Wireless Adaptors: USB or Fixed?

[steve-o]

Hi All,

I'm about to get wireless broadband in a week or so. I get a free wireless router with the broadband and I'm wondering: regarding the computer end, is a fixed (PCI?) adaptor any better, in terms of sensitivity, etc. than a USB one?

The ones I'm looking at are the Belkin N ones - similar price and I can see the advantages of the USB version (portability). However, I can also see the three (relatively) big antennas on the fixed version and I'm wondering if the fixed version would perform any better? The PC that I'll be fitting it to is fixed so performance is more important to me than portability and/or convenience.

Also, some of my friends that know about networks and things reckon that the freebie router that I get will probably not be good enough for my house (router downstairs, signal has to go through three brick walls, over a distance of about 20 feet, to get the my PC. I'd like to send music files around the house and access the internet) and have said that I may be better off with a more upmarket one. I've been looking a the Belkin N1 (F5D8631UK4A) 'cos it seems to get good reviews regarding signal transmission. Am I in the right ballpark with this? I notice that the D Link DIR-655 seems to be well thought of on these forums too. Thanks in advance for your input.

 

[thiggins]

Will the freebie router be 802.11b/g or draft 11n? If it's b/g, then get a b/g USB adapter.

If it's draft 11n, then select your adapter to match the chipset used in the router. There is no advantage in using a draft 11n adapter with a 11b/g router.
See:Can a New WiFi Adapter Change Your Wireless Performance?

You don't need much bandwidth for music streaming, so you don't really have to go for draft 11n.

 

[steve-o]

Thanks Tim. I'm not sure, until I get it, what standard the router uses but I get the message about matching the adaptor to the router.

 

[Norcross]

Thanks Tim. I'm not sure, until I get it, what standard the router uses but I get the message about matching the adaptor to the router.

My personal experience (and I have no data to back this up) is that I've gotten a little bit better performance on the PCI card wireless over the USB, mainly consistency.

In addition, that leaves more USBs open for other devices.

 

[prijikn]

USB Adaptor Walmart prices are reasonable.

 

[arias]

Hi All,

I'm about to get wireless broadband in a week or so. I get a free wireless router with the broadband and I'm wondering: regarding the computer end, is a fixed (PCI?) adaptor any better, in terms of sensitivity, etc. than a USB one?

Since you mentioned that performance is of more importance for you since it will reside in a stationary machine, then yes you will receive better performance with a PCI based wireless card, especially one that allows you to upgrade the antennas from stock.

 

[thiggins]

then yes you will receive better performance with a PCI based wireless card, especially one that allows you to upgrade the antennas from stock.

Arias, can you provide some specific examples of a PCI wireless adapter performing better than a USB adapter?

 

[arias]

PCI vs USB

My experience is limited to the first linksys usb adapters that came out compared to their PCI brethren. It was no comparison.

These days, I wouldn't expect it to be any different considering usb 2.0 has a max theoretical transfer rate of 480Mbits which comes out to about 60MB/s, this would be the limiting bottleneck of any usb wireless client.

PCI these days has a peak transfer rate of 533 MB/s with PCI-X almost doubling that at 1014 MB/s. N wireless cards can be purchased for either PCI standard.

Now, I expect a whole different ballgame with the release of the usb 3.0 standard wireless N clients, which purports a maximum theoretical speed of 4.8 Gbits aka 600MB/s which is supposed to be out any day now. Or maybe you've already covered usb 3.0 wireless N clients and I missed the release date. But outside the possibility that you were referring to usb3, I can't see how a usb2 wireless N client could even come close to its PCI counterpart.

 

[yellowfever9271]

difference in usb

Arias, can you provide some specific examples of a PCI wireless adapter performing better than a USB adapter?

(I know I'm not arias)

I've been coming to this website for a long time, and this is my first post. I'm not very knowledgeable network wise, and always assumed bandwidth is severely limited in usb adapters versus pci-x due to usb specifications like usb versus ide transfer rates in hard drives. I have the netgear WGU624 router which is dual band dual radio based on atheros super ag. I remeber switching from an airlink? usb adapter to my current blank board atheros 5002X minipci (both running super g) and everything seemed to speed up, which reinforced this way of thinking. After reading this post I figured I better do a little research; however the only website I could think of that tests all format adapters is CNET, but even they have very few comparisons with the same chipset and their network tests are pretty abysmal. I did find these to compare and although the usb version does not say which router is used, I assume by CNETs lab specifications they used the same:

Netgear WG511T 108 Mbps Wireless PC Card I know it's express, but they didn't review pci-x

Netgear WG111 USB Adapter

When the two adapter are close to the router they have almost the same max thoroughput, but when far away USB thoroughtput drops immensly, and if you notice the other usb adapters are all way less than 10. The express cards on the other hand are almost double at 200ft. My guess is that it's not thoroughput that's being reduced, but the power specification of usb 2.0 (which also has to conform to the smaller forn factor) reduces the radio's power which is exemplified in a mimo type operation (super g) over distance. The express card is a good comparison since it takes the antennas out of the equation.

If this is right, the next questions would be whether draft n 2.0, whose theoretical bandwidth which is almost 3 times as much as super g (108 vs. 270-300) is throttled on usb versus express card/pci-x, or whether distance is reduced even farther due to power demands.

Either way I like minipci/pci-x better since I can physically plug it into the motherboard, it makes me feel like I'm reducing latency. Sorry for the long post, I hope someone can find a better comparison.

 

[thiggins]

Connecting a small pipe (draft 11n) into a larger pipe (USB 2.0 or PCI / PCI-x) doesn't make the bits flow any faster.

The max phy rates for draft 11n, which right now are 270/300 Mbps for 3T3R systems, are wireless rates. By the time the bits hit the wired interface, the required rate is at best half that.

Yellowfever's post has a better review of the reasons why USB adapters might have lower performance than Cardbus or mini-PCI. One more factor to add for draft 11n is that all USB adapters that I have seen so far have two antennas. CardBus and mini-PCI usually have three. The # of spatial streams will also reduce throughput.

 

[thiggins]

Welcome yellowfever. Nice job on the research and post.

If this is right, the next questions would be whether draft n 2.0, whose theoretical bandwidth which is almost 3 times as much as super g (108 vs. 270-300) is throttled on usb versus express card/pci-x, or whether distance is reduced even farther due to power demands.

See the response to arias above. USB 2.0 has plenty of bandwidth to handle the physical PHY rate.

Either way I like minipci/pci-x better since I can physically plug it into the motherboard, it makes me feel like I'm reducing latency.

Don't worry about latency as a factor in client form-factor choice. The dominant factors for latency are the ranging algorithms and other client driver effects.