I recently sent this to MikroTik support, but I thought I'd copy it here in case anyone else is having similar issues with the new MikroTik Audience. Basically, I recently bought four of these to test them out, and I've been struggling to get consistent wi-fi performance from them. I've tried a couple of different units in different locations, and last night I tested one out in a rural cabin with no other 5Ghz access points within radio range. Here's what I'm seeing:
1. The speeds are highly, and seemingly randomly, variable, from <1Mbps up to 400Mbps. This seems to be affected more by time than location, as I can repeat the same test in different parts of the room, or different rooms, and get largely the same results, but I can walk away for an hour then retest, and the speeds will have changed significantly. Of course, this sounds like a poor radio environment, but this test was carried out in a cabin with no other 5Ghz interference. I tried these scenarios:
i) Auto control channel, no secondary
ii) Auto control channel, auto secondary channel
iii) Fixed control channel, no secondary
iiii) Fixed control channel, auto secondary channel
...but this made no difference to the performance. Also, even when selecting a centre frequency for an 80Mhz band I had already scanned and knew to be empty, the device would continue scanning for frequencies anyway. I checked the log and there was no mention of radar being detected.
The performance problems occur both with the IPQ4019 and the QCA9984 radios. All the time, devices are shown as connected at MCS22-23. CPU on the Audience never reaches higher than 5%. The Audience was connected back to the network via a solid 1Gbps ethernet. Afterwards, I tested a Meraki MR33 in the same location, and it was giving me 300Mbps+ solidly, all the time. I appreciate this is a slightly more expensive unit, but I'm just mentioning it to give evidence to the fact that it was possible in that environment to produce a stable 5Ghz, but the Audience didn't seem to be doing that successfully. I have tried using both the stable and latest development software versions and firmware, but there is no difference.
2. Upload speed is usually significantly faster than download speeds from a device (e.g. 10Mbps down vs 250Mbps up). This is the same behaviour I have often observed with the hAP AC, and there are a few forum threads asking about this issue which weren't ever able to be resolved. I didn't see this with a Meraki MR33 when I tested it in the same location.
2. The 4th chain on the QCA9984 is never used. I noticed that MCS24-32 are not available to be configured; perhaps the software does not support using all four streams at the moment? Test device was an iPhone 11 Pro. This was a nice-to-have and I was just curious, but the device is sold as being capable of 4x4, so I think it's important that this works.
3. The QCA9984 spends an enormous amount of time scanning for radar, even in bands which do not require the mandatory 10-minute listening period. I'm guessing this is why you chose to use it only for meshing by default, so it's not necessarily a problem, but just thought I'd mention it.
I'm attaching my test configuration, but it's pretty default. I fear that some of these problems are echos of the issues people were having with the hAP AC. Has anyone else had a similar experience?
You do not have the required permissions to view the files attached to this post.