Announcement

Collapse
No announcement yet.

Wireless throughput in A band (N300)

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Wireless throughput in A band (N300)

    Hi everyone,

    I am having a strange issue that is the reverse of what I had a couple of weeks ago. Now I can't seem to get decent wireless throughput on my D-Link router (which previously worked fine) OR my Netgear router with DD-WRT (though that same router is fast with the stock firmware).

    If you saw the thread about recapping my WNDR3700, you might want to skip the next four paragraphs as I recap (no pun intended).

    I recently discovered that my wireless throughput on my Netgear WNDR3700v1 was not what it should be. I'd been using the 3700 without issue for years, but never really had any reason to consider wireless throughput (I was using it for internet only) until I decided to set up a computer to act as a NAS for backups for the laptop (wireless) and desktop (wired).

    As part of the troubleshooting process, I hooked up my D-Link DIR-825 A1, and it instantly outperformed the Netgear by a factor of 4 to 1 in actual throughput with the same reported link speed (300Mbps). No combination of settings would make the Netgear any faster.

    The Netgear firmware has long been reputed to be full of bugs, even though I personally have never had any problems with it, so I was thinking that's where the issue was. I posted about it on the Netgear forum, and the person there clued me in on what should have been obvious-- hardware failure.

    I was not a registered user here, but I've been reading this forum for years, so immediately my mind went to capacitors. Lo and behold, upon opening the unit, I found three bulged electrolytics. With the help of some great people here, I got some replacements ordered, and as soon as they arrived, I installed them. As soon as I turned it back on, the router was back up to the speed where it should have been.

    Just yesterday, though, I had a recurrence of an issue I thought had been resolved. My laptop uses the Intel 4965AGN wireless card (mini PCIE), a card Intel EOL'd years ago. The latest driver for it is about five years old, and apparently even then contained a bug that Intel has written about being fixed in the newest drivers. I can't confirm it is the same bug, but the symptoms I am experiencing are identical to what Intel describes, and fixing it in drivers that no longer support my wireless NIC does not help me (though I have wondered if editing the .inf might work as it does with some mobile graphics cards).

    The issue is that during periods of high bandwidth use on wireless, like copying a large file or doing a backup, the wireless link loses connectivity. The tray icon on the laptop gets the little ! indicating that it has no internet access, and Windows keeps trying to copy the file, even though the throughput has dropped to nil. It's not just the internet that has lost any throughput... it is the whole wireless connection. I am not able to access the router's configuration interface or that of the DSL modem from the laptop when it is doing this, but I can see them from the wired desktop (so I know the router is still working). If I disable the wireless card and re-enable it via the laptop button, it sometimes gets some throughput back, but it is slow and flaky until the laptop is rebooted.

    I never had that happen with the D-Link in the week or so I was using it (though I have also not had it happen with the Netgear post-recapping until now), so I decided to try it again and let it download some massive files (backup files on external HDD) and see if I could make it fail. The driver bug on the laptop is supposedly only triggered when paired with certain routers, though that remains to be verified.

    The thing is, even though that router is still configured exactly as it was when it was outperforming the Netgear with the bad caps, it no longer has that great throughput it did before. Actual throughput dropped down to 2MB/s, about 1/7th or 1/8th of what it was before, even though the link speed was reported as 300 Mbps (the maximum) as it was previously.

    I opened the D-Link, expecting to find more bulged caps, but nope! They were Chemi-Cons, looking just as nice as the ones I just put in the 3700. Kudos to D-Link for using good caps... unlike the Ltecs in my Netgear and the Su'scons in my LG monitor!

    I decided to try DD-WRT again on the 3700, with the new caps. It delivered the same performance as the D-Link! I changed settings all over the place, but nothing I did got it to perform as the D-Link had before, or as it did with the stock firmware.

    I booted a Live USB thumb drive with Linux Mint on it on the laptop, and the throughput was the same there as it was in Windows.

    I verified that mine is still the only network on the A band, so if there is interference, it's not from a WLAN.

