Lycamobile Data: No More VPN Shenanegans!

I’ve been with Lycamobile for my data services for a while, mainly due to price and coverage reasons, but one of the biggest drawbacks was the wacky way in which their transit arrangements routed your traffic, resulting in high latencies and limited throughput.

Mid last month, I noticed something unusual – for a while, my phone was able to connect to data without the roaming icon coming up, and then after about 30 minutes, was kicked off and back to the roaming set up. I suspected there were some network changes going on at the time, but could not determine what was going on.

I haven’t been carefully keeping an eye on the service, but just yesterday, I discovered that Lycamobile no longer terminates the data connection in the UK. No configuration changes are required at your end, and the roaming indicator is still active, but the data connection is now terminated in Melbourne, Australia.

I discovered this when I decided to do an impromptu speed test, having the feeling that the service was no longer as choppy as it was before. Speedtest.net’s app chose to use the South Australian test server, and turned in a much more respectable result:

To South Australia

The ping time is slashed by a factor of close to 7 and the throughput is over 10 times higher for download and about 4 times higher for upload than in the past. Manually selecting the Telstra server in Sydney, I get the following:

Sydney Server

The latency has gone up, but the results are fairly similar. There is evidence of some form of buffer bloat somewhere, as the download tests take an age to finish, and the upload tests have a strong leading-peak. The latency has gone up a bit, but that’s normal range for 3G service. The download shows a strong “TCP slow start” effect, meaning downloads will take a while to ramp up to full speed (probably due to throttling or packet loss).

In an effort to try and improve the results, I tried giving Telstra’s Melbourne server a test.

Melbourne Server

Disappointingly, the test still resulted in similar figures, with the upload a little less, and strangely, the latency even increased. It’s still not Telstra-grade internet, but it’s being “carried” over the Telstra 3G network. I wonder whether the speed limits we see are artificially imposed in a way similar to the previous reseller agreements with Kogan/Aldi Mobile and their upstream supplier (i.e. 7.2Mbit/s).

A quick check of the IP address still reveals that they use some carrier grade NAT, so this service isn’t going to be something to use when you’re looking to serve something.

Screenshot_2014-10-01-15-22-43

Maxmind’s GeoIP2 demo page suggests that this IP block belongs to Lycamobile itself, so in essence, it is its own ISP.

Screenshot_2014-10-01-15-23-49

Unfortunately, the data service is still only accessible when roaming is enabled for data services. This may change in the future, if they reconfigure their network again and new carrier settings are pushed to phones.

What does it all mean?

This is likely to mean a much better service for Australians on Lycamobile who are accessing local content. They won’t experience the long latencies before, which will make the service feel snappier, and they won’t be subject to the transit congestion of 10+ hops. That’s a good thing. Throughput is much higher, and competitive now with many other “value” providers which should improve customer satisfaction.

Users won’t be “redirected” away and warned that their services are in the UK and they should try their UK version of the page. Likewise, local services such as banks and educational institutions are likely to no longer treat logins from your Lycamobile service as suspicious as they will now appear to be originating from Australia.

On the downside, people who have been relying on it as a proxy into the UK can no longer do so – no more BBC iPlayer for you!

Posted in Computing, Telecommunications | Tagged | 1 Comment

USB Cable Resistance: Why your phone/tablet might be charging slow

Tablets and smartphones are now ubiquitous, and almost universally, they rely on some form of USB connection to provide the power to charge the device. This is, in no doubt, because of a desire to reduce waste and meet EU regulations and has made it rather convenient for end users as cables and chargers are easy to find and are, to some degree, interchangeable. However, it does cause problems sometimes – namely slow or incomplete/inconsistent charging.

The Problem

When USB was conceived, the maximum current available was 500mA at 5V for a total of 2.5W. This is enough to power most smaller devices, but with smartphones and tablets, this is often only enough to power the device without any surplus left for charging.

In order to get around this, many manufacturers increased the current delivered through the same port/cable to higher levels, beginning at 700mA, 800mA, 1A, then 2A and now even 2.4A. But the devices needed to know when they were attached to a high-speed charger, and hence an array of incompatible, vendor specific communication methods utilizing voltages on the D+ and D- lines were developed, as well as a later USB Dedicated Charging Port specification.

This is why, when an iPad is attached to a regular USB port, it will display “Not Charging”, and when devices are attached to chargers of “other” products, they can sometimes charge at reduced rates or display “Not Charging”.

However, an often overlooked issue relates to the cable used when charging. It’s all too common that cables eventually wear out, get lost or damaged and users end up replacing them with an aftermarket cable. A cable is just a piece of wire, right? Almost.

