combined latency in IEM with 01v96

[brian maddox]

okay, not sure this qualifies as varsity since it involves 01v96s...



i'm running a hybrid IEM system using 3 01v96s to generate 15 stereo mixes with a 4th used as a drum submixer and vocal mic processor/router.



so today on my ride home it suddenly occurred to me that i might be dealing with some latency issues and just hadn't heard it yet.



so here's the question. how much latency? all the vocal mics go through one 01v96 [analog-adat in, adat out-analog out] and into another [analog-in, adat out-analog out] before hitting the IEM transmitters. any idea how much latency the double trip is causing?



also, all my other inputs are split across the three consoles by simply looping through each console using the adat digital ports [i've got the dual adat card in each desk]. so what difference does it make if it's looping through, but staying digital...



not sure i can much about this regardless, but it'd be a nice thing to know...



thanks...



brian 'delayed reaction' maddox

 

[jeffsw6]

Re: combined latency in IEM with 01v96

There is latency introduced by converting from the console's native digital format to ADAT, as well as D/A and A/D delay for the sub-mixes which are riding analog patches between the consoles. If Yamaha specifies the latency for ADAT->console or console->ADAT, I cannot find it.



The LS9 manual makes mention of this only once, on page 209, There is no limit to the number of units that can be cascade-connected, but the signal delay from each cascade-slave will increase according to the number of units between it and the cascade-master.



Certainly the problem does exist with your configuration. Inaudible problems are often not worth solving.

 

[Tim McCulloch]

Re: combined latency in IEM with 01v96

Hi Brian-



The multiple AD/DA conversions will add to the latency. The ADAT i/o will add a tiny bit, on the order of a couple of samples in each direction.



Yamaha says latency is <1.6ms from analog input to stereo (presumably analog) output at 48khz clocking.



You can measure these things with Smaart, SysTune or other dual channel FFT analyzer, too. Do you have, or have access to such things? I only have 1 01v96 to work with, so I can't replicate your set up.

Have fun, good luck.



Tim Mc

 

[brian maddox]

Re: combined latency in IEM with 01v96

Hi Brian-



The multiple AD/DA conversions will add to the latency. The ADAT i/o will add a tiny bit, on the order of a couple of samples in each direction.



Yamaha says latency is <1.6ms from analog input to stereo (presumably analog) output at 48khz clocking.



You can measure these things with Smaart, SysTune or other dual channel FFT analyzer, too. Do you have, or have access to such things? I only have 1 01v96 to work with, so I can't replicate your set up.

Have fun, good luck.



Tim Mc

well, my copy of smaart is on floppies and says 'JBL smaart' on it. so it's a bit dated...

i've been lazy lately and haven't kept up with the FFT tools, especially since i switched to all fruit based computer products.



but i know folks that have such things. i think i'll see if i can swap a favor...



brian

 

[jeffsw6]

Re: combined latency in IEM with 01v96

You can measure these things with Smaart, SysTune or other dual channel FFT analyzer, too. Do you have, or have access to such things? I only have 1 01v96 to work with, so I can't replicate your set up.

I thought for a moment that you actually would be able to replicate it using only one 01V96 and looping ADAT right back into the console. This wouldn't work because the ADAT input data would not be synchronized to the word clock, right?

 

[Tim McCulloch]

Re: combined latency in IEM with 01v96

Dunno about the WC, but it's the multiple AD/DA conversions looping their analog way around that concerns me most.



I think digital to digital should have a latency of exactly 1 sample in each direction, but I hedged my bet

I'm probably wrong, too... but I'll stick my neck out of this one.



One way to measure the digital transmission/cascade latency would be to measure from the analog in of the first ADAT AD converter to the analog out of the last console's ADAT DA. The ADAT devices have published latency specs that can be factored in. There are probably better ways, but I'm having a brain cramp figuring out how to get an ADAT pipe into my laptop...



Let's see what the forum has to say!



Tim Mc

 

[brian maddox]

Re: combined latency in IEM with 01v96

Dunno about the WC, but it's the multiple AD/DA conversions looping their analog way around that concerns me most.



Tim Mc

yeah, i should add that nearly all my AD/DA conversions are being done by the venerable 'B-word ada8K'. not sure how quality of converters factors in to the overall equation either...



brian

 

[Christian Tepfer]

Re: combined latency in IEM with 01v96

Hi Brian-



The multiple AD/DA conversions will add to the latency. The ADAT i/o will add a tiny bit, on the order of a couple of samples in each direction.



Yamaha says latency is <1.6ms from analog input to stereo (presumably analog) output at 48khz clocking.



You can measure these things with Smaart, SysTune or other dual channel FFT analyzer, too. Do you have, or have access to such things? I only have 1 01v96 to work with, so I can't replicate your set up.

Have fun, good luck.



Tim Mc

Not to forget the matrix, that uses a good bit of time to process all the audio data from input to output.



When having an audio interface with ADAT I/O, one should be able to measure the 01V96 latency using ADAT I/O and the 01V96 latency using analog I/O with one desk.

So there's 2 desks, 1x analog I/O, 1x digital I/O -> The sum should be the wanted end result.



Without measuring, I would say you have a combined latency of about 2.5ms @48kHz, 1.3ms@96kHz



brian, at what sample rate do you run the desks?

 

[Douglas R. Allen]

Re: combined latency in IEM with 01v96

7 years ago when I first purchased my 01v96 I measured 1.2 ms input to output.48khz. This was with no eq anywhere. 2.4 to 3 ms would be a good guess.

The latency of 1 unit is in the manual way in the back. I don't have my manual right here but I believe it shows different latency according to what you are running the board at.



Douglas R. Allen

 

[brian maddox]

Re: combined latency in IEM with 01v96

Not to forget the matrix, that uses a good bit of time to process all the audio data from input to output.



When having an audio interface with ADAT I/O, one should be able to measure the 01V96 latency using ADAT I/O and the 01V96 latency using analog I/O with one desk.

So there's 2 desks, 1x analog I/O, 1x digital I/O -> The sum should be the wanted end result.



Without measuring, I would say you have a combined latency of about 2.5ms @48kHz, 1.3ms @96kHz



brian, at what sample rate do you run the desks?

you know, this forum makes me feel dumber and dumber all the time...

i have no idea what rate i'm running the mixer at. how lame is that. i never thought to check. just figured you plug it in and it does it's thing at 96K and converts everything going in and out. which in hindsight seems incredibly naive' of me...



so, now that i've sufficiently flogged myself, i think i'll actually try to find out. that'd be a nifty thing...



brian

 

[Scott Helmke]

Re: combined latency in IEM with 01v96

well, my copy of smaart is on floppies and says 'JBL smaart' on it. so it's a bit dated...

It's not like the definition of a millisecond changed or anything. If you can find a computer with a floppy drive that version of SMAART should be able to tell you the latencies involved.



I think the real question here is, ''is anybody complaining''. Most of the issues you'll find with latency and IEMs will have to do with vocalists, and you won't be able to hear them until you start singing through the system yourself and hearing the mix between IEM signal and ''leaked through the skull'' signal.

 

[Christian Tepfer]

Re: combined latency in IEM with 01v96

Not to forget the matrix, that uses a good bit of time to process all the audio data from input to output.



When having an audio interface with ADAT I/O, one should be able to measure the 01V96 latency using ADAT I/O and the 01V96 latency using analog I/O with one desk.

So there's 2 desks, 1x analog I/O, 1x digital I/O -> The sum should be the wanted end result.



Without measuring, I would say you have a combined latency of about 2.5ms @48kHz, 1.3ms @96kHz



brian, at what sample rate do you run the desks?

you know, this forum makes me feel dumber and dumber all the time...

i have no idea what rate i'm running the mixer at. how lame is that. i never thought to check. just figured you plug it in and it does it's thing at 96K and converts everything going in and out. which in hindsight seems incredibly naive' of me...



so, now that i've sufficiently flogged myself, i think i'll actually try to find out. that'd be a nifty thing...



brian

Brian,



I personally think that when I stop learning things I am practically dead, so don't feel bad but good instead.



Now I'm curious what the measured latencies will be. I try to find 2 01V96 that I can measure.

Let's see if my equation is correct (01V96 ADAT I/O + 01V96 analog I/O = analog input -> 01V96 -> ADAT output -> ADAT input -> 01V96 -> analog output)

 

[Christian Tepfer]

Re: combined latency in IEM with 01v96

I just stumbled over 2 01V96, so here you are with the latencies:



01V96 analog channel in/Aux out

44.1kHz: 1.31ms

48kHz: 1.19ms

88.2kHz: 0.63ms

96kHz: 0.56ms



(group out adds a little latency, not measurable, around .01ms)



01V96 analog in -> channel -> aux -> ADAT -> 01V96 channel -> aux out -> analog out



44.1kHz: 1.9ms

48k: 1.73ms

96kHz: 1.38ms



01V96 analog in -> channel -> aux -> analog -> 01V96 channel -> aux out -> analog out



44.1kHz: 2.6ms

48kHz: 2.4ms

96kHz: 1.15ms



So what's interesting: the analog double conversion is faster than ADAT at higher sample rates, slower at lower rates.



EDIT: a lot of people are using external A/D to raise the usable channel count. Be careful when using internal and external A/D, as there is a difference in latency:



Focusrite @48kHz: 1.31ms

internal A/D @48kHz: 1.19ms

 

[Art Welter]

Re: combined latency in IEM with 01v96

well, my copy of smaart is on floppies and says 'JBL smaart' on it. so it's a bit dated...

It's not like the definition of a millisecond changed or anything. If you can find a computer with a floppy drive that version of SMAART should be able to tell you the latencies involved.



I think the real question here is, ''is anybody complaining''. Most of the issues you'll find with latency and IEMs will have to do with vocalists, and you won't be able to hear them until you start singing through the system yourself and hearing the mix between IEM signal and ''leaked through the skull'' signal.

