Inspired by the topic "WE826-T2 out performed by mini $12 router?".
I must admit recently I also bought a GL router (decided to get the best one) since I wanted to see how the latest OpenWrt 19.X works on a 600 MHz device prior I start make drastic changes to my dear MOFI (read WE826) router.
I read a lot of wonderful reviews at Amazon about GL-AR750s dual radio mini router (109$ at Amazon.ca), people were saying how small and lightweight it was, how great was WiFi coverage, how simple and easy was to setup VPN connection - simply put this is the best 2020 travel router (those words are not mine).
So, I bought it, setup VPN (after spending a day, I use my own VPN server running on pfSense ( Intel(R) Celeron(R) CPU N2930 @ 1.83GHz 4 CPUs: 1 package(s) x 4 core(s) and have some exprience). I don't know if I can post here a link to GL forum but I would like to do that to save myself some time from typing the same stuff twice - just scroll to the bottom of the thread (where I publish some results of my test) if interested:
https://forum.gl-inet.com/t/ar750s-open ... e/10238/19
Now my question: OpenWrt allows setting different transmit power levels. But this particular router seems to locked to certain level, plus one user mentioned that consumer grade routers cannot go higher 23dB (in Canada you are allowed to use up to 1W for 5Ghz upper channels like 149+). And some other things make me believe that a USB Wifi powered device 5V/0.5A advertised as 2W wifi adapter is something which is not very realistic?
And just to be clear - when I say 1000 mW I don't mean that device should output exactly 1000mW, this is the total: transmit power + antenna attenuation" (I read about it in some wiki.dd-wrt.com article)
Then I dare to say that no, mini 12$ router is not better than WE826 (I believe Ralink Wifi chip) when it comes to 2.4 Ghz Wifi radio, because WE826 with 2 good Wifi antenna seems to have better sensitivity (according to my observations).
And now my question for knowledgeable folks : do you know any consumer grade wifi adapter capable of outputting 1000-2000 mW? In confirmed way, like using a SDR to measure the relative signal strength or at least to monitor the changes the signal levels while changing transmit power on the router? Or does everybody simply rely on those transmit power numbers in the router GUI / documentation/advertisement/ reviews written by confused users hoping that different numbers lead to different results?
Thanks!
is WE826-T2 really outperformed by mini $12 router?
Forum rules
Please assure there is not an existing forum and topic related to your post
Please assure there is not an existing forum and topic related to your post
- Didneywhorl
- Posts: 3646
- Joined: Fri Mar 23, 2018 5:37 pm
- Location: USA
- Has thanked: 1370 times
- Been thanked: 764 times
- Contact:
Re: is WE826-T2 really outperformed by mini $12 router?
Over my head. I usually see on here that when people get into needing top power WiFi they go to dedicated access points like the Ubiquiti's and such. I honestly think the WE826s wifi is decent for me but its far lacking. My Surf SOHO MK3 router has damn good wifi power and I honestly only use it now as an access point in my garage.
Re: is WE826-T2 really outperformed by mini $12 router?
I connected both We and GL to my FIOS 150 up /120 down line, set both 2.4 GHz radios to the same channel and began running speed tests and simply using them (but not simultaneously) in real life from different client. We (MOFI in my case) slightly wins because ( I think) GL uses 2 built in antennas and my MOFI uses an 2 external 4dB antenna. Both devices while using 2.4 GHz radios are not the top performers, just average and this is to be expected.
But what I found really interesting is that GL 5Ghz radio supports bands 100-116 normally used by radar systems. It's legal to use those channels in Canada
https://www.semfionetworks.com/blog/5gh ... -in-canada
I have a couple of 3-4 year old expensive routers with AP (MU MIMO and such) but none of them allow to select those channels, they seem to be locked to US country code. In GL if I choose CA for the country code the channels 100-116 become available.