It turns out that the wire thickness used inside the cable impacts on the resistance of the cable assembly – this resistance causes energy loss inside the cable when an attached load draws a current, and causes a voltage drop which can reduce the voltage to the end device to a point where it is not possible to charge quickly or completely.

Users can often exacerbate this by attaching USB extension cables for convenience or buying longer cables.

This will often cause the device to indicate it is charging, but be charging very slowly, especially towards the end. It may also cause the device to finish charging, but only at an intermediate state (e.g. 94%). If it gets really bad, using the device while attached to the charger will cause continued depletion of the battery rather than charging it.

But how bad is it really? Lets do some basic calculations to find out.

The Calculations

The thickness of USB cable conductors is normally specified in American Wire Gauge (AWG). In this system, larger numbers indicate thinner wires. As the USB specification makes reference to AWG numbers of 20, 22, 24, 26 and 28, calculations were performed for these AWG figures.

USB Cable Resistance Data in Calculations

The data used in calculations is summarized above. The conductor resistance is taken from Wikipedia and assumes a copper conductor. The USB contact resistance of 30mohm is a “middle of the road” figure – many connectors specify 10mohm when new, and 30mohm through their lifetime, although 50mohm is acceptable for micro connectors.

The voltage drop is derived from ohm’s law – Voltage = Current * Resistance.

The resistance itself is calculated by the route length which is twice the length of the cable (as current travels from the charger to the device and back). For the one with contact resistance, four times the contact resistance is added to compensate for the charger positive, device positive, device negative and charger negative contacts.

This figure is in mohms, and is divided by 1000 to get to ohms, and then finally multiplied by current in amps to get the voltage drop.

The Results

The results are colour coded as to their voltage loss. For a 5v output, the USB specification demands that the voltage remains within 5% (i.e. an acceptable voltage drop of 0.25v). All voltage drops less than 0.25v are colour coded green. From there, voltage drops of between 0.25v up to 0.5v are colour coded yellow. This is because, while the USB specifications are stringent, most devices only require 4.2v – 4.35v at the battery to fully charge. Most chargers are of the linear or buck (step-down) type, and thus the supply voltage must be greater than the battery voltage for charging to happen. Accounting for losses in the charging circuit, it is determined that approximately 4.5v is required to ensure a full charge. Any losses greater than 0.5v are coloured red, as they are likely to cause problems.

With Contact Resistance

USB Cable Resistance with Contact Resistance

With contact resistance taken into account, it can be seen that it is difficult to meet requirements at high currents of 2A and 2.4A. As a result, cables only up to about 50cm can be used with 24AWG, or maybe even 1m with boosted source voltage. This is most significant for tablets.

For smartphones, with a requirement closer to 1A, 24AWG wires up to 2m could be sufficient, or 1m at 26AWG, and 50cm at 28AWG.

However, at 500mA (the original spec), it can be seen that it is possible to meet the stringent USB voltage requirement at every length with 20AWG wire, and 2m with 24AWG (probably by design).

The disadvantage of pushing more current at low voltages is very clear – the resistance causes power loss very quickly!

A corollary of this is that if you have a 3m cable, of 24AWG, then it’s likely your charge current is not going to exceed 1A (it’s probably 500mA-1A) purely because of the voltage drop that is caused by the cable.

Without Contact Resistance

USB Cable Resistance Negating Contact Resistance

If we assume that connector resistance isn’t part of the equation, and look at the wires itself, the situation is a little less stringent, at high currents, but still illustrates the difficulty in keeping voltage drop under control.

One caveat of this, is that some vendors have realized the issue and decided to push the output voltage up to 5.1v or 5.2v, which is still within USB specification but allows for an extra 0.1-0.2v voltage drop. This is a potentially nice feature as it means the requirements on the cables are slightly relaxed (i.e. maybe even 0.7v voltage drop is tolerable).

It is also possible to see higher charge currents at the expense of voltage drop when the cell is fully discharged at 3v, as a ~3.2v input would be enough to start charging. Likewise, as the cell reaches full charge, the current tapers off, thus reducing the voltage drop. This might be happening already with some cables “pushing” what is possible. You know when it’s pushed too far when it doesn’t consistently charge fully (i.e. stuck at 94%, when eliminating all other issues).

Checking Your Cables

Depending on who makes your cables, it might be possible to determine what sort of wires your cable is made out of. Ever seen the text on the side of your cable? Well, here’s a quick primer on what to look for.

DSC_8209 DSC_8210

The two cables above are pretty much equivalent. The top cable says 28AWG/1PR AND 24AWG/2C, and the bottom cable says 28AWG/1P+24AWG/2C.

