changes were approved and I would be awarded. Just in the nick of time. When I had almost thought that all hope was lost. I would be graduating this year.

I hastily filed my digital thesis submission to the library on the same day to make sure there wouldn’t be any hold-ups in regards to conferral and graduation booking. Just yesterday, that was all approved.

As a result, I was able to book in my ceremony. Today is the day of conferral …


… so technically speaking, I am now Dr. Gough Lui, after four years of hard work, and nine years of being at UNSW including my undergraduate degree. To celebrate, I’ve updated the logo with the news:


To that end, I still have to update my bio amongst other things. I’ll get around to that, someday.


The assessment process proved to be more complicated and stressful than I had imagined. I thought I would miss this year’s graduation ceremonies and conferral thanks to administrative delays, but luckily, they pulled through merely hours before the graduation window closed.

While now I have my degree and title conferred, there are still things to be done. For one, I’ve now got to organize the clean-up and hand-over of archival goods, and get the printed-and-bound copies of the thesis organized for myself, supervisors and the school. The thesis has been submitted to the library but it will be subject to a 12 month embargo while I continue and finish all of the necessary work to get more papers published.

That being said, I feel much more at ease to think about the future – namely next year, I will be spending a lot of time travelling in Asia for a well-deserved holiday, so the focus of the site might change more towards personal posts. In the interim, I’ll be busy as always.

Good luck to all the others studying for their PhD!

Posted in Opinion | Tagged | 10 Comments

Review, Teardown: Askborg ChargeCube 20800mAh Powerbank (M021)

When it comes to power banks, while some users are content to have a moderate-sized unit that charges their phone a few times, there are other users who really demand as much capacity as practical. For those users looking for 20,000mAh and above, the number of options are more limited and the price rapidly escalates.

This post looks at the Askborg ChargeCube 20,8000mAh power bank, a power bank featuring two USB outlets, one with “SpeedID”, an LCD display and a reasonable price. The unit was supplied directly by Askborg under review challenge terms for a thorough test.



The unit was packed inside a wood-pulp coloured cardboard box with a white cardboard wrap-around. The logo and branding is on the front.


Support information is provided on the rear.


The specifications are provided on the side. The unit has a model number of M021, with a 5V 2A input and a claimed 20800mAh capacity.


Inside the fold-open cardboard box, the unit sits inside a folded cardboard frame for protection in shipping. Unfortunately, due to the weight of the unit and the stresses of international travel, the cardboard was badly tortured. That being said, the unit did survive without any damage.

2016082316588286 2016082316588289

The unit has a black aluminium body with rounded sides, as is common with other power banks. The size and shape suggests the presence of eight 18650-type cells, each with a 2,600mAh capacity for the 20,800mAh capacity. This unit features an LCD to indicate charge status as a percentage, as well as charging or discharge status. The LCD is backlit with a blue LED when the button is pressed. The logo is etched into the front, with the specifications on the rear.

2016082316588287The top side has a plastic facia with a single button to power-up and check the unit’s capacity. The unit auto-senses loading on the ports and starts up automatically. Two outputs are available, with the one nearest the power button featuring the SpeedID capability, and the one further away claiming “2.1A” output. The charging input sits in-between the two.


Aside from the unit, a fold open leaflet is included, with a small blurb on the SpeedID feature and a short microUSB cable with unknown wire gauge. The cable does have all pins connected and can be used for data synchronization as well.

Performance Testing

This section will look at both results of testing and subjective opinions from using the power bank in a day-to-day situation.

Delivered Capacity

Tests of the delivered capacity were made with the same apparatus used to test the majority of the power banks.

Load (mA) Run Capacity (mAh)
500 1 18683.89969
500 2 18663.34003
500 3 18677.61891
500 4 18624.39555
500 5 18623.74813
Mean 18654.60047
Range 60.15156036
StDev 28.84820926
Load (mA) Run Capacity (mAh)
1000 1 18046.5039
1000 2 17984.35378
1000 3 18116.31793
1000 4 18187.75547
1000 5 18194.9656
Mean 18105.97933
Range 210.6118231
StDev 90.88850731
Load (mA) Run Capacity (mAh)
2000 1 16658.9234
2000 2 16752.38394
2000 3 16756.05647
2000 4 16642.02521
2000 5 16766.02669
Mean 16715.08314
Range 124.0014705
StDev 59.49114356

Based on the delivered capacity in the testing, the unit is quite likely a genuine 20,800mAh power bank, delivering an average of 18,655mAh at 500mA, 18,106mAh at 1A and 16,715mAh at 2A on a 3.7V nominal voltage basis. Assuming the cells are 20,800mAh results in an efficiency of 89.6% at 500mA, 87.0% at 1A and 80.4% at 2A which are reasonable figures, considering that there may be 1-2% extra losses in the cabling of the test rig. The range of the results is a little higher than usual in the case of the 1A results, which were the first run, which may reflect controller uncertainty around the cell capacity.

