Does USB vs PCIE wifi adapter make any difference?

[Bleedingyamato]

Same question as the title but also is this any good?

 

I am currently looking specifically for a USB wifi adapter to try to solve wifi issues on my old laptop but I asked about USB vs PCIE wifi adapters just so I understand and downsides to USB if there are any.

 

 

https://www.amazon.com/dp/B00HM0K61Y/ref=olp_product_details?_encoding=UTF8&me=

The AC750 one.  Idk why it's called that since its AC600 speed whatever that means.

 

I'm tempted to get this but I'm concerned about it being a nub size adapter vs being larger like the AC1300 model that can be found with the selector on that item page.

 

 

[NumLock21]

USB lets you connect with a extension cable so it can be moved around to get a better reception. The other advantage is, they can be used on both laptops and desktops.

[pwn_intended]

PCIe ones tend to come with better antennas, but largely theyre about the same speed wise. The only thing I can think of is that the USB one might have a slightly worse ping due to the extra overhead. (not an expert in this though)

[Mira Yurizaki]

USB ones have the advantage of being discreet and more portable if needed and can be used on any computer with a USB port. Just make sure if you're getting an 802.11ac adapter that it's USB 3.0. I've seen 600Mbps adapters run on USB 2.0, which is capped to 480Mbps without USB overhead (so it's more like 300Mbps on a good day).

 

Add-in cards usually have better antenna configurations, but you can find them as well on USB ones.

[Bleedingyamato]

5 minutes ago, NumLock21 said:

USB lets you connect with a extension cable so it can be moved around to get a better reception. The other advantage is, they can be used on both laptops and desktops.

It's AC so I think range isn't an issue but in this case I am a little worried that the tiny AC750 model might be iffy.  

 

I mean those tiny USB adapters don't always have the best reputations do they?

 

The 1300 model has an extension cable and an external antenna to attach to it if need be but it's $55 vs the 750 costing about $35.  I'm tempted to save money by getting the 750 and its smaller size would be better for a laptop but I just want to be sure it'll work well.

 

1 minute ago, pwn_intended said:

PCIe ones tend to come with better antennas, but largely theyre about the same speed wise. The only thing I can think of is that the USB one might have a slightly worse ping due to the extra overhead. (not an expert in this though)

Yeah, the ASUS AC1300 PCIE version I have has 2 what I consider to be almost annoyingly large antennas that attach to it.  ?

 

What do you mean by overhead?  

[pwn_intended]

Just now, Bleedingyamato said:

What do you mean by overhead?  

Protocol overhead. Instead of the NIC (network interface card) talking directly with the CPU through the PCIE lanes, it has to go through the USB controller first. The only reason I mention this is because when using a thunderbolt to ethernet adapter on my MacBook air, my speedtests aren't as high as they are on my desktop. This may not actually be an issue in your case, but it is something that I thought i'd share.

[Bleedingyamato]

7 minutes ago, M.Yurizaki said:

USB ones have the advantage of being discreet and more portable if needed and can be used on any computer with a USB port. Just make sure if you're getting an 802.11ac adapter that it's USB 3.0. I've seen 600Mbps adapters run on USB 2.0, which is capped to 480Mbps without USB overhead (so it's more like 300Mbps on a good day).

 

Add-in cards usually have better antenna configurations, but you can find them as well on USB ones.

Is it common for Internet to be faster than 480-600Mbps?

 

Can you tell me how to check my connection speed in Windows 10? 

 

I figure it would help to be able to tell you my connection speed to give you an idea of whether or not I can even take full advantage of 480Mbps or if I'd be able to make full use of 600 or 1300.

 

I am connected using Ethernet currently but I could switch to wifi if that would be more accurate for this. 

 

The AC750 (AC600 speed) ASUS adapter in the link is USB 2.0 according to ASUS's website.  I think it might only be AC on the 5GHz band.

 

The ASUS AC1300 model is USB 3.0 but it's $20 more so I'd mainly consider it if my internet is fast enough for USB 3.0 to matter here.

 

Is all wireless AC 5GHz only or is that only for some wifi adapters that manufacturers choose to only make AC on 5GHz while deliberately leaving 2.4GHz as N?

