Repair: BenQ DW1640 DVD±RW Drive Tray with Chinese Belts

In an era where IDE and DVDs have been left behind in preference for SATA and Bluray, it might seem strange that I’m choosing to repair my retail BenQ DW1640 DVD±RW Drive. Like many optical drives of any vintage, the rubber belt that powers the tray has degraded much like as what happened to my Bluray player. This time, I was going to replace the belt entirely and do the job properly as this is a drive that I am very fond of for a number of good reasons.

The Drive

My first DW1640 was a retail model that I asked a relative to buy for me from Hong Kong and bring into Australia. Strangely enough, I eventually came to own four DW1640s, with the remainder purchased locally as “OEM” drives without the fancy bezel, some of which have failed and been disposed of. Despite this, my first DW1640 was still the best of them all for burn quality and high-speed burning/ripping stability, and was quite affordable.

Despite BenQ being a subsidiary brand of Acer, the DW1640 was a big break from their (normally) rather unremarkable and pedestrian peripherals. My first DVD writer was a “plus-only” BenQ DW400 which was very plain but fared well for about half a year. It then succumbed to laser failure and owing to a lack of replacements, I got an upgrade to a LiteOn SOHW-812S which had its own benefits and drawbacks. Despite this, it seems BenQ DVD writers had a good reputation and the DW1640 was where a combination of BenQ-exclusive innovations came to fruition. As a result, it won CDRinfo’s editor’s choice award.

The most obvious when looking at the drive was their dual cooling system – the channels on the top of the lid are part of a system that helps funnel heat out of the drive while remaining (somewhat) dust resistant and quiet, all without the need of a dedicated cooling fan such as in early CD-recorders.

The other counterpart to it can be seen on the right side where a set of holes allow for heat to exit the drive through the side. Of course, there are a number of other nifty features which you cannot see.

Possibly due to the different top-lid design, they decided to label the drive on the underside. The drive is dated 14th May 2005, and shipped with BSGB firmware. The final version released was BSRB, showing quite active firmware development throughout its lifetime.

On the rear, everything is as regular as with most IDE optical drives with SPDIF, analog audio out, master/slave/cable-select jumpers, IDE interface and 4-pin Molex power connector.

The drive has survived being transported between a number of computers, and even survived a minature flood. The case has rusted in some spots as a result. Undoing four screws allows the rear plate to be removed, where a few thermal pads are installed to conduct heat to the drive casing as a heatsink.

The drive itself is based around a Philips NXP PNX7860E chipset. To continue disassembling the drive requires an emergency eject with a paperclip, so as to remove the front fascia panels.

It’s at this point we can see the belt and wheels – there’s some discolouration which may be due to the rubber losing its elasticity and degrading. The kink in the belt shows that the period when it was stored without use has caused the belt to permanently deform. If the tray has enough clearance, it may be even possible to use some needle-nose tweezers to do the replacement without disassembly, but I decided to tear it apart first.

The lubrication on the tray end looks fine – hidden labels under the tray are not uncommon.

The upper lid has the spindle clamp kept captive within it, as is common for PC drives.

The remainder of the mechanism is built around a plastic frame. While it’s not obvious, we are looking at the special lens focusing mechanism which is (supposedly) better at steering the beam and compensating for disc warp. We are also looking at the turntable spindle clamp mechanism which has been designed to be more precise.

To fully release the tray, a small tab needs to be pushed down, then the tray can be completely slid out.

The offending belt is a much longer belt (about 138mm circumference) than the average drive has (which is about 80mm circumference).

The Cheap Belts

In my search for belts, I found some listings on eBay for individual belts, but at fairly high prices but of an inappropriate dimension for this particular drive. As a result, I got desperate and decided to cheap-out on belts, ordering a “60+ mix 1mm square belt 40-130mm” packet for just AU$1.89 including postage. It was so cheap that I didn’t care much if it was going to work or not.

When I received it, the package smelt heavily of rubber. This suggests that the stuff is probably not particularly high-quality or stable in the long term. It came in a sealed clear plastic bag with a product information card in Chinese that had not been filled out at all.

A barcode inventory label was affixed to the other side.

The belts came in about five different sizes (roughly sorted). The number of shorter belts seemed to be higher than the number of long belts. I didn’t count the number – for AU$1.89, it was cheaper than even ordering just one belt from some other sellers.