So, 5 GHz channels 38-48 are packed with different routes, 149-165 are very busy as well and there is only my GL using channels 100-106, (80 MHz). That is a nice surprise since looks like I don't see any interference and can get very decent speed like 150 down and 110 up from different places in the house. But GL does not support MU MIMO therefore the best results can be achieved only with one client connected.
What is also interesting is that GL have 2 connectors on the PCB and use short pigtails to connect 5GHz radio to the small/non removable external antennas. So, if one really wants he can install SMA connectors and use more powerful external antennas for 5Ghz radio. The problem is that the device is so small (probably 3/4 of standard sig. pack length) that its hard to find room for those SMA connectors.
But what I found really interesting is that GL 5Ghz radio supports bands 100-116 normally used by radar systems. It's legal to use those channels in Canada
https://www.semfionetworks.com/blog/5gh ... -in-canada
I have a couple of 3-4 year old expensive routers with AP (MU MIMO and such) but none of them allow to select those channels, they seem to be locked to US country code. In GL if I choose CA for the country code the channels 100-116 become available.
So, 5 GHz channels 38-48 are packed with different routes, 149-165 are very busy as well and there is only my GL using channels 100-106, (80 MHz). That is a nice surprise since looks like I don't see any interference and can get very decent speed like 150 down and 110 up from different places in the house. But GL does not support MU MIMO therefore the best results can be achieved only with one client connected.
What is also interesting is that GL have 2 connectors on the PCB and use short pigtails to connect 5GHz radio to the small/non removable external antennas. So, if one really wants he can install SMA connectors and use more powerful external antennas for 5Ghz radio. The problem is that the device is so small (probably 3/4 of standard sig. pack length) that its hard to find room for those SMA connectors.
Re: is WE826-T2 really outperformed by mini $12 router? (VPN speed test)
Over last few days I was playing with setting up OpenVPN on different devices. I am not using a commercial VPN provider (in my view it has very limited usage and people normally greatly overestimate the benefits of using a third part VPN provider but this is a different story).
For tests I used a Qotom 5 NICs mini-pc (1800Mhz 4 Core CPU) playing a role of a router running FreeBSD/pfSense 2.35/OpenVPN server (single core application) and a lot of other useful stuff. I also used 2 FIOS connections on the same circut/using the same hub (ISP side). That means that I have quite reliable connection with very low ping between end points. The idea is to set up a VPN connection from IP1 to IP2 (pfSense/OpenVPN server) and use IP2 as an internet gateway. For measuring speed I used ATT speed test (http), speedof.me, speedtest (in my view the least reliable way to measure speed in "scientific way").
Speed test results from both IPs is about 90-100 up/90-100 down. Speed could go up to 180 Mbps but not less than 90Mbps - this will be our reference point.
I used the following VPN clients:
a) pfSense 2.35 in VMware running on a powerful laptop, 1 core, 2700 Mhz;
b) GL-AR750S with stock firmware build on the top of OpenWrt 18.06 / plain vanilla 18.06 firmware (btw, it did not make any difference other than eating a lot of my time to set it up and have a false sense of security);
b)Slightly customized (by disabling all what I don't need) MOFI4500-4GXeLTE-V2(read WE826), OpenWrt 14.7 (if my memory serves me well);
VPN config: Crypto: AES-128-CBC/SHA1 D-H Params: 0 bits / peer to peer / TCP tunnel/
RESULTS (Max steady speed I could achieve). For GL-AR750S / MOFI4500-4GXeLTE-V2 I used 2 types of connections - WAN and Wifi repeater, by connecting them to another, very fast router (FIOS modem/router with NAT) - i.e I had multiple NAT but this is not a problem for OpenVPN and me personally.
a) pfSense 2.35 in VM, 1 core, 2700 Mhz - WINNER - 50/50 up Mpbs, very reliable and well manageable connection;
b)GL-AR750S with stock firmware build on the top of OpenWrt 18.06 / plain vanilla 18.06 firmware - 20 /18 up Mbps up - very surprising result!
c)MOFI4500-4GXeLTE-V2(read WE826), OpenWrt 14.7 - 13/13 up Mbps - well, this is what it is.