 

 

 

[Mira Yurizaki]

Just now, Bleedingyamato said:

Is it common for Internet to be faster than 480-600Mbps?

No, but if you're using your network for anything else, having a wider channel helps immensely. Also there was something about if you have slower devices on the network, they can dumb down the speed of that channel's network, but I forgot where I heard that.

 

Plus why pay for a 600Mbps adapter when you're limited to 480Mbps at best?

 

Just now, Bleedingyamato said:

Can you tell me how to check my connection speed in Windows 10? 

http://www.totusoft.com/lanspeed1

 

Just now, Bleedingyamato said:

Is all wireless AC 5GHz only or is that only for some wifi adapters that manufacturers choose to only make AC on 5GHz while deliberately leaving 2.4GHz as N?

802.11ac is purely a 5GHz band and since 802.11n doesn't have any throughput increases on 5GHz, it's left at 2.4GHz most of the time.

[Bleedingyamato]

28 minutes ago, M.Yurizaki said:

No, but if you're using your network for anything else, having a wider channel helps immensely. Also there was something about if you have slower devices on the network, they can dumb down the speed of that channel's network, but I forgot where I heard that.

 

Plus why pay for a 600Mbps adapter when you're limited to 480Mbps at best?

 

http://www.totusoft.com/lanspeed1

 

802.11ac is purely a 5GHz band and since 802.11n doesn't have any throughput increases on 5GHz, it's left at 2.4GHz most of the time.

I think AC600 is the minimum speed for an adapter isn't it?

 

Heres what the test showed.  It had asked for me to set a destination folder for something.  It's labeled as "folder or server IP" in the window for the speed test.

I just set it to my desktop if that matters.

 

Edit: it also asked for a file size which I left set the default of 20MB.  

 

I also took a picture of the wifi router box that shows what speed of AC it is.

 

My father wasn't sure but he thinks we have 15Mbps down.  No idea about up but we have Time Warner if that helps.

[ProLogic]

Try speedtest.net for testing your speeds

[Brightglaive]

 

20 hours ago, Bleedingyamato said:

I think AC600 is the minimum speed for an adapter isn't it?

.....

I also took a picture of the wifi router box that shows what speed of AC it is.

.....

 

LOL. AC600 is meant to define a level of throughput given a certain compression scheme and encoding scheme and a particular guard interval and channel width and number of radios on the AP and antennas per radio and number of antennas on the client and assuming one client on the AP with one spatial stream. <BIG BREATH> In truth, 802.11ac IS backwards compatible with 802.11a which means instead of using AC600 (256-QAM with a 5/6 coding rate, 400ns Guard interval, and an 80 Mhz wide channel) it can drop back to 802.11a 6Mbps rates if needed (BPSK modulation with 1/2 coding rate, with an 800ns guard interval, and a 20MHz channel). Throw in another client on that and it has to share bandwidth if you aren't using MU-MIMO. Of course on many APs you can specify what data rates are acceptable. So this may allow you to maximize your wireless throughput. However it may also cause client disconnects if you drop below the acceptable data rate. Similarly you can adjust the guard interval and channel widths, but if a legacy adapter comes in range this could throw a wrench in your carefully crafted wireless network settings. This also does not take into account Joe Blow next door who is running an 802.11a/g AP from 2001 that may be interfering with your 160MHz channel and 400ns guard interval and screwing with your 2 or 3 spatial streams.

 

AC2600 means a maximum of 1733 Mbps. This assumes a 256-QAM, 5/6 coding rate, 160MHz Channel, 400ns guard interval with TWO spatial streams. This also assumes at least TWO 5GHz antennas on the AP and two 5GHz antennas on the client card. (It is likely the antennas on the EA8500 are 2 for 2.4GHz and 2 for 5GHz ranges) If your client only has a single antenna (1x1 MU-MIMO) your throughput drops to 867 Mbps with the channel, QAM and coding rate maxed and the short quard interval. If any of those other contributing factors change due to interference from older 802.11a technologies or non-802.11 sources (e.g. radar, 5GHz cordless phones, WiMax etc.) then your throughput will drop. Add more clients and the throughput to each client will drop. There are SO MANY variables. A site survey with a tool like WiFi Analyzer can help iron out some of those problems.Unfortunately it does not sniff out non-802.11 interference sources. 

