Dell 1500 WLAN - Broadcom 94321MC WLAN review

[mobilenvidia]

Model as tested, Dell Wireless 1500 WLAN in mini PCI-Express format, RevA00

This was bought of Ebay to try out Draft-n.

To test the capabilities I use a Buffalo WZR-AG300NH router.

Broadcasts in both A and B/G channels up to 300Mbps

Has Gigabit LAN so no bottle neck that 10/100 would cause.

BCM94321MC 2.4/5-GHz Intensi-fi draft-802.11n PCI Express Mini Card reference design

? First IEEE 802.11n draft-compliant Wi-Fi® solution

? BCM4321?MAC/baseband controller for draft-802.11n and

802.11a/b/g devices (PCI/PCI Express®/CardBus)

? BCM2055?2x2 dual-band single-chip radio?3x3 and 4x4

antenna options for increased range and performance

? All CMOS implementation of the world's most advanced

system architecture for draft-802.11n networks

? Fully backward-compatible with IEEE 802.11a/b/g legacy

devices

? Allows cost-effective system design using low-cost PCB and

single-sided assembly

? High-performance features across the Intensi-fi? product line

? High-performance whole house wireless coverage

? OneDriver?

? SecureEasySetup?

? SmartRadio?

? WPA?/WPA2?

? Cisco® Compatible Extensions (CCX4)

? Full-rate Advanced Encryption Standard (AES) engine in

hardware

? WMM? for quality of service

Interface/System bus support PCI Express

Standard IEEE 802.11g, 802.11b, 802.11a, Intensi-fi draft-802.11n

Data rate (Mbps) 54, 48, 36, 24, 18, 12, 9, 6, 11, 5.5, 2, 1

 20 MHz BW: 130, 117, 104, 78, 52, 39, 26, 13

 40 MHz BW: 270, 243, 216, 162, 108, 81, 54, 27

Modulation PSK/CCK, DQPSK, DBPSK, OFDM

Network architectures Infrastructure and ad hoc

Operating frequencies 2.4 GHz?2.5 GHz, 4.9 GHz?5.85 GHz

Operating channels 2.4 GHz: 11 for North America, 13 for Europe, and 14 for Japan

5 GHz: 12 for North America, 19 for Europe, and 4 for Japan

RF output power 2x2: 2.4 GHz 16 dBm ±1.5 dB, 5 GHz 14.5 dBm ±1.5 dB

Antenna connectors Two or more antennas for simultaneous transmission and reception

Power requirements 1.8V (3.3V for reference designs)

Power consumption Average standby <130 mW

Security IEEE 802.1x, WEP/WEP2, WPA/WPA2, TKIP, Weak-key avoidance, CCX, CCX 2.0, CCX 3.0, CCX4.0, 128-bit OCB mode

 AES, IEEE 802.11i

Software support Microsoft® Vista, Windows® XP, Windows® 2000, Linux®, VxWorks®

Certifications Worldwide regulatory, Wi-Fi 802.11a/b/g, WHQL, CCX

System as Tested for WLAN card(s):

Dell Inpsiron 9400/E1705

CPU Core 2 Duo T7400 2.16Ghz

RAM 2GB 667Mhz Dual Channel

Hard Drive 120GB Hitachi 5400RPM

Video nVidia go7900GS (unlocked BIOS)

Screen WUXGA (1920x1200)

Windows Vista x64 Ultimate

Router used to test the Buffalo WZR-AG300NH

Marvell 88W8363 Chipset

300Mbps on both A and G bands simultaneously.

WPA2 PSK AES encryption same on both Bands

Multicast enabled

 IVP6 Pass through enabled

 Multicast rate A 12Mbps

 Multicast rate G 11Mbps

High Throughput mode enabled.

A Band on Channel 36 - 40Mhz Bandwidth

B/G Band on Channel 6 - 40Mhz Bandwidth

File server Dell Latitude D820

CPU Core 2 Duo T7200 2Ghz

RAM 2GB 667Mhz

Broadcom Gigabit LAN

Connection to router 1000Mbps.

100GB 5400RPM Hard Drive

Windows XP Pro x86

File size transfered 298MB HP 3300 Series printer driver in 1 file.

Dell Wireless 1500 WLAN driver version v4.150.22.0 as found here

I've tried to setup the test scenario for typical situations that might occur.

So both Channels were enabled on the router and WLAN card.

Also I used WPA2 encryption which also has a performance hit, but who is going to run their router with no security.

Firewall and intrusion alerts were also enabled.

All the above have a small performance hit on the system, but chances are 99% of users will have the above enabled in real world situations.

I could have disabled all the above and got a highest possible speed, but this way you guys get a true speed in a normal situation.

The numbers will start to make more sense once I start to test more WLAN cards.

In the pipeline is the GIGABYTE GN-WI06N 802.11N a/b/g/n draft 2.0 Atheros AR5008 chipset and possibly the SparkLAN WPEA-214N 802.11 a/b/g/n draft 2.0 2.4/5GHz Dual-band Mini PCI-Express Module, Atheros AR5008E-3NX AR5418 chipset

I'll also work on the Intel 3945 based WLAN even though only 54Mbps this chipset is most notebooks out today.

Performance:

Below is a table of performance of the WLAN card.

Distance is measured in Meters, first number after A and G band is connection speed in Megabits per second, 2nd number is actual throughput in Megabits per second, measured with time taken to transfer a certain sized file.

