Hook up PCMCIA card to internal antenna?

[patfla]

Anyone ever thought about how one might go about hooking up a PCMCIA networking adapter to the internal antenna?

pat

[bill bolton]

Anyone ever thought about how one might go about hooking up a PCMCIA networking adapter to the internal antenna?

There's no practical way to do that.

Cheers,

Bill

[cmarti]

What would be the use?
The cards found in the T4x are better that the ones found in most computers and way too much better than pcmcia cards..

[patfla]

The card I'm trying to get to work better (possibly if it were hooked up to the very nice IBM antenna to be found in the LCD) is a WPC54GX4. This is Linksys's SXR400 technology. I also bought the WRT54GX4 router. (you obviously need both to get the full benefit).

This is 802.11n-like based on Airgo's 3rd gen chipset. (which of course lost out in the first 802.11n draft; but then that Draft didn't get the votes it needed).

Here's my situation. I'm at 50 ft on a T43 talking to the WRT54GX4 (with its 3 antennas among other things). Going through a number of walls, including a kitchen wall covered in cabinets and a refrigerator.

The SRX400 technology is rated at 240 Mbps. To get real-world numbers I use iperf: http://dast.nlanr.net/Projects/Iperf/. Run it in server mode on my router-hard-wired machine. Then run it in client mode on my wireless T43.

Three scenarios (using iperf):

1. wrt54gv3 router (my old) router talking to Intel 2915 (with the conditions above). 3-5 Mbps.

2. wrt54gx4 router (new) talking to Intel 2915. 12-18 Mbps. (big diff!).

3. wrt54gx4 talking to wpc54gx4 card in my T43. 22-28 Mpbs.

For 'real-world' #2 is pretty darn good and #3 is blazingly fast (about as fast as my Comcast broadband connection which in theory can go as high as 6 MBps down but has never gotten more than 3.5 MBps down).

#1 and #2 connections are rock-solid - never a drop. #3 (with GX4 on both sides) drops intermittently. But when it does it's a series of drops 5 minutes, say, apart.

I normally stay with scenario #2. (and just put the PCGX4 card away).

My thinking is that if the PCGX4 card had a better antenna, that could make all the difference in connectivity.

One problem though. SRX400 wants adaptive antennas (beam forming in real time [and more I'm sure]). What's in the T43 LCD are probably dipole. Don't know how much adaptive antennas depend on the physical shape (and inner electrical characteristics) of the antenna. Potentially may be quite important (in which case, don't even think about hooking it up to the internal antenna).

I'm of course also looking at how I might open up the card and add a better, external antenna.

But again there's this issue of adapative vs. dipole.

So those are my reasons.

Besides it would be fun.

My understanding is that none of the new Draft-N stuff (maybe one exception), which manufacturers have pushed immediately to market, measure up in either range or speed to SRX400.

Airgo's technology (which didn't make it into 802.11n Draft 1 [but then Draft 1 was voted down]) goes back in a direct lineage to Bell Labs where MIMO was worked on for 10 yrs before there were any viable commercial possibilities. Greg Raleigh was one of the Bell Labs researchers and is now the head of Airgo.

It appears that the 'big boys' in this field (and those who'd see Airgo as a threat are Broadcom, Marvell and Atheros [although Atheros isn't that old either and they've had maybe the best 802.11g technology - it of course took Intel some time to 'get up to speed').

pat

[jaybert]

uhmm...16-18mbps that you are seeing in the 2nd example is WAY more than your cable modem will ever be able to do. 16-18mbps is about 2MB/sec, and I doubt your cable modem can even do 1/3 of that (600kB/sec is pretty [censored] good for a cable modem).

your cable modem can in theory do 6megabits/second and not 6MB/sec. I think the only service that would come even close to this is FIOS, and you would have to have the most expensive package and even with that, it is doubtful you are going to find many servers where you'll be able to take advantage of those speeds


EDIT: unless you have some crazy comcast broadband that I havent heard about....then I apologize and take my comments back

[patfla]

Duh. Yes you're absolutely right.

I've used www.broadbandreports.com/speedtests often enough that I should have remebered better off the top-of-my-hd. (I was at work when I first wrote).

