Phil's corner

iPlayer streaming test

2009-05-27T09:52:00.000-07:00

How does the BBC's iPlayer get on streaming video on a connection limited to various speeds ? Below are traffic graphs showing how the iPlayer uses bursts of download where possible, but as the available bandwidth reduces it has to take up everything it can.

Graphs are 2M, 1M and 0.5Mbits/s download rate.

BT i-Plate filter

2008-07-06T11:27:00.000-07:00

Samknows reports the introduction of BT's Patent Pending iPlate. This is another take on filtering of the bell wire and works by interposing a plate between the master socket wall plate and detachable faceplate. On this plate is a circuit board which provides a 22 mH choke in series with the bell wire to minimise interference pickup from both wired and plugin extensions.

The iPlate is intended for end user fitting using only a screwdriver, without disconnecting and reconnecting wires. The iPlate plugs into the wall plate and the faceplate into the iPate, the extension wires on the faceplate are passed through a slot in the bottom of the iPlate to save the effort of remaking the connections.

In addition to the series choke on the bell wire both of the signal lines 2 and 5 pass through another twin coil device with a ferrite (?) core. Here's a photo of the device:

The coil device looks like a transformer (to a relative layman) however the DC continuity of both signal lines is maintained so it appears to inductively couple the two lines while at the same time inserting a 4.6 mH inductance into each. The device is marked B302H.

Here are some more general shots of the iPlate

The iPlate is intended for a specific set of ADSL users - those who don't want to get involved in wiring changes and who have a BT NTE5 master socket with a T Telecom or BT Piper logo. The new Openreach logo NTE5 plates (as shown in the photo) already have the bellwire choke built in. Non-NTE5 faceaplates can't accept the iPlate.

An iPlate is probably best suited to users who want to use ADSL on an extension, as it will minimise the impact of extension wiring. It does not replace the microfilters you need on each phone / fax / Sky box.

If you can use the Broadband adjacent to the NTE5 master socket then a full filtered faceplate is a better choice as it removes the need for microfilters and removes the ADSL signal from the extension wiring.

If you have a DIY bent then simply removing the ring / bell wire from terminal 3 of your master socket will achieve what the iPlate sets out to do - isolate the ring wire as a source of interference.

Here's a simple circuit diagram of the iPlate internals :-

If anyone can offer an explanation of the function of the twin coil device please post a comment below. Thanks.

LLU coverage

2008-03-21T15:25:00.000-07:00

One hears a lot about Local Loop Unbundling (LLU) and its contribution to the UK broadband market. It is worth reflecting on how widespread its coverage is :-

With thanks to SamC of samknows.com for the graph (current as of posting date).

A green dot is a telephone exchange with one or more LLU providers. A red dot is an exchange with none ie BT Wholesale only, or no ADSL.

MaxDSL speeds in practice

2007-11-04T16:01:00.000-08:00

When BT brought out MaxDSL "up to 8M" broadband the naysayers and cranks with sand in their vagina blurted on about how nobody would get an 8M connection. We can now see how wrong they were thanks to Entanet's BRAS profile distribution. This reveals that 21.6% of their users have a 7150 profile, the maximum you get with an 8128 connection. a further group 10.4% have a 6500 profile which means the speed is at or above 7392 kbits/s. If we add in the 2.3% with a 7000 profile we can see that one third of Entanet's users are at or above 7392 and over 20% are at the full 8128 speed.

The Enta display is dynamic so here's a snapshot copy for posterity :-

Bandwidth update

2007-09-23T08:34:00.000-07:00

Over at Entanet the BT Central capacity has been sweating this weekend, it was running at capacity from 10am Saturday through 2am Sunday, then the late risers of Sunday morning brought it up to capacity at 11am where it has been sat up to 4.30pm when I'm writing this. A perfect illustration of the demand expanding to meet the supply when there is no constraint - Entanet's 300 GB off-peak allowance is effectively "no limits" as only a few % of their users could ever get 300 GB/month in the available off-peak hours.

Broadband bandwidth

2007-09-14T04:07:00.000-07:00

It is interesting to observe the total load on Entanet's ADSL broadband connections via their on-line monitoring. This also shows the expected maximum rate for users on each of the BT Central connections feeding ADSL traffic into Entanet.

Entanet (via their resellers) offer a range of packages with fixed GB/month limits for example 30 GB/month peak / 300 GB/month off-peak. To get 300 GB/month would require the user to average 2 MBits/s through the off-peak period so in reality this is only available to a very small percentage of subscribers. "Off-peak" starts at 10pm and the surge in load is clear from the graph, usually taking until 2am to return to its pre-10pm level :-
There is clearly a lot of unused capacity in the 0200-0800 window and its a pity the downloaders that fire up at 10pm don't use scheduling software to exploit this slack period. As things stand Entanet use an "anti-loss tool" to throttle the individual connections down to 2 Mbits/s in order to avoid packet loss when the 10pm surge occurs.

Another interesting aspect is the fairly consistent upload (green) trace, running at about one third of the maximum downstream capacity. If this upstream traffic is capacity limited, as seems like with Asymmetrical DSL,a change to faster upstream speeds or to symmetrical technologies like FTTH would see a large increase in demand - possibly exceeding the download demand.

Windows Security - damned if you protect yourself, damned if you don't.

2007-09-08T13:09:00.000-07:00

It seems widely acknowledged that Windows has some design issues and Microsoft has some QA issues that result in a fairly significant stream of security problems and fixes with Windows operating systems. Combine that with its dominant role in the market place and the loathing of Microsoft that can be found in some quarters and the net result is a fairly toxic security environment with legions of spammers, hackers and malware creators doing their best to get into your system.

No problem, you might say, simply get all the Microsoft automatic updates and use good security products and your problems will be over. Yeah, right !

Automatic updates generate an unstable computing environment where things that did work suddenly stop working. A wireless card perhaps, or a software application. Worked yesterday, won't work today. Hmm, I needed that. Oh well, never mind. Perhaps another update will fix it.

In addition to Windows updates the antivirus and security software suites rely on frequent updates, patches and fixes to maintain their defences. This may mean that a dial up user goes on line and half way through downloading his email the antivirus has an update to install so it stops the email scanning to update itself. The result is a dropped POP3 session requiring a restart of the email client and another download of all that stuff you were part way through receiving. Oh well, at least I know my spam has been filtered with the latest virus signatures.

Internet Explorer has stopped working again, and I see the Trend Micro PCCillin icon has vanished from the systray, so an update of something has broke that. Does this mean I have no security now ? I have two other browsers (and another operating system) to go at, so life goes on.

