External antenna positioning for best results

So glad I've found this forum - we're in the UK and I've been searching for 3G forums for ages with knowledgeable people to try to help me get the best from what we have. I hope I might be able to get a few tips here:

We live in a rural part of the UK. We can get ADSL2+ but it's very slow (2Mbps) because the phone lines are poor quality. So we've been using 3G for years. Speeds are much better, but it's all a bit "hit and miss" and I was looking for a tool which would enable me to see the cell information and signal strengths.

I downloaded the tool pinned in this forum and it's brilliant.

Current setup is an Edimax 3G-6200n router with an E367 modem on the 3 network.

Using the Ofcom sitefinder database I've been able to locate 7 cells on the 3 network which are nearby.

Two of them are about 2.5km away both in rural areas. Base station #1 and #2. Two more are also rural but further away and the others are in the nearby town (about 5km away).

I have an antenna on the roof - this one - which, given that I wasn't able to measure signal strength before, is pointed roughly at base station #1.

Plugging that into the E367 does improve signal strength and usually performs like this:



Unplugging that and putting the E367 out of the window and rotating it gives a variety of different speeds. Now that I have the app helpfully provided on here, I can see why: even the slightest movement of the modem causes it fo flick from one base station to another - I can see the ID changing. And different cells seem to have different capabilities and probably different levels of congestion.

With exact positioning, I can get this:



.. and sometimes this...



We live at the bottom of a very slight "dip" - not hills, as such but the landscape sweeps upwards away from us. With the roof mounted antenna I'd say we probably have near line of sight to #1. But #2 provides the best speeds.

Now I'm getting the guy back round to remount it higher up and pointing at base station #2 which seems to be the best one. I did investigate if it's possible to have some kind of motor arm attached to it so I could control it from here, but apparently not. So I need to get the position just right and will work with him testing signal strength with the app.

If you have read all that, thanks Any tips at this point?

Should the antenna point out at a right angle, or be angled up/down?

If I get the positioning just right, will it naturally latch to the cell that I want and stick there?

The adapter which converts the SMA to CRC9 socket..





.. doesn't really stick in the modem very well. It does go in, and you can see the performance difference as measured, but it just falls out again easily. It has to be stuck on with sellotape. Is that normal?

Thanks very much - for the forum, the application, and any advice in advance.

Hi and welcome .
That antennae looks like a log periodic (http://en.wikipedia.org/wiki/Log-periodic_antenna)
It should be pointed directly at the transmitting antenna for best results.
Your speeds there seem quite decent compared to what we get here although some places we get decent 3G.
A dedicated antennae cut for the exact frequency that you want to receive and send will give you a better result than a log periodic antennae .
The adapter i cant see on those pictures,but if it dont fit properly you will get losses to some degree.
What frequency is your 3G there ??

Greetings from South Africa - Some general thoughts that may be useful. Your antenna is highly directional so it should be correctly lined up both in azimuth and in elevation - being in a dip you would near to have it raised a suitable amount above the horizontal. Your adapter is incorrect if it needs to stuck in place - replace it with a correct one. I do not have enough understanding of cell hand-off in detail but suspect that whenever your chosen tower gets too busy it will try and palm you off to another one! Hope this helps

You are doing well. The weakest point is your pigtail connector. It should fit firm, but not to tight, it would damage modem socket. Check maybe you have already lost outer shield in the modem socket, it is why is so lose. I guess, it previously was tight, and now loose? Right?
Inspect this socket with magnifying glass. I can't see on the picture, it is out of focus.
EDIT: If the outer metal barrel is missed, it probably sticks inside cable connector.

DTMark said:

Should the antenna point out at a right angle, or be angled up/down?

Click to expand...

I normally mount the antenna at a right angle, or tilt slightly up if the tower I'm aiming at is higher up.

DTMark said:

If I get the positioning just right, will it naturally latch to the cell that I want and stick there?

Click to expand...

Yes.

DTMark said:

The adapter which converts the SMA to CRC9 socket..

.. doesn't really stick in the modem very well. It does go in, and you can see the performance difference as measured, but it just falls out again easily. It has to be stuck on with sellotape. Is that normal?

Click to expand...

I think you should try and get something like this pigtail:
http://www.poyntingdirect.co.za/pCAB-038/Adapter-Cable-for-HUAWEI-3G-Card.aspx
The connector clicks into place and doesn't tend to fall out. Solwise seem to be reselling Poynting equipment, perhaps they can order a CRC-9 pigtail for you?