OpenVPN on GL-AR750S and MOFI4500-4GXeLTE-V2 seriously loaded the CPU (up to 98%), pfSense in VMware loaded the CPU up to 70%. The CPU speed of GL-AR750S is about 10%-15% higher than MOFI4500-4GXeLTE-V2 - may this explains the difference in the results.
PS: when I was debugging the VPN connection I set on GL-AR750S VPN client the debug verbosity to max (9 when 3 is default) - the speed went from 20 down/18 up Mbps to 1.5/1.5 up Mbps and spent several hours trying to figure out the reason of such speed deterioration.
For tests I used a Qotom 5 NICs mini-pc (1800Mhz 4 Core CPU) playing a role of a router running FreeBSD/pfSense 2.35/OpenVPN server (single core application) and a lot of other useful stuff. I also used 2 FIOS connections on the same circut/using the same hub (ISP side). That means that I have quite reliable connection with very low ping between end points. The idea is to set up a VPN connection from IP1 to IP2 (pfSense/OpenVPN server) and use IP2 as an internet gateway. For measuring speed I used ATT speed test (http), speedof.me, speedtest (in my view the least reliable way to measure speed in "scientific way").
Speed test results from both IPs is about 90-100 up/90-100 down. Speed could go up to 180 Mbps but not less than 90Mbps - this will be our reference point.
I used the following VPN clients:
a) pfSense 2.35 in VMware running on a powerful laptop, 1 core, 2700 Mhz;
b) GL-AR750S with stock firmware build on the top of OpenWrt 18.06 / plain vanilla 18.06 firmware (btw, it did not make any difference other than eating a lot of my time to set it up and have a false sense of security);
b)Slightly customized (by disabling all what I don't need) MOFI4500-4GXeLTE-V2(read WE826), OpenWrt 14.7 (if my memory serves me well);
VPN config: Crypto: AES-128-CBC/SHA1 D-H Params: 0 bits / peer to peer / TCP tunnel/
RESULTS (Max steady speed I could achieve). For GL-AR750S / MOFI4500-4GXeLTE-V2 I used 2 types of connections - WAN and Wifi repeater, by connecting them to another, very fast router (FIOS modem/router with NAT) - i.e I had multiple NAT but this is not a problem for OpenVPN and me personally.
a) pfSense 2.35 in VM, 1 core, 2700 Mhz - WINNER - 50/50 up Mpbs, very reliable and well manageable connection;
b)GL-AR750S with stock firmware build on the top of OpenWrt 18.06 / plain vanilla 18.06 firmware - 20 /18 up Mbps up - very surprising result!
c)MOFI4500-4GXeLTE-V2(read WE826), OpenWrt 14.7 - 13/13 up Mbps - well, this is what it is.
OpenVPN on GL-AR750S and MOFI4500-4GXeLTE-V2 seriously loaded the CPU (up to 98%), pfSense in VMware loaded the CPU up to 70%. The CPU speed of GL-AR750S is about 10%-15% higher than MOFI4500-4GXeLTE-V2 - may this explains the difference in the results.
PS: when I was debugging the VPN connection I set on GL-AR750S VPN client the debug verbosity to max (9 when 3 is default) - the speed went from 20 down/18 up Mbps to 1.5/1.5 up Mbps and spent several hours trying to figure out the reason of such speed deterioration.
- BillA
- Posts: 1230
- Joined: Sun Dec 01, 2019 6:46 pm
- Location: USA
- Has thanked: 219 times
- Been thanked: 327 times
- Contact:
Re: is WE826-T2 really outperformed by mini $12 router? (VPN speed test)
56387f wrote: Wed May 06, 2020 11:31 pm Over last few days I was playing with setting up OpenVPN on different devices. I am not using a commercial VPN provider (in my view it has very limited usage and people normally greatly overestimate the benefits of using a third part VPN provider but this is a different story).