What this means is that the cable is made of one data pair of 28AWG thickness, and two conductors (that carry positive and negative) of 24AWG thickness. Remember, the larger the AWG number, the thinner the wire.

The data pair thickness really is unimportant when it comes to charging, so just look for the AWG attached to the conductors. The 24AWG thickness is pretty common stuff.

DSC_8211

That being said, many of the cheaper cables utilize 28AWG/1P+28AWG/2C. This isn’t the best stuff for charging.

DSC_8212

It’s possible to be surprised – I had to look hard to find this one which is 28AWG/1P 22AWG/2C. I’ve also seen 26AWG used, but 20AWG is almost impossible to find but is referenced by the USB standards.

Unfortunately, many newer cables are not marked at all as to their composition, and in the case of some cheaper cables, it can be suspicious as they often utilize non-type-approved cables of custom construction, or leverage “audio” grade cable without the appropriate twisting or shielding for USB operation. These cables still often work, but poorly in RF/EMI harsh environments, and often have thinner conductors, but this is not a given as many OEM cables are unmarked as well.

You cannot tell the thickness of the wire within just by measuring the external diameter of the cable either. In the case of the cables examined above, all three have extremely similar external outer diameters, but the wire thickness inside is obviously different. The printing itself could be faked as well, you never know.

Other problems can include low quality connectors which don’t fit well and exhibit high contact resistance because of poor quality plating.

Unfortunately, direct measurement of the resistance of the cable is complicated. This is because of the low resistance of the assembly itself – below 2 ohms – and resistance contributions of the connectors themselves which may make it less easy to quantify the cable resistance accurately.

An attempt can be made by making a custom rig with a USB A Female end and a compatible mating connector for the other end of the cable, and running a known current (say, 2A) through it to measure the voltage drop.

However, for most concerned users, an easier method can be to try measuring the charging rate of the device by noting the amount of charge accumulated from flat after an hour or hour and a half of charging. In the case of problematic cable, there will be a significant difference.

Conclusion

If you want to avoid any charging problems, it’s always easiest just to stick with the OEM charger and cable. Replacement of chargers with other models may result in incompatible signalling that can cause slow charging.

However, if you are going to replace your cable with an aftermarket cable, it would be best to see if you can find a cable with the thickest possible conductors for the power. If that isn’t possible, stick with short (to very short) lengths, as that always works.

It’s not possible to be sure of the thickness of the wire just by the external appearance – printings can be forged, insulation can be made thicker (quite common). Likewise, it’s not possible to easily directly measure the resistance by using a regular multimeter due to the complication of the connectors, and the very low resistances (sub 2 ohms) involved.

If you really need to have a longer cable, consider extending the primary side using a mains extension cable instead.

However, I think it’s clear that by pushing the current levels through USB to higher levels than originally intended, we are losing efficiency and reaching the limits of low-voltage power distribution. That’s why, the USB power delivery standard is opting for higher voltages.

Posted in Computing, Electronics, Tablet | Tagged , , , , | 2 Comments

Quick Review: Tenma SMD Rework Station Bundle

Have to say I’m on a bit of a roll when it comes to bargain chasing, because my shopping at element14 continues. One tool which I don’t have, and had always wanted, was a hot air rework station. I’m no certified BGA reballer, nor have I got any SMD skills at all, but having one would let me pluck components off boards and give me an easier time trying to solder some of the SMD parts to adapters. It might even give me a chance to salvage and repair some things I wouldn’t otherwise have any chance of doing.

For that matter, I also wanted a decent soldering iron.

DSC_8183

I’m sure many people have one of these nasty buggers lying around somewhere. This one was a Super Cheap Auto $7 60W iron I got at 15% off on a sale. This iron probably uses some nichrome wire heating element hooked up to the mains, with no thermal regulation whatsoever. I know this because it takes an age to get to temperature and if you use it for longer than about 30 minutes, it gets hot enough to burn you through the handle!

The tips on those were plenty terrible – so this one I improved by shoehorning in a Weller tip – originally conical, but now worn to resemble a chisel shape. I’ve desoldered the capacitors from over 20 motherboards and replaced them with this iron, and I’ve built many circuits on veroboard and kits as well. Without thermal regulation, it can be a little hard at first to get it right – but I’ve gotten used to its foibles over time, knowing when to give it a little time to recover, and soldering just enough to keep the tip from getting too hot.

This sturdy, primitive iron has done all of my soldering and desoldering since 2004. It’s always been handy, but it’s probably about time I upgraded.

