Consistent Wireless Speeds - Reasonable Expectations

[FadgewackeR]

I've been chasing this around for a while, and though I'm confident with the manipulation of different firmware and basic Putty manipulation, actually snooping the root cause of my issues has been a bit of a fumble in the dark.

Specs:

• 86U - Latest Merlin FW
• 80mhz - ch 56 seems to be best so far on speeds / load solidity (I'm part way through testing all channels on 5ghz with file transfers over wifi from and to my NAS)
• UK ISP - Zen Internet - 150dn/30up FTTP - I get just >150dn and >30up absolutely rock solid, no drop in transfer rate at all when tested either direct from the modem to the laptop, or through the 86U
• I'm in clear line of sight of the router, approx 2m away.
• There is nothing competing with me on the 5ghz network (INSSIDER)
• All tests last night were done on a Test SSID and I had only my Dell Latitude E7240 Ultrabook (AC network card) and an iphone XR connected on the 5ghz. No other devices on ethernet, or any other wireless band.
  
• Both of the above devices are fully up to date with drivers.
• In the current test state, I have no QoS at all in play.
• Router has been processed as per the following - https://www.snbforums.com/threads/major-issues-w-rt-ac86u.56342/page-4#post-495710
• No modifications have been made, other than Merlin latest FW flashed.
• Connection to the laptop on Tx /Rx 866.7

I suppose first question is this - What sort of 5ghz wifi performance to these devices should I expect to see from my connection, based on the circumstances above? Am I chasing something that isn't ever going to be consistently there?

I was getting anything from 65dn to 120dn before I performed the surgery last night.

I'm getting up to 152mbps down at present, but this is almost never constant (using dslreports speedtests). There is almost always a period where the transfer speed drops to 80-100 ish. This can happen in the first third, the 2nd 3rd or the final 3rd of the tests. So Some tests are concluding at 152mbps (img1), and some are concluding at around 100 (img2)... img3 is sort of what I would exoect, but in the past I have been able to achieve a very steep ramp to max and maintain that (albeit when fq codel QoS was in play).

The images below are of tests ran last night, which are the best I've had for about 3 weeks. Nothing could maintain max speed over wifi for the duration of the test, once max was achieved post ramp. Ethernet, no problem.

The other speedtesets that I use are speedtest.net and thinkbroadband.

Think broadband has me more randomly in and around 100mbps...

Is this ^ completely normal? Am I going mad chasing something that doesn't need to be chased (since last night's improvements)?

Cheers

 

[FadgewackeR]

FWIW< here are some results from Think, at similar times. Upload is constant. Pings are consistent. Quality (not shown) was very good.

 

[FadgewackeR]

For completeness, the results from speedtest.net, over a similar time period - the one at 8pm was hard wired.

 

[L&LD]

A few points that I see as potential issues.

1. 2M away from the router is much too close to be doing maximum speed tests with on today's routers. I would recommend minimum 10' (3M) and preferably 15' to 20' from an AC powered laptop set to 'Performance' and not 'Balanced' or 'Battery'. Also, leaving the laptop in the exact space, orientation and screen tilt (where the antennae are) is very important between testing. Any changes, even ones as few as a couple of inches in any direction, may again negate the test results.
2. INSSIDER and other similar apps are not very useful programs for determining anything beneficial for tweaking in your network environment. If you are leaving this program open when doing your speed tests, throw them out the window as this may greatly affect the speeds achievable.
   1. The WiFi environment usage/setup is determined mostly by the router, not the client. Use the scanning tools built into the router to determine what 'it' sees and therefore bases its setup on. Seeing what the client sees may or may not be what the router is seeing (and acting on).
3. With 150/30 down/up Mbps ISP speeds, I would expect full speed on 5GHz within a 10' minimum to 40' to 80' radius of the router, (depending on obstructions, or not). But not without QoS.
4. I have a hard time with 'latest RMerlin firmware'. Which version are you using specifically? There are 5 or 6 possibilities right now that are a couple of months old, and three that are less than 3 weeks old
   1. Using old SSID's may or may not give you the best connection possible between your router and your clients.
   2. Not having performed a full reset to factory defaults followed by a minimal and manual config to secure your router and connect to your ISP may be the reason for getting inconsistent results.
   3. Not optimizing the location, orientation and antennae of your router may also be contributing to erratic speeds, in addition to using the wrong Control Channel for your environment.
   4. What other settings in Wireless General and Professional have been changed past defaults?