I'm home now; went back to the wired computer; and ran a broadbandreports speedtest.

It lists the speeds as follows:

2553 kbps
319 KB/sec

Well I was totally off base. (and this was even slower than normal; I guess normally I get something like [upper figure] 3500 kbps).

Which even then would be something under a hlf of a MB/sec.

So what then is the need for SRX400? Well I bought it and would like to get it working proplery. That doesn't seem a bad reason to me.

In addition, I not infrequently do large file transfers from my wireless laptop to the wired desktop. For the largest transfers, I've taken to carrying the laptop into the study where the desktop is located. I turn off the wireless on my laptop and plug in the second Ethernet cable that I leave hanging off the router.

pat

[kam_]

If the transfers are that large definately use a cable - maybe upgrade your desktop to gigabit if it doesnt have it already.

The problem with all Wireless today is that the frequency band is VERY VERY narrow. Another router or device (Ghz phone, Ghz video sender) on the same band will cause interferance. They say pick a frequency for your wireless from 1 to 13, at least 5 away from any other router. In reality any other router on ANY frequnecy 1 to 13 will affect your connection. The reason is down to reflections and harmonics. At Ghz frequencies, the signal is never going to be pure, and leakage will be huge.

The further interfering routers are, the less it will affect your connection. However walls etc between you and your router will also decrease the SNR of your own connection.

The reason you're getting drop outs at the higher speed is interferance, and not exactly signal level. Sure a stronger signal (better antenna) will help a bit, but at those speeds it doesnt take much interferance to drop the connection. I've known air conditioning units to cause wireless drops, and they're for sure don't run near Ghz frwquencies!

[patfla]

Thanx kam_, that's useful.

The narrowness of the frequency band. Yes the bands come in 20 MHz chunks. One of the points though of MIMO is that, for the first time, it can use 2 of these (40 MHz). And they don't have to be contiguous. You're in the UK where you have 1-13. In the US we have 1-11 (2 less). And the way it works is that typically the two (20 MHz) bands are very separate. Say 1 and 5. Or 6 and 11.

Interference, yes I've thought about that some but maybe not enough. This is a less dense suburb of San Francisco (in the East Bay). We live in one of the neighborhood/developments. Single family homes typically on 1/4 of an acre. When I search for APs I may sometimes see that of the guy (also a programmer) who lives directly across the street from us. Sometimes and sometimes not. And no other APs.

That's outside. Inside we have 1700 sq ft and, as I said before, I'm about 50 ft from the router.

There is a 2.4 GHz mobile phone in the room where I'm getting the disconnects. At one point, I turned it off (unplugged it) but that seemed to have no effect. Of course there's the main unit in the kitchen broadcasting and so maybe I needed to turn that off as well.

We have a microwave, also on 2.4 GHz I believe. It's used sometimes. But I've seen no correlation between when it's used and when my connections dropped.

Those are the only things in the house that I can think of that are also in the 2.4 GHz band.

Lastly let me repeat a point I made earlier. I have the disconnect problems with using the WPC54GX4 pcmcia card. Which uses the more complex SRX400 technology. And given the card's form factor, the antennas must be tiny. MIMO likes large antennas and widely spaced. I splayed the ones on the router as much as possible.

I have no disconnects whatsoever, at the same distance etc, with the built-in mini-PCI standard G Intel 2915 a/b/g that came with the T43. From the exact same position where the WPC54GX4 frequently disconnects.

One of the features of MIMO is that multi-path, which normally makes reliable connections more difficult, actually becomes an advantage in MIMO (because of DSPs I believe behind the radios - DSPs running very sophisticated algorithms sorting through and adding up [hopefully correctly] the multitude of bouncing waves that arrive at the laptop).

All of which should help range and (I would think) make MIMO more interference tolerant.

pat

[bill bolton]

uhmm...16-18mbps that you are seeing in the 2nd example is WAY more than your cable modem will ever be able to do. 16-18mbps is about 2MB/sec, and I doubt your cable modem can even do 1/3 of that (600kB/sec is pretty [censored] good for a cable modem)

My cable modem, a Motorola SB5101i, can and does do it. It depends where in the world you are and what cable provider you are connected to.