 

I think Linksys said it best on the box in 2 places. "Actual performance may vary." and "Actual data throughput will be lower and may depend on the "type and number" (I assume this is what it says as it is cut off in the picture) of wireless products used and external factors." In fact it seems the right side of the box contains all the fine print that says (and I'm paraphrasing here): "the actual throughput you get will be lower and vary from the specification due to variables that can't be predicted for each user's environment (not the least of which are client hardware, configuration and radio frequency/structure interference.) Don't hold us responsible."

 

Something else you should keep in mind is that Firewalling and Intrusion prevention services (IPS) and routing are likely going to be the limiting factors when going from wireless to wired connections and when going from LAN to WAN/MAN (that is to say your internet service provider)

 

So in short, your coverage and speeds will vary. My suggestions to get the most bandwidth out of your wireless setup are as follows:

1. try and use clients that are 802.11ac. don't use older 802.11a client cards if possible.

2. Keep your lower bandwidth devices on 2.4GHz (where most 802.11b/g/n interference lives)

3. Site survey, site survey, site survey!!!! Be comprehensive when you do. 

4. Use the widest channel width that does not have any overlapping interference that you can, use a short guard interval (400ns) and the highest QAM and coding rate possible (these are often dumbed down to Mbps or ACxxxx rates in the setup on consumer routers) 

5. Make sure the clients and AP are not up against stone, concrete or metal surfaces. (hint: Ceiling mounting an access point or it's antennas in a central, open, point in your house may get you the best coverage and rates possible)

 

Lastly remember even if your wireless is as fast as it could be you are STILL limited by your internet connection. So don't expect a gigabit wireless connection to be able to download at gigabit speeds or even at hundreds of Mbps if you are running firewalling/IPS and Routing and only have a common 16, 40, or 100Mbps download speed from your ISP. Most ISPs do not even provide a connection at 480Mbps or 600Mbps unless you are a business paying for a portion of a gigabit fiber internet link. (or for some reason your ISP or Google forgot to cap your download speed on google fiber ;-) )

[Bleedingyamato]

58 minutes ago, Brightglaive said:

 

 

LOL. AC600 is meant to define a level of throughput given a certain compression scheme and encoding scheme and a particular guard interval and channel width and number of radios on the AP and antennas per radio and number of antennas on the client and assuming one client on the AP with one spatial stream. <BIG BREATH> In truth, 802.11ac IS backwards compatible with 802.11a which means instead of using AC600 (256-QAM with a 5/6 coding rate, 400ns Guard interval, and an 80 Mhz wide channel) it can drop back to 802.11a 6Mbps rates if needed (BPSK modulation with 1/2 coding rate, with an 800ns guard interval, and a 20MHz channel). Throw in another client on that and it has to share bandwidth if you aren't using MU-MIMO. Of course on many APs you can specify what data rates are acceptable. So this may allow you to maximize your wireless throughput. However it may also cause client disconnects if you drop below the acceptable data rate. Similarly you can adjust the guard interval and channel widths, but if a legacy adapter comes in range this could throw a wrench in your carefully crafted wireless network settings. This also does not take into account Joe Blow next door who is running an 802.11a/g AP from 2001 that may be interfering with your 160MHz channel and 400ns guard interval and screwing with your 2 or 3 spatial streams.

 

AC2600 means a maximum of 1733 Mbps. This assumes a 256-QAM, 5/6 coding rate, 160MHz Channel, 400ns guard interval with TWO spatial streams. This also assumes at least TWO 5GHz antennas on the AP and two 5GHz antennas on the client card. (It is likely the antennas on the EA8500 are 2 for 2.4GHz and 2 for 5GHz ranges) If your client only has a single antenna (1x1 MU-MIMO) your throughput drops to 867 Mbps with the channel, QAM and coding rate maxed and the short quard interval. If any of those other contributing factors change due to interference from older 802.11a technologies or non-802.11 sources (e.g. radar, 5GHz cordless phones, WiMax etc.) then your throughput will drop. Add more clients and the throughput to each client will drop. There are SO MANY variables. A site survey with a tool like WiFi Analyzer can help iron out some of those problems.Unfortunately it does not sniff out non-802.11 interference sources. 