    I flashed my Netgear back to the stock firmware and the speed is back up again. It really does not make any sense to me!

    The only thing I can even hypothesize is that hauling the old D-Link out of storage and using it intensively for a week or so (while I waited for the new caps to arrive) did something to the long-unused caps in the D-Link that didn't show up at that point, but after I let it cool again, they failed without bulging (even good ones can dry out)... and coincidentally, DD-WRT in its current build just didn't work well on my router and caused it to have the same symptom. It doesn't seem likely that two completely dissimilar problems would just happen to come together like that (and match the same issue I had with the Netgear that had me recapping it)... but what else is there?

    I don't have any other wireless devices that operate in the A band to test. I have an Android tablet and an old laptop (keep in mind that my "new" laptop was manufactured in 2008) that are both b/g items, so their throughput is terrible at best anyway.

    Does anyone know of a diagnostic utility that could tell if I am getting a lot of retransmissions of packets, which could be causing the slow throughput despite the maximum link rate?

    Thanks!
    Last edited by Ascaris; 12-03-2015, 11:48 PM.

    #2
    Re: Wireless throughput in A band (N300)

    ping the router over the wifi, that should tell you latency & packet-loss

    Comment


      #3
      Re: Wireless throughput in A band (N300)

      Have you ruled out any kind of interference? Some cheaply built electronics can really mess up wireless performance. Try moving the router at least a meter or more from any other electrical device. Turn of as many electronic devices as possible just to check.

      Another thing to try is to change the orientation of your router, just tilting it on its side can make a massive difference if there is interference around. If your router have external antennas, try moving them around at different angles.

      Comment


        #4
        Re: Wireless throughput in A band (N300)

        Thanks for the replies.

        It looks like there might be some interference. Pinging the router continuously shows that there are periods where packet loss is pretty high, interspersed with longer periods where everything is working... that's to the Netgear with the stock firmware, which has thus far been more capable of reaching higher actual throughput at a given connect rate.

        The Netgear has internal antennae only, so I tried placing it on one end instead of flat on the table; packetloss went up quite a bit. It did have an effect, but it was not the one I'd hoped for.

        The only other thing that could be causing it now, I think, would be something wrong with the antenna in the laptop, which I already know is partially true at least. It's a 3-antenna card, but I managed to sever one of the leads reconnecting the display panel a while back, so it has been running on only two antennae. I have the broken one disconnected from the card and taped off so it won't short anything. The fact that I managed to break one antenna lead makes me wonder if I didn't ham-handedly damage the others too, so now I am in doubt about all of them. I don't know enough about them to even know if it is possible to test them other than by trial and error. And with that last thought in mind...

        I've ordered a mini PCIE to PCIE 1x wireless adapter (with 3 external antennae) to test it. I will be able to run the laptop's wireless adapter in a desktop PC with completely separate antennae. If the same issue recurs, it has to be interference or the wireless card itself.

        Thanks again for getting me on the right track. I must be getting old-- these times where I keep focusing on one thing while looking for the trouble (like caps in this case, or the firmware the time before), whatever it may be at the moment, and not seeing other things that should have been evident (how obvious is RFI interference as a potential issue in a wireless connection? I've done troubleshooting of this type before, and that was what led me to adopt the 5 ghz band in the first place... I even used the same ping tool as I did years ago) has not always been a trait of mine... sigh. I've gotten dumber over the years!

        Anyway, when that item arrives, I should have more information. I wonder if I will get higher signal strength numbers with the presumably better antennae. That might be something I would have to control for when testing for interference, since the greater signal strength (obviously) would mean a greater SNR with all else being equal.

        Thanks again guys!

        Comment


          #5
          Re: Wireless throughput in A band (N300)

          Some of you may know that many dual-band wireless routers have video streaming optimization options only for the 5GHz band.
          My first choice in quality Japanese electrolytics is Nippon Chemi-Con, which has been in business since 1931... the quality of electronics is dependent on the quality of the electrolytics.

