Signal Strenght 374.38.1-em vs 374.35_4-sdk5

[Iceberg]

Maybe it can be of some interest.

I have tried some firmwares to see if there was any difference between versions about wireless range and strength.

NVram was cleared and setting were inserted manually after firmware change.

Everything else was left at default.

Tests made using Android Wifi Analizer repeated 2 times.

Values in -dBm max and min values observed during a period of 2 minutes for every point of observation.

Of course this is just a test, made with an entry level Android Smartphone, that may give same results for everyone.


----38_em-35_4

1 - 47/49 35/37
2 - 54/57 41/43
3 - 68/71 66/69
4 - 64/70 69/71
5 - 54/62 55/58
6 - 54/64 52/55
7 - 70/72 57/59
8 - 68/70 68/72
9 - 88/91 85/88
10 - 77/79 67/68
11 - 79/85 80/81
12 - 82/85 77/79
13 - 82/83 78/79
14 - 71/73 68/70
15 - 71/78 63/67
16 - 77/85 77/84
17 - 62/63 55/58
18 - 60/63 62/66

 

[reddice]

I too tried different firmwares even using stock but I settled on 374.35_4-sdk5. I will stick with it as it gives me the best wireless performance. Unless a new sdk5 build comes out I will stick with it for the life of the router. Why break something that is already working.

 

[jlake]

When I tested EM vs sdk5, I just took my ipad outside my house to see which one had the greatest range (while still being able to provide 6Mbps ookla speed test). Sdk5 won by about 6 feet in my unscientific, but real world tests on 2.4ghz.

 

[L&LD]

Iceberg,

thank you for the data points, but using a smartphone as you mention is not a good indicator of performance. Nor is signal strength an indicator of throughput (performance).

It is a measure of range though and some may find it useful.

To counter your results; in my testing (RT-N66U, correct?) the 38_2-em and v2239 Beta db results were lower with the latest firmware - yet they also provided the highest throughput I have seen in my network.


Just mentioning this because a connection without any 'meat' behind it is as good as no connection.


jlake, I also appreciate your informal testing too - but I have to ask, do you normally use your ipad outside?

 

[L&LD]

I too tried different firmwares even using stock but I settled on 374.35_4-sdk5. I will stick with it as it gives me the best wireless performance. Unless a new sdk5 build comes out I will stick with it for the life of the router. Why break something that is already working.

No more sdk5 builds as per RMerlin.

The latest v2239 build is an eye opener for the 5GHz band.

What you'll be missing from using an older build is more than the 'cost' of 'breaking' it (once in a while) to find out.

 

[reddice]

No more sdk5 builds as per RMerlin.

The latest v2239 build is an eye opener for the 5GHz band.

What you'll be missing from using an older build is more than the 'cost' of 'breaking' it (once in a while) to find out.

I don't care anyway. As long as my wired and wireless is strong as it can be and stable is all that mattes. When it comes to drivers and new versions are not always better. I use a two year old wireless driver for my laptop because the all the newer ones I get constant wireless dropouts even dropping out completely but the last good old version works fine.

 

[Iceberg]

Iceberg,

thank you for the data points, but using a smartphone as you mention is not a good indicator of performance. Nor is signal strength an indicator of throughput (performance).

It is a measure of range though and some may find it useful.

To counter your results; in my testing (RT-N66U, correct?) the 38_2-em and v2239 Beta db results were lower with the latest firmware - yet they also provided the highest throughput I have seen in my network.


Just mentioning this because a connection without any 'meat' behind it is as good as no connection.


jlake, I also appreciate your informal testing too - but I have to ask, do you normally use your ipad outside?

My test is not scientific, I know that there is no perfect correlation between dBm and throughput.

I'm more interested in increased range because I need internet outside so I made a test to find which firmware seems to be better for this.

 

[L&LD]

My point is that if you don't test for throughput, the dBm number may be irrelevant - even if it shows you connected.

 

[RogerSC]

I too tried different firmwares even using stock but I settled on 374.35_4-sdk5. I will stick with it as it gives me the best wireless performance. Unless a new sdk5 build comes out I will stick with it for the life of the router. Why break something that is already working.

Seems to do better with IPv6, as well as having great wireless...I agree, 374.35_4-sdk5 is a good firmware version to stick with. The only reason that I can think of to switch off of it would be if security fixes are needed.

 

[CJDias]

My point is that if you don't test for throughput, the dBm number may be irrelevant - even if it shows you connected.

What would be a good way to test for throughput? Would a an ookla speedtest run from a laptop be reasonably indicative?

I too have been on the fence, hesitating from upgrading from 374.35_4-sdk5 which just works!

 

[L&LD]

It depends on your ISP speeds. If they're high enough to test your throughput, sure, that may be enough.

But even with a 100Mbps/100Mbps up/down ISP connection, I would be testing with a real file copy off your fastest (RAM disk or SSD based) secondary system connected with an Ethernet cable. If for nothing else as a sanity check against the Ookla speed tests (and the real test of your wireless network, imo).

When I was testing the Asus beta v2239 and saw the dBm values sink in inssider, I too almost reverted back to 374.38_2-em right then. However, I am so glad I decided to test the throughput anyways - the results spoke for themselves.


Edit:
And, after reading something new and then taking my own advice (test/challenge each setting, no matter how much you think you 'know' what works best), I have more than doubled my 5GHz throughput (all while inssider and Ookla speed tests showed, essentially, nada).

http://forums.smallnetbuilder.com/showpost.php?p=101034&postcount=51



.

 

[RMerlin]

When testing Wireless throughput, I recommend sticking to LAN-only tests, to eliminate any additional variable associated with WAN, or the CPU load on the router if you don't have HW acceleration working. What I usually do here is I use either Lanbench (easy to use, with a GUI) or iperf (very good benchmarking tool but command-line only). I run one end on my desktop wired over Gigabit Ethernet, and the other end runs on the laptop over wifi.

 

[hang10z]

Quick Question, how do I check my sdk version? wl ver?

 

[RMerlin]

Quick Question, how do I check my sdk version? wl ver?

Tools -> Sysinfo page. Driver version is shown there, 5.xxx = SDK5, 6.xxx = SDK6.

 

[L&LD]

When testing Wireless throughput, I recommend sticking to LAN-only tests, to eliminate any additional variable associated with WAN, or the CPU load on the router if you don't have HW acceleration working. What I usually do here is I use either Lanbench (easy to use, with a GUI) or iperf (very good benchmarking tool but command-line only). I run one end on my desktop wired over Gigabit Ethernet, and the other end runs on the laptop over wifi.

I agree that this would be the most consistent testing. But for me, it would also mean running my file copy tests too as this is what I primarily use my wireless network for (besides internet, of course).

http://forums.smallnetbuilder.com/showpost.php?p=101027&postcount=11


If I'm testing changing any parameters (as I did in the link included in the link above; changing and testing different 5GHz channels), then the relative differences are all I need to base my decision on.

The absolute highest transfer speeds that Lanbench indicates may not match or even correlate to the actual file copy tests that I do daily.

 

[CJDias]

I installed the 374.38.2-em build on my RT-N66U. Did some before and after testing using robocopy to transfer a 500MB file off a shared drive on a PC connected via physical LAN to a laptop on wireless. Speeds on 2.4G and 5G didn't drop (after I re-tweaked the channels after factory reset) and I would daresay improved slightly based on my un-scientific testing (10-15%)!

Are there any new features of this build apart from the Openvpn stuff? Not that it matters -I just get the good feeling of running the latest stuff

Thanks Merlin as always for the great builds!