 

I think Linksys said it best on the box in 2 places. "Actual performance may vary." and "Actual data throughput will be lower and may depend on the "type and number" (I assume this is what it says as it is cut off in the picture) of wireless products used and external factors." In fact it seems the right side of the box contains all the fine print that says (and I'm paraphrasing here): "the actual throughput you get will be lower and vary from the specification due to variables that can't be predicted for each user's environment (not the least of which are client hardware, configuration and radio frequency/structure interference.) Don't hold us responsible."

 

Something else you should keep in mind is that Firewalling and Intrusion prevention services (IPS) and routing are likely going to be the limiting factors when going from wireless to wired connections and when going from LAN to WAN/MAN (that is to say your internet service provider)

 

So in short, your coverage and speeds will vary. My suggestions to get the most bandwidth out of your wireless setup are as follows:

1. try and use clients that are 802.11ac. don't use older 802.11a client cards if possible.

2. Keep your lower bandwidth devices on 2.4GHz (where most 802.11b/g/n interference lives)

3. Site survey, site survey, site survey!!!! Be comprehensive when you do. 

4. Use the widest channel width that does not have any overlapping interference that you can, use a short guard interval (400ns) and the highest QAM and coding rate possible (these are often dumbed down to Mbps or ACxxxx rates in the setup on consumer routers) 

5. Make sure the clients and AP are not up against stone, concrete or metal surfaces. (hint: Ceiling mounting an access point or it's antennas in a central, open, point in your house may get you the best coverage and rates possible)

 

Lastly remember even if your wireless is as fast as it could be you are STILL limited by your internet connection. So don't expect a gigabit wireless connection to be able to download at gigabit speeds or even at hundreds of Mbps if you are running firewalling/IPS and Routing and only have a common 16, 40, or 100Mbps download speed from your ISP. Most ISPs do not even provide a connection at 480Mbps or 600Mbps unless you are a business paying for a portion of a gigabit fiber internet link. (or for some reason your ISP or Google forgot to cap your download speed on google fiber ;-) )

Wow that's a lot of detail.  Are you some sort of wifi deity?  ?

 

Pretty sure at least half or more went over my head sorry.  ? 

I appreciate the high level of detail though.   

 

Why did you say "LOL. AC600..."?  Is AC600 funny to you or was it something I said?  ?

 

 

What do you mean by site survey?

 

Unfortunately I have zero control over what devices are connected to which wifi band. (2.4 GHz or 5GHz)  The best I can do is manage my own devices.  

 

I think we have some devices that have G wifi but most of our devices are N I think with a few being AC. 

 

 

[skywake]

@Bleedingyamato

To put it simply, and I don't know why you wouldn't, AC750 does not mean "750Mbps". It's actually a lot slower than that and therefore won't be bottlenecked by USB 2.0. In real world use you're probably going to get speeds of around 90Mbps or so out of it. But if you were to get a higher spec wireless NIC? Something that was AC1200, AC1750 or something? It might start to become a real bottleneck. But probably not because of USB. More likely because of the physical limits on the size/power of a USB dongle that don't exist with pcie cards.

[Brightglaive]

LOL.   I was actually a wireless engineer for Cisco systems for 10 years. You learn a lot.

 

Simply put,

USB adds overhead and delay.

wireless speeds are theoretical and subject to real world interference and variables

Other processes like routing/firewalling/IPS introduce some real bottlenecks when transmitting out the WAN port on the Linksys that can slow the data to MUCH lower speeds (10's of Mbps to Low hundreds of Mbps)

WAN/MAN/internet connections are generally 100Mbps or less unless you have Gigabit fiber or a business gigabit or multigigabit internet connection (like OC-48 or 10GbaseX or 1000baseX fiber or copper)

 

By way of explanation: 

AC600 is not really a speed per se. It's a combination of speeds based on the 2.4GHz and 5GHz bands, rounded up or down to the next 50 or 100. It's a rating that encompasses all those pesky variables like channel width, frequency, coding rate etc.