Normally, such SMD rework stations are pricey and out of the reach of a student. I was definitely in for a surprise when I spotted this bundle which includes the Tenma 900W SMD Rework Station, a special in the Connect magazine, going for AU$142.60. It includes a soldering iron and a hot air rework system in the one unit.

The bundle itself seems to include the following (and the corresponding regular prices are):

2062633 Rework Station 900W 220V UK/EU 226.58
2102733 Multicomp YP-35/YC-12 H05VV-F – Mains Plug IEC C13 1.83M Black 2.19
5090854 Multicore Solder 2096125-M 60/40 1.2mm 500g 68.04
3125634 Duratool 8PK-366D-F Desoldering Gun Antistatic 9.68
1156001 Duratool PL-501 Micro Cutter 5″ 8.6
1616334 Duratool D00837 Tweezers Type 13 SMD SA ESD 120mm 26.62

That’s about AU$340 of stuff going for $142.60. In fact, just getting the cutters, desoldering gun and reel of solder would be plenty useful for me.

Of course, when you see specials like these, you have to be a little careful – quality is probably not one of their strong points, and with Tenma, things can be a little hit and miss. In fact, just looking at the pricing, even if the station itself was useless, I wouldn’t be losing by much (AU$27.47) – worth a gamble! That’s probably why they’re all out of them at this present time!

Unboxing

The bundle itself came in a box of its own, with the rework station inside the product box.

DSC_8073

The box itself was a little crumpled, but hey, it doesn’t look too bad.

DSC_8075

The features are shown on the side of the box. There seems to be some evidence of “chinglish” expression in the grey box, referring to the product as “it”, and the strong personification of the product.

The unit inside requires some assembly – the side bracket for the hot air rework nozzle is not attached and must be screwed on. There are mounting points on the left and right side, so you can choose either one. You will also find the bracket itself is magnetic, so it will tend to try and “grab” to the side of the unit when screwing the bracket on – no doubt this is used by the nozzle to sense when it is placed into the holder to go to sleep.

DSC_8081

Another assembly step – the soldering iron pencil has to be plugged in and screwed in as well. The soldering iron comes with a thin conical tip, metal stand (impressive) and a regular sponge cleaner (slightly less impressive). Unfortunately, iron doesn’t fit nice and snugly into the stand, and has a habit of backing off a little and tilting upward, resulting in the metal locking collar ring making contact with the stand. It’s not what I would expect to be the best outcome – it would be nicest if the surrounds were nicely snug so that the iron was held in place at the plastic edge so that the heat isn’t being transferred to the stand.

The length of the cable between the rework station and the pencil (and hot air rework gun) is only about 1m, making it a bit short for my liking. Unfortunately, the cable to the soldering pencil doesn’t seem to be burn-proof cable either (it doesn’t feel like silicone rubber, but I’m not going to try and burn it). The hot air rework gun has three buttons on it which allow for changing the temperature and flow from the handle itself, although with the way the tubing is fitted, the gun always wants to have the buttons facing the holder where it can inadvertently be pushed.

DSC_8078

The unit looks like a piece of sound and lighting equipment for the stage, black and with a (plastic) carry handle. The front is adorned with a hardware power switch, two LCDs (one for the hot air rework, and the other for the solder iron) and four push buttons. There are individual push buttons to turn on and off the soldering iron and the hot air rework station, and shared buttons to control the temperature, airflow, etc.

The system is microprocessor controlled and employs feedback to ensure temperature regulation. The unit is capable of reporting heater error and sensor error via the LCD in case of failure.

Not pictured – the unit has a fused IEC connector at the rear for power and is supplied by default with only the UK and Euro IEC cables. There is a separate IEC cable for Australia included in the bundled products it came with. The unit itself seems to be rated for 220v by the nameplate – this seems a little odd.

Most products are rated by their harmonized voltage – i.e. 230v. Because it claims to be 220v, I think it’s likely to be (really) a 220v product made for the Chinese market, and its lifetime could be shortened when used in 240v countries. Otherwise, it could be a 230v harmonized product incorrectly labelled.

The quality of the construction of this unit can only be described as marginal to poor. The soldering iron connector is off at an angle, there are outward dents on the top panel, and the power switch is a bit crooked as well.

DSC_8080

Anyway, I’ll be happy if it works. Lets take a quick look at the other stuff that it came with – for example, this selection of nozzles for concentrating the hot air when working with particular sorts of SMD components.

DSC_8093

Also included is this manual … notice how they misspelt rework on the front cover …

DSC_8094

Unfortunately, this is where the “chinglish” sets in, because this manual is really hard to understand, and it makes my head hurt. I still haven’t gotten around to working out how all the presets work … but at least I can change the temperature!