Unfortunately, the quality of the belts did vary, with the belts being rather under-dimensioned for thickness, and some belts being thinner than others with occasional rough edges.

I took one of each size, cut it and measured the length of the belt unstretched. It came out to 84mm, 109mm, 177mm, 248mm and 260mm (circumference). Many times, the belts are specified as their half length, which is more like 42mm, 55mm, 89mm, 124mm and 130mm. Unfortunately, the smallest ones are a bit too long for most optical drives which need about 40mm half-length, and the slippage results in it not working properly. In the case of my Philips Bluray player, none of these belts proved helpful.

But I’m glad to report that I did manage to get a belt to work with my DW1640 – namely a stretched one of the smallest size, as the 109mm (while expected to work) stretched too much when in use and slipped. The smallest one is a bit tight, but it does work, so that’s better than having to get the emergency-eject-paperclip every time.


The BenQ DW1640 is a special drive that holds a place in my heart. As a result of trying to fix it, I found a very cheap packet of rubber belts from eBay and decided to give it a go. While it did fix the problem, the belts themselves were not quite correctly dimensioned and they smelt very rubbery, which suggests to me that they might not last too long in service. It’s still nothing to complain about, but I hope to be able to show why the DW1640 is such a special drive in some follow-up posts.

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

Tech Flashback: Telecom Australia 600-series Adapters & Extensions

In the present day where landlines are a dying breed and mobile telecommunications has become ubiquitous, the beige-cream coloured Telecom Australia 600-series socket remains a common sight in Australian households. While I have no idea when the 600-series socket came into common usage, it would have likely existed around the same time (or even before) the inauguration of Telecom Australia in 1975, making it over 42 years old.

A Telephone Extension Reel

When I was young, these telephone cord extension reels were quite popular and could be bought from any variety shop for a reasonable price. Similar in concept to a power extension cord reel, these were fitted with a 605 plug on one end and a 610 socket on the body of the reel. The reel had about 10 meters of flat telephone cable. The design with the holes mimicked the design of a rotary pulse dial of a telephone of the era. The whole unit was the same “beige” colour that typified telephone accessories, and was clearly labelled with the Telecom Australia Authorization Number of C85/1/45. As it was illegal to connect unapproved devices, this was a necessary marking to demonstrate to the consumer that the accessory was approved. Despite being fitted with Australian 600-series sockets, the unit was made in Hong Kong. This one was actually ours and had survived a number of house moves.

The 605 plug can be seen to have three blades with two contacts per blade. A wider non-conducting spigot is provided. As the extension leads are intended mainly for temporary use, the spigot has a “filled” core – some others intended for semi-permanent installation had a hole, allowing a screw to be fitted in the mating 610 socket to hold it in place. The plug itself has an approval RA85/121 as well.

Removing the single slotted screw allows the cover to pop off. Six contacts are clearly visible, to which three lines (each requiring a pair of connections) could be connected. That being said, only contact 2 (tip) and 6 (ring) are carried by this extension lead, which will limit the functionality of some devices, as noted later in the section on Mode 3 Adapter. The configuration of connecting 2 and 6 is most common, and is the minimum necessary to support Line 1 operation.

The design of each of the blades is quite simple – an insulating plastic form has a channel in which the contact pin sits. The end of the pin is flared out, to allow a crimped spade connector to be connected to. The whole pin-form-pin sandwich sits inside a cut-out in the socket, across a channel at the bottom which prevents the assembly from sliding back into the plug. It’s likely that this is a later design, owing to the thin profile of the pin contacts.

A partner 610 socket is used in the centre of the reel, but as I didn’t want to disassemble the reel and risk breaking it, I didn’t take any photos of it. The concepts are easily conveyed with the next few adapters.

Telecom Australia Mode 3 Adapter

A Mode 3 adapter is not something that is too often seen, but when it is, it is often incorrectly mistaken as a double-adapter that doesn’t work. On a recent trip to a local thrift shop, I came across these.

These are Telecom Australia branded accessories, item number 557/3, still in its original packaging. This really speaks back to the days where you’d rent a telephone from the telco. I like the tagline – “The Vital Connection”. The adapter is used to connect various types of communication equipment (Viatel terminals, older fax machines, alarm diallers, answering machines) which have a special Mode 3 connection. This type of connection passes the telephone line into the equipment through the line 1 terminals, and returns the line through the line 2 terminals. This allows the device to “pre-empt” any downstream devices and take them off the line, so as to seize the line without any potential interruptions, similar to how a dial-up modem may have a line and (switched) telephone socket. This is “neatly” packaged into an adaptor, for where a Mode 3 socket doesn’t already exist and Mode 3 functionality is desired.

