ADSL routers compared

Or, how the TI AR7 chipset isn't a problem. At least, not for me.


I've been on ADSL since 2004. I signed up with Demon Internet for a "wires-only" service, and bought a Netgear DG834G router. This is what is now known as version v1 - i.e. larger case, 15V AC power supply, and the TI AR7 chipset.
This initially gave me 512kb/s service, and then BT introduced the 2Mb/s service - although I only achieved 1Mb/s because my line attenuation (46dB) was too high for BT to attempt 2Mb/s.
Later BT introduced ADSL Max, and I achieved 5Mb/s.

In the summer of 2006 I had a fault on my line - an underground chamber had flooded. When the fault was fixed, my sync speed dropped slightly to 4.5Mb/s.

A few years later I was lent a couple of Speedtouch routers (an old ST546, and a newer TG585) so I thought I'd try them out. These consistently synced at 3.5Mb/s. With these I lost 1Mb/s, and gained a tortuous user interface - so I carried on with my Netgear.

In 2009 Demon upgraded me to ADSL2+, but I didn't see any increase in sync speed. Later in the year I switched ADSL supplier to Be Broadband, and they sent me their Be Box - another TG585. Sync speed was the same - 4.5Mb/s on the Netgear, 3.5Mb/s on the TG585.

In the spring of 2010 my line developed an intermittant fault, and my sync speed was all over the place. I'd heard good things about the Netgear DG834GT router, so I bought an ex-Sky unit from Ebay and flashed it with the DGTeam 1018 firmware. This worked OK, but the sync speed was still extremely variable.
Eventually the line failed completely. Once again the same underground chamber had flooded and needed to be pumped out. With that fixed and my running the ex-Sky DG834GT I had solid ADSL at 4.1Mb/s.


With a stable ADSL speed again, I realised that each time the underground chamber was pumped dry I seemed to lose some ADSL speed. I then remembered that I wasn't running the same router as I'd done before. Time for some tests. These tests were done mainly over a 1 hour period around midday. The results are repeatable.

To recap, I have 3 routers:

The DG834G has a TI AR7 chipset. The TG585 and the DG834GT both have a Broadcom chipset, and we all know that Broadcom is the best chipset...
Be Broadband lets a user select his/her profile - I have mine set to "standard", i.e. a target SNR of 6dB.
My line stats are slightly worse at night - presumably because of radio interference. My line is fed overhead from a pole, and my drop cable isn't a twisted pair.

Router Sync speed Attenuation Noise margin (day) Noise margin (night)
DG834GT 4.1 Mb/s 49.0 dB 7.4 dB 6.4 dB
DG834G 5.0 Mb/s 44.0 dB 6 dB
TG585 4.3 Mb/s 48.0 dB 7.0 dB

Hang on a mo. So the DG834GT, with its Broadcom chipset and reputation for good performance, is the worst performer. And the DG834G, with its TI AR7 chipset which is supposed to be poor, is the best performer.

Now the DGTeam firmware on the DG834GT lets me adjust the target SNR on the router. The table above shows performance "out of the box" with the SNR variable at 100%. If I reduce this to 70% I get:

Router Sync speed Attenuation Noise margin (day) Noise margin (night)
DG834GT 4.5 Mb/s 48.5 dB 6.3 dB 5.1 dB

That gets the DG834GT outperforming the TG585, but it's still some way behind the DG834G. How low can I go with the SNR variable ? Or do I carry on with the older DG834G ?

A few tweaks of the DG834GT SNR variable...

Note that this table duplicates two of the above tests.

SNR Variable Sync speed Attenuation Noise margin (day) Noise margin (night)
100% 4.1 Mb/s 49.0 dB 7.4 dB 6.4 dB
70% 4.5 Mb/s 48.5 dB 6.3 dB 5.1 dB
60% 4.7 Mb/s 48.5 dB 5.3 dB 4.3 dB
50% 4.8 Mb/s 48.5 dB 4.3 dB 3.9 dB

It looks like the returns are starting to diminish, but I'm now getting very close to the DG834G's performance.