DSC_8095

A much closer look at the unit reveals this is a rebadged Atten AT8502D Multi-Function Rework Station. Interestingly, the power is rated as power consumption, and the claims for the rework station is only 550W consumption for hot air and 50W consumption for soldering iron. The numbers don’t even add up. Tenma seem to claim 800W hot air + 40W diaphragm pump + 50W soldering iron, which is 890W despite their claim of 900W. Eh, it’s all marketing …

That being said, you still get more – you get the other stuff as promised in the bundle.

DSC_8096

The desoldering gun is of decent quality, although the knob has a tendency to ride out of the rails, easily fixed with a pair of pliers. The tweezers, are like regular tweezers, but have a flat-rectangular shaped end at an angle, to help you grab onto SMD chips, as well as a enamel-like coating on the handles. The flush side cutters work pretty well – I’m quite pleased with the cut. Best of all, the solder is the Multicore stuff, which is pretty much great quality stuff. The 60/40 “leaded” solder seems to be an odd inclusion for a “lead free” workstation, but it is easier to solder with and probably the best stuff for through-hole users with the 1.2mm diameter. In fact, I’m about 90% way through a 250g reel of the exact same thing.

Does it Work?

In a word, yes. It does work. After a bit of fiddling and sighing at the keypad beeps, it’s actually not a bad station. Soldering with it is plenty easy, although the fine tip makes it suitable really only for PCB level work. While there appears to be other tips for the unit according to the manual, their availability seems to be a problem. The hot air rework gun was a bit trickier to get going – once it was operating, the unit hummed much like an aquarium pump and it was possible to desolder some BGA packages just fine … (yes, I ruined a useless RAM stick …)

DSC_8086

Setting the temperature and air flow is a bit tricky, and so far it’s really been a bit of button mashing until I got the right settings. The unit does detect when the gun is in the cradle, and will continue to blow air until the temperature drops to 100 degrees C and then turn it off into sleep mode.

How long will it last? I have no clue, but we’ll see. At least it lasted a weekend so far …

What’s Inside?

Of course, my next curiosity was just what was running the unit and why it was so comparatively big. Taking off the screws around the top shell allows it to be completely lifted off to reveal the insides.

DSC_8160

There’s a bit of empty space in here. The big unit at the mid right is the diaphragm pump which provides the airflow to the hot air gun. The pump is the black unit at the bottom, which feeds its flow to a white plastic chamber attached to the top with cable ties (likely to smooth out the airflow). The whole pump and chamber is mounted on rubber suspension and is free to move, to absorb vibration, hence the metal guards attached below and to the left to limit the amount of travel.

Power appears to come in from the IEC connector and goes straight to the PCB line conditioning components. Then it passes through the hardware switch back to the PCB, where it is controlled and directed to the transformer (might run the soldering iron) and the airflow pump and heater element of the hot air gun).

DSC_8161

The main PCB is seen here – the power enters on the left side. Interestingly, it’s all controlled by a PIC microcontroller in the centre. The unit does have some puzzling construction choices – middle top, you see two red wires which carry primary power somewhere are supposed to be attached to a connector and are instead soldered to the connector and heatshrinked. That’s not great construction. Elsewhere, the proper connectors seem to be used.

DSC_8162

Closer inspection of the transformer seems to give the impression that the unit does employ components designed for 220v, such as this Chinese transformer. The output is a dual winding of 12v at 0.3A and a 19V at 2A winding … not too sure if it’s running the iron, because if it is, then it’s really only giving out about 38W (RMS of course, although it is 53.7W peak due to the AC waveform). I suspect there might be some lies here.

DSC_8163

The pump itself seems to be marked AP-251 and is for 220v, and probably about 13W as well, but it’s not awfully clear due to obstruction by cable tie. I didn’t want to risk damaging the device getting the cable tie off, so I’ll just leave it there.

Conclusion

When I saw this bundle, I thought it was the bargain of the century. I’ve never seen any hot air rework station at this price – even most decent temperature regulated soldering irons start at this price. As a result, I couldn’t resist jumping in with both feet.

The product itself seems to be of mixed quality – the documentation is poor, the build quality is marginal, getting to grips with the front panel controls and incessant beeping is a pain and the parts seem to be 220v. This might mean a shorter lifetime at 240v, and isn’t really the best especially if you’re going to be using it a lot. The cables are a bit short, and it seems like there might be some cheating with the ratings here and there.

But it is cheap, and when you’re cheap, you expect these things. And at the end of the day, it does work, and it solders decently. I can’t say I’m not impressed by the price, but the jury is still out when it comes to lifetime and durability.

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