Broadcom BCM94321MC, Dell 1500 WLAN

Distance from Router--------------00-----------03----------10---------15----------20----

A Band 40Mhz Bandwidth--270/056--243/056--162/054--108/054--000/000-

A Band 20Mhz Bandwidth--130/046--117/046--078/052--039/024--026/017-

G Band 40Mhz Bandwidth--270/056--270/056--270/056--270/050--040/019-

G Band 20Mhz Bandwidth--020/040--078/029--130/048--117/038--104/031-

Gigabyte AR5008, Gigabyte GN-WI06N-RH Tested with Vista x86 and Win XP Driver v6.3.0.94

Distance from Router--------------00-----------03----------10---------15----------20----

A Band 40Mhz Bandwidth--300/063--300/058--300/065--300/021--300/021-

A Band 20Mhz Bandwidth--

G Band 40Mhz Bandwidth--300/063--300/061--300/063--300/019--300/023-

G Band 20Mhz Bandwidth--

15m was through 1 wall

20m was through 2 walls, A band refused to stay connected to the router so no result.

G Band 20Mhz Bandwidth was very strange the closer I got to the router the slower the connection became :)

To make it even more odd the signal strength reads excellent (5 bars)

The best performance was @ 10m.

I should test G band 20Mhz bandwidth at futher range it should do a few more meter, will mean going outside as the house is not that big :P

Conclusion:

With this testing I have also come to realize that A Band routers could/should be making a comeback.

Even though the connection speed to the router was way less than G Band at at the same distance, A band still managed to keep maximum throughput.

It only lost connection at 20m with 2 walls, this is only when the connection is stresed as I can happily surf from this distance and stay connected.

To setup a good test bed, I choose channel 6 on G band to get away from both Channel 1 and 11 which my neighbours are already using.

A band on the other hand is free from users near me (I'm probably the only with it where I live).

G band congestion will become more of a problem with 11n as the routers use 2 channels to transmit data.

Only channels 1, 6 and 11 are non overlapping so this limits the choices.

By using the same chipset router as your WLAN card I would assume faster throughput can be achieved.

I would ofclourse love to test this, but finances will not allow it (sponsorship would be appreciated hint hint :) )

A Band, great for speedy close up networking if your in the same room or 1 room away as the router then this is ideal WLAN.

A band is not congested and throughput doesn't change with distance with no interference.

A band and it's very short wave lengths means that the signal is more easily absorbed into hard surfaces (dry walls etc)

So range is far less than G band at the same distance.

G band, has it's plusses it's far better range means if your router is somewhere on a different floor or room(s) away this will be the way to go.

But to get past 130Mbps you need 40Mhz Bandwidth which in turn means using 2 channels making congestion even worse.

Using 2 channels is 'Not WIFI' friendly' some routers (not the Buffalo) are beginning to auto change the 20/40Mhz bandwidth to be more WIFI friendly.

For best range and very little difference in throughput run the router in 20Mhz single channel mode, this has the least impact on others and gives you the best throughput if G Band is congested.

I'm quite surprised at A band, and would recommened this to other users as a better option for WIFI use.

56Mbps max real life throughput is not what I expected with the Broadcoms 270Mbps hookup to the router.

That is only 1/5th of the available bandwidth is beng used, quite dissapointing.

I'm pretty sure a setup with same chipset would probably get a better throughput to enable their special bursting technologies.

I hope I haven't baffled you with all the above.

Other than Video drivers, WIFI has the most impact on how I use my laptop, these 2 components are the most used on a laptop and have the greatest potential to be tweaked by the user to get the most out them.

More updates to come will do 20Mhz bandwidth tests as well.

Also am looking forward to the other WLAN cards coming to test these as well, to see if the router or the WLAN card is the bottleneck

[uPDATE 14th September 2007]

The Broadcom based WLAN card was not as fast as the Atheros based card up close to the router.

But the broadcom was a lot more reliable in the testing, see Atheros based WLAN review for details on this.

At 15m+ the Broadcom card performed much better, the Atheros will probably go further but will be so slow as not be worh the effort.

I would at this point probably choose the Dell 1500 WLAN over the Gigabyte due to the fact I'm running Vista x64 where the Dell works just fine and the Gigabyte will not run stable.

If file transfering is your thing and you don't mind Vsta x86 then the Atheros is for you, but after 10m definitely go for the Broadcom.

[somms]

http://www.notebookforums.com/showthread.p...286#post2723286

Nice Review... Recently upgraded to the Broadcom 1505 from the 1500 and latest driver 4.150.29.0.

Would be interesting to see how much improvement this more recent hardware/software would make to your testing...

Edited September 9, 2007 by somms

[mobilenvidia]

Review is a work in progress, would be good to try a Dell 1505.

I see the 1505 has a max speed of 300Mbs vs the 1500's 270Mbps.

No to find this even newer driver.

[mobilenvidia]

I think that the Dell 1500 and 1505 are the same thing.

They have the same part numbers.

Me thinks Dell gave them a new name for the new range of Santa Rosa based Lappies.

%BCM1500M_DeviceDesc% = BCM1500M_NT60, PCI\VEN_14E4&DEV_4328&SUBSYS_00091028

%BCM1505M_DeviceDesc% = BCM1505M_NT60, PCI\VEN_14E4&DEV_4328&SUBSYS_000A1028

The 2 use the same chipset in the driver, and look identical.

I'm 99% sure the 1505 is the 1500 RevA01 card.