UniFi AP AC Lite - OpenWrt or stay with stock firmware?

I'm looking to install a single UniFi AP AC Lite which I already have, sitting unused. I just want to use its 5GHz radio w/ multiple SSIDs and VLANs and also 802.11r for quick transitions between APs (the other "AP" is actually a Netgear R7800 running OpenWRT in AP mode).

I have a few questions please:

  1. Should I stay with stock firmware or flash it to OpenWRT? For a single AP, dealing with the UniFi controller is overkill (and the mobile app is terrible just very basic settings). But that might be the true and tried firmware version that has many deployments. Not sure how well OpenWRT works on it and what version to use. The attraction of OpenWRT is that I can just use the WebUI/ssh which I'm familiar with from my R7800.

  2. I'd like to use the same SSIDs/VLANs as on my R7800, to improve coverage in a corner. Any ideas if 802.11r will work well between these 2 devices? The backhaul is wired, no mesh or repeater, though not sure if a mesh helps in any way with a wired backhaul and what I'd need to do, if it even has a chance with these 2 different devices.

  3. I assume that I want different channels, e.g. use ch 36 on the UniFi and ch 149 on the R7800. I'm also thinking of turning down the power on the UniFi, I just need some better coverage in a corner (which luckily happens to have an Ethernet drop).

Try openwrt, you can always go back to stock through tftp method. I have the AC LR and i prefer it with openwrt, i think it works better with it than with stock.

you are asking in openwrt forum if you need stay on stock or flash openwrt. i can 't believe that.
this is OPENWRT

2 Likes

I'm asking for real life experiences with this device in particular, in a perhaps not common setup. It doesn't mean that OpenWRT is best for all devices that it can boot on, see Broadcom devices, for example. Or that it supports all features stock firmware can, for example hardware acceleration on some chipsets.

Which OpenWRT version do you use? Just staying with latest stable or are there useful fixes in snapshots? Does anyone maintain any optimized builds for these APs? Thanks.

I used both master and stable and performance is similar, so just try them and use the one you prefer

1 Like

What's the proper procedure to install OpenWRT on these devices? I have the latest UniFi firmware on it and it sounds like I have to downgrade it to some older version? Any hints please?

First you need to downgrade the stock firmware to a certain version within the controller (or manually with sftp and ssh) before you can even flash OpenWRT with it. You should find plenty of information about that on the net.

I use two of them with a recent master build and the new DAWN controller. The performance is great (same as stock as far i can tell) and you can use wpa3, something that Ubiquiti only implemented recently in a closed beta.

I moved them away from stock because Ubiquiti tend to release broken firmwares. I nearly bricked both of them in a recent update...

So go with a stable and you should be fine.

2 Likes

What kind of throughput do you get with OpenWrt? On stock without doing any NAT it can shift north of 300 Mbps, can it do the same on OpenWrt?

I believe I've seen a bit above 300Mbps in iperf tests with the client close to it, but it goes down quickly at some distance. I don't consider these to be powerful APs, more like 1 room coverage.

1 Like

I installed openwrt on my Unifi AC LR device. everything is fine but I can reach a maximum speed of 400mbps. I can go up to 500-600mbps with stock software. I tried with different ath10 drivers, nothing changed. for this reason I went back to the stock software.
The openwrt version I have installed is: 19.07.7 ath79
Stock software I have installed: 4.3.28

What do you think is the reason why the performance of the stock software cannot be reached with openwrt? Is it due to a shortage of CPU? Is it because the Atheros driver is bad?

Generally, I always use OpenWRT stock firmware, not closed source software full of bugs.

Beware that if frequencies are overlapping with neighborhood APs, your AP might switch from 40Mhz size to 20 Mhz to avoid interference. So any testing needs to be conducted in the same situation at the same place. OpenWRT might follow standards more closely. In one of my houses with no interference, I can connect at 866 Mb/s without problem with a GliNet B1300.

1 Like

I run 21.02-SNAPSHOT, r15863-3feef9c555 on mine with the mainline ath10k dirver and firmware (not -ct). I can reach 400-500Mbps close to the AP, but it drops quickly as I move away from it. Good enough for my use cases, stable, FT works to roam between APs, simpler to manage than the UniFi controller.

I can connect at 866 Mb/s without problem with a GliNet B1300.

I believe he's talking about actual throughput not link speed.

yes i talked about the actual download speed. I'm testing it by downloading from my nas server. i am using intel ax200ngw wifi card. Connecting 867mbps. I tried the near the unifi. Never exceeded 400mbps with openwrt. With stock software, it fluctuates above 500mbps. I will try again when openwrt 21 stable version is released.

I did not understand what is better with the stock software. Wifi driver better? Or is it doing cpu acceleration?

I set wifi settings to 80mhz, and disable 802.11b legacy devices. I'm disabling short-gi and using wpa2-aes. What should I do for better performance with openwrt?

For comparable, real throughput measurements, lets agree on a measurement. I recommend:

Server side should be a sufficiently fast device with at least a wired 1G Ethernet connection at the LAN port of the AP:
$ iperf3 -s

Client side, Upload measurement:
$ iperf3 -c <IP or name of server> -t0 --parallel 5

Client side, Download measurement:
$ iperf3 -c <IP or name of server> -R -t0 --parallel 5

Otherwise, without agreeing on a certain measurement, our results are not comparable.

For mobile devices there are Apps that bundle iperf3 for iOS and other platforms as well, which allows us to even compare those devices.

Here are some numbers using iperf3 as above, about 6 ft from the AP:

Upload
[SUM] 0.00-22.24 sec 1.28 GBytes 496 Mbits/sec receiver

Download
[SUM] 0.00-19.90 sec 1000 MBytes 422 Mbits/sec sender

I call it good enough for what a single device needs. But the AP doesn't have range or does well through walls. From a different room it drops to 50-80 Mbps. That's fine, you just use multiple APs.

It's because ubiquiti has access to qualcomm sdk for enterprise which means they have their own daemon for the wireless stack and L2 bridge bypass. Qualcomm added value is this software stack, not necessarily hardware. If you go up one gen to ath10k, the hardware gets integrated even more into their own daemon and L2 bridge bypass with the NSS cores. Mac80211, which openwrt uses, is very compatible and enables neat features such as fq codel but chokes on mips and even weak arm cores.

I too have setups where i run stock ubiquiti because it can support much more clients and benchmark well. However, targeted DSCP tagging in CS1 or CS2 can make bufferbloat go to 250ms or more. "Bad wifi" as the leyman term but 2.4ghz can support 50 clients all hammering with mix traffic load.

Similarly, there are setups like home environment where I need bufferbloat on wifi to be the minimum and even DSCPs in CS1 and CS2 range wont incur 200ms bufferbloat. In fact, everythere is smooth 15-60ms range thanks to "make wifi fast" on ath9k and ath10k from the fq codel in mac80211. Good wifi. BUT, and the big but, is it chokes with client load with mix traffic around 20 clients and it'll start dropping @ 2.4ghz, with 5ghz around 45.

Tradeoff is the name of the game.