The more I think about this the more I am convinced it can't continue. I cannot live with an Operating System that is dependent on frequent updates to keep itself safe - the result is just too unstable.

Equally I can't be doing with Internet Security products that slow the machine down noticeably and break themselves and other things with monotonous regularity. A customer spent £60-odd quid on Norton 360 and it doesn't auto-update, so it declares itself insecure and spooks the guy - LiveUpdate appears to be broken and there is no fix available. Are Symantec a bunch of total Fuckwits ? selling a security product with a broken updater is perhaps worse than selling a vulnerable operating system that is dependent for protection on security products.

So try buying a PC without Windows - you can forget an HP, Acer, Samsung, Sony or Asus for a start. All of these will ship desktop or laptop retail PCs only with a Windows OS installed and paid for, even if your first act will be to format the HDD and put a Linux CD in the drive. Unbelievable.

Hacking a Livebox, Pt.1

2007-06-23T15:05:00.000-07:00

I acquired a Wanadoo Livebox made by Inventel - model DV4210-WU - and set about trying to get inside its head. My motivation for this was to attempt to change the ring voltage on the phone port for the VoIP service, more on that later. After the inevitable Googling around I concluded that I needed to downgrade to an earlier firmware to ensure the hacks would work - so I went from 5.06.2-uk to v5.04.3-uk using a "repair CD" downloaded from here and followed the instructions from that site. Basically you put Linux commands into the Router name field and a "backdoor" executes those commands and installs a telnet server for you. Much of the original work was done by Andy Potter - respect.

The hardest bit was getting the Livebox to download the file from my laptop, in the end Filezilla FTP server managed it as I could define the default directory that the Anonymous FTP user would use.

Once I had Telnet access I was able to roam around and explore the config files. For my dorthcoming VoIP activities it was interesting to look in /etc/invoip.conf and see a line IFACE=ppp0 #add - this looks like an opportunity to point the thing to use eth0 instead and allow it to connect via another router ( common request from users of the similar BT HomeHub).

To unlock the Livebox and allow it to connect to a different ISP (ie not Wanadoo / Orange) you have to generate an 16 byte key and plug it into a converter on Andy's site to generate the unlock code. The equivalent feature on Scumperson's site generated a wrong code (three characters) but it turned out this also happened on Andy's site when using IE6 on the laptop rather than Opera9 on the desktop !

For the record, the correct answer was "Your unlock key is: wyq0je1wctkbz8j2" from a 16 byte code of "bqjtyewzkw8201jc". With that entered into http://10.0.0.1/brdgoff.html the Livebox was able to connect to Demon happily. http://10.0.0.1/brdg.html confirms the current lock status.

Next I need to research how to stop it "phoning home" and updating itself to a firmware version with no hacking loopholes.

So the story so far is that I have telnet access to a Livebox, can use it to connect to a different ISP, but it isn't configured for VoIP.

New BT faceplate - BFFP

2007-05-07T01:26:00.000-07:00

I previously speculated that BT might abandon the 3rd wire (bell wire) but it appears they are taking a different tack and have come up with a new type of faceplate for the NTE5 master socket. This is the BFFP (Bellwire Filter Front Plate) also marked in some cases as "Induction Face Plate". I believe this may become standard issue shortly.

The BFFP inserts a single component - an inductor or choke - between the ring circuit capacitor in the NTE5 backplane and terminal 3 on the faceplate socket and extension wiring connectors. A simple and elegant solution to reducing interference pickup from the bell wire. An inductor has a low resistance to DC current but an increasing resistance (strictly speaking increasing reactance ) to higher frequencies. This allows it to pass the low frequency ringing voltage but to attenuate high frequency ADSL-type signals or noise. The DC resistance of the inductor measures at about 35 ohms, presumably because it consists of a long length of fine wire wound round a ferrite core.

A nice man sent me some of these early BFFP's to try. I installed two on domestic installations where the ADSL may have been affected by extension wiring. Case A was a new ADSL customer with a wired extension that was not in use at the time. Case B was a customer with problems getting ADSL to work reliably, suffering low speeds and unable to connect their two Sky boxes to the phone line. Case B had several extension legs connected to the NTE5 master and a further extension plugged in, the ADSL modem was at the end of one of the extensions.

Case A.

In this case the ADSL router was plugged into the faceplate, so the BFFP was only needed for the extension wire. Removing the extension would have been better, or to preserve it for future use I could have fitted a full filtered faceplate.

Using a BT Voyager 105 USB modem the test socket in the NTE5 gave a downstream line rate of 4544 kbits/s with an attenuation of 48.5 dB. Restoring the faceplate with its extension wiring gave a line rate of 3520 kbits/s, a loss of 1 Mbits/s. Switching the standard faceplate to the BFFP with the extension connected gave a line rate of 4256 kbits/s - 288k less than the test socket but 736k faster than the standard faceplate - a useful improvement from a single electronic component !

From the bit plots below we can see that the ring wire reduces the number of bits carried in higher frequencies and prevented use of the "scatter" of higher channels that were capable of carrying 2 bits. The BFFP restored the use of many, but not all, of the high frequency channels out to the right of the plot.

Clearly the BFFP makes a substantial improvement, reducing the impact of the extension wiring by over 70%. This may not be enough for a perfectionist, but its a very quick and low cost fix.

Case B.

This was a more complex case, an existing broadband user with their "Home Hub" on an extension but only connecting at 288k downstream, 448k upstream. A reboot lifted this to 1152 kbits/s downstream. The Voyager 105 test modem managed a line rate of 960 kbits/s with a fairly poor looking bin plot (see below). Connecting to the test socket behind the NTE5 upstairs gave a big increase to 2688 kbits/s which was more like the right speed for the location (over 2 miles from the exchange).

With a BFFP fitted and back downstairs on the extension the Voyager 105 gave a downstream line rate of 2560 kbits/s which is only 128k less than the test socket. This was a good outcome as it avoided wiring a long unfiltered extension or moving the customer's router upstairs. After hooking up one of the Sky boxes and retraining the Voyager managed the same 2688k as it had in the test socket.

Connecting the Home Hub back up gave a surprisingly good connection at 448 / 3,392 so the BFFP accounted for a threefold increase in downstream sync speed !

Finally, a quick look inside the BFFP to see how it works. Note that this may be a prototype and the final product may look different, which would be handy as currently they are nearly identical to look at apart from the "bulge" covering the inductor.

FUP - Why all the fuss ?

2007-03-16T10:25:00.000-07:00

Lots of hot air is currently in circulation on the subject of Fair Usage Policies (FUP) on some UK broadband services. Typically services using an FUP are advertised as "unlimited" as they do not have an explicit rigid data quantity limit such as 20 GB. An ISP like Demon with an FUP may take action against subscribers that transfer more than 50 or 60 GB per month.

ISPs use an FUP as one method to limit peak demand on their services. Demon's ADSL was becoming near-useless at peak times (09:00 through 23:30) until they invoked the FUP and restricted some high users to 128k in peak times, thereby reducing the load and improving speeds seen by the vast majority of customers.

Other methods are in use, such as traffic shaping or throttling certain protocols (FTP, NNTP and P2P are common targets), capped accounts which stop working at a pre-defined GB limit, metered accounts charged per GB and so on. With over 100 ISPs to choose from there is plenty of variety and nobody need have a service that works in a way they don't like, if they bother to do some research first.

Why are these methods necessary ? in essence they are there to protect the interests of the majority from the excesses of the few. Residential broadband is a contended or shared service which means the total end user link capacity is way in excess of the capacity further down the line. If a few people set out to use their line flat out, as opposed to the bursty use the system is designed and sized for, then congestion occurs and everyone gets stressed.

As an example, a 622M connection to an ISP can carry at least 30,000 connections however it only needs 100 of them to be 6M connections running full blast and everyone will see a slow down.

Some idealists argue that the ISP should just provide more bandwidth. The problem here is economic, the 622M connection costs £120,000 per month and our end user subscription of £20 contributes less than £10 per month towards it. So logically we are buying about 52 kbits/s on average or if you prefer 6 kbytes/s or about 15 GB per month if used consistently 24/7.

If we want accounts where everyone can pull 50 GB per month then we need to be paying £20/month more under current pricing regimes. Nobody wants to do that, of course, they just want to have their cake and eat it.

So ISPs have to hold the peak time demand at a workable level, and the average demand at something around 15 GB/month or less. Plusnet choose to use heavy traffic management and shaping, some ISPs have GB limits as low as 2 or 5 GB/month, others have FUPs. Plenty of choice is out there in the market.

The advantage of an FUP is that it allows an intelligent being to take a view on the use of a connection. It will be obvious to them that a 24/7 BitTorrent addict trying to download the internet is different to a typical user with variable usage. The baying mob however demand "clarity" of the FUP limits so they can operate at 99.9% of the limit thereby missing the point entirely - to manage the average does not require that you limit everyone to the average, merely that you lop off the top end demand to hold the average at the right level.

Some seeking clarity of a contract term that incorporates leeway for reasonable behaviour is more than likely setting out to be less than reasonable. Nobody worries about the exact trip point of a speed camera unless they are bent on breaking the speed limit and trying to get away with it.

So for the majority of users an FUP is a perfectly reasonable way for an ISP to manage demand. The action level isn't fixed as it depends on the usage patterns of the user population. I believe it is better for most users to have an FUP kicking in at 40, 50 or 60 GB than a rigid cap at 20 GB.

The only people it doesn't suit are the roaming band of heavy users who aren't prepared to pay the economic cost of their habits and move from ISP to ISP as their presence either results in an FUP, traffic management, unusable service or the ISP goes bust. One day they'll figure it out and start paying their way.

Vista sorted

2007-02-14T05:55:00.000-08:00

Have now got my multibooting Vista sorted. Once I got rid of Skype with its 30 kbytes/s upload and download I set about exploring the rules around location of files etc for the booting of XP and Vista in a dual (or more) boot situation.

Turns out that XP's boot.ini, ntldr.exe and ntdetect.com have to be on the active partition which in my case is the Vista one, so I end up with these three files in the same partition as Vista's /boot directory. Once that was in place everything "just worked" with Vista's boot loader told that Vista and XP are on the same partition / drive. Obviously they aren't on the same partition but the entries in boot.ini define that with absolute reference to partitions.

With hindsight I didn't need EasyBCD and I didn't need to tell the Vista bootloader that XP was on a different drive. All that achieved was getting rid on the "ntldr bad or missing" error without actually allowing Vista's boot manager startup routine to boot me into XP.

So, to summarise :-

1. Format the partition you want to use for Vista using a Vista install CD / DVD, don't use 3rd party formatting tools yet as NTFS has changed with Vista.

2. Reboot into XP and make the target Vista partition the Active partition from within XP so that Vista will actually load into it. Vista install can format a partition but not make it active !

3. To dual boot succesfully put ntdetect.com, ntldr.exe and boot.ini from the C:\ directory of the XP installation into the root directory of the Vista partition where the /BOOT folder will also be. This should be the active partition. It may be wise to do this as part of step 2 so that Vista has things in the right place from the word 'go'.

A 3rd party utility like EasyBCD may be required to add XP to the boot menu if it isn't there automatically, but it won't help you with the files in (3.) being in the wrong place.

OSS to the rescue

2007-02-12T09:06:00.000-08:00

EasyBCD restored XP to the Vista boot manager in a way that avoided the "corrupt installation" error, but left me with a situation where selecting XP just takes me back to the boot manager :-(

Stuck the Freespire ISO in the CD drive and rebooted, using Gparted to select either XP or Vista as bootable allowed a reboot into a working version of either, although the "front door" method remains broken. Added Freespire to the 3rd partition and the resulting GRUB menu offers me Freespire, Vista and XP as a triple boot menu - all of which work - RESULT !!! OpenSource rides to the rescue and allows me to boot into XP or Vista via a menu.

After checking out the sound card in XP (updating its drivers to identify exactly which version of Soundblaster Audigy I have) I was able to download 40 MB of Vista beta drivers for the sound card - now we have sound output and the microphone works too.

Lessons learned :-

1. Check out each item of hardware and look for Vista drivers, don't trust the Vista system scanning tools they may give you wrong information about compatibility. Do this while you still have XP and can specifically identify the model and version of each component.
2. Don't use anything other than Vista to format the disk or partition Vista is going on - NTFS has "evolved" under Vista and most if not all of the 3rd party tools are currently broken.
3. If dual booting Vista and XP expect to be told XP is broken when it isn't. Keep the XP Setup CD in its box and fix the boot manager references with EasyBCD
4. Vista isn't any more complete than previous versions of Windows, within minutes I was having to download utilities to perform tasks that other Operating Systems do out of the box.

I'm still in information overload seeking a solution to the circular booting situation with XP off the Vista boot menu.

A little bit of Wow!™

2007-02-11T05:39:00.000-08:00

Thought it was time to look at Vista so ordered the OEM version of Home Premium for £70 off www.ebuyer.com . The Vista compatibility wizard didn't like the RAID controller or the printer at home so decided to dual boot the XP machine in the office as that seemed OK.

DVD duly arrived, partitions setup with PartitionMagic and booted to the DVD. Vista installer saw all the partitions and I asked it to format the target one, which it did. It would then not install as it could find "no available supported hard drives to install Windows Vista on" or words to that effect. Presumably that excludes the one it just formatted :-) Many dual-booters have hit this problem and it is well documented on the web - the partition you intend to use has to be Active in order for Vista to install so I rebooted to XP and set the target partition as active with Control Panel/Administrative Tools/Computer Management/Disk Management. Rebooted to the DVD and started the install.

After installation the machine rebooted straight into Vista. I had been intending to listen to a radio show while doing this so I was somewhat miffed to find I had no sound available. "No sound devices are installed" or words to that effect - Vista doesn't recognise my Sound Blaster Audigy card - so much for "Living your digital lifestyle in full" - more like "in silence" !!!

Creative look to have Beta drivers so I'll have to get back to XP to explore which one of those I need, as the software update utility "does not yet support Vista".

While faffing with Vista I took the opportunity to download a Freespire ISO as the 3rd OS on the machine is currently Linspire. Managed to do this OK but then discovered Vista doesn't know how to burn an ISO to a CD it simply created an iso file on the CD - another coaster !

Fortunately someone is on the ball and Iso Recorder v3 is Vista compatible. CD burned OK :-)

In an attempt to add an XP / Vista boot menu option (which I had expected to appear automagically) I downloaded EasyBCD to tame the all-new Vista bootloader manager, added the XP drive and rebooted.

Got the boot options menu up but it said XP was corrupted, went into Vista, set the XP partition to Active, rebooted straight into XP. I see a pattern emerging !

So Vista gives me no sound and can't burn CD ISOs out of the box. It can partition and format a drive but not set the partition to Active and hence it can't
install into that which it just created - nice, not.

Not a good start. Now in XP trying to find Vista sound card drivers.

Skype stole my PC :-)

2006-11-24T07:35:00.000-08:00

The office PC seemed sluggish, looked at the network stats and it had hundreds of connections. Firewall shows it has the preset maximum of 2000 data streams currently running through it. Skype is the culprit, using 6 kbytes/s up and down for its P2P services :-

Cisco ATA186 SIP & BT Broadband Talk

2006-09-14T14:43:00.000-07:00

The ATA186 works OK with Sipgate, given the vagaries of Sipgate. I found the logging utility useful, 'prserv -t' in a command window gives you a diagnostic chatter from the ATA186 :-

060914|22:42:58 >> SIP << 060914
22:42:59 Refresh DNS for sipgate.co.uk d90a4f17 0 0 0 0 060914
22:42:59 Trying 1st IP ADDR d90a4f17 060914
22:42:59 proxy=sipgate.co.uk:5060 060914
22:42:59 Stun Svr[0]=217.10.79.2:10000 060914
22:42:59 [0]REGISTER Retry 0 060914
22:42:59 [0]Reg Resp 401; Unauthorized 060914
22:42:59 [0]REGISTER Retry 0 060914
22:42:59 [0]Reg Resp 200; OK 060914
22:42:59 [0]Reg OK (270) 060914
22:42:59 NextReg in 206 060914
22:43:20 0:30;0,0,0,0,

(note the typical Sipgate registration failures). I changed the audio mode setting to 0x00050005 in order to get in-band DTMF to use with IVR systems, the default did not send tones out onto the PSTN. An ADSLguide poster had been fishing in his Voyager 10V for the settings for BTBBT and posted them on their VoIP forum.

Telnet into the Voyager 10V (admin/admin) and with the command 'sip show reg' you can see what's going on :-

Connection Info:
-----------------------------------------------------

TEL number : 445601484xxx
Expire Time : 3600
Status : Registration OK
Registrar : btsip.bt.com:5060
Proxy : btsip.bt.com:5060
Service Domain : btsip.bt.com
Outbound Proxy : sip.btsip.bt.net:5060
Enable : Yes

TEL number :
Expire Time : 3600
Status : not ready or no need to Send Registration
Registrar : btsip.bt.com:5060
Proxy : btsip.bt.com:5060
Service Domain : btsip.bt.com
Outbound Proxy : sip.btsip.bt.net:5060
Enable : No

-----------------------------------------------------

So the 10V looks to be 2 line capable. More of that another day..
the 44560 number is the incoming phone number of my BTBBT service and this becomes the UserID UID0 on the ATA186. Plug in the password, the Proxy and Outbound Proxy settings from above and try it.....

Got a dial tone, seemed to register, but can't call out (engaged type tone). Tried dialling into the 0560 number from BT line, silence all round :-(

Changed the RTP / Media port from Sipgate's 5004 to the 49152 revealed by the 10V :
[admin]# sip show rtp_port
SIP : RTP Port Number = 49152

Still no good. Mental note that there is no STUN server configured. Tried Sipura default RTP setting of 16384 - same. Twigged that the 10v was still on line, unplugged it.

Still no outgoing calls, but made an incoming succesfully - voice and all !

Time to go to sleep on a high note, meantime here's the failure log :-

060914|23:37:54 Calling 1571
060914|23:37:54 SCC Cmd[0:0]CLIP
060914|23:37:54 SCC Cmd[0:0]CALL
060914|23:37:54 [0:0]Start RTP Rx
060914|23:37:54 RTP Rx Init: 0, 0
060914|23:37:54 [0]INVITE Retry 0
060914|23:37:55 [0]INVITE Retry 1
060914|23:37:56 [0]INVITE Retry 2
060914|23:37:58 [0]INVITE Retry 3
060914|23:38:00 [0]INVITE Retry 4
060914|23:38:02 [0]INVITE Retry 5
060914|23:38:04 change next hop[0]: c = 2
060914|23:38:04 Trying IP ADDR 51906a28
060914|23:38:04 INVITE TO; TryAltProxy
060914|23:38:04 Response=400; Bad Request
060914|23:38:04 INVITE failed

Cisco ATA186 SIP image.

2006-09-10T01:53:00.000-07:00

BT's Broadband VoIP service BTBBV was originally supplied with Cisco ATA186 ATA hardware using the MGCP profile. In the beginning mine arrived before the service was up and running so I flipped it into a SIP profile and used it with FWD. BT have now moved their rebranded Broadband Talk service onto SIP and sent me a BT Voyager 10V ATA. Time to get the 186 back into service on SIP....

First problem was finding the ATA on the LAN, pressing the red button on the 186 with the connected phone off hook brings up the IVR voice menu and keying 80# has it read out the IP address. I was then able to see the 186 web interface at http://ataip/dev

Next to locate the SIP image and upgrade tool. This is something you can get from your VoIP provider, buy from Cisco or Google for ata186us.exe and use your imagination. ata186us.exe is the loader and the image file is *.zup

Make a note of the current firmware version at the bottom of the web page or use IVR 123# and IVR 123123# - in my case Version: v2.16.1.ms ata18x (Build 030814a)

You then run an upgrade TFTP server on a local PC with a command like :-
sata186us ATA030100SIP040211A.zup -d1 -any2
where any2 means do the upgrade even if its a different version to the old one.

the onscreen instructions tells you to..
"Using dialpad of your telephone (attached to your ATA box),
press ATA button to go to main menu, and enter:

100#10*0*0*23*8000# (to upgrade code)"

there is then a long and agonising pause, followed by the IVR lady announcing "upgrade succesful" - Result !

Back to the web interface to check....
Version: v3.1.0 atasip (Build 040211A)

Off to sipgate to look at the example ATA186 configuration. Loaded up my details then hit a "password change" screen from the ATA. Google reveals the default to be '0' so I change it to a 4 digit PIN (in case at any point I have to put it in the IVR). That works, so now I have to get the settings loaded up and debugged.

Update on phone wiring for ADSL.

2006-06-01T03:34:00.000-07:00

BT have introduced "MaxDSL" services which give "up to 8M" depending on line conditions. These services are "rate adaptive" which means the speed achieved by the modem depends on the line condition - a good short line will achieve 8128 kbit/s downstream and the longest poorest line will either drop to 160 kbits/s or fail to connect. In general this means people are getting faster speeds with Max than before, as the previous limits were conservative.

Getting your wiring in order with Max could give you faster download speeds and higher reliability (less errors), and it is worth doing this before changing to Max so the initial training period reports your line at its best. If your attenuation is 35 dB or less you should get 8M, falling to 2M around 60 dB although some have achieved 4M at this distance.

The best option for wiring is to have a filtered faceplate from Clarity.it or ADSLnation.com fitted on your NTE5 master socket. If you don't want your modem or router next to the master socket you can run an unfiltered extension line from the back of the filtered faceplate.

If you don't have an NTE5 master with removable faceplate it should be possible to get BT to fit one for about £30. Some people have achieved this for free, others have been quoted over £100 - its at the bottom of this page at £30. Probably best to call 150 and say you "want to wire up a permanent extension and you don't have a linebox to connect it to". Don't mention broadband, for simplicity. They won't fit a filtered faceplate, just an NTE5 for you to fit one to.

If you don't get an NTE5 then removing the ring wire / bell wire from terminal 3 of your master socket will get most of the benefit. Since writing the original item I have found a couple of installations where it was necessary to remove all of the ring wires from terminal 3 of all the extension sockets to get the results. Bear this in mind if the first one doesn't do it for you.

Please note that the colour codes on cables are for convenience, the blue and white pair is conventionally used for the phone signal on terminals 2 and 5 but you may find the green/white pair being used or the orange/white pair. Other colour schemes may be encountered, including single (solid) colours. The important thing is that you only need 2 and 5 for the phone to work and that the same colour should be used on the same terminal of each socket.

Finally, remember that removing the ring wire reduces noise on the line but you still have to filter every phone/fax/sky box/alarm/56k modem to avoid them interacting with the ADSL signal. A filtered facaplate removes the need for separate microfilters.

More on phone wiring

2005-08-26T23:21:00.000-07:00

Or should that be Moron phone wiring :-)

After the success with taking the ring wire off and improving the SNR I was blessed with a free speed upgrade to 2M by my ISP Demon which was facilitated by ADSLguide getting my entry changed in BT's line prequalification number checker.

This speed upgrade took the downstream SNR margin down to 25 dB, still healthy but I was pleased I did the ring wire thing to improve it. Last night I hooked the ring wire up again and the ADSL modem lost sync, it re-trained to an SNR of 12 dB and ran OK but clocked up errors. The reason I put it back on was to look at the frequency effects of the interference, using a Sagem F@st 800 modem's diagnostic mode.

The first plot is with the ring wire disconnected, showing a wide range of useable frequencies. The second plot shows the effect of reconnecting the ring wire - it loses virtually everything above frequency band 140 with a consequent loss of ADSL data carrying capacity. The lower frequencies are a lot more ragged too. It would still operate at 2Mbits/s but was clearly less capable than before.

With the higher frequencies rendered unuseable by the ring wire interference, the modem is forced to cram more data into the lower frequency bands. This in turn makes it more sensitive to interference as a higher SNR is required in order to use more bits per channel.

So at least we can see why the ring wire impacts upon the SNR, it is adding interference in the 600 kHz - 1 MHz range. This is AM radio territory, so perhaps this should not be a suprise - for all I know the ring wire may be a nice wideband AM radio antenna, built into my house :-)

For my final experiment I hooked up a filtered socket from ADSL Nation in place of the master. This provides a filtered and unfiltered socket on the front, and an IDC terminal block for filtered extensions on the back. So I used it to provide me with a filtered ring wire that could supply ringing voltage to all the extensions without compromising the ADSL. This works well and maintained the 25 dB SNR and over 10 hours my router only clocked 3 CRC errors.

LocalSNRMargin = 25.0 dB
LocalLineAttn = 32.5 dB
LocalTxPower = 10.5 dB
LocalFastChannelRxRate = 2272000
LocalFastChannelTxRate = 288000
LocalFastChannelFEC = 0
LocalFastChannelCRC = 3
LocalFastChannelHEC = 0
RemoteLineAttn = 22.0 dB
RemoteSNRMargin = 29 dB

The only point in doing the filtered ringwire bit, as opposed to simply disconnecting the ring wire, is to leave the extension wiring fully functional, such that if you aren't using ADSL any more or leave the property the phone sockets will behave as expected. One wonders if at some point BT will abandon the ring (3rd) wire, it may have made sense 20 years ago for electromechanical bells but its a real nuisance with ADSL broadband.

The images below show the "before" bin plot with the original ringwire connected, the filtered ringwire connection and the "after" bin plot using the filtered ringwire.

Finally, just a reminder that the above refers to star or spur wired systems where an NTE5 master socket can't be fitted. The NTE5 filtered faceplate is the preferred solution where it can be used, it incorporates filtered extension wiring and, if needed, an unfiltered extension connection for a modem distant from the NTE5 linebox / master socket. Having said that, one ADSLguide user has reported improved SNR through removing the ring wire from his filtered faceplate - so if you are trying to squeeze the last few kbits/s of speed or are trying to make a dead or unreliable line work it may be worth a try.

ADSL tweaking.

2005-08-14T05:09:00.000-07:00

Most broadband in the UK is ADSL supplied over phone lines. This involves a cunning plan to send frequencies of 100 kHz up to 1 MHz (AM through Short Wave radio) along the line superimposed on the low frequency voice signal. This allows speeds of several MBits/s to be transmitted over twisted pair copper cables only ever intended for analogue voice, quite neat really.

Nowadays ADSL is supplied "wires only" ie BT put the signal on the line and the user gets to make it work. When the twisted pair phone wire arrives in the user's house it is converted by a "master socket" to a 3-wire system where the 3rd wire provides a ring circuit, originally designed to power clunky bells in old steam phones, see this explanation. This 3rd wire is a "bit of an issue" as it makes the nicely symmetrical balanced twisted pair into the equivalent of a 3-legged ballerina.

In my case, the ring wire runs around my house as part of the feed to at least 6 extension phone points, in doing so it acts as a nice big aerial (antenna) to collect any radio transmissions and random electrical noise that may be passing through. It then feeds this mush back into one side of the twisted pair through the capacitor that is there to provide the ring signal and consequently degrades the ADSL signal.

This "degradation" appears as a reduction in the signal to noise ratio (SNR) and as CRC errors counted up by the modem and in some cases causes disconnections or prevents the end user getting the speed their line should be capable of. As we move to faster speeds and to "rate adaption" where the ADSL goes as fast as the line allows it, the dear old ring wire will slow down our connections.

So, what to do with the 3rd wire. The Rolls Royce solution comes in the form of a filtered faceplate (photo on right) which fits onto modern BT NTE5 lineboxes and provides ADSL and voice on separate sockets with filtered extension wiring connectors at the back. This keeps mush from the extension wiring, including the ring wire, out of the ADSL signal. It also avoids needing to use microfilters on each telephone appliance, which is A Good Thing™. Also available from ADSLnation.com

However, many of us don't have a nice NTE5 terminating our incoming phone line.

My phone line comes into a junction box in the roofspace from which the wires run in a "star" or spur configuration, the "master" socket is on one leg of the spur system and is identified by having a ring capacitor and a surge arrestor fitted.

As I have no NTE5 I can't use a filtered faceplate without a re-wire, so a bit of innovation is called for.

The photo on the right shows the back of my master socket with the ring wire disconnected from terminal 3, the other half of the orange/white pair is also disconnected from terminal 4. A higher resolution photo is at the bottom of this article.

By disconnecting the ring wire from the master we remove the 3rd leg and restore the balance / symmetry to the line.

Well, so much for the theory, what difference does disconnecting the ring wire actually make ? The stats from my router below show the results :-

Before, with ring wire (1 hour stats)

LocalSNRMargin = 20.5 dB
LocalLineAttn = 32.5 dB
RemoteLineAttn = 23.5 dB
RemoteSNRMargin = 25 dB
LocalFastChannelCRC = 22
LocalFastChannelHEC = 6

After, ring wire disconnected (20 hours)

LocalSNRMargin = 32.0 dB
LocalLineAttn = 32.5 dB
RemoteLineAttn = 23.5 dB
RemoteSNRMargin = 29 dB
LocalFastChannelCRC = 2
LocalFastChannelHEC = 0

So we see an increase in local SNR margin of 11.5 dB, meaning the router sees 11.5 dB more of the useable signal from the exchange after subtracting the noise. Going the other way the exchange gets to see 4 dB more of useable signal from the router. These figures after rebooting the router so it could adapt to the new line conditions.

20 hours after disconecting the error counter hadn't shifted from its initial value of 2 errors, compared to 28 CRC & HEC errors in just one hour with it connected. Result !!!

The phones still ring because they are using micro-filters and each of these has its own ring capacitor on the phone outlet side of the filter. If you disconnect the ring wire it would be prudent to re-connect it if you move out of the property, so the next occupier's phones work as expected.

Enjoy !

VoIP USA

2005-07-11T20:22:00.000-07:00

I took the covered wagon from Minneapolis up to Fargo, North Dakota (it was a DC9 actually, a bit late and very hot as it was 90 °F and sunny).

After visiting Walmart - 8pm on a Sunday - BestBuy, OfficeMax, OfficeDepot and finally Radio Shack (Tandy as was in the UK) I accumulated enough bits of wire to power and connect my BT Broadband Voice box (Cisco ATA186). I am connecting the laptop wirelessly to the hotel hi-speed internet and a crossover ethernet cable connects to the ATA186. The laptop is on an external IP address and the ATA gets a 192.168.0.x address from the laptop. I took a photo but seem to need yet another bit of wire to get it onto the laptop - there is no GPRS signal for the Treo600 to email me the photo it took. Watch this space. Picture here shows laptop on left connected by crossover cable to ATA in centre with hotel phone on right. Can't get the in-line image upload to work :(

The system works a treat, I accidentally called someone in the UK at 03:30 (woops) but then saved an arm and a leg by collecting my mobile voicemail at BT's offpeak Orange rate of 11.4p/minute rather than making a roaming Orange call at £1/min or whatever.

This was my first "hotspot" experiment - how to hook a dumb ATA up to a service that requires authentication via a web page. The laptop handled the authentication and hid the ATA from the hotspot - but how would you do this with an 802.11b SIP phone ??

I phoned across town via the UK to my friend in Fargo who has cable internet and a Vonage VoIP service (2 lines) with no analogue phoneline. There was a marginal delay but not the sort that makes conversation difficult. His ATA is the Linksys PAP2 but locked down to the Vonage service.

In Best Buy there were several routers or ATAs on offer with bundled VoIP service. Generally you get a mail-in rebate or account credit equal to the hardware cost when you sign up for the service. The kit included ATAs and routers with VoIP phone ports in both wired and wireless flavours.

Clearly the market focus is on getting cable internet users (the vast majority) to use VoIP in preference to analogue PSTN phone lines. The monthly cost is much lower and a bundle of services like voicemail are included for free.

From South Witham to Minneapolis airport

2005-07-10T14:03:00.000-07:00

This comes to you via the wireless internet access at MSP airport, a pay-to-use service which I access via IPASS roaming thanks to a prepay facility run by the folks at http://www.roamintl.com/

Yesterday's VoIP-a-thon at S Witham was interesting. A collection of many different ATAs, IP phones and Skype over bluetooth headsets. Basically it all "just worked" via Tom's wireless network, which runs on beer and tobacco. http://www.wireless.southwitham.net/

My PAP2 worked good on two accounts, also the Cisco BTBBV box fired up nicely on the SWBB net once their MAC codes were granted access by Tom. NO port forwarding or firewall fiddling required. We also dipped our toes in the confusopoly of charging rates to different types of number / different countries.