Very true. Stage musicians have been dealing with time of flight latency from stage monitors and side fills that are anywhere from 5 milliseconds to 30+ without much issue. Put a couple ms delay in ears or headphones and sing and it is a different story.

Makes cymbals ''swishy'' too

.

 

[brian maddox]

Re: combined latency in IEM with 01v96

well, my copy of smaart is on floppies and says 'JBL smaart' on it. so it's a bit dated...

It's not like the definition of a millisecond changed or anything. If you can find a computer with a floppy drive that version of SMAART should be able to tell you the latencies involved.



I think the real question here is, ''is anybody complaining''. Most of the issues you'll find with latency and IEMs will have to do with vocalists, and you won't be able to hear them until you start singing through the system yourself and hearing the mix between IEM signal and ''leaked through the skull'' signal.

to your first point, true enough. except i long ago 'drank the koolaid that steve was selling' and the thought of touching a pc without at least thick rubber gloves makes me very queasy.



to your second point, true enough x 2. however, unlike most of us here, i actually DO get a chance to sing into the monitor rig i'm running, since i'm the music director of the band AND the sound guy. and until recently i was having significant pitch issues and have been looking for the culprit anywhere i could. including the always possible 'lack of talent' place... [ thread on 'another forum']



at any rate, i don't think that my pitch issue has[d] anything to do with latency. i think it was just level based. with a little mix tweaking, it's all but disappeared. so yeah, this is largely an academic exercise...



but i do like working my brain muscles too...



brian

 

[brian maddox]

Re: combined latency in IEM with 01v96

I just stumbled over 2 01V96, so here you are with the latencies:



01V96 analog channel in/Aux out

44.1kHz: 1.31ms

48kHz: 1.19ms

88.2kHz: 0.63ms

96kHz: 0.56ms



(group out adds a little latency, not measurable, around .01ms)



01V96 analog in -> channel -> aux -> ADAT -> 01V96 channel -> aux out -> analog out



44.1kHz: 1.9ms

48k: 1.73ms

96kHz: 1.38ms



01V96 analog in -> channel -> aux -> analog -> 01V96 channel -> aux out -> analog out



44.1kHz: 2.6ms

48kHz: 2.4ms

96kHz: 1.15ms



So what's interesting: the analog double conversion is faster than ADAT at higher sample rates, slower at lower rates.



EDIT: a lot of people are using external A/D to raise the usable channel count. Be careful when using internal and external A/D, as there is a difference in latency:



Focusrite @48kHz: 1.31ms

internal A/D @48kHz: 1.19ms

this is awesome stuff. thanks a million for taking the time to do this.



with these figures i can pretty much calculate any latency i have in my system regardless of the signal path. given the relatively small numbers, i doubt that any of these are causing a problem for me. but it's great information to know if for no other reason than i can eliminate a possibility from the list whenever an issue crops up regarding vocalists and their 'my mix sounds weird' comments...



owe you one. or two. thanks again...



brian

 

[Brian jojade]

Re: combined latency in IEM with 01v96

Sound delay is about .9ms per foot. If using floor monitors, that means there is about a 5ms delay between the speaker and the singers ears. I've never had anyone complain about that.



Just a point of reference whether it matters.

 

[Christian Tepfer]

Re: combined latency in IEM with 01v96

Sound delay is about .9ms per foot. If using floor monitors, that means there is about a 5ms delay between the speaker and the singers ears. I've never had anyone complain about that.



Just a point of reference whether it matters.

When using conventional monitoring I rarely hear complaints, when I do it's mostly drummers that are annoyed by too much spill from the too wide sidefill systems so they have to ask for more volume in their drumfill and become even more detached.

Note to festival providers: please think twice about sidefills. I can understand using gear that is sitting in the shop but a 120 degree line array? Please...



Its different when a singer is using IEM because there are 2 paths that mix in the head:

- bone path

- actual monitor signal



Depending on the latency the result can be very different to what the engineer is listening to. Well there are a lot more variables but latency really counts when singers are on IEM.

 

[Christian Tepfer]

Re: combined latency in IEM with 01v96

I just stumbled over 2 01V96, so here you are with the latencies:



01V96 analog channel in/Aux out

44.1kHz: 1.31ms

48kHz: 1.19ms

88.2kHz: 0.63ms

96kHz: 0.56ms



(group out adds a little latency, not measurable, around .01ms)



01V96 analog in -> channel -> aux -> ADAT -> 01V96 channel -> aux out -> analog out



44.1kHz: 1.9ms

48k: 1.73ms

96kHz: 1.38ms



01V96 analog in -> channel -> aux -> analog -> 01V96 channel -> aux out -> analog out



44.1kHz: 2.6ms

48kHz: 2.4ms

96kHz: 1.15ms



So what's interesting: the analog double conversion is faster than ADAT at higher sample rates, slower at lower rates.



EDIT: a lot of people are using external A/D to raise the usable channel count. Be careful when using internal and external A/D, as there is a difference in latency:



Focusrite @48kHz: 1.31ms

internal A/D @48kHz: 1.19ms

Side note: @96kHz the ADAT channel count is only half. So only 4 channels on one ADAT port.