 

Originally there was one frequency band, one channel width, one guard interval, one data stream, 2 coding rates and 2 compression modulations so they would just say 11 or 5.5 or 2 or 1 Mbps for speeds (back in the day of 802.11b introduction.) 

 

Since 802.11 uses CSMA/CA (Carrier Sense Multiple Access/ Collision Avoidance), it basically listens for anyone transmitting and then if the channel is clear the client or AP will transmit. If it "hears" a transmission occurring it will back off and wait until the transmission is done. The time it waits to hear a transmission before starting to transmit is called a guard interval. I believe it is also used as a multiple to determine how long it will wait before checking again to see if any client or AP is transmitting if it hears a transmission.

 

Channel width was originally 22 MHz wide. If you look at the center frequencies on a frequency chart of 802.11b you can see the frequencies vary by 5 Mhz. Channels 1, 6, and 11 are non-overlapping (and 14 where it is supported)  

5 Ghz uses slightly smaller channels, 20MHz. it uses different coding rates and modulations to compress more data into transmission. That's why it could transmit up to 54 Mbps when it came out. 802.11n allowed you to cut the guard interval in half (to 400ns). It introduced newer, higher compression to the SAME 2.4 or 5GHz channels and allowed you to use channels that were twice as big as before (40MHz). if you put all those together you could theoretically achieve 150Mbps using a 40 MHz channel and the highest compression rates.

 

802.11n also introduced a revolutionary technology called MIMO or multiple in multiple out. Normally when you use a single antenna on a client and single antenna on the AP you get something like this:

 

The left shows the direct transmission and the reflections or echos of the signal. This was called multipath distortion.  The right shows the distortion of the signal that occurs. Antenna diversity (using 2 antennas) helped to get a clearer signal by switching the receiving antenna to whichever antenna had the better, more accurate, less distorted signal i.e a signal with less echos. MIMO used the idea of triangulation with multiple antennas to be able to determine when an echo was arriving by measuring the delay between when each antenna recieved a signal. Knowing what the signal is like between a transmitter and receiver on a given channel is called channel state information or CSI.  Through the magic of algorithms and silicon the radio could now use the CSI to reconstruct the original transmission by combining the direct transmitted signal and it's echos. It could also use the CSI info to transmit a signal using a filter that adjusts the signal and takes into account all the multipath distortion in a given direction and which would cause it to arrive at the client as a clear transmission. This was called spatial beamforming. 

 

Yeah I know....It's a lot to take in, and a LOT of tech speak But stick with me, I'm getting to my point. If you then take some data, break it up into 2 separate streams and use that CSI stuff to transmit both of the streams as separately encoded transmissions at the same time on the same channel with 2 different antennas so that a client with 2 or more antennas can hear both of the different transmissions and reconstruct the original data, you have the idea of how spatial multiplexing works.

This allows you to send 2 or more transmissions at the same time using the same channel and get more throughput if you use the same compression. Using beamforming and MIMO antennas with those special signal processors/algorithm allows each transmission to arrive as a clear distinct signal. More antennas = more streams. (the above diagram depicts 3 transmit antennas, 3 receive antennas and 2 streams or 3x3x2 MIMO)

 

Multiple data streams increased the theoretical throughput to 300 Mbps with 2 streams and up to 600Mbps with 4 streams.

 

802.11ac expanded the channels to 80 and 160MHz (in some cases more than doubling and quadrupling throughput) and added the ability to address a stream to a particular client. POOF! you have Multi User-MIMO (or MU-MIMO) and 1733Mbps up to 2340Mbps per 5GHz radio. (ok that may be an oversimplification but this is getting too long)

 

On a side note the various modulations and coding rates and guard intervals and channel widths and the correct combinations of each (and their resulting data rates) were defined and documented in the modulation and coding scheme or MCS. For 2.4GHz 802.11n defines 9 MCS levels for 1-4 streams, For 5Ghz 802.11ac defines 9 MCS levels for 1-3 streams.  you can see them at: 

 

https://en.wikipedia.org/wiki/IEEE_802.11n-2009

and 

https://en.wikipedia.org/wiki/IEEE_802.11ac