Power Bank Endurance Test – Hillo Power Jin Gangxia (Part 7)

If you’ve been following along with the experiment from the beginning, then these words don’t really need to be here, as you can already guess that another 50 cycles have been added to the battery pack.

But for those who haven’t, this is just an experiment using the power bank capacity measuring rig to try and see how an “unbranded” Chinese power bank handles charge and discharge cycling. I wanted to see just how long it would last, as I had previously bad experiences with some lithium-polymer power banks, so I committed to trying to destroy this power bank.

Of course, I expected something exciting, but so far … I’ll let you be the judge.


The power bank has now experienced 350 cycles since the beginning of the experiment, or 366 cycles from new which represents a cycle life consumption of 73.2% to >100% based on the 300-500 cycle expectation for cells.


From the graph, it seems that the degradation in the later runs are at a slower rate, with a higher variance. However, the degradation is still happening, just slowly and buried in the noise. I expect that the cell may experience more sudden degradation as it gets towards the end of its lifetime. It doesn’t quite “reach up” into the 3700+mAh territory as it used to, but it’s nowhere near the 3100mAh “line” which represents a 20% capacity degradation, normally used as the end of life indicator.


As you can see with a zero-scaled graph, there’s a long way to go to the bottom. Based upon the curve fits, very little change in expected cycle life is observed, with a projected life of 1122 to 1324 cycles based on power and linear fits.


This particular cell seems to be holding in much better than expected. Another update will be provided when we reach 400 cycles since the beginning of the experiment – maybe we will begin to see some accelerated degradation soon. I wonder how long I will keep this up for … you can’t imagine how much time it takes to collect and analyze the data which currently totals 396Mb in four separate Excel spreadsheets …

Posted in Electronics, Power Bank | Tagged , , , | Leave a comment

Warning: Don’t store Blu-ray discs in Sleeves & Heat Gun Revival Technique

Towards the end of last year, I posted a warning about not stacking Blu-ray blanks in sleeves, after being devastated at losing a good number of “fresh” blank TDK BD-R DL discs I paid a good amount for.

It seemed that storing them in sleeves, under pressure, may have caused damage to the coating causing them to fail burning and come out of the burner with a spotty pattern.

Comments included not using PVC sleeves, which is sensible advice since the CD days as PVC outgasses and has a tendency to “stick” to the top foil data layer on CDs and “peels” them off causing permanent destruction. Of course, most sleeves since then aren’t made with PVC.

Albert, a senior moderator at MyCE forums warned:

The sleeves were known to make things go bad after the discs had already been written; good to have confirmation that it can affect things before burning them, too.

The folks who have had issues with storage of previously written discs have been able to revive them with a heat gun to undo the pattern left by the sleeves. I’m not sure if the same revival technique could have helped, or if there was more damage done than just that.

cd pirate, a MyCE resident replied:

I’ve also noticed this myself back in the DVD days.

You don’t need to have anything pressed on top either. Sometimes just the pressure of having two discs back to back in a cd/dvd wallet with the whole wallet filled with discs can cause it.

Some wallets are tighter than others too. I’m currently using a sleeve system but the discs are stored upright with no pressure from other discs so we’ll see if that causes any indentation.

BD-R discs “hard coat” probably gets more messed up than cd/dvd too. It’s probably just a lacquer or something cheap. Would have been nice if the plastic itself was scratch proof.

I never had such issues back in the DVD days myself, so I didn’t pay much attention to it. I had already read back over 90% of my BD-R collection to hard disk because of a lack of trust in the discs, so I only had about 30 discs to do. Unfortunately, I had some important things to do immediately, so I left the discs on a shelf and forgot all about them until just recently.

Finishing the Copy Operation

All of the discs that needed to be copied were about 6-year old burns from my LG GGW-H20L. They were stored in PP sleeves (not PVC) and were mostly stored vertically so as not to have pressure pushing against them by stacking. Most of them were TDK Made-in-Japan discs, with only a few CMCs mixed in.

Starting the copy operation, I got read errors on quite a few discs. Looking at the bottom, the dreaded issue which showed with the blank discs reared its head again.


The pattern is hard to observe, especially with a camera, and can only be seen at the edges of bright direct light. It tends to be a cross-hatch pattern corresponding to the patterning on your sleeve material, which cannot be cleaned off the disc using microfibre cloth, or even with high-purity ethanol.