If you haven't done so already (but consider that you may want to wait until 384.12 goes final first before doing the following), I would flash the 384.12 (currently Beta 2) firmware that is currently the most secure, stable and up-to-date firmware possible on our routers.

Follow that with a full reset (M&M Config).

Along with a Sanitize Network process (please see my signature below for links to all the guides I'll mention here).

At this point, before any further customizations, I would spend a few minutes finding the optimum location for the router in its 3D space, including at least 10' above ground level, and 3' minimum free space all around with the antennae all pointing straight up (but play with the orientation and also how the router is facing too). This would be with the laptop plugged into AC power in Performance mode and left stationary at least 15' to 20' away from the router.

At this point, I would choose the best Control Channel for both 2.4GHz and 5GHz bands. You may find that you need to slightly adjust the changes you made above. I would not bother using an app to set the channels with. What they indicate is mostly irrelevant. Simply test 1, 6 or 11 at 20MHz width for the 2.4GHz band and choose the best channel overall (there will not be a 'perfect' channel). For the 5GHz band, using 80MHz width, test the channels available and you'll quickly find the best one there too. Don't look for simply the highest throughput, rather, choose the Control Channel that gives your network the most responsiveness at the highest throughput too.

With this base setup complete, I would follow the amtm Step-by-Step guide and using a spare USB drive, format and create a swap file on it for the router to use.

Now, I would be installing FreshJR's script and setting it up minimally (first two or three posts in his thread) and then tuning the up/down speeds in the Adaptive QoS section to optimize your connection. Don't forget to wait 10 minutes after making any changes so that the script can be properly applied.

At the end of this process, your network, router, and clients will have been optimized for your WiFi environment.

Unless you get better hardware, just enjoy.

P.S. I don't know how much of the above you may have done, but doing so in the order presented builds on the previous steps in a synergistic way.

 

[Grisu]

yes that has been the greatest problem when I tested speeds and always had Inssider open in a background window or any other Wifi-app!
Speeds went up and down all the time till I once luckily closed the window and got aware of it.
Another problem has been 2 routers not far apart both with 40MHz 2G channel 3(1+5) and 11(13+9), so should be fine for EU but anyway too much interferences with Asus routers.

 

[FadgewackeR]

A few points that I see as potential issues.

1. 2M away from the router is much too close to be doing maximum speed tests with on today's routers. I would recommend minimum 10' (3M) and preferably 15' to 20' from an AC powered laptop set to 'Performance' and not 'Balanced' or 'Battery'. Also, leaving the laptop in the exact space, orientation and screen tilt (where the antennae are) is very important between testing. Any changes, even ones as few as a couple of inches in any direction, may again negate the test results.
2. INSSIDER and other similar apps are not very useful programs for determining anything beneficial for tweaking in your network environment. If you are leaving this program open when doing your speed tests, throw them out the window as this may greatly affect the speeds achievable.
   1. The WiFi environment usage/setup is determined mostly by the router, not the client. Use the scanning tools built into the router to determine what 'it' sees and therefore bases its setup on. Seeing what the client sees may or may not be what the router is seeing (and acting on).
3. With 150/30 down/up Mbps ISP speeds, I would expect full speed on 5GHz within a 10' minimum to 40' to 80' radius of the router, (depending on obstructions, or not). But not without QoS.
4. I have a hard time with 'latest RMerlin firmware'. Which version are you using specifically? There are 5 or 6 possibilities right now that are a couple of months old, and three that are less than 3 weeks old
   1. Using old SSID's may or may not give you the best connection possible between your router and your clients.
   2. Not having performed a full reset to factory defaults followed by a minimal and manual config to secure your router and connect to your ISP may be the reason for getting inconsistent results.
   3. Not optimizing the location, orientation and antennae of your router may also be contributing to erratic speeds, in addition to using the wrong Control Channel for your environment.
   4. What other settings in Wireless General and Professional have been changed past defaults?

If you haven't done so already (but consider that you may want to wait until 384.12 goes final first before doing the following), I would flash the 384.12 (currently Beta 2) firmware that is currently the most secure, stable and up-to-date firmware possible on our routers.

Follow that with a full reset (M&M Config).

Along with a Sanitize Network process (please see my signature below for links to all the guides I'll mention here).

At this point, before any further customizations, I would spend a few minutes finding the optimum location for the router in its 3D space, including at least 10' above ground level, and 3' minimum free space all around with the antennae all pointing straight up (but play with the orientation and also how the router is facing too). This would be with the laptop plugged into AC power in Performance mode and left stationary at least 15' to 20' away from the router.

