Hello,
We all know, that Dante does not work over traditional Wifi networks but how about PtP or PtMP bridges in the 60GHz range?
Looking for a solution to bridge some 300-500m distance with a handful of Dante-Channels.
Am I up for a disaster or can this be done reliably? Thanks for any form of feedback.
Best Regards
HST

I'd see this done once with some UBNT with some AVIO adapters. Venue suggested its work fine. While I believe them... the use case was not something like FOH to a stage where failure be disasterous. Theoretically, 60Ghz should work.

You also do have Dante's latency setting to tweak to help. So there is some margin for error to increase the audio latency setting. So if you want to run Dante at 1ms, that would seem risky. Or in bad weather...

We use dante over 60ghz PtP, it works very well, the only thing we can't carry is the clock. We tried two audio mixers and a 60ghz link but the slave mixer often lost sync with the master clock (and was quite slow to get back into sync, about 30-60sec.). We solved it by using GPS to manage the clock sync.
As a test to verify the limits we managed to transport the audio between two locations using the resident internet connections. (about 25ms latency between the two locations)

Couple questions:
- Were you in AES67 mode?
- Did the Dante controller have a specific error on the clock and/or see a lot of jitter in clock's graphs?

One thought is by default 60GHz PtP Mikrotiks use a bond with 5Ghz. I'd remove the bond if it was being used, since failover to 5Ghz is not going to work for Dante & it one more thing in the network stack.

- Were you in AES67 mode?

No

- Did the Dante controller have a specific error on the clock and/or see a lot of jitter in clock's graphs?

Yes, the slave mixer get muted and the clock unlocked. In the clock jitter graph it was very variable (and with red columns), unfortunately I don't have a screenshot.

I was using a pair of UBNT AF60LR (with a 300mt link), so no 5Ghz backup. Latecy point to point was 0.8-0.9 ms. Do you think a pair of RBLHGG-60ad or nRAYG-60ad could perform better?

Re UBNT vs MT.... I think it be more physics, than hardware since it's a standard protocol. Only note be that Mikrotik uses a bond on 60Ghz PtP products with 5Ghz backup... For Dante, 5Ghz be worthless as backup, so using bond would add a smidgen of latency that could be avoided.

I'm pretty sure Dante's AES67-mode would fail over 60Ghz, but never tested – why I asked. AES67 uses PTPv2. Maybe it work perfectly with clock with PTPv2-aware switch in-between the 60Ghz PtP link – dunno....

But using GPS for clock with 60Ghz seems like a great call... given standard Dante's PTPv1 clocking.

I was using a pair of UBNT AF60LR (with a 300mt link), so no 5Ghz backup. Latecy point to point was 0.8-0.9 ms. Do you think a pair of RBLHGG-60ad or nRAYG-60ad could perform better?

Generally I've seen more jitter on MikroTik's and Ubiquiti's Qualcomm-based radios than on Tachyon and Ubiquiti's Peraso-based radios. That said, I thought I'd run a couple of ping tests over my various links to put my money where my mouth is.

A short test from router through AP, CPE, to remote router at short (<600m) distances gave the following results:

CRS310 -> Tachyon 301 -> 302 -> AX3 (<400m):
AX3 -> Wave Micro -> Wave Pico -> AC2 (300m):
CRS310 -> wAP 60 -> LHG 60 -> hEX POE (<600m):
CRS310 -> Gigabeam Plus x 2 -> hEX S (<600m):
CCR2116 -> AF60HD -> AF60XG -> CCR2116 (600m):
For comparison, I added a Siklu 8010 80GHz link. It did not disappoint.
CCR2116 -> Siklu 8010 x2 -> CCR2116 (600m):
Summary:
Some of the links are lightly loaded with customer traffic, probably more than the audio traffic you'd see. Likewise, my sites have a lot of 60GHz radios around, so there could be some interference causing the jittery symptoms. An indoor venue or temporary outdoor setup with no other radios around are not likely to see interference.

I didn't include the ping lists here due to brevity. But from my observation:
- Gigabeam and MikroTik 60GHz had the worst jitter (variation in latency)
- MikroTik, Tachyon, and Wave hovered just over 1ms, with the Peraso radios showing better jitter
- The Gigabeam Plus link was slightly lower, at .7-.8ms with spikes to 2-3ms
- Of the 60GHz radios, the AF60HD + AF60XG combination (2160MHz-wide channels, split 30/60) had the lowest latency at 300us-600us
- 80GHz full-duplex links can get you down to 1/8ms (125us), but at a price tag of $3-6K depending on bandwidth (and links have to be licensed in most regions)

Based on these results, I would wager a pair of AF60HD's at <500m would be your best bet. It appears the higher-bandwidth Qualcomm radios are much better quality than their lower-speed counterparts.

(Another factor in slightly higher latency on some of the links could be the difference between CPU loading on the CRS310's and AC/AX routers vs. the CCR2116's.)

Thanks @sirbryan. My knowledge of 60GHz is limited.

I do know that it's jitter that kill you for Dante, so great data.

The thing that seem limiting is there are not any MCS-like knobs to tweak – since I'm not sure negotiating a higher MCS is helpful for stability. e.g. Changing MCS has to introduce some jitter to achieve higher bandwidth, but that bandwidth isn't useful for some fixed 10-200Mb/s audio stream.