This seems to affect the discs I had stored in sleeves both vertically and horizontally, generally being worse for discs which had been stored for long periods without removing and rotating.

My hypothesis as to the cause goes as follows:

  • The data layer sits much closer to the bottom in a Blu-ray disc, covered by a 0.1mm polycarbonate and hard-coat cover. This is barely the thickness of two sheets of paper, or a sheet of laminator plastic.
  • The thin cover layer has some pliability, as most thin pieces of plastic do. You can prove this to yourself by carefully flaking off a portion of the bottom layer and trying to pull on it or push into it – it will stretch and dent.
  • The sleeves are made of plastic which may chemically react or degrade over time causing outgassing which condenses on the surface and may react with the hard coat to form more “opaque” spots which affects the readout laser.
  • The uneven surface may exacerbate this as the pliability would result in a pattern being “written” into the thin layer, affecting its refractive index (potentially) and the focus of the laser (more likely), resulting in a “pattern” of errors in readout. If the errors are dense enough, data is lost.

I really wanted the data back, so being the desperate person that I was, I decided to follow some advice given by Albert – after all, I didn’t have much more to lose.

Heat Gun Treatment

For the heat gun treatment, I decided to just experiment and see what happens. I used my Tenma SMD Rework station, which has a temperature controlled gun. I set the temperature to 180 degrees Celsius, ensured no concentrating nozzle was installed and set the flow to 75/100.

Before beginning, it’s worthwhile to clean the surface of the disc to remove all contaminants, and then blow using a photographic blower bulb to remove all the remaining dust.

I placed the disc with the data side facing up on my flat wooden desk, and waved the tip of the heat gun about 5cm above the data surface in a circular motion, making sure to move around the disc continually to avoid concentrating the heat too long at one particular spot.

Remember that we want to get the plastic hot enough to become a little more pliable to hopefully restore its shape and state, but we don’t want to expose the data layer to too much heat because that will destroy the data. The heating may have a secondary benefit in evaporating any outgassed contaminants that might otherwise have adhered to the surface.

Generally, you won’t need to use the gun for more than 20 seconds at a time as the substrate will begin to warp very quickly, causing it to “rise” near the centre hub ring. If the substrate sets “deformed”, this can affect readability, so I tend to stow the gun and then immediately put my fingernails on the centre of the hub and push down towards the desk to ensure the disc “sets” relatively flat.

You will notice a slight improvement in the surface condition, but it will not restore the surface to perfect flatness. That is not the aim of the process – the aim is just to restore enough readability to recover the data.

In order to gauge the improvement – I took the scans of six bad discs before the treatment and immediately after the treatment.

Disc 1

Disc1-SleeveReadErr Disc1-AfterHeating

After heating, the scan doesn’t look particularly appetizing, but it is readable. I was a little cautious, so I decided to go relatively quick with heating. The improvement is apparent, with a more than halving of the error rate.

Disc 2

Disc2-SleeveReadErr Disc2-AfterHeating

This one I gave a little more heat because of the bad state of the before scan. Unfortunately, it seems that the outer edge didn’t respond so well to the treatment, and retained its high rate of error. This may have been because of the original burn, thus giving us a clear reason as to aim for better burns as that increases the margin available for degradation. The first half of the disc after treatment is barely even passable for a fresh burn after 6 years.

Disc 3

Disc3-SleeveReadErr Disc3-AfterHeating

This disc was not as bad as the others in the middle, but the edge seemed to have issues yet again. Despite the horrid scan, it was possible to read back all the data, although not at full speed.

Disc 4

Disc4-SleeveReadErr Disc4-AfterHeating

This one seems like another particularly bad disc, even after heating it with the gun, but it restored it all to under 80 BIS which seems to be the safe threshold for readability in my experience. The reduction in error rate is pretty dramatic however.

Disc 5

Disc5-SleeveReadErr Disc5-AfterHeating

Again, improvements towards the edge seem more limited, suggestive of disc degradation, dirt or other influences such as warpage which may affect readability. I did apply additional heat for this round to try and see whether this could be improved upon over the previous samples where I presumed the outside may not be getting enough heat, but it didn’t seem to make any difference. On the upside, the disc did not bubble, smoulder or get damaged despite the rougher treatment.