At this point, I would choose the best Control Channel for both 2.4GHz and 5GHz bands. You may find that you need to slightly adjust the changes you made above. I would not bother using an app to set the channels with. What they indicate is mostly irrelevant. Simply test 1, 6 or 11 at 20MHz width for the 2.4GHz band and choose the best channel overall (there will not be a 'perfect' channel). For the 5GHz band, using 80MHz width, test the channels available and you'll quickly find the best one there too. Don't look for simply the highest throughput, rather, choose the Control Channel that gives your network the most responsiveness at the highest throughput too.

With this base setup complete, I would follow the amtm Step-by-Step guide and using a spare USB drive, format and create a swap file on it for the router to use.

Now, I would be installing FreshJR's script and setting it up minimally (first two or three posts in his thread) and then tuning the up/down speeds in the Adaptive QoS section to optimize your connection. Don't forget to wait 10 minutes after making any changes so that the script can be properly applied.

At the end of this process, your network, router, and clients will have been optimized for your WiFi environment.

Unless you get better hardware, just enjoy.

P.S. I don't know how much of the above you may have done, but doing so in the order presented builds on the previous steps in a synergistic way.

Thanks for the detailed response! Just to be clear, I was able to manage 147mbs dl over wifi with A+ grades across the board around 3 weeks ago.

1: much more than 10 to 15ft away from the router will introduce obstructions, sadly. Laptop was in performance mode, and plugged in.

2: INSSIDER was not on during tests. Only to determine exactly what was competing with the channels, prior to my tests on the channels performance individually.

3: OK

4: I see little drop off in performance within 30ft of the router, regardless of obstructions - it's the base performance that is the issue - please can you explain how I'm likely to get full speeds with QOS. This is essentially what I had 3 weeks ago, but I can't manage that at all now

5: My firmware is the latest actual release, not beta. 384.11_2

6: The SSID's that I used last night, specifically for test, were fresh. I then changed the names to the known ones I had been using today. Performance was the same. I have now changed the SSID names to ones not previously used. Is that enough, or do I need to actually set up the SSIDs manually?

7: I reset to factory defaults by following your procedure in the link attached above, steps 1-9 and did the M&M setup. I got as far as sanitising the network last night. I obviously need to do more work on the control channel set up.

8: This could contribute... however, I've never had issues with full speeds previously, with the same router and location, albeit before I got fibre so at lower speeds. The upload speed isn't affected, only the download speed.

9: Nothing set as non standard - I followed your instructions.

So next - go to Merlin beta, then factory reset, and go through the M&M I did last night, plus sanitise? Regards placement, I cannot get the router 10ft off the floor with 3ft all around, but I can be looking directly at it, with vertical antennae, 15 - 20ft away.

Please can you advise the best way to select the best channel? Before you mentioned the routers tools could identify this... Assume I start there and then set up multiple wifi file transfers and log data?

I'm well versed with AMTM and Fresh's adaptive QoS implementation, so no issues there.

 

[L&LD]

You're welcome.

1. At that distance, should not be an issue unless your dwelling has lead lined walls.
2. Test each channel individually, regardless of what any app indicates.
3. Consistency may be a factor of your WiFi environment, nothing can 'fix' that except moving to a better one.
4. QoS may not give you full ISP paid for speeds, it's the consistency we're going for using it (and the lower bufferbloat, of course)
5. Okay.
6. If you set up new SSID's, that is doing it manually. However, depending on what you've changed or haven't changed past defaults, it may not be enough (look over the M&M Config steps).
7. The Controls Channels used can take a network from usable, to wow, depending on external factors in the WiFi environment, of course.
8. Location is dependant on the router model used (along with the RF design/theory in place and the firmware options used too). I don't ever take that for granted.
9. Okay.
10. The 10' off of the ground is the outside ground level, not the floor level. It is not a hard and fast rule, but play with the z-axis of the router and find the sweet spot for your location.
11. The router tools can give you an idea of what it 'sees'. But I don't recommend using them to pick a channel. Test all. Select the best, overall. Like I've already mentioned, the maximum throughput may or may not match your ISP paid for speeds, but what I tune for is network responsiveness, not maximum throughput for a single client in a specific location.

 

[Klueless]

WiFi variability. Maybe it's not you? You do make a compelling case in that your Ethernet tests have been rock steady yet I wonder.