It comes with a fold out instruction leaflet which details how to install the adapter, including the potential issue of meeting a semi-permanently installed plug.

The adapter is in the beige colour, as expected. The front has the Telecom logo, the rear has the text describing the function.

The adapter has a subtle, tapered shape. Some others were more angular, but I suppose this could be a hint as to its age. Each of the blades has a rounded profile with a large contact area.

A closer look at the front shows that the spigot has a stepped profile. I believe this was to key the connection to avoid inappropriate connections, although I’m not particularly clear as to how that would work.

Support for this is seen on the “input” side of the adapter. Notice the spigot holes are different shapes for the two ports. The lower port is used for the telephone, the upper port is used for the Mode 3 device – inserting them in their correct sockets is vital for correct operation.

The unit can be disassembled by prying out the front cover.

Internally, two different types of pins are pushed into the plastic holder.

Two of the pins at the ends are of the sort that I will term seriesing pins. For the upper device, it has its pin 2 and 6 wired connected directly. The pin 1 and pin 5 connect to pin 2 and pin 6 of the lower device. The lower device has its pin 1 and pin 5 connected directly.

This has the effect of routing line 1 through the upper (Mode 3) device, which returns the line through line 2 connection which is then given to the lower (telephone) device. The completion of the connections to the telephone allows the telephone socket to also work on line 2 connections.

To facilitate the downstream telephone working on Line 1 while no Mode 3 device is plugged in, the contacts for pin 1,2 and 5,6 can be bent toward each other so they short together when no device is inserted. I presume the device may have been designed in this way, but whether the springs have retained their original position is not known.

The pin used for the middle connection is what I term a straight through pin. This basically has things parallel wired. This means that line 3 connections are duplicated to both devices in corresponding fashion (i.e. pin 3 to pin 3, pin 4 to pin 4) without interference.

It can be seen how this lends itself to easy production of double adapters and Mode 3 adapters through the use of similar parts.

If this adapter was plugged onto the end of the extension lead above, it would work just fine with the exception of having no line 2 or 3 service. However, if this adapter was plugged into the wall, and the extension lead above was used to connect a Mode 3 device, then everything breaks, because the Mode 3 device will attempt to return the line via the line 2 connections, which are not wired through on the cable. As a result, a four-core extension cord with line 1 and line 2 wired is necessary to extend a Mode 3 device (assuming it doesn’t use line 3 at all).

AWA Telephone Double Adapter

In a case of being in the right place at the right time, I found this in the same thrift shop in the same pile. I suspect they may have come from the same place, but this one is branded AWA (another bygone Australian brand). This is a double-adapter intended for connecting two telephones to a socket. Note how as a third party brand, it displays the necessary authorization numbers (C87/1/252). In this case, all the instructions are shown on the rear of the package, but note the similarity of the font between this and the adapter above.

This adapter has a hole through its spigot, allowing for semi-permanent installation. This has an AWA logo moulded into the same place as the Telecom logo was in the above (which implies the same manufacturer may have been involved). Because it is a double adapter, there is no distinction between the two outlets, both of which have a square spigot hole.

It should come as no surprise that it comes apart in the same way and consists of three straight-through modules, duplicating each pin in perfect correspondence.

RJ12/RJ45 Adapter

The RJ12 adapter is a common device which is often found bundled with any telephone, modem, fax machine sold in Australia. The RJ45 version is a little rarer. This one I found “dumped” outside during a council clean-up, so I decided to take it home just to take it apart.

Being a much more modern device, the full plug body is now omitted, and no screw is provided as the adapter is glued shut. An Austel approval is present on the side, and the pin for line 3 operation is omitted entirely as well.

With cost-reduction in full swing, the contacts for line 2 are omitted as well, making this a line 1 only adapter – the most common type and sufficient for most uses as the RJ12 connector is rarely used for two or three-line service.

The adapter itself accommodates an RJ45 style plug natively, but works with RJ12 plugs through a reducing sleeve. Only the centre two contacts are used.