Disc 6

Disc6-SleeveReadErr Disc6-AfterHeating

Another outcome, pretty similar to Disc 5. This makes six out of six heat gun successes – all discs did not read correctly and had CRC errors prior to the heat gun treatment, and all of them were read successfully afterwards.

This shows the value of optical disc scanning. Even if you can’t relate the readouts of the test to industrial test equipment, it still provides a relative gauge of how easy or hard a disc is to read using a particular drive assembly. With digital error correction, a “cleanly” viewed signal after error correction could be moments away from failure. This is the same reason why you have signal quality bars on your digital TV receiver, and noise margin stats on your ADSL modem.

While the discs were read and recovered successfully, they did not exhibit the same readability margins expected of good fresh burns or even degraded properly-stored burns. As a result, these “recovered” discs should be considered expired and due to be disposed, and this is just a last ditch effort to recover the data.

However, it also lends support to my observation that the test thresholds we use are too stringent – in all cases, the after scan was readable on the LG GGW-H20L with no data errors, so if your BIS peaks are below about 80, your disc is quite likely to be readable. Generally, after fresh burns, we demand BIS peaks around 9 maximum, which is very stringent. It didn’t seem repeated treatments provided much benefits – a single round of treatment was sufficient.

In all cases, the heat gun treatment, used judiciously, did not cause any harm and instead improved read error rates. However, I do not take responsibility for anything you do with this information – if you melt your precious disc, or set your house on fire, too bad!


Whatever you do, do not store any type of Blu-ray disc in any form of plastic sleeve – be it written, unwritten, pressed, rewritable. It seems that the sleeves will cause damage to the thinner and more fragile surface of the disc which will cause potentially irreversible data loss. If it’s too late, a heat gun treatment seems to allow for recovery of the data, but does not restore the disc to perfection. Consider yourselves warned.

As a result, I have officially abandoned all use of Blu-ray media for archival purposes. It is not stable enough, or reliable enough depending on the storage method. It’s also not fast enough, and too expensive compared to the cost of hard drive archival. It seems the heyday of optical media is well behind us.

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

Genuinely Confused: Another Samsung Galaxy SIII Battery (EB-L1G6LLU)

Around a year ago, I needed a replacement battery for my Samsung Galaxy SIII, which started a long series of posts encountering fake batteries. This provoked some responses, and a lot of interest, given the quality of the fake cells. I finally found a (believed to be) genuine cell and life was merry.

Or so I thought. Just a year on, the genuine cell I had decided to balloon out, just like the old one, pushing against the casing, leaning out so far as to occasionally lose contact and reset the phone.


The Curse of the Swollen Batteries

Swollen batteries are generally a big deal in the industry, and is one of the failure modes which have previously prompted recalls, as they can sometimes be the precursor to more destructive failure modes, such as an outright explosion. Most incidents are noticed by the user before they become too dangerous, and a small cut in the cell acting as a pressure relief vent often opens before something disastrous occurs, but it is a reminder of the volatility of Li-Ion chemistry.

The causes for swollen batteries are rather numerous. One of the main explanations involve the quality of the materials used in manufacturing. If the materials contain impurities, or excess humidity, this can cause gassing to develop during the charging process which causes the cell to balloon.

Another cause is overcharging. In Li-Ion, overcharging is not permissible, and many properly engineered systems take great lengths to ensure overcharging does not occur. This includes monitoring the temperature and voltage continually during charging and controlling the current accordingly. A failure of the charge controller to terminate the charge at the correct voltage can cause shortened cell life and swollen batteries.

Some seem to believe that swollen batteries tend to be due to storage at high states of charge. While it isn’t conclusive, generally, high states of charge cause voltage stress on the separator which could contribute to unwanted corrosion-style reactions. Unfortunately, I don’t know enough about the cells to explain this any further.

So, why did this battery end up swelling? Upon further reflection, it seems likely that manufacturing quality is probably a big part of the problem. The original cell I got with the phone was made 22nd June 2012, and the replacement claims to be manufactured 20th May 2012. This likely points to similar batches of material used in manufacturing. The original cell swelled to this state in about 1.5 years of use, with the replacement taking a year to reach the same state – again, a fairly similar amount of time.

Could the phone be sabotaging the battery? This is a distinct possibility, so I got out a multimeter and checked the charge termination voltage. The charge terminates reliably at 4.300v, as measured by an Agilent U1241B multimeter.

