Issues with ITB mixing with outboard

[eirikur]

Hi guys!
In a mixerless setup, I will have to route signals out through my AD/DA to hardware units; fx/comp etc, and then back in through the same AD/DA.

What worries me is an issue probably related to my AD/DA, but I am curious how people have solved this... I do the following with a stereo track (All inputs/outputs of all equipment unbalanced):

1. Route a signal (L) out through AD/DA output channel 1, into a compressor, set compressor to bypass, signal from compressor into AD/DA input channel 1, summed together with another signal (R).

2. Right channel is left ITB, i.e. no outboard fx/comp applied, summed directly together with L and output to monitors.

The "issue" is that the signal, although the hardware unit is in bypass, becomes audibly quieter, and also appears to loose quite a bit of information in the upper mids/highs.

Why would I want to do this test, you might ask? Well, it was simply to test if there was a notable amount of latency when including hardware units into my digital setup. And the latency was just fine, but the other artifacts I discovered put me a bit off. Of course, there will be degradation when AD/DA'ing a signal multiple times, but I didn't think the degradation would be that apparent. Also, that still doesn't explain why the L signal became quieter than the R signal?

I am using a creamware card with a 2496 AD/DA... Could the level difference be due to differences in the AD/DA input/output level capabilities? And am I supposed to hear the degradation in signal quality this apparent?

Thanks for any help/thoughts!

[cold c]

Hi,

You could try routing the analogue out on your creamware card directly back into the analogue input (without the compresor) to see if the signal degrades 'significantly'.

(By the way, you can also test your audio card's loopback quality/frequency response with Rightmark audio analyzer )

If the direct loop back shows no problems then the problem is unlikely to be the audio interface or converters (though they will of course affect the signal, it should not be that 'drastic').

Another possiblility is that the bypass on the compressor is not a true bypass and and is affecting the signal when 'bypassed', but you didn't mention which compressor is being used (some poorer implementations of guitar pedals suffer a similar problem).

One further possible cause is the latency of the sound as it is converted to and from an analogue signal. This could mean that the returned signal is 'offset' by some samples which could cause a perceived drop in volume particularly if the tracks being summed are left/right of the same source. You can test if there is an audio offset issue by using a loopback again and simply playing a track on one channel out through one output and recording it back on another track in you sequencer, then look at where the recorded samples are in relation to the original.

If this is the problem, then note that some DAW software such as Cubase SX3 has a built in latency compensation for such hardware offsets, so check your software manual to see if it can do that.

Hope that helps.

[5down1up]

the ad and da stage need to be " calibrated " right . the operating levels of the units need to match each other . even when set up right , theