Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Gideon Kahan]

Hi folks,

This is my first post on the forum, although I have been doing some reading up on here before I joined, including trying to find a solution for this particular issue.

I work in an auditorium and we have been having an intermittent issue with our wireless mic system. Last summer we installed a set of 6 Shure ULXD4Q units, along with an array of ULXD1 and ULXD2 receivers/microphones. They are all connected to a Yamaha mixer via Dante and generally speaking it all works exactly as it is supposed to - good, clean sound and great coverage throughout the event space.

The issue we are having is that every so often, one or several of the mic/transmitter channels will sound broken, crackly in a sort of 8-bit, lo-res, crunchy kind of way and occasionally cutting off the ends of words. It seems to be an issue that only occurs during events, when we have a number of people in the room as attendees, and not at rehearsals for said events.

When it was happening during the run of the play, we had 24 body pack transmitters on, running through the same system. Most of them worked fine, sounded fine, no issues. The one or two that did exhibit the problem were not noticeably different in terms of the signals on the meters of the ULXD - either for audio level or RF signal strength, and neither of the problem units were overloading.

The only way we have solved the issue is by restarting either the receiver unit for the mic that seems to crackle, or restarting the mixer, or just generally restarting the whole system. The problematic channels will then be absolutely fine. The downsides of this short-term fix are obvious, and are of no help to us if it starts to happen during an event. We have had this happen during a run of a play using the body packs and headsets, but it wasn’t a gain structure issue because it only happened on two transmitters during the play itself while the rest of them worked fine and we tested the body pack and the headset before and after and they all worked correctly and cleanly.

It’s not a gain structure thing, because all of the handheld mics are setup the same way and all of the body packs are setup the same way and they don’t have any issues. We don’t get any audio issues with any of the other inputs going in or coming out of the mixer itself, including music or video playback or any wired inputs, and it’s only been with mics connected to the Shure ULXD receivers, connected via Dante.

I’ve used the Shure ULX units fairly extensively prior to this and alongside this on other jobs in other locations and am intimately familiar with the various gain stages on the mics, the receivers, etc. As far as that goes, I have personally checked and set the offset of each transmitter, the gain on the receivers, the input gain on the desk, etc. The majority of the time we have no issue. The units were brand new and so far I have no reason to believe that any of this is hardware failure. I’m sure this is some sort of user error that we just haven’t got to the bottom of.

Any help and advice would be greatly appreciated.

[Russell Ault]

{...} This is my first post on the forum, {...}

Welcome! Thanks for actually following the rules and using your real name as your display name!

{...} They are all connected to a Yamaha mixer via Dante {...}

First question: what are the exact models of switch you're using in your Dante network (particularly the one that the ULXD RXs are plugged into)?

{...} The only way we have solved the issue is by restarting either the receiver unit for the mic that seems to crackle, or restarting the mixer, or just generally restarting the whole system. {...}

Second question: is Dante Controller showing any errors? (Sounds like a clocking issue, which would throw errors.)

{...} I’ve used the Shure ULX units fairly extensively prior to this and alongside this on other jobs in other locations and am intimately familiar with the various gain stages on the mics, the receivers, etc. As far as that goes, I have personally checked and set the offset of each transmitter, the gain on the receivers, the input gain on the desk, etc. {...}

I doubt this is your problem, but FWIW ULXD doesn't follow any of the normal rules of RF gain-staging. TX offsets should be left to 0 dB (except in the very specific case of alternating back-and-forth between a pair of TXs with different level sources on a single channel of RX), and all gain-staging happens at the RX (I know this sounds wrong; I promise you it isn't). Using ULXD in a Dante system (where 0dFS is the same on both sides of the digital transport) the only actual gain knob in the whole system is the ULXD RX's gain control (which, conveniently, can be controlled from your Yamaha desk); every other level control (including, IIRC, the TX offset) is effectively a digital trim.

-Russ

[Helge A Bentsen]

Sounds like a clock issue, I’ve had that happen with wireless over Dante.
Make sure your mixer is set to use Dante as clock master.

[Gideon Kahan]

Thank you for the responses so far.

First question: what are the exact models of switch you're using in your Dante network (particularly the one that the ULXD RXs are plugged into)?

The switch in the control room rack is a Cisco SG300-10. It’s a 10 port gigabit managed switch. We need an upgrade/expansion because currently almost all the ports are taken up, but that’s another story. The ULXDs are daisy chained and the first one is connected to the switch, the Tio is also connected directly to the switch. We have another two Tios on the stage that are connected via ports to the control room switch, but they don’t seem to be part of the issue.

Second question: is Dante Controller showing any errors? (Sounds like a clocking issue, which would throw errors.)

Haven’t found any in Dante controller yet. We’re going to keep an eye and an ear out for when the issue next arises and then go through all these suggestions to check.

I doubt this is your problem, but FWIW ULXD doesn't follow any of the normal rules of RF gain-staging. TX offsets should be left to 0 dB (except in the very specific case of alternating back-and-forth between a pair of TXs with different level sources on a single channel of RX), and all gain-staging happens at the RX (I know this sounds wrong; I promise you it isn't). Using ULXD in a Dante system (where 0dFS is the same on both sides of the digital transport) the only actual gain knob in the whole system is the ULXD RX's gain control (which, conveniently, can be controlled from your Yamaha desk); every other level control (including, IIRC, the TX offset) is effectively a digital trim.

All the ULXD gain practices you mention are ones I’m already familiar with and have been through. All of our transmitters are set to 0dB offset. We have entirely separate scenes on both the mixer and Dante controller and channels setup depending on the number of hand held or body packs we need to use. The receivers are all set with some mild gain added, but that varies depending on other factors. Regardless, the signals coming in on the meters are all clean and not even close to clipping.

Sounds like a clock issue, I’ve had that happen with wireless over Dante.
Make sure your mixer is set to use Dante as clock master.

In Dante controller under clock status the Mixer is set as Preferred Leader. Are there any other settings or preferences to look for?

[Mac Kerr]

Thank you for the responses so far.

The switch in the control room rack is a Cisco SG300-10. It’s a 10 port gigabit managed switch. We need an upgrade/expansion because currently almost all the ports are taken up, but that’s another story. The ULXDs are daisy chained and the first one is connected to the switch, the Tio is also connected directly to the switch. We have another two Tios on the stage that are connected via ports to the control room switch, but they don’t seem to be part of the issue.

First suggestion is do not Daisy chain the receivers. Run individual CAT5 cables from each receiver to the nearest switch. If you have to add another switch for the receivers, that will only add one network hop vs the 5 hops for the farthest link in the daisy chain.

Mac

[Helge A Bentsen]

Thank you for the responses so far.

The switch in the control room rack is a Cisco SG300-10. It’s a 10 port gigabit managed switch. We need an upgrade/expansion because currently almost all the ports are taken up, but that’s another story. The ULXDs are daisy chained and the first one is connected to the switch, the Tio is also connected directly to the switch. We have another two Tios on the stage that are connected via ports to the control room switch, but they don’t seem to be part of the issue.

Haven’t found any in Dante controller yet. We’re going to keep an eye and an ear out for when the issue next arises and then go through all these suggestions to check.

All the ULXD gain practices you mention are ones I’m already familiar with and have been through. All of our transmitters are set to 0dB offset. We have entirely separate scenes on both the mixer and Dante controller and channels setup depending on the number of hand held or body packs we need to use. The receivers are all set with some mild gain added, but that varies depending on other factors. Regardless, the signals coming in on the meters are all clean and not even close to clipping.



In Dante controller under clock status the Mixer is set as Preferred Leader. Are there any other settings or preferences to look for?

Yes, make sure that sync source in the mixer is set to clock from the Dante network.

[Gideon Kahan]

First suggestion is do not Daisy chain the receivers. Run individual CAT5 cables from each receiver to the nearest switch. If you have to add another switch for the receivers, that will only add one network hop vs the 5 hops for the farthest link in the daisy chain.

Mac

We had it so that there is a direct connection to the switch at both ends of the chain, so only 3 hops at most but I get what you mean.

As I say, we need a new, bigger switch.

[Russell Ault]

We had it so that there is a direct connection to the switch at both ends of the chain, {...}

Sorry, do you mean you've created a loop? Or did you break the "chain" in half, with each half connected to the switch? If the former, try the latter (Ethernet loops are a bad thing, and will either result in total network failure or cause an appropriately-configured switch to automatically break the loop in order to prevent total network failure).

Either way, given the number of hops involved you might also want to check your Dante latency settings (particularly if you did create a loop that the SG300 has "Spanning-Treed" back into a straight chain).

{...} As I say, we need a new, bigger switch.

I'll echo Mac's suggestion to just mount a second Dante-compatible switch in the RX rack; it'll keep the cable runs neat and should cost significantly less (especially since switch prices tend to increase exponentially with number of ports).

-Russ

[Brian Jojade]

Your description of having both ends of the chain connected to the switch is most definitely a problem, if you're describing it correctly.  ie, port 1 of the switch into mic 1, connected from mic 1 to mic 2, mic 2 to mic 3, mic 3 into mic 4, then mic 4 back into the switch.  That's now created a network loop and is very bad.  Your switch will try to handle it to prevent a catastrophic network meltdown, but that may disrupt Dante traffic.  That's the first fix.

Dante can be VERY picky on networks.  As previously suggested, daisy chaining is less ideal. While it works for smaller networks, as things grow little details can make or break the network. If you've got 2 ports available and 4 receivers to work with, pair up 2 receivers to each port.  2 is better than 4.  Again, NO LOOPS! (that's part of why I don't like daisy chaining anything, as it's tough to create a loop if you only use one jack on each receiver)

You mention that your control room switch is almost full. Is this connected to another network, or is your dante network completely isolated from everything else? Because, again, Dante can be very picky with other network traffic.  If something else is on the network and decides to have a broadcast storm, that too will mess stuff up.  Completely isolated is the easiest.  Setting up separate VLANS or QoS when you must live alongside other traffic isn't difficult, but you've got to have a basic network understanding to go there.

[Gideon Kahan]

Sorry, do you mean you've created a loop? Or did you break the "chain" in half, with each half connected to the switch? If the former, try the latter (Ethernet loops are a bad thing, and will either result in total network failure or cause an appropriately-configured switch to automatically break the loop in order to prevent total network failure).

Either way, given the number of hops involved you might also want to check your Dante latency settings (particularly if you did create a loop that the SG300 has "Spanning-Treed" back into a straight chain).

Yeah it was looped - Switch port A - Receiver 1 Primary, Receiver 1 Secondary to Receiver 2 Primary, and so on until Receiver 6 Secondary to switch port B. I’ve removed that now.

I'll echo Mac's suggestion to just mount a second Dante-compatible switch in the RX rack; it'll keep the cable runs neat and should cost significantly less (especially since switch prices tend to increase exponentially with number of ports).

I’ve now temporarily redone the network connections so that all 6 ULXD receivers have an individual ethernet connection to a Netgear GS116 switch, which then has another connection out to the Cisco switch. It doesn’t strike me as being the ideal best practice, but seems to be working for now.

I think somebody got the Netgear one to replace the cisco one but it didn’t have the rack mount parts for it so it hasn’t made the transition yet. I digress.

You mention that your control room switch is almost full. Is this connected to another network, or is your dante network completely isolated from everything else? Because, again, Dante can be very picky with other network traffic.  If something else is on the network and decides to have a broadcast storm, that too will mess stuff up.  Completely isolated is the easiest.  Setting up separate VLANS or QoS when you must live alongside other traffic isn't difficult, but you've got to have a basic network understanding to go there.

Our network for our AV equipment is closed from everything else and just used for Dante. There are one to two ports that were occasionally used for a chauvet dj hub for lights but that isn’t currently connected and going forward is going to have its own switch. There are whole areas of the control room layout that predate my working here and I am (and have been) trying to simplify and tidy everything.

There are just a fair few items on the network that need to be connected:
- Mixer
- Mac Mini for Dante Controller + Virtual Soundcard
- 3x Tio Stageboxes
- 6x ULXD4Qs

11 items on an 8-port switch. That’s part of the issue. Any recommendations for a switch upgrade?

[Russell Ault]

{...} I’ve now temporarily redone the network connections so that all 6 ULXD receivers have an individual ethernet connection to a Netgear GS116 switch, {...}

The GS116 will probably make things worse (potentially a lot worse): it has 802.3az Energy-Efficient Ethernet enabled on it (with no way to disable it), which is basically a Dante-killer (unless you have a GS116E, in which case you'll be fine, just make sure to disable EEE in the switch's settings).

{...} which then has another connection out to the Cisco switch. It doesn’t strike me as being the ideal best practice, but seems to be working for now. {...}

Having more than one switch in a Dante network is basically the norm, especially on larger shows (e.g. a switch at FOH, a switch at the stage rack, a switch in the RF rack, etc.), and won't cause any problems if Dante's latency settings have been configured correctly for the maximum number of hops; is there a specific reason you'd prefer to use only one switch?

-Russ

[Brian Jojade]

I’ve now temporarily redone the network connections so that all 6 ULXD receivers have an individual ethernet connection to a Netgear GS116 switch, which then has another connection out to the Cisco switch. It doesn’t strike me as being the ideal best practice, but seems to be working for now.

STOP.

https://service.shure.com/s/article/disqualified-network-switches-for-shure-dante-devices?language=en_US

Note that the GS unmanaged switches are specifically DISQUALIFIED for use by Shure.  You WILL have problems with that configuration.

Instead of changing multiple things at once, and adding additional potential problems, simply split the receivers you have, daisy chaining just one pair, using up the 2 ethernet ports you were already using.  By making this ONE change, that will tell you if it's the culprit.

If you change 2 things and the problem still exists, you don't know if you just put in another device that can cause you additional problems.

[Gideon Kahan]

Having more than one switch in a Dante network is basically the norm, especially on larger shows (e.g. a switch at FOH, a switch at the stage rack, a switch in the RF rack, etc.), and won't cause any problems if Dante's latency settings have been configured correctly for the maximum number of hops; is there a specific reason you'd prefer to use only one switch?

-Russ

The reason for using just the one switch if possible is that apart from all the Dante kit in the control room, the two tios on the stage are in separate racks on different sides of the stage, each connected to the control room by direct ethernet ports which connect to the control room switch itself. There doesn’t seem to be any point in having an additional switch on the stage for anything with our current setup.

STOP.

https://service.shure.com/s/article/disqualified-network-switches-for-shure-dante-devices?language=en_US

Note that the GS unmanaged switches are specifically DISQUALIFIED for use by Shure.  You WILL have problems with that configuration.

Instead of changing multiple things at once, and adding additional potential problems, simply split the receivers you have, daisy chaining just one pair, using up the 2 ethernet ports you were already using.  By making this ONE change, that will tell you if it's the culprit.

If you change 2 things and the problem still exists, you don't know if you just put in another device that can cause you additional problems.

Well I eliminated that netgear trash shortly after that (although it worked without issue for events that day) and returned the entire setup to how it was, apart from the receiver layout. I have units 2, 4, and 6 daisy chained to units 1, 3, and 5 respectively, with the 1, 3, and 5 each being connected directly to the switch.

I ran into the IT guy the day after I posted last and casually asked if they had any spare switches lying around. Some building and renovation had been going on and he had just taken a couple of gigaport switches out of circulation and just gave them to me. I’ve left the original switch in place for the time being but after testing with that I switched the same layout to the ProCurve 1800-24G Switch. It’s only go 9 different connections going to it as of right now, but eventually I will connect each of the ULXDs directly to the switch.

I am trying to run everything, including the receivers into one switch which will be identical to the original network configuration except for a) the switch itself b) the receivers will not be daisy chained.

The thing is, as this issue only crops up every once in a while it’s hard to test for it other than changing one thing and waiting to see if it happens again. Everything has been fine so far.

[Jordan Wolf]

The thing is, as this issue only crops up every once in a while it’s hard to test for it other than changing one thing and waiting to see if it happens again. Everything has been fine so far.

With a nicer switch, you can probably set up SNMP alerts for certain errors (lack of network traffic, network traffic using a particular port, etc.) so that you can check the logs and see what is going on inside the network.

[Russell Ault]

{...} I switched the same layout to the ProCurve 1800-24G Switch. {...}

I'm fairly certain that switch supports 802.3az, so you'll have to make sure it's disabled for Dante to remain happy.

-Russ

[[email protected]]

The reason for using just the one switch if possible is that apart from all the Dante kit in the control room, the two tios on the stage are in separate racks on different sides of the stage, each connected to the control room by direct ethernet ports which connect to the control room switch itself. There doesn’t seem to be any point in having an additional switch on the stage for anything with our current setup.

I would highly recommend to have a switch in ALL the installations (doesn't matter if it is fixed or rental) with Dante network for few obvious reasons:
- You get less hops, which means less latency and especially less variance in latency
- If one of your equipment fails, it does not affect the other (if you daisy chain and one equipment looses it's power, fails... you'll loose all the equipment behind it)
- Using switches enables reduntant topology. Of course to have it real redundancy you'll need separate switches to both lines (Primary and Secondary)
- Using big enough switch makes it easy to add i.e a computer with DVS to system, or some other Dante device, without re-cabling.

And the main thing to look at switches when you use unicast is that they are not enabled to EEE (most of the dummy ones unfortunately are, but not all, and on all the programmable ones you can disable EEE).
If you need to go to multicast (don't go if not absolutely needed) you need to take into account things like IGMP snooping, QOS...

But basically if you run just unicast, you'll be good as long as you have
- at least 1GB switch
- EEE disabled

Then, it you start to use different brand switches etc. just remember that trunking and LAG might be handled differently and the safast way is to go "port-to-port" between different manufacturers, especially with Dante which typically does not jam the network with it's load.

Br,
-Minka-

[ProSoundWeb Community]