Modern cell phone batteries really push the boundaries of Lithium Ion chemistry, and in order to get a little more capacity out of the same sized cells, they tend to charge to slightly higher voltages than regular cells, and discharge deeper slightly deeper. This is why I really didn’t want a counterfeit cell as they are generally made to operate in the 3v to 4.2v “regular range”. The higher quality, high capacity cells, achieve their capacity by extending this range to about 2.8v to 4.35v. As the charge controllers are generally inside the phone, a counterfeit cell may be overcharged and overdischarged on every cycle resulting in a very short lifetime.

Unfortunately, such boundary-pushing also puts a lot of demands even on genuine cells, and if they aren’t made quite perfectly, then they will fail. It seems like this is the case with this one.

Sourcing Another Battery

Given that I did have a source of a (believed to be) genuine cell, why wouldn’t I purchase from them again? The phone is getting on a bit in terms of age, and the AU$50-ish cost to buy another genuine cell from them just wasn’t worth it.

Instead, I gave eBay another twirl, expecting to get a fake. Instead, I found a seller which claimed to have only genuine cells with working NFC functionality for a very reasonable AU$12.

This proved to be quite the mysterious item. Is it real, or is it a very very very good fake? I’ll let you be the judge – compare with this post.

Analyzing the Battery

The battery came in a thin plastic bag with a resealable edge and recycling logos printed in white. The battery inside seems to have a new style of print that doesn’t match either one of my former batteries. I suppose changes to packaging style is not unexpected over time.

20150413-1632-4558 20150413-1632-4559

Lets take a closer look at the cell printing itself.


The cell printing actually looks reasonable, with the label having very slightly rounded edges at the seam as expected. The battery claims to be “Manufactured by SDI”, which refers to Samsung SDI which is their energy solutions division. As nice as the label is, the text seems to have a slight thickness difference for the word China. Hmm. I’m not too sure.


The rear of the cell seems to have a mixture of font sizes, for the numbers in the brackets and for the phone numbers. It seems to be lacking a space after the word temperature/altas and the open-parenthesis, which I would expect if I was formatting the text. Inconsistently, between the word properly and call, the colon has a space on both sides, but that line omits a period at the end. I don’t have a genuine cell with this text to compare it with however.

On the upside, there is no obvious spelling mistakes on the label.

The printed datamatrix barcode does give some cause for concern. This datamatrix is low-density, and only encodes a simple string of GH43-03699A+AA1F920AS. My original battery encoded GH43-03699A+EB-L1G6LLU+C6NTX04261. The original battery string includes the model number and a alphanumeric string which is not related to the serial number (AA1C622TS/2-B). At least it does have some format resemblance and is not entirely garbage. I just hope that the serial number is unique. Anyway, this new cell does away with the manufacture date which seems a little strange.

Regardless, I have tested the cell and the NFC feature does work. This could be an indication that counterfeiters are getting much better and are able to include these features.

Lets go one step further, and start peeling back the label.


The label comes off rather smoothly, and reveals the NFC antenna on the cell, with a slight kink on the left side, mid-way up the cell. Looking at the end caps of the cells, we can see the “overlap” of the end into the battery area whose shape matches my original pretty much perfectly.


On the other side, we see a datamatrix code, some batch numbers and the correct cell type (i.e. an ICP515161A). Unusually, it claims to be made by Samsung SDIEM – I have no idea where the “EM” comes from, and I haven’t seen it elsewhere. If this is counterfeit, they’ve gone a long way.

I didn’t want to pull it apart any further, because I intended to use it, but for AU$12, this appears to be likely genuine in some way, but inconclusive based on the information I have. Good enough for an older phone, I suppose, provided it lasts.

Taking Apart the Old Swollen Replacement

As I have no use for the old swollen battery, I decided to peel it apart just to check.


The NFC antenna has that similar kink (the ripples are due to delamination in peeling the label back). The diagnostic text seems fairly similar, but the logo placement above the text differs.


The protection PCB has a single chip and a hardware fuse on a black PCB.


Again, it looks identical to my original battery that was my first swollen battery for the SIII. As a result, our conclusion that the Kogan supplied battery was a genuine was correct.

Posted in Computing, Electronics, Telecommunications | Tagged , , , , , | 1 Comment