          Comment


            #6
            Re: Wireless throughput in A band (N300)

            japlytic,
            I'm back to DD-WRT on the 3700, and with all of the options it has, I think it probably would have the options on the B/G band too. I am not familiar with it enough to say one way or another.

            Unfortunately, my laptop's Intel wireless card (or the driver for it) comes with the "feature" of only being able to use channel bonding in the A band, so B/G is always going to be slower. I love it when the features of my hardware are designed to benefit others more than the person spending the money for said hardware. It's wholly impossible to avoid impinging on the same channel other networks use anyway, with 10 or more networks detectable on the BG band at any given time (and over thirty reported by inSSIDer if I scan over time), so there's going to be frequency contention and packet collisions no matter what.

            I have learned a little more about what is going on, but I will put that in another post so that it doesn't get TL;DRd any more than necessary.

            Comment


              #7
              Re: Wireless throughput in A band (N300)

              Ok... I learned some more about what is happening, I think.

              I opened up an old (still perfectly functioning, but obsolete) Zyxel router that has external antennae. They attach to a Mini PCI (not PCIE, it's the big one) card inside the router via the same little connectors that are on my mini PCIE card.

              That router is a 2.4 GHz only, so the suitability of the antennae to 5ghz operation is questionable. Even so, I tried disconnecting both of the remaining internal antenna leads from my laptop's wireless card (recall that has 3 antenna connectors, but I broke one of the leads, so that wire is taped off and not connected) and connected the router's leads, with the top shell of the router sitting on the desk next to the laptop.

              It made no difference. Throughput, packet loss, and reported signal strength were the same as they had been with the two internals.

              Next, I tried reconnecting the two internals to the antenna connectors on the card marked 1 and 2, and connected one of the router's external antennae to the connector on the card marked 3, so it once again had three (though one of them was again of questionable quality for the A band).

              Again, no difference in any of the areas above. Not even a 1dB increase in signal strength in either direction (DD-WRT reports the signal strength from the laptop; inSSIDer reports it from the router). Having all three connected appears to make no difference at all. Now if I connect only one antenna to the wireless card, it halves the reported connect rate from 300mbps to 150mbps... so I gather that means it's a 2x2 MIMO setup, even with three antennae, right?

              As for there being interference... it's a definite maybe. Unlike the stock firmware, the DD-WRT has an auto channel function in the A band (the stock stuff only has it for BG). It has switched the channel around a few times... I am not sure if it is doing that because it sensed something or if its just trial and error, looking for something stronger, but that's what it has done.

              The signal strength reported by inSSIDer (from the router to the laptop) is -50 dB; the signal from the laptop to the router, as reported by DD-WRT, is -59 dB, which DD-WRT reports as 43% signal quality. It is very nearly line of sight from the laptop to the router... the armrest of my sofa blocks it a little bit. I tried putting the router on the armrest temporarily, so there was true LOS, and the numbers got better in both directions by 5dB... I will have to find something to put under the router (raising it above the armrest) to get better LOS! The distance is 12-15 feet, I would guess.

              My Mini PCIE to PCIE adapter could arrive as soon as tomorrow... more info to come when it does.

              Comment


                #8
                Re: Wireless throughput in A band (N300)

                Ok...

                The issue is in the laptop, not the AP. I've reproduced it in the desktop using the PCIE-Mini to PCIE adapter.

                Something happens and the wireless throughput drops to half or less of what it should be. When I look at the network monitor in the task manager, the throughput alternates between peaks of good throughput and zero, every few seconds. The graph ends up looking like this:



                The continuous ping results match what you see there... several pings that time out, several that are returned (but with longer times than when its working well), then several more timed out... repeated infinitely.

                Rebooting the router does not help, but rebooting the PC (laptop or desktop) does. Eventually, it happens again... seems to happen more often after the PC has resumed from standby.

                If it's a driver issue, I am out of luck unless Intel decides to go back and make a driver for a card they've ignored since 2010 (yeah, not holding my breath).

                Comment

                Working...
                X