BigPond Broadband Cable

Cheers,

Bill

[kam_]

Thanx kam_, that's useful.
The narrowness of the frequency band. Yes the bands come in 20 MHz chunks. One of the points though of MIMO is that, for the first time, it can use 2 of these (40 MHz). And they don't have to be contiguous. You're in the UK where you have 1-13. In the US we have 1-11 (2 less). And the way it works is that typically the two (20 MHz) bands are very separate. Say 1 and 5. Or 6 and 11.

Yes i'm familiar with it, but i've never designed such a router yet. From what i remenber reading each channel works independantly from the other, and so if one drops you don't loose the connection.
I don't recall which specification this manufacturer uses, but if you're sure the frequencies are distinct, then there is no corection between the two transmitters, and they act as two distinct transmisions.

The idea with the frequency spacing was suposed to be that with 20Mhz gaps in the Ghz range, the next harmonic would end up way outside the frequency range. This means you only need to handle 'cross talk' between two channels close to each other, and interferance caused by reflections.

There is a 2.4 GHz mobile phone in the room where I'm getting the disconnects. At one point, I turned it off (unplugged it) but that seemed to have no effect. Of course there's the main unit in the kitchen broadcasting and so maybe I needed to turn that off as well.

These cordless phones use a beacon interval to poll for new handsets. It may be the beacon thats disconnecting you. I would definatly disconnect both ends of the phone and try it.

We have a microwave, also on 2.4 GHz I believe. It's used sometimes. But I've seen no correlation between when it's used and when my connections dropped.

Microwaves are heavily sheilded and the box creates a quite efective farady shield around the whole thing. I wouldn't expect it to be the cause of the problem.
Those are the only things in the house that I can think of that are also in the 2.4 GHz band.

Lastly let me repeat a point I made earlier. I have the disconnect problems with using the WPC54GX4 pcmcia card. Which uses the more complex SRX400 technology. And given the card's form factor, the antennas must be tiny. MIMO likes large antennas and widely spaced. I splayed the ones on the router as much as possible.

I have no disconnects whatsoever, at the same distance etc, with the built-in mini-PCI standard G Intel 2915 a/b/g that came with the T43. From the exact same position where the WPC54GX4 frequently disconnects.

From what i remember they use antenna diversity to extend the range. This means the router uses widely spaced antenna's but the card can use a single antenna.
I wonder do you have any bluetooth devices around?

One of the features of MIMO is that multi-path, which normally makes reliable connections more difficult, actually becomes an advantage in MIMO (because of DSPs I believe behind the radios - DSPs running very sophisticated algorithms sorting through and adding up [hopefully correctly] the multitude of bouncing waves that arrive at the laptop).

I'd be VERY surprised if a consumer unit can do active filtering for reflections at those frequencies. It's probably running a soft DSP on an FPGA decoding the modulation, and probably not much else.

I guess i could download the spec and double check how it works. I would try to eliminate any interferance first and see what happens.
I had similar problems with my wireless at home, so i brought home a Ghz DSO from work and did a little search! It turned out to be some messy burst coming from my neighbours house.. probably some cheap phone.

[patfla]

Hi kam_

thanx again. Will look over what you've written several times.

Very briefly since I'm in the middle of my morning routine.

• 1. What's a GHz DSO? (almost certainly not something I'm going to buy for myself, but anyway ...)
  2. Here's an on-line chapter from an O'Reilly book on 802.11. The chapter is on 802.11n. It's reasonably technical but may not have the details you need. http://www.oreilly.com/catalog/802dot11 ... r/ch15.pdf

[patfla]

And (my mistake) 20 MHz channel, not band.

The channels subdivide the band.

Helps if one gets one's terminology right.

[kam_]

1) DSO is a digital storage osciliscope. It samples upto 4 Ghz and costs about $60,000

2) i'll see if i get time to read up on this later tonight.

The band is 2.4 Ghz and each channel in the band is 20Mhz wide. You can use this to calculate the theoretical max bandwidth, then multiple that up based on the modulation.