You mention you want to mount the antenna higher. Make sure first the higher position gives a much better signal. Every meter of cable is roughly 1 dB of signal loss, if not more. So mounting higher could end up giving you a weaker signal.

Thanks so much for all the (very quick!) replies.

A bit more detail:

Current antenna fitting:



I printed out a map of the area (the Sitefinder database maps can't be printed easily) and noted all the cell positions:



.. and then, the data available for each one:



This shows the frequencies, all 2100Mhz.

Just to clarify/confuse, actually, the antenna is currently pointing at #2 "Bentley" and the best speeds seem to be with #1 "Peakham Copse". (I quoted the numbers the other way around before)

I'm making an assumption - but one which tallies with the performance - that when it hits the cells in the nearby town of Alton (lower left of the map) they may have better capacity, but also, more congestion. There doesn't seem to be any one of those cells that the E367 alone "naturally" favours, with the antenna connected, it favours #2.

At least, I think so. Because the Site ID given by the app is presumably what the base station "broadcasts" to identify itself with e.g. cell #2 Operator Site Ref "GU0427" is the operator's site 52537 which the app quotes, whereas the data from the Sitefinder database uses "global" IDs (the GUxxx ones) across all operators. GU is the postcode area.

In other words, I speculate based on the evidence and direction that Site ID 920 as reported by the app, is cell #1 "GU0350", the one which works best. It isn't visible from here because of the way the landscape slopes up away from us then down the other side. Mind you it might be just about visible to the guy on the roof if I supply him with binoculars

Finally, as regards the CRC9 fitment, sorry about the photo quality... it's really hard to get a detailed shot of that socket. This is a brief video showing the fitment or lack thereof:

http://s1113.photobucket.com/albums/k510/dtmark/3G HSPA/?action=view&current=IMG_0082.mp4

You can see the "sloping" landscape in the background. The direction you see, is (helpfully) towards cell #2 - it's on the "right side" of the house.

Many thanks all. The detail on this site is exceptional - we don't have anything like this in the UK, while ADSL is widespread, the phone network is very old and ADSL performance is pretty dire once the line is about 3.5km long - ours is 3.6km long - but crucially the phone network is a mix of copper and aluminium and any of that, as our line has, basically kills ADSL as an option so that's very hit and miss as well.

FNfal said:

http://opensignalmaps.com/coverage-maps/UK/

Click to expand...

Now I've begun looking into this, everything makes sense now

Where we are, mobile call quality is appalling on all networks (tried them all). When you look at operators' coverage maps they all say "poor". The little circle representing "poor" is exactly where our house is as it's in a slight dip.

Our mobiles are on O2 which is dreadful but about the best - call quality appalling and drops calls now and again. O2 mobile broadband coverage is also fairly poor. Poor mobile reception is a widespread complaint in the village not least because it's a bit "hilly". That site you linked to does recommend O2 and it's spot on.

3 have more cells, and works OK outdoors, but unlike the MBB antenna the phone can't live outside and our cottage has 1ft thick stone built walls. And because 3 operates on higher frequencies (IIRC) building penetration is poor. So 3 barely works at all for calls here - "No network" downstairs. Likewise if I take the dongle downstairs, no network at all. Bring it upstairs and put it on the window sill and 4Mbps. Put it outside = 6 to 8Mbps. Get the positioning exactly right = 10Mbps+.

Thankfully VOIP over the 3 MBB connection works fine.

Price differences for data are huge. We use about 10GB/mo (MBB is our only/primary BB connection, we also stream TV with it - will even do HDTV) and that's about £30/mo with 3. The same thing on O2 costs £75 and it's not as quick/won't do HDTV.

Line of site with a good Yagi antennae should sort your problem ,that is if there is no congestion on the tower you choose.
The antennae you have may do the job if you get LOS.

OK - an update. I've had the guy back around to adjust the antenna.

Swinging it around as far as it can go in both directions, the best I can get is a strength rating of -71 to the only cell we have line of sight to. The previous best signal was -79 with the antenna.

That was the one we originally took a stab with and pointed it at, the Bentley one, #2. It is actually visible with the naked eye from the roof but binoculars confirmed it (there are a few other objects in the area and it's hard to be certain without that confirmation).

