E122 Modem (three) will only work with 3G, not HSDPA

lucernarian

Hey there,

I'm trying to fix up a three ireland connection, got under the NBS but right now in use in Dublin. There have been serious speed issues for the past couple of days, and when I checked as to what was going on, the modem was stuck in 3G mode. Signal strength was a constant 2 to 3 bars throughout all tests. I always got the full 384 kbps on testing websites but there were webpage loading issues and other stuff that "3" users are probably familiar with:(

On one occasion, the modem did change to HSDPA and when I managed to run the speedtest for it, I got an easy 3 mbps download speed and about 300 kbps upload speed. The modem was in an upstairs room at this point with 3 bars of signal so maybe that made a difference. The modem dropped down to 3G near the end of the upload part of the test, and has not risen to HSDPA since.

So what's going wrong here? The mast is clearly capable of delivering HSDPA speeds when it briefly gave the connection. I tried with a Windows XP and Vista laptop, and the behaviour was replicated in both.

If anyone can give me any pointers on how to fix this, that would be great!

Thanks!:D

watty

Capacity.

As signal drops or number of users increases the signal isn't good enough or the capacity isn't enough. A sufficient number of phone calls can cause this. You can easily have 0.12Mbps and be the ONLY user on a Mast near edge of cell, or get 0.05Mbps close to mast due to number of users (phone & Data).

It's a Mobile phone system. Not Wireless Broadband.

lucernarian

I know that 3 are hardly the best in terms of contention, but when I managed HSDPA for a brief period of time, speeds were ten times faster. If there were insufficient transmission codes available for the unit to connect at HSDPA speeds, like when too many users are connected to a mast, I woudn't have managed 3 mbps download speeds.

I've come across other UMTS modems before, and they all worked at HSDPA when in use and dropped to 3G when transfer speeds were low.

That's why I don't think this is a generic capacity/contention issue. Why else would I get full transfer speeds on 3G and (for that brief time) HSDPA? I have been around here long enough not to expect much from three Ireland

Does anyone know where I could find some sort of updated connection software for the E122? I've had no joy finding anything on the internet. The 3connect software is 2.0.0.27. I'm also going to try an unlocked phone to connect with instead, and I'll report on how that goes.

jor el

The software won't help. If you can't connect at HSDPA, then the modem will fall back to 3G. Most software have an option to force the connection to 3G or 2G, but you cannot make it give you a HSDPA connection if it's not available.

Capacity changes all the time, this could be why you did get a HSDPA connection at one point, but then it went away.

My guess would be that you are at the edge of the cell, and your connection is going to bounce up and down as the cell breaths and you get a better and poorer signal.

jay93

id say what was happening is that the network was over congested at the times you where trying to get online!!this happens me the odd time at peak times on o2 E1752 modem it locks onto WCDMA (3G) then at later times i will get HSPA singal again :D

lucernarian

These tests were carried out at 1am last night, and at 10 am again this morning with the same results.

I tried a HSDPA-enabled phone with the same outcome, so that certainly ruled out the modem software. The phone was somewhat more helpful in stating when I was using 3G or HSDPA. It changed to HSDPA every 30 seconds or so, and then went back to 3G after only a second. It changed to HSDPA more often when I had full signal. Transferring data over the phone's internet connection did not make it change to HSDPA any more often.

As for the capacity issue, a mast cannot go from having only one HSDPA code available (aka plain 3G) to having 5 or more codes available in the space of a second. That is simply impossible. And there wouldn't be 3 mbps download speeds possible without at least 5 codes being employed on the UMTS transmission level. It couldn't be contention between the mast and customers. And this is ignoring how I tried at quiet times of the day, and outdoors too which gave me full signal on the phone and 4 bars on the dongle. No difference when I tried this morning.

watty

Unless you have diagnostics SW you have no real idea how many codes are in use. You can only guess.
With ALL codes and no other users but poor signal (Range or Interference): Min Speed is maybe 1.8Mbps

With one Code and best signal (no interference) and no other users best speed might be near 3Mbps. Possibly 2Mbps.

Speed tests are not accurate.

Downloads are often using "on the fly" compression. Occasionally I get downloads at apparently twice my package speed because the download speed of browser is not the link speed but bytes written to disk.

Phone and Modem "signal bars" are only indicative. You have no idea what the phone/modem software is measuring.