I agree on using your own VPN server for security reasons, that is if you really have to use a VPN.
Nowadays with most sites using "https" everything is encrypted, a VPN is just an overkill in my opinion and slows down the connection significantly. To hide even the web address, one could simply use a secure DNS server in the router, without the additional/double encryption overhead. Moreover, if a public VPN server gets hacked, there goes all your data (hackers are much more interested in a central server than individual IP's). It was great before "https", but I just don't see it that useful anymore... just my 2cents.
Re: is WE826-T2 really outperformed by mini $12 router?
I use it when I travel abroad.BillA wrote: Thu May 07, 2020 8:30 pm It was great before "https", but I just don't see it that useful anymore... just my 2cents.
One reason is to use VPN while is in a different country is to access my local network (documents, the printer, the camera to check the cat etc) in simple and secure way.
Another reason is to make different payments and such - nowadays banks (and even google - it really drives me crazy when it says that the computer is not recognized - the computer is same but it moved 10000km in 2 days !!). And every time when one has to make a bank payment etc he has to answer several security questions, solve different puzzles, prove that he is a human and this and that. With VPN everything stays the same and this is important.
Plus, I recently noticed that sometimes my speed over VPN (T-Mobile) is higher than without VPN but running VPN over mobile network while roaming and using dynamic IP addresses is not easy - the external modem IP changes very often comparing to cable/DSL/FIOS.
- BillA
- Posts: 1230
- Joined: Sun Dec 01, 2019 6:46 pm
- Location: USA
- Has thanked: 219 times
- Been thanked: 327 times
- Contact:
Re: is WE826-T2 really outperformed by mini $12 router?
56387f wrote: Fri May 08, 2020 11:08 pm I use it when I travel abroad.
One reason is to use VPN while is in a different country is to access my local network (documents, the printer, the camera to check the cat etc) in simple and secure way.
Another reason is to make different payments and such - nowadays banks (and even google - it really drives me crazy when it says that the computer is not recognized - the computer is same but it moved 10000km in 2 days !!). And every time when one has to make a bank payment etc he has to answer several security questions, solve different puzzles, prove that he is a human and this and that. With VPN everything stays the same and this is important.
Plus, I recently noticed that sometimes my speed over VPN (T-Mobile) is higher than without VPN but running VPN over mobile network while roaming and using dynamic IP addresses is not easy - the external modem IP changes very often comparing to cable/DSL/FIOS.
To get a static IP anywhere in the world, you could use a regular cheap proxy with a lot less headache setting it up (no encryption keys and configuration to worry about... occasionally I use a proxy from BestProxyAndVPN.com to select any geographic location I want on the fly). Coincidentally, a regular proxy can also bypass YouTube video throttling on all carriers, no need for a full blown VPN (which was a pleasant surprise). You can even use a proxy manager plugin for the browser to set up rules in order to specify certain addresses only (ex. *.GoogleVideo.com for YouTube).
Nowadays virtually all sites are using https connection, and by using a VPN you're just double encrypting the connection, which is an overkill in my opinion, both in terms of cost and performance hit. Each encryption layer uses an unbreakable AES256 (or stronger) key, so before https/TLS, a VPN was the only way to insure privacy, however those days are over, course VPN companies would like to convince you otherwise. The reason why it seems you're getting higher speed tests on Tmobile with a VPN, is because they specifically filter speed tests and video streams. But hey, you can use whatever you prefer. lol
Re: is WE826-T2 really outperformed by mini $12 router?
I should respectfully disagree about VPN in general - it was used in the past, is being used now, will be used in future.
In other words VPN is used to _join_ 2 local networks (LAN). As the result and as the direct benefit the remote VPN client can use the local gateway / the local computers and other local network resources.