The signal strength was -79 so it's improved to -71, this has brought the ping times down to a consistent 30ms to 40ms (the odd spike up now and again, but the effect in the latency reduction is noticeable) and the speed checks in at a pretty constant 6Mbps down 2Mbps up which is about as good as I think I'll get with that cell. It does go higher, but doesn't go lower. The adjustment wasn't that great, actually - the antenna was pointing just to the left of the tower and now it's pointing at it.

The signal strength was nowhere near as good in any other direction.

Maybe someone could try and help with a few more questions...

q1. The most performant cell - the one I got the 11.5Meg from - isn't in line of sight. The area is too hilly and the hill it's over the other side of, is just too high to see it. The directional antenna doesn't even pick that cell up at all. Yet, the dongle (alone) favours it most of the time. Why would that be? Note: that I don't *know* that is in fact the cell in question. I speculate because it is the next nearest, and, the dongle will still favour that one if you bring it indoors.

q2. The "Site ID" can flick between two "sites" with the directional antenna in the same location. Is this because the "Site ID" is one of the cells on the tower it points at, and that tower can have two or more "Site IDs" (cells) which are uniquely identifiable? Apologies if I got some of the terminology wrong, hope that makes sense.

q3. Is a figure of -71 reasonable for that antenna pointing directly (eye-aligned with binoculars) at the cell about 2.7km away reasonable? (Thanks for the tip on the "pigtail" connector above - I have one on order which may make better contact and affect this)

q4. The highest ever attained speed, around 12Mbps, was with the dongle alone. The dongle never attains the same sort of signal strength by itself. The speed test was repeatable several times in succession, but not, for instance, the next day - more like 9Mbps. Is that likely to simply be contention/lack thereof? The top speed here is 21Mbps but I don't think every tower has 21Mbps capable cell capacity.

Thanks so much, I'd love to "give something back" in return for your help, but, I'm not in the same league as you lot!

Mark

DTMark said:

q1. The most performant cell - the one I got the 11.5Meg from - isn't in line of sight. The area is too hilly and the hill it's over the other side of, is just too high to see it. The directional antenna doesn't even pick that cell up at all. Yet, the dongle (alone) favours it most of the time. Why would that be? Note: that I don't *know* that is in fact the cell in question. I speculate because it is the next nearest, and, the dongle will still favour that one if you bring it indoors.

Click to expand...

You should be able to confirm which cell you are attaching to by the Cell IDs. You have the tower locations, so take a drive with the dongle and a laptop and see which Cell IDs you pick up close to the towers.

DTMark said:

q2. The "Site ID" can flick between two "sites" with the directional antenna in the same location. Is this because the "Site ID" is one of the cells on the tower it points at, and that tower can have two or more "Site IDs" (cells) which are uniquely identifiable? Apologies if I got some of the terminology wrong, hope that makes sense.

Click to expand...

You might be able to tell from the Cell ID whether you are seeing two cells on the same tower. The operators in South Africa normally use the first few digits of the Cell ID to represent the tower and the last digit to represent the particular cell on that tower. e.g. Cell ID 5221 might be the North-facing cell on tower 522, 5222 might be the Southeast-facing cell on the same tower and 5224 might be an additional North-facing cell, perhaps at a different frequency, on the same tower.

Unfortunately, there is no standard, some operators use the first digit for the cell, some use the second-last digit. You can find additional information and a useful program for Symbian here: http://www.afischer-online.de/sos/celltrack/.

DTMark said:

q3. Is a figure of -71 reasonable for that antenna pointing directly (eye-aligned with binoculars) at the cell about 2.7km away reasonable? (Thanks for the tip on the "pigtail" connector above - I have one on order which may make better contact and affect this)

Click to expand...

-71 seems reasonable to me.

I have in the past with a signal of -87 dBm achieved several mbit/sec on a tower 11 km away. It really is difficult to predict what performance you will get from a tower.

But it could be possible that if you have line of sight from a more distant tower, it could give better performance as multipath reflections don't come into play. I really don't have that much knowledge of the modulation techniques and their robustness, but common sense dictates to me that if you get signal arriving at your destination from more than one direction at different points in time, it would degrade throughput. Think of using a bunny ears set top aerial for analog terrestrial television. Ghosting is usually a big problem, unless you're close to the tower.