I suppose first question is this - What sort of 5ghz wifi performance to these devices should I expect to see from my connection

Oh, I dunno, around 500 Mbps? Of course you're limited by your Internet service so 150 x 30. Which is what you got, until you didn't, about 3 weeks ago.

I got a new Internet service about two years ago. I signed up for 60 and got a solid 70. Then for Christmas they gave me a free upgrade to 100 and I get a solid 120. Last year speed tests started bouncing around with some as low as 70 and 80 but usually, around four in the morning, I still get my 120 : -)

For me, I think my ISP's recent success has left them slightly oversubscribed, too many customers contending for the same resources. For you? I don't know.

Maybe when I get a chance I'll compare my WiFi with Ethernet and then I'll run some speed tests internal to my network rather than over the Internet and see what I see. Maybe ...

 

[FadgewackeR]

I was away at a family funeral yesterday, but when I got back I ran a quick check on a 1gb file transfer to and from my NAS, and over wifi, thru the 86u, my upload was about 235mbps (IIRC) and my download was 399 (I remember this specifically). I will obviously do much more testing of this, and will schedule some transfers periodically, but internally the speed, to a simple man such as I, seems to be more than sufficient. I don't have the ability currently to see the speed of transfer graphs to see how that speed was maintained.

I upgraded to the beta RMerlin 384.12-beta2 and reset to factory defaults (2nd time in as many days), wenth though the minimum config, renewed the SSID's, created a swap and added Fresh's QoS before I had to hit the hay.

No change to the performance over wifi whatsoever. Download scatty, Upload better, more bloated than it used to be, but generally getting up to 30mbps.

I as recently able to get A+ across the board by limiting my DL to 147 and just putting a cap on my UL at 30, but I'm miles away from anything like that at present.

 

[FadgewackeR]

Well... I think I've fixed it. More testing to come, but, after a LOT of configuration change and test:

• I uninstalled INSSIDER, then...
• I repaired the installation of my Intel wireless client (I had previously done this a number of times with no success)

Full speed maintained throughout the tests, A+ A+ A+ over on DSL reports - Minimal QoS (145/28) which I will tweak moving forward.

I'll factory reset once more (now on 384.12) and start from scratch again. I'll continue to do my full range of 5ghz channel checks to determine the best available channel, but it looks to me like INSSIDER was dragging my connection down, despite not being open during any testing.

 

[L&LD]

Well... I think I've fixed it. More testing to come, but, after a LOT of configuration change and test:

• I uninstalled INSSIDER, then...
• I repaired the installation of my Intel wireless client (I had previously done this a number of times with no success)

Full speed maintained throughout the tests, A+ A+ A+ over on DSL reports - Minimal QoS (145/28) which I will tweak moving forward.

I'll factory reset once more (now on 384.12) and start from scratch again. I'll continue to do my full range of 5ghz channel checks to determine the best available channel, but it looks to me like INSSIDER was dragging my connection down, despite not being open during any testing.

I really, really do not like these 'apps' when I'm setting up a network. Much more trouble than they're worth.

This is one of the reasons why I use a 'secret' SSID to do the tuning with (because I can't trust that the customer or their kids, won't be using something like this without my knowledge, somewhere in the home).

 

[FadgewackeR]

That's good enough for me... Bloody INSSIDER has cost me some time.

 

[Klueless]

I uninstalled INSSIDER

Interesting conclusion ... Thanks for sharing!

 

[Grisu]

INSSIDER is way to old and not updated, no 160MHz (or was it 80/80 or both???) support and some other misses.
Use https://lizardsystems.com/wi-fi-scanner/ instead much better. You need to register for free download.

 

[FadgewackeR]

Ok, bit of an update - this evening the performance reverted to garbage - no INSSIDER installed.

After a few other failed changes, I’ve repaired the installation of Intel Wireless Pro Set Manager and it’s gone back to normal.

I’ll uninstall that program and see if it falls over again when I get he chance, then if it is the culprit, I’ll pick it up from there.


Sent from my iPhone using Tapatalk

 

[Klueless]

Wow; if it weren't for your bad luck you'd have no luck at all ...

 

[FadgewackeR]

Wow; if it weren't for your bad luck you'd have no luck at all ...

I'm delighted that you're finding this so interesting...

After initially thinking that INSSIDER was causing the issue, it was actually the repair of the installation of the Intel driver/software that fixed it...

Hopefully I can determine exactly what the problem is this particular driver/software installation and I could let people know about it, if both they and I are so lucky.