Because of the large capacity of the power bank, testing a single run can take two to three days to complete. Now I know why very few people do reviews like these …

Output Voltage Profile


The output voltage was maintained within the USB requirements of 4.75v to 5.25v throughout, with the exception of towards the end of discharge where the voltage dropped somewhat gradually over a span of time. This was, however, displayed on the LCD as a flashing 0%, so it probably is intended to extract the last possible bit of charge before the unit shuts off, and being out of spec is not unexpected. The unit does disconnect the cells on end of discharge and the output falls to zero. During discharge, the voltage does have slow sawtooth-style movements, likely due to the limited granularity of duty cycle adjustment in the switching converter.

Output Ripple and Noise


At 500mA loading, ripple frequency was 122.6khz with a peak-to-peak voltage averaging 123.1mV with a transient “blip” every switching cycle. This is roughly the same level as many “stock” chargers.


At 1A, the switching frequency was 149.2khz with ripple reached a peak-to-peak voltage of 149.8mV. This is around the same level as stock phone chargers, and is considered safe although is not the “quietest” output around.


At 2A, the switching frequency was 161.4khz and the ripple voltage reached 233.5mV peak-to-peak, with the peak transients being the main offender. Without the transients, the ripple would have been similar to the 1A case. This is somewhat borderline – it’s higher than that offered by many stock chargers, but even some branded power banks have been seen putting out higher ripple voltages. It is probably quite acceptable, although ideally, I would prefer staying to <=150mV.

Charging Current Profile

Charging current was measured using the home-made USB current shunt and Keysight U1461A meter combination.


As the unit claimed a 2A input, I first tried charging the unit with a generic Qualcomm Quick-Charge 2.0 compliant charger with a 2A output at 5V. Unfortunately, charge current topped out at 850mA resulting in a long charge time of 23 hours and 48 minutes.

Because of that, I tried using an Apple iPad 10W (2A) charger, seeing as this is probably a common alternative it may have been optimized for. While charging was indeed faster, it didn’t use the full 2A capacity. It took 16 hours and 36 minutes to charge, and this may have been because of sensitivity to cable/contact resistance.

As a final test, I chose to try my Xiaomi 2A mobile phone charger, which successfully allowed the unit to draw just above 1800mA (the expectation for a 2A charger) and complete charging in 13 hours and 47 minutes.

Charge termination occurs at around 200mA input, and charging appears to be based on a switching converter. This would correspond to a termination current of about 34mA per cell, which is a little lower than the 100mA normally expected for most 18650 style cells.

Subjective Opinions

In my opinion, the Askborg PowerCube seems to be quite an attractive product based on its reasonable price. It is no-frills when it comes to inclusions and manuals, and it seems to have some contradictions – for example, the webpage seems to claim:

Exclusive SpeedID Technology: Detects your device to deliver its fastest possible charge speed up to 3 amps per port or 4 amps through three ports.

This is obviously not possible as the unit only has two ports and not three. It also claims in its manual that it weighs 455 grams, but in my own testing, it weighed 492.7 grams. This is consistent with the website claim that it weighs 493 grams – however, understating the weight in the manual might mislead some consumers, especially travellers with limited carry-on baggage who might need every last gram. The manual is brief but contains useful hints – such as recharging the power bank periodically to maintain performance. It also has a 18-month warranty which is a little longer than other products in this class.

Travellers might be slightly less satisfied with the 8-cell design using 2,600mAh cells. This leads to a higher weight and larger physical size than designs with more premium 3,400mAh cells from Panasonic (where six cells would give 20,400mAh). Accordingly, such cells command a significant price premium in return, so this is likely a compromise to reach such a competitive price-point.

While in use, I suffered no compatibility issues with my devices, namely several Android mobile phones and tablets, an Apple iPad and a Sony Playstation Vita. Charging rates monitored by a charger doctor were able to see about 2-3A total across the two ports – I wasn’t able to see 4A even with two tablets connected, possibly due to limitations in voltage drop. Under heavy loads, the unit did warm up slightly, but this is expected.