Nowadays some people run on their laptops many different OS at once (virtualization, you just need enough RAM, storage and CPU power), such as Windows, Linux, FreeDBS, Solaris (Intel) and even MacOS (Apple restricts running virtualized MacOS on certain hardware), Android and OpenWRT. Can you imagine 5 or 6 OS running simultaneously on your laptop? And when one does it is has a fully fledged virtualized network which should be managed as such.
I am not talking about the people who use 3 party commercial VPN to "protect their privacy" while they are browsing internet from Starbucks - this is a totally different phenomenon.
HTTPS has it own problems (for example, openssl, 2014, Heartbleed):
https://en.wikipedia.org/wiki/Heartbleed
Even now it has a list of found vulnerabilities so this is not 100% perfect. But at the same time nothing that normal /typical internet user should be extremely worried about.
That's not what I do - I do the following: "use VPN while is in a different country is to access my local network (documents, the printer, the camera to check the cat etc) in simple and secure way."BillA wrote: Sat May 09, 2020 12:00 am Nowadays virtually all sites are using https connection, and by using a VPN you're just double encrypting the connection, which is an overkill in my opinion (both in terms of cost and performance hit).
In other words VPN is used to _join_ 2 local networks (LAN). As the result and as the direct benefit the remote VPN client can use the local gateway / the local computers and other local network resources.
Nowadays some people run on their laptops many different OS at once (virtualization, you just need enough RAM, storage and CPU power), such as Windows, Linux, FreeDBS, Solaris (Intel) and even MacOS (Apple restricts running virtualized MacOS on certain hardware), Android and OpenWRT. Can you imagine 5 or 6 OS running simultaneously on your laptop? And when one does it is has a fully fledged virtualized network which should be managed as such.
I am not talking about the people who use 3 party commercial VPN to "protect their privacy" while they are browsing internet from Starbucks - this is a totally different phenomenon.
HTTPS has it own problems (for example, openssl, 2014, Heartbleed):
https://en.wikipedia.org/wiki/Heartbleed
Even now it has a list of found vulnerabilities so this is not 100% perfect. But at the same time nothing that normal /typical internet user should be extremely worried about.
- BillA
- Posts: 1230
- Joined: Sun Dec 01, 2019 6:46 pm
- Location: USA
- Has thanked: 219 times
- Been thanked: 327 times
- Contact:
Re: is WE826-T2 really outperformed by mini $12 router?
56387f wrote: Sun May 10, 2020 7:42 pm I should respectfully disagree about VPN in general - it was used in the past, is being used now, will be used in future.
That's not what I do - I do the following: "use VPN while is in a different country is to access my local network (documents, the printer, the camera to check the cat etc) in simple and secure way."
In other words VPN is used to _join_ 2 local networks (LAN). As the result and as the direct benefit the remote VPN client can use the local gateway / the local computers and other local network resources.
Nowadays some people run on their laptops many different OS at once (virtualization, you just need enough RAM, storage and CPU power), such as Windows, Linux, FreeDBS, Solaris (Intel) and even MacOS (Apple restricts running virtualized MacOS on certain hardware), Android and OpenWRT. Can you imagine 5 or 6 OS running simultaneously on your laptop? And when one does it is has a fully fledged virtualized network which should be managed as such.
I am not talking about the people who use 3 party commercial VPN to "protect their privacy" while they are browsing internet from Starbucks - this is a totally different phenomenon.
HTTPS has it own problems (for example, openssl, 2014, Heartbleed):
https://en.wikipedia.org/wiki/Heartbleed
Even now it has a list of found vulnerabilities so this is not 100% perfect. But at the same time nothing that normal /typical internet user should be extremely worried about.
Well, now that you've specified your exact intent, that makes sense.
I'm all about simplicity, K.I.S.S. to more exact, because life is already complicated enough. lol