The construction follows very similar principles to the wiring in the extension lead, with some corrosion appearing on the lower pin. This is probably because it had been exposed to water or chemical vapours in the past.


It’s interesting to think how we ended up with this “odd” connector that isn’t really used anywhere else (to my knowledge). Despite being a British “outcrop”, we could have just as easily adopted the BT socket just as New Zealand (across the pond) did, but we didn’t. This may have something to do with the Postmaster General or Telecom Australia’s commercial interests. That being said, the US RJ12-style and RJ45-style sockets eventually became the preferred means of telephone device connection, being more compact, cheaper and quicker to terminate but also pesky with the tab that likes to snap off.

Despite the old appearance and the limited market, it seems that the 600-series connectors will live on in some households as RJ12 adapters are ubiquitous, and the landline may be used to supply VDSL2 services over the new “mixed-technologies network” NBN that has been a disappointment all round. That being said, I don’t think such old connectors are optimized for impedance or high frequency operation, and all those connections would only serve to impede the connection quality, so maybe it’s a good idea to change over your wall plates and get rid of the beige.

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

Review, Teardown: LaCie 2200mAh Promotional Power Bank

Earlier this year, I attended the SMPTE exhibition and was lucky enough to receive a power bank as part of the bag of goodies that I snagged, thanks to LaCie and Silicon Memory Technologies. As a result, I thought I’d give it a thorough test just to see what it’s like inside.


Before I begin, for all avoidance of doubt, LaCie does not make this power bank as far as I can tell. It is merely a promotional item that is used to promote the brand, and doesn’t reflect the quality of the products in LaCie’s range.

The power bank is packaged in a white matte finish cardboard box with an off-white LaCie sticker on the front.

All other sides of the box are blank, with the exception of the rear, which states that it was Made in China.

Inside the box, there is an instruction leaflet. This leaflet seems to imply the unit is good for a 5V 1A output for 300 cycles, and contains a 2200mAh “A-grade” Li-Poly cell which is enough to charge a device once. It claims to be unsuitable for tablets (which is sensible advice to give).

Unlike most other power banks, the leaflet is marked with the Australian Regulatory Compliance Mark (RCM) that means it meets our standards.

The body of the unit has a roughly-credit-card sized footprint. The unit is about as thick as a regular 2.5″ laptop hard drive. The whole unit forms a neat rounded rectangle, with a status indication LED window in the middle left. The branding is printed in the center, with the output available at the top left.

The rear of the unit is not particularly exciting, but it does give the specs again – namely 2200mAh capacity, 5V 800mA input and 1A output. The Australian RCM makes an appearance again, along with a recycle logo (which seems a little strange).

Charging is through a microUSB-B port on one side of the device.

Output is through a flat-flexible cable terminated with a microUSB-B connector. An Apple Lightning adapter is also included which tucks into a small hole, although whether this actually works with the most recent iOS was not tested.


The unit is held together with no screws, and merely clips together. Unclipping the two halves is almost certainly going to result in breakage of clips, so is not advised.

The cell is not marked on this side. The power conversion PCB takes only a small space in the top corner, and is free of components on its rear.

The top side of its PCB shows just how few components are involved in making a power bank. This one uses a DW01A with an 8205S MOSFET to form the Li-Poly protection circuit. A TP4056 linear Li-Ion charger is used to charge the battery, whereas the D6GG7K is an integrated switching converter that provides the output using an unshielded 4.7uH inductor and a few ceramic capacitors. Two LEDs are used to provide indication. This is very much a cost-optimized design, and in fact, is very similar in principle to the other “free” power bank I received in Korea.

The rear seems only to have a few interconnecting traces, otherwise, is completely ground plane. If they could get rid of it, maybe they could have gone for an even cheaper single-sided board!

Looking towards the output, we can see that the data pins on the USB side are unconnected – not even soldered for mechanical support, probably to save a little money and avoid shorts. The output, however, is concerning. The wire used is “headphone style” enameled copper which is tinned at the ends. The length of the tinned segments makes shorting a possibility if the wire is not assembled quite right into the casing (as it is a tight squeeze). Because of the squeeze, the copper colour wires have been wrapped into a very contorted mess, and some strands (not visible in this image) are broken and “free”. Should those make contact with any of the others, then a short circuit could result producing some heat (current limited by the converter) which might end badly depending on the duration and resistance of the short. Not the best sort of construction – some sort of heatshrinking would be a good idea.