Also all the operators only have 3 x 5MHz channels. For best cell isolation you need 9 x 5MHz channels. There should be only ONE mobile infrastructure not four.

A cell usually has 3 sectors that overlap slightly like petals. In an ideal world with flat terrain and even population distribution, the cells would be laid out in a honeycomb "hex" structure so that each cell is "rotated" compared to neighbour so adjacent sectors overlapping on nieghbour cells are a different channel and thus can use the same codes.

But masts are where you can get them. Some are on higher ground etc, so some sectors on separate masts using same channel unfortuately overlap. This can lead to very rapid changes in data speed as phone calls are made or phones poll for email and users load web pages (all bursts).

Some places the same channel from three masts might overlap. That's why recommended channels (N) is 9 not 3.

But the mistaken belief is that competition is always best for Consumer. In this case it isn't. The 3G spectrum efficiency would be much higher (capacity, speed and lack of interference) if there was a single regulated infrastructure wholesaler with all the masts and all the operators were MVNOs like Tesco and Post Office. You could even have a model where the operators all have 10% to 30% share each of ownership. Or it could be an independent company or a State body (given how we do stuff and regulate, the 2nd and 3rd option might be poor).





There are various things:

* Voice traffic
* Signal level Eb/No (SNR)
* Codes
* Number of users
* scheduling

Your view taking into account just codes is simplistic


This one compares W-CDMA/HSPA with Fixed Wireless that is NOT CDMA based


It excludes effect of phone users
(both from http://www.radioway.info/comparewireless/)

Have a read of this one too (see 3rd figure)
http://www.techtir.ie/comms/fixed-wireless-broadband-better

General background on 3 and Mobile
http://www.wattystuff.net/issues/mobile-and-nbs/
http://www.wattystuff.net/issues/issues/mobile-and-nbs/is-mobile-in-ireland-destroying-infrastructure/

lucernarian

watty wrote:

Unless you have diagnostics SW you have no real idea how many codes are in use. You can only guess.
With ALL codes and no other users but poor signal (Range or Interference): Min Speed is maybe 1.8Mbps

Indeed I can only guess, but as based on the table you showed above, it's extremely unlikely that less than 5 codes were used given that I was not working in ideal radio conditions like in the 500 metres range scenario referred to above.

watty wrote:

Speed tests are not accurate.

Could you elaborate on this for me please? I tried several methods to verify speed differences between 3G and HSDPA. If a speed test gives a 7x faster result with HSDPA, and loading a couple of webpages was much faster with HSDPA also, surely the speedtest is hardly plucked from air? I understand there are theoretical differences between speed tests and real world usage, but if this is backed up by (briefly) faster browsing on a page with very mixed content, I see no reason to discredit a speedtest. The speedtest data was also fully backed up by the usage meter of the 3connect software which gives a crude speed reading also. I don't think the 3connect software would be aware of the apparent speed that a browser is transferring data with, I would imagine it's simply acting as a measure of TCP/IP througput. In any case, are three.ie known to be using compression routines and if so how do they work specifically for three Ireland?

watty wrote:

Phone and Modem "signal bars" are only indicative. You have no idea what the phone/modem software is measuring.

That's all well and good, but I fail to see how pointing out relative signal strengths is invalid. To say they give "no idea" of signal strengths is somewhat far-fetched. You know well that a phone needn't be giving out dBm readings to show that you have poor signal quality and with W-CDMA signal quality is substantially expressed through cell breathing and capacity constraints, once signal level is reasonalbe to begin with. My experience with dongles is that 2 bars is enough for adequate connectivity. I don't see why that threshold would vary significantly between various Huawei dongles. If I was able to obtain high speeds while working on HSDPA, it indicates that capacity was not a primary issue and that signal quality must have been at least reasonable.


I'm only trying to figure out why using three would peg me at plain 3G, and instead I'm being given general articles on 3G, which is not helping in any way to deal with the problem? I wouldn't mind, but it's not like I'm a total newbie to this and I wouldn't have come here if I had any of the usual reasons for thinking it was a capacity issue.

From looking at other threads, there seems to be no answer given to problems except "accept it, there's nothing to it except UMTS is an inadequate technology for internet access, blame the govt for picking them for NBS etc". If people had a choice, I know they would not use three in particular for mobile "broadband". They don't need to be told how terrible they are...