I am looking to replace an old N wireless bridge to my workshop. I am seeing the Archer 7's available real cheap on Ebay. I mostly see A7 V5.x and C7 V2.
Comparing the versions I think I am seeing:
- The C7 V4 being the fastest processor
- The C7 V1 only having 8MB flash
- The A7 & C7 V5 being reported as hard to flash.
There are lots of threads, not read them all. I prefer newer hardware all else being the same. I've done TFTP flashes before, not really intimidated. Not clear if the issue applies to older Versions that are on later firmware (latest V4 is 12/19)
Are any of these versions better or worse than another?
unsupported 5 GHz wireless.
v4 has a faster stock clocked processor. Problem is that it has 1 less CPU port (ethernet).
v2 can be overclocked with a custom bootloader to 1ghz.
Thanks, missed that one.
As best I can tell they all have 4 LAN Ethernet ports. Am I missing something?
CPU ethernet port, not switch ethernet port.
https://youtu.be/wf9JvfF7bro?t=669 is an example. That one shows two CPU ports for the switch, same as the v1-3. v4 and above has one CPU port only.
Also C7 V5 has only one USB port, instead of two ports found in earlier versions.
@neheb OK, so does this mean that the CPU has 1 port that feeds all the LAN ports and another that feeds the WAN? What is the theoretical and practical implications as opposed to a V2 or V3. Trying to learn here, thanks you for this detail and taking the time to highlight the specific portion of the video.
What you're talking about is port allocation. How it's actually done I'm not clear on. Examples:
eth0/eth1 with link aggregation.
AFAIK, swconfig uses either option 1 or option 3 depending if the switch supports it. The DSA patchset (still WIP) uses option 2.
Basically, the difference is how much bandwidth the switch can handle. With an extra CPU port, it's 2 gigabits compared to 1.
There are better options today than an Archer C7/A7. The Linksys EA6350 V3 (only the V3 is supported by OpenWrt) can be had for $25-$30 on ebay. Despite its quirks, I think it's a far better choice than an Archer C7/A7. I replaced an Archer C7 V2 with one to improve my WiFi speed.
The EA6350v3 supports faster 5 GHz than the Archer A7/C7 despite having only two antennas (and how many clients have more than that anyway?), because it has Wave 2 WiFi (Archer A7/C7 is Wave 1) support. The four core 710MHz ipq4018 SoC ARM CPU on the EA6350v3 is considerably faster than the single core Archer MIPS CPU despite its slightly slower clock rate. It also has double the memory and flash of the Archer A7/C7.
There are a couple quirks.
The radio calibration file that comes with OpenWrt for the EA6350v3 delivers terrible 2.4 GHz performance. This is easily fixed by replacing the file /lib/firmware/ath10k/QCA4019/hw1.0/ board-2.bin with /files/etc/calibration/pwr/board-linksys_ea6350v3.bin (rename it to board-2.bin) from NoTengoBattery's optimized build. I last pulled the replacement file from his openwrt-1.11.oc.tar.gz custom build archive (it's an older version of his custom build). Or just use NoTengoBattery's optimized build.
Do not use luci to edit the VLAN set up (it will save it wrong on this device) and do not touch or change the default LAN eth0 and WAN eth1 assignments. If you set up vlan's, put your vlan's on eth0.20, eth0.30, etc. and do it by editing the network file directly. There is something about the switch in the ipq401x SOC that requires its default vlan assignments to be left alone.
That said, the Archer A7/C7 is not without its issues:
The Archer C7/A7 is still a usable, albeit dated all in one router design if you already have one and its performance meets your needs, but if I were starting out with a new (or new used) device, I would look elsewhere. The EA6350v3 is but one alternative.
I don't agree because its the openwrt official build that ships unstable ct wifi drivers with its image instead of the stable non-opensource non-ct wifi drivers. non-ct has proven stability to me, especially since 21.02.0 and offers more throughput (ca. 50 MBytes/sec. max.). I've bought c7 v5 devices a few months ago and all work fine. v2 may have had some reboot requirements to fix problems after > 40 d uptime.
ref State of TP-Link Archer C7v2|v5 in 2021
That was already one of the two links I referenced. My intent was mainly to point out the A7/C7 benefited from some care and feeding as well. Though swapping out an ath10K package is certainly easier than changing up files from the command line, so there is that.
I don't disagree the Archer A7/C7 can be made serviceable with OpenWrt. It has been a very good all in one WiFi router design for a very long time and tons of them have been sold over the years. If its performance meets the requirements, it will do the job as well as anything else and quite inexpensively too.
However, tons of them are also being replaced. It would not be routinely available used on ebay in large numbers for $20 shipped otherwise. No ath10k software package swap out will ever give it Wave 2 WiFi hardware support and there are faster CPU's available today for handling SQM, VPN, file server, etc. duties in comparably inexpensive hardware.
This is easily fixed by replacing the file /lib/firmware/ath10k/QCA4019/hw1.0/ board-2.bin with /files/etc/calibration/pwr/board-linksys_ea6350v3.bin (rename it to board-2.bin) from NoTengoBattery's optimized build. I last pulled the replacement file from his openwrt-1.11.oc.tar.gz
Can you confirm the board-linksys_ea6350v3.bin calibration file works on any release of 19.08, 21.02?
When I previously tried, the 2.4Ghz after 1hour slow to 4mbps, is this still the case?
Yes. The board-linksys_ea6350v3.bin calibration file works with 19.07 and recent versions of OpenWrt just fine. I am using the file now on snapshot r17482 (more recent than 21.02) and have previously used this same file on 19.07 without issues.
It's been some time since I tested performance. The EA6350v3 has been up for ~3 weeks. Nothing special on the WiFi setup beyond turning the power down from 30 to 28 dBm on both radios so that they aren't "shouting" so much louder than the clients in the house can transmit back to them anyway. The 5GHz for the EA6350v3 is on channel 149 and 80 MHz width. The 2.4 GHz is on channel 11 and 20 MHz width. I just tested 2 locations with iperf3:
A) A thin client with a monitor display blocking the desktop wifi antennas and with a couple walls or a mostly open, but tortuous "U" hallway between the desktop and EA6350v3 some distance away (unscientifically measured as around 17 paces). I know the thin client has an Intel AX200 WiFi card, since I put it there I would call this a fairly challenging medium to long distance location.
B) From an Acer Swift 1 laptop right next to the EA6350v3.
A) 187 Mbps
B) 423 Mbps
A) 68 Mbps
B) 73 Mbps
I would not recommend it generally (a wired connection would be best), but I use location "A" over 5GHz to tune my ER-X gateway SQM settings to tame buffer bloat - mostly because I'm lazy and that's the computer I use most often, so I'm already there. I don't notice much difference between location A and a wired set-up elsewhere in the house the few times I've bothered using the wired set up, so the EA6350v3 WiFi seems to be doing just fine without adding any noticeable latency. I'm using the "stock" ath10k-ct that comes with OpenWrt by default.
Thank you for the insight and heads up