This unit has a bit of a strange quirk. As it seems the blue activity LED is driven by the boost converter, it lights up very dimly in brief pulses when no load is present. The duty of the pulses vary depending on the load, and attempting to measure the LED pulses even with an oscilloscope probe at 10Mohms resulted in some variations.

It seems that the LED pulses are about 25% on time at a rate of about 500khz, implying that the converter is basically running all the time, relying on the capacitors and other components to have such low leakage (and the converter to have such low quiescent current) that the unit doesn’t self-discharge too quickly. While it simplifies the design somewhat, it’s also somewhat inefficient.

As for the claim of an “A-grade” cell, I’m not so sure. This cell manufacturer really wants to stay anonymous and most A-grade cell manufacturers don’t.

Performance Testing

As the unit didn’t claim to be able to do more than 1A, I only tested the discharge characteristics at 500mA and 1A using the B&K Precision 8600 DC Electronic Load. At the 500mA rate, it could deliver 1460mAh at 5V or about 1973mAh at 3.7V implying 89.7% efficiency if the cell was truly 2200mAh. At 1A, this fell to 1332mAh at 5V or about 1800mAh at 3.7V for an 81.8% efficiency. The voltage remained relatively stable, although on the lower side of the allowable USB voltage. This can act as a “signal” to loads not to draw excessive currents, but will also decrease charging speeds in some circumstances. Cut-off occurred at the end of discharge, protecting the cell from over-discharge.

At 500mA, the ripple was averaging just 7.452mV peak-to-peak on the Picoscope 2205A, which is exemplary performance. The larger ripple component was at 486khz, although it seems there may have been a much higher frequency ripple component which would make sense as higher frequencies generally minimizes the converter and capacitor size, at the expense of higher inductor losses.

At 1A, the ripple was averaging 12.08mV peak-to-peak, an excellent value. The frequency was about 528khz, although again, a higher frequency component can be seen.

As promised, the charge current (as measured by a Keysight U1461A connected to a modified Charger Doctor) was just shy of the claimed 800mA at 782mA. This fell as charging progressed, as per the standard “linear” charging IC characteristic. Charging took about 3 hours and 23 minutes, delivering 2045mAh. This is close to the claimed three hour charge. As some of this current is lost to powering the LED, it seems unlikely the cell is 2200mAh, and is probably closer to 2000mAh. The cell voltage was measured on the Keithley Model 2110 5.5 digit multimeter as peaking at 4.23775V. This is a bit higher than the 4.2V termination voltage that a standard battery would expect by 0.8% which may contribute to slightly shortened cycle lifetime, but is within the datasheet claimed accuracy of 1.5%.

An output I-V curve was developed using the B&K Precision 8600 DC Electronic Load. This shows the output with a nearly constant slope as current increased – output was taken via 18AWG wire and crocodile clips at the solder joints, thus lead resistance is much lower than usual, and most of the slope is due to the converter itself. The unit was run open, with five seconds at each current level. The output was falling continually by 1.9A and was completely disabled at 2A. This suggests that the over-current protection may be related to the thermal performance of the converter IC’s over-temperature protection, thus currents of 1.8A may only be possible for short durations. When placed inside the casing, it’s unlikely that it could be sustained, and this may be the underlying reason behind the 1A claimed output current. It is short-circuit protected as a result, and did not suffer any permanent damage from overload.


I’m a sucker for free stuff and it’s always great to get some. However, that being said, it still deserves some scrutiny, just to be sure it’s “safe” to use and to learn a little about what it’s made of, how it’s made and how it works. In this case, the power bank pretty much met all of its claims with exemplary ripple performance, with the exception that the capacity is probably closer to 2000mAh rather than 2200mAh.

The unit does have some strange behaviour in the sense of the flickering blue LED in idle which probably needlessly increases quiescent current consumption. The use of enameled wire on the output is probably not the best choice either.

However, at the low-costs such an item has to target, it’s not a bad effort at all and definitely something people might get some use out of. Thanks LaCie!

P.S. You might note that the suite of power bank tests are changing gradually with each unit – I hope to further refine this as my suite of equipment improves and other issues become important.
Posted in Electronics, Power Bank | Tagged , , , , , , , | 2 Comments