Best go now - off to queue behind a "line" of Americans taking their shoes off to go through a metal detector (?).

Linksys PAP2 - out and about.

2005-07-04T16:47:00.000-07:00

I thought I would try to see how little configuration I could get away with. A couple of posts by others suggested port forwarding to the PAP2 wasn't necessary and indeed after turning it off in my router (Belkin wireless ADSL jobbie) I can confirm this. Equally I don't have any ports for RTP streams open, and no STUN server settings.

I rebooted the beast this morning and it still worked, then rebooted the router too and that didn't affect it. Then I took it to the office and hooked it up behind an Intertex IX66 router without doing any port openings or forwarding - and it worked there too ! Granted some ports may have been open from previous VoIP efforts. So pretty robust and portable really.

All in all a success. The office DECT phone (BT Diverse 5310 - Siemens) would ring on incoming calls without the microfilter, it also showed the caller identity so CLID works fine too both in and out.

One thing that can catch you out is the lack of refresh on the Info page of the web interface - clicking on Info doesn't refresh it, it needs a browser refresh F5 to bring it up to date and show the current time and call / line status.

Its a bit clunky having a local 01780 incoming number as you can dial the VoIP number from BT locally without the 01780 code but to dial out over VoIP always needs the full dial code and number. I think a tweaking of the Dial Plan could fix this - insert 01780 in front of any number dialled that starts with a 7. Something else to learn how to do.

For VoIP to work in consumer land I think the ATA needs to be pre-configured to the account and tested before shipping. NAT routers could cause enough headaches without the poor user having to learn to setup the ATA.

I can see why people are shipping ADSL modem/routers with VoIP functions built in - better Quality of Service and no NAT workaround issues. Sipgate are selling a router / ATA from AVM for £120 that provides two analogue phone VoIP ports as well as the usual ADSL modem/router functions. Appears to have some integration of the analogue PSTN phone line too.

Linksys PAP2-NA - Day 2

2005-07-03T15:54:00.000-07:00

After inspecting the firewall logs I could see incoming traffic bouncing off the firewall that was from sipgate IP addresses and on SIP ports, so I put the Linksys into the DMZ on the router exposing it to the ills of the internet.

Calling it from the landline I got ringing tone but the DECT phone on the Linksys didn't ring, and there was no flashing LEDs etc to indicate activity. On inspecting the Info page I could however see that it was ringing. Inserting an ADSL microfilter in between the DECT phone and the PAP2 brought it to life - clearly the output needs an RJ11-BT adaptor that includes a ringing capacitor (provided in my case by the microfilter). So it is actually working with Sipgate on Line 1 while in the DMZ !!

Without the ring capacitor you can answer the call if you know its ringing, but that is only evident from the web interface :-(

Reading up about the Linksys PAP2-NA I discover that it is allegedly the same unit as a SIPURA SPA2000

The forums at http://voxilla.com/PNphpBB2.html were very helpful, and include configuration tools for common ATAs and poplar VoIP service providers, very useful !

I turned off the DMZ on the router and it still works. Not sure it will survive its hourly re-registration though. I have port forwarding of 5060 and 5061 to the Linksys but no other router tweaks. May need to test its resilience.

Inspired by getting the first line to work I set about connecting the second up to beta VoIP service from 18866.co.uk - this sat saying "can't connect to registration server" and declined to do anything - no blue LED :-(

After messing around with Line 1 for a while I looked at the settings in Line 2 and saw that they were corrupted - what should have been "sip.call18866.co.uk" in the Proxy field had morphed into something like "985@-16,1428@-16,1777" which is a tone definition or dial plan config - so the moral here is to check what is actually saved when you hit "Save" by revisiting the same page.

Corrected the settings and Line 2 leapt into life, made a call out and all worked well. I tested it by ringing my answering machine, to ensure 2-way audio was present.

Cool - 4 blue LEDs all shining nicely at me. Lets see if it lasts. Still not in the DMZ.

I found a load of NAT and STUN settings at the bottom of the SIP page when viewed in Advanced mode as Admin. Currently they are blank however the voxilla config wizard gives the settings for sipgate. I'm never quite happy with these 2-line devices having one set of parameters for something, nagging doubts that it'll work for both providers. For reference I have Firmware Version: 2.0.12(LS)

Off to buy some cake to celebrate.

Linksys PAP-2 experiences

2005-07-02T11:16:00.000-07:00

An "as it happened" walkthrough of setting up a Linkysys PAP-2 :-

The Linksys PAP2 is an analogue telephone adapter (ATA) to allow a normal phone to make Voice over IP (VoIP) calls. It has an ethernet socket to connect to a broadband router and two RJ11 phone sockets. Wall mounting screw holes are provided, as is a plastic foot to allow the unit to sit vertically on a surface. At 4 inches (100mm) square by an inch (30mm) thick it isn't large.

I bought mine from http://www.broadbandbuyer.co.uk/ for just over £40 inc VAT and p&p. The price was the main reason for selecting it over the Sipura models.

The box says model PAP2-UK but the device PAP2-NA, as the quick installation sheet says to dial "1 + area code + number" it would appear to be a USA model localised for the UK by adding a clip-on 13A plug for the power supply and a few sheets of CE compliance information.

On plugging in to the LAN and powering up I had one intense blue LED pulsing away with the letter D in morse ie -.. -.. this is the power LED. Not quite the "all LEDs will be solidly lit" of the documentation. I had to find an RJ11 to BT adapter so I could plug in the DECT phone I planned to use.

After a bit of thought I realised that the LAN LED was not lit and nor was the other end on the router, so tried a different cable and got the LAN LED up and the power LED stopped flashing - one step forward ! Reverted to original cable after a bit of wiggling plug in socket.

I see from the router that I can ping the PAP2 on the IP address it collected by DHCP. Popping the relevant IP address into Firefox I get to the Linksys PAP2 configuration page. Under Line 1 it correctly shows the Hook State: as "On" and it toggles in response to me keying the DECT phone. So far so good. There is however no dial tone, so I will RTFM and set it up as guided.......

First tried the phone keypad tone IVR menu - punch **** into the phone keypad and it responds verbally, hang up to exit. So I know the phone connection works, and where to go if I needed to set up the IP address manually. 110# discloses the current IP address, 100# says if DHCP is enabled or not. 732668# (reboot#) and 73738# (reset# - a factory reset) might be handy later.

Back to the web interface - this has Basic View and Advanced View modes, and runs in Admin and User modes (the upper right of the screen says User Login when in Admin mode, and vice versa). Default is to have no passwords, though the CD manual refers to setting passwords via the IVR system. This looks a bit clunky - "To enter A, B, C, a, b, or c — press 2" - however its actually just using a PIN as to put in Phil you would enter 7445, its not like texting and pressing the 2 key 5 times to get a 'b'.

Clearly the assumption in the manual is that the unit is shipped pre-configured with a voice service, as the last chapter of the manual concludes
"Configuring the Settings for Your Internet Phone Service
If you want to change the settings for your Internet phone service, visit the website of your Internet phone service provider and make configuration changes online. For more information, refer to the instructions provided by your Internet phone service provider."

The IVR menu command 7932# is to enable/disable the web interface, this is where a password would be required if a service provider had shipped the unit locked down to their service.

So, another cup of coffee then let's try to figure out the setup :-)

In the User mode, Basic or Advanced view, there are 4 screens on the web interface titled Info System User1 User2. Info is a Status screen, System covers TCP/IP network settings and password and the User screens are full of Speed dial, Suplementary services, Distinctive ring and othe ruser features.

Change to Admin mode and extra screens become available - SIP, Regional, Line 1 and Line 2. Sounds like I'm going the right way. In the Advanced view version a further screen - Provisioning - appears. Checking back at some of the other screens I can see that Advanced really is when it comes to things like defining dial tones and codec pre-selection strings. Let's stick to basic for now....

Under Regional I select GMT +1 for the timezone and put in the date in mm/dd/yyyy format and the time in hh/mm. The screen actually specifices mm/dd but that left me in 2003 ! There's a drop down box for setting FXS impedance - this is set at 600 (ohms ?) but the dropdown box has a myriad choice of ohms + capacitance values. If I find out what the UK phones expect I can change it later.

Line 1 looks to be the place for setting up the account details. I'll start with a Sipgate account from www.sipgate.co.uk as this gives me a free regional incoming number that matches my local dial code 01780. The PAP-2 isn't listed on Sipgate's help pages, so I look at the "other devices" page, then revert to the X-ten page where you can get a personalised setup screen pre-filled with your details (bit of a crutch for a newbie). Taking the minimalist approach I entered :-

Proxy: sipgate.co.uk
Display name : Phil Thompson
User ID : 140xxxx
Password : *****

the default SIP port (5060) and registration time (3600s) seemed to match sipgate's needs, so I hit "Save settings", get a "please wait page - bit of red and blue flickering from the power LED and, lo, the Line 1 blue LED is lit :-) and the Info page shows Registration: Online under Line 1 Status. Flip the DECT phone open, punch the green button and YES ! we have a dial tone.

The sipgate account page shows me online, so I dial up from the BT landline and the X-ten softphone on my PC leaps into action. Oh pooh. Two things registered on the same account.

Shut down X-lite, redial, "the service cannot be connected" message from BT. Let's try outbound - dial the sipgate 10000 test number and yes, a German guy talks back at me. So I have an outbound phone service but not incoming - sounds like a job for "Router Man".

The only port forwarding to the desktop is port 4569 which is IAX2 protocol, but I do know the X-lite was setup to use STUN in some way.

After hacking around I can get silence from BT when I try to call myself, but nothing else happens. Try again in the morning.

EU sugar regime

2005-06-23T14:41:00.000-07:00

Well the EU has managed to upset everyone with its proposed reforms of the sugar regime. It has to do something as the exports currently involve a subsidy that exceeds WTO commitments. Too much money is being spend subsidising export of surplus sugar.

The chosen path is a dramatic 40% price cut, which is expected to put several EU countries out of the sugar business. Unfortunately it will also impact heavily on developing countries that have preferential access to the premium price EU market - either former colonies in the ACP (African Caribbean Pacific) group or the Less Developed Countries (LDC's) who were progressively gaining access to EU markets for everything but arms (EBA) exports including sugar.

So the poor folks in Guyana, Barbados, Swaziland, Fiji, Mozambique et al will get to share the pain with the EU farmers and processors. I guess they're just collateral damage from Oxfam's intensive lobbying efforts, so I hope they know who to blame. Similarly if I were a beet farmer with a trailer load of beet it would be tempting to clamp it in the doorway of an Oxfam shop :-)

The price cut is of course just one way to achieve the WTO obligations. Oxfam is now in a tyre smoking U-turn crying out for prices to be maintained so that its customers in the developing countries keep their access to the premium price market. Well tough shit guys, you should have thought of that when you were mouthing off about the price so much. Put "EU sugar price" into Google and see who's to blame for raising the profile of the price issue.

Something we haven't heard much about is the detail of the subsidised exports. The EU regime is complex, but doesn't involve handing wads of cash to beet growers or to beet processors, so its not a direct subsidy. Instead its a rigged market with a high price for sugar and for sugar beet, so the farmers and processors get their rewards from selling at artificially high prices.

To keep a lid on this there are production quotas, limiting how much can be sold within the EU as these prices. Surplus sugar produced outside of quota is sold at the world market price, quota sugar that is surplus to consumption requirements is also sold onto the world market but in this case the seller gets an export subsidy funded by a levy on all the quota sugar beet grown and produced. The idea is that the levy pays for all the subsidy and its self financing.

In addition to this mechanism the EU imports over 1m tonnes of raw cane sugar and some of this is also surplus to consumption requirements and so gets exported. The export subsidy on this is paid by the EU budget out of taxpayers money. Tate & Lyle received £100m in one year from the EU for subsidised exports of surplus cane white sugar.

So part of the current problem is not just the price but the surplus of quotas over demand. The French, to pick one example, have a beet sugar production quota that exceeds their national consumption by 1.7 million tonnes. The UK by comparison produces about half its needs from beet. So the UK beet farmers & British Sugar have traditionally paid for the export subsidy on French surpluses without getting any export subsidy themselves, but nevertheless enjoying the fruits of the high internal prices.

So the French sit there with a production quota of 3.8 million tonnes, hoping we won't notice, waiting for Ireland's sugar industry to go bust so that its 200,000t quota can be grabbed by the French as an outlet for its huge excess of production over consumption.

Germany, The Netherlands and Belgium are other countries with ample surpluses of production over consumption - total 1.2m tonnes - that get exported with a subsidy. Strange isn't it, another Old Europe vs new Europe situation. Rather than addressing the institutionalised excesses of Old Europe the EU come up with a severe price cut that will hit many coutries in the EU hard, including the new entrant Eastern countries and as a side effect dump on our overseas friends struggling to make ends meet.