A TP-Link CPE510 (v1.1, client mode, openwrt 19.07.3) negotiates only 54 MBit/s when it talks to
another CPE510 (v1.1, access point, latest TP-Link Pharos fw). Some client data:
Please test it under load (iperf, iperf3, flent, a big file copy between wired and wireless devices, etc), don't rely on the stated speed (and keep in mind that an idle link will reduce the link speed and that you'd get ~half of the theoretical throughput at most).
Antennas are well (I know this device -- I have 7 APs and 21 Clients up and running :-)).
And as I said: All other stuff left unchanged, I get 300 MBit/s links and 100 MBit/s real data speed when using PharOS.
There is little information on PharOS + MAXtream technology, we cannot tell. You are mixing PharOS and OpenWRT, so the bottleneck could be anywhere.
It does not seem clear whether you are using Ethernet cables + switch for your main backbone or if you are using only radio. I guess you are using only radio.
If you are using only radio (no wired links) make sure to use different radios for main backbone of APs (ex : 5Ghz) and wireless clients (ex : 2,4 Ghz). You may also look at mesh settings, but your device has only 8MB flash from memory and you may need more space to install wpad-mesh-openssl.
My impression is that a well-configured mesh could be faster than FarOS. Of course, if you want to increase speed to 300Mb/s, use only wires as your backbone. It is the ultimate solution for speed, far better than PharOS. You only need to use stock OpenWRT and enable fast roaming.
If you plan to reach 100 to 300 MBits, you need at least a wired network for some stations to be used as a backbone.
Any setup based on radio only (with server AP/client AP, retransmission or mesh) will devide your bandwidth by two or more, which is normal. As written previously, make a real speedtest under load using for example iperf3. Don't trust announced speed. To make it simple, keep in mind that all clients and AP using the same ESSID and channel share the same bandwidth. The CPE510 has only one radio and does not support ac mode, so speed is very limited. PharOS claims to limit data collisions, but this is done natively as APs and clients are listening to each other and not emitting at the same time. With PharOS, you share 300Mbits on all APs and clients running on a single wireless channel. This is not much under load.
If you are planning hundreds of simultaneous users, the only solution under load (for example in a camping or in a public space) would be a wired backbone and APs with same ESSID operating on different channels with roaming enabled and low emission intensity. You don't have a choice. Please note that PharOS can probably operate on wired network, which seems normal.
Of course, for personal or family use, you may use wireless as your backbone. No problem.
We can skip backbone issues -- the AP is connected by cable to the LAN and the client (openwrt) is connected by cable to a PC. No other clients on the AP. MAXtream is shown as N/A on the AP. Please see above regarding a "real speedtest".
Don't want to sound rude but I am doing networking for 30+ years so we can assume "the rest" to be OK. If you think that running the AP with openwrt as well should help I can try this. It is definitely some issue with OpenWRT in combination with an PharOS AP (not judging whom to blame :-)).
I've been quite happy with the CPE510 and PharOS over years (given the fact that it's just consumer crap)
but I've some minor things I'd like to improve....
Thanks. You are telling us the important information : you are using wired as a backbone.
Don't use client AP mode if you have a wired network (as I can read from your settings above). Your backbone should be only on wired network. Client/master APs devide bandwidth by two, this is math.
Try using OpenWRT only APs in Master mode with same ESSID operating on different 5Ghz channels without overlapting and enable fast roaming in LuCi. Lower emission to avoid APs and clients listening to each other (this might be the tricky part). Disable DHCP server on each AP.
Based on experience over load, you might in some places use two CPE510 configured on different non-overlapting channels pointing in the same direction.
This will give you maximum speed over load.
To test real speed, you may use iperf3:
Somewhere near your main router : iperf -s
On your wireless client: iperf3 -c server_address
The advantave of PharOS is that it might help you organize the sharing of radio bandwidth. In a normal AP master setup, each AP operating on a frequency listens to each other and waits for no emission to begin transmit. This is not the case when using PharOS as each AP is informed by PharOS when to begin and stop transmission. Therefore wireless bandwidth is better used. There is no such wireless controller mode in OpenWRT.
I have just one AP (at least here in the lab) and one client. As I said, the AP is currently PharOS. There is no "client AP mode" -- it is just called "Access Point". However, I'll try an OpenWRT AP next week if time permits...
You don't need PharOS with only two APs operating on different non-overlapting frequencies. PharOS is only usefull when runnig several APs on the same frequencies, i.e. in a large and upscalable design.
Just use two OpenWRT APs in master mode, with non-overlapting radio channels (no collision), same ESSID and wire them to the main router. If one of your APs is your router, link it to the second AP using wire. Disable DHCP server on your second AP. Bridge LAN and WLAN on each AP. Enable Fast roaming on each AP.
Be careful with choosing the AP channels. If your neigbourg is on channel 100, you share the same bandwidth...