The LCD display was helpful in diagnosing remaining capacity and is more granular than most LED displays while consuming less power in operation. It’s not perfectly linear, but it’s pretty good. The biggest disadvantage is its limited viewing angle, which makes it hard to read when it’s placed down somewhere without tilting it to the right angle to maximise contrast.

While using it, I also found that the unit is not capable of synchronous charging and discharging. This may be a desired feature in some rare cases, but it seems this is not supported with the outputs shutting down as soon as a charger is plugged in.

It would be nice if the unit offered Qualcomm Quick Charge 2.0/3.0 abilities to more rapidly charge from the wall, as its large capacity makes overnight charging only possible where the adapter and cable are well matched. If it offered this ability, it could boost charging rates significantly and cut recharge times. Alternatively, in the case of QC 2.0/3.0 output, it could charge connected devices quicker, however, it seems likely that such a design would be more costly.

The output USB ports did feel a little “loose” and didn’t grip onto the connectors as firmly as I would have expected. On some cables, this resulted in slightly intermittent contact which could “wiggle out” if placed in a bag while walking around. Bending the USB connector shells slightly improved this.


Taking the unit apart is not particularly easy. As it turns out, both ends had their plastic fascias glued on in some way – the top using super-glue, and the bottom using something akin to liquid nails.

2016082317008290 2016082317038291

I started by removing the bottom, which revealed that inside, the cells are very firmly siliconed into the body. This keeps the cells from moving around, which is great, but complicates taking the unit apart and identifying the cells. Even with copious use of a thin knife to cut through the silicone, I was not successful in extracting the batteries because there’s even more silicone inside.


Extracting the circuitry at the top required removing the LCD protective window first and then sliding out the capsule at the top.


A look at the PCB shows that it has some flux residue and a solder splash near the rear of the right USB port. More attention to manufacturing might be warranted. The wires to the battery pack are also slightly thinner than optimal. The batteries are well insulated with cardboard, and it seems they have been faced so that negatives of all eight cells are facing towards each other at the middle, and the positives are facing top and bottom, bussed together with a cable.

Because they were siliconed in place, I was not able to identify the make and model of cells used in the power bank. That being said, it appears that all eight cells are connected in parallel with no additional protective devices, which could be problematic if one cell were to fail as the other cells would dump into the failed cell limited only by (potentially) the PTC on the cell. There was no pack thermistor or additional one-time fuses visible. This is a common arrangement in lower-end power banks.


The controller is identified as a Shenzhen Legendary Technologies LDR5409 integrated power bank controller. According to the Chinese datasheet, it appears to be an LCD driver and power bank controller in one, and was designed to produce a 5V/2.1A output with under 100mV ripple at 2A. Other key specifications include a 3A output cut-off, >=92% efficiency at 1A and >=88% at 2A. That being said, it is possible that this design has modified the current sense and component values to squeeze more current out of the converter.


The other side of the board has the backlit LCD module, with a few ICs including the regular 8205A MOSFETs, DW01A Li-Ion protection IC and 9926A/DN1519A (MOSFET?) packages. A single un-enclosed inductor is used, suggesting a single rail design which the two ports share. As a result, the actual achievable current is not clear and will depend on just how the converter has been configured.


There is virtually no circuitry hiding underneath the LCD screen.


The Askborg ChargeCube 20,800mAh power bank is a very reasonably priced unit that appears to deliver the promised level of capacity at a sufficient voltage, with an informative LCD display, physically robust build, and trouble-free compatibility with my tested devices. The unit is backed up by an 18-month warranty for peace of mind, and comes with a charge cable.

Rather unfortunately, the unit doesn’t support Qualcomm Quick Charge capabilities, which is something it could really benefit from due to its large capacity and hence longer charge/discharge times. Its USB ports were slightly loose, and internally, the soldering quality left a little to be desired. The absolute maximum current delivery across the two ports was not established, but it appeared to be a single-rail design. The ripple voltage at 2A loading was slightly elevated due to the presence of switching transients at the peaks. I was unable to confirm the supplier of the cells, and the long-term durability of the unit as that would really depend on many factors.

That being said, it would not be fair to expect more especially when it retails for US$30.99 on special, as you really are getting a lot of usable capacity for your dollar. Readers interested in purchasing the Askborg ChargeCube 20800 can do so via Askborg’s Amazon stores (US, DE). Thanks to Askborg for supplying this sample for review.

Posted in Electronics, Power Bank, Tablet | Tagged , , , , , , , , | 2 Comments