Maybe my English is just too bad... I have only ONE AP and ONE client. I use(d) PharOS since it is the default fw that comes with the CPE510. I don't compete with any other 5 GHz AP in my test envirnoment.
I replaced PharOS (the default fw) on the client in order to try OpenWRT. The speed went down from 300 to 54 MBit/s (or 100 to 22 Mbit/s actually usable bandwith).
A TP-Link CPE510 (v1.1, client mode, openwrt 19.07.3) negotiates only 54 MBit/s when it talks to another CPE510 (v1.1, access point, latest TP-Link Pharos fw).
So you have two CPE510 or not? From my understanding, the first CPE510 is used as your main wireless controller using PharOS as controller and AP in master mode. The second CPE510 is using AP client mode. So you are not using wires but only WIFI to link the controller and the client AP. Are you using wired links between the two APs? I guess not.
You may use the first AP in master mode. Set the second AP in client mode. Install iperf3 on each AP and mesure wireless network speed.
This kind of setup will never give you good network speed and is not expandable. The two APS (server and client) share the same radio freqencies and the same bandwidth. It is better to link the two APs using wires and have them operate of different frequencies with same ESSID and roaming enabled. This is a basic setup supporting multiple wireless clients.
If your only clients are connected using wires to the Master AP and the client AP, then your setup is correct. On the converse if you are planning to provide WIFI, it is a wrong topology.
Also, in a professionnal environment, if you are planning to link site A with site B, don't use only WPA2 as it has been broken and no solution is working, you may only slow down attackers (crack attack). Use a VPN to secure the link over the APs.
You don't get me. I have ONE AP. And I have ONE client (or call it station or whatever). BOTH are CPE 510. I have FULL speed (300/100 MBit/s) between them when using stock TP Link fw PharOS. Speed dropped to 54/22 MBit/s when replacing PharOS with OpenWRT on the client CPE 510.
Of course I DON'T have a wired link between both CPE 510. If I had a wire, I could sell both CPE510 on ebay.
PharOS uses a wifi driver running a proprietary TDMA scheme which may have limited compatibility with standard clients. That is a question for TP-Link not here. If you connect a smartphone etc. to the PharOS AP do you get higher than 54 Mb?
Tried with two clients which are 802.11 n/ac compatible. Both do 54 only. So you are probably
right assuming the original FW does something incompatible.
I will switch the AP to OpenWRT as well and see if things are better if AP and client are OpenWRT...
I have two factory / stock CPE510 v1.1s talking to each other in a PtP configuration. The max I can get on the download (AP -> Client is what I'm calling the download) is 30mbps which is very likely a 54mbps link. It advertises a 240/300mbps on TX/RX, but it is clear that it does not support anywhere near that much.
The AP is connected to my LAN. My LAN has 200+ mbps coming in. The cable going to the AP is the same kind of shielded CAT6 that I'm using elsewhere on my LAN that supports full gigabit speeds locally as well as the 200+ mbs coming in from my provider.
The Client is 298 feet away, has a similar signal and noise reading, and this link is forced forced to 802.11n.
I have taken my spare (third CPE510) which is a V3 and used it to perform a wide variety of testing. Conclusively, I can say that these garbage CPE's will NOT download to a client PC connected directly to the Client CPE at anything more than 30mbps. The built in speed tests on the radios claim to be downloading at between 120 to 150 mbps, but when I actually load the link up, it's a whopping 30mbps every single time.
Yes all the basics are covered. The antennas are aligned and pointed at eachother. The noise is at around -98 and the signal is at around upper -30's to low -40's depending on transmit power I choose. The VERY simple PtP configuration should be getting a massive amount of speed between the two points. The build in speed test function seems to thing there is 214 mbps download and 166 mbps upload on the radio link, but I can't even get the 80mbps that the physical interface supports.
Amongst all the various testing I did, I found that the physical interface on the V1.1 does not support the full 100mbps, it doesn't support the ~90mbps that it should even with TCP overhead. It comes in at less than 40 mbps on the physical interface. I replaced the V1.1 AP with the V3 CPE and repeated a wide battery of tests. Conclusively, I still cannot get more than about 30 mbps (often less) 298 feet away. The V3 supports about 80mbps on the physical interface connecting it to my LAN. The other end of the link still only downloads at 30mbps or less.
I'm really glad I found the OP's thread here. I'm VERY sorry I don't have any potential fixes or things to suggest, but I have done extensive testing with the stock firmware and found these CPE510's to be garbage. Even V1.1 -> V3 or V3 -> V1.1 connections are not any better. I apologize for bumping an old thread, but I hope others considering these terrible CPEs will instead consider something much better and not that much more expensive.
Well, I am getting 100 MBit/s (measured with iperf or FreeBSD's netstat 1) with the stock fw between AP and Client. I've mostly a mix of 1.0 and 1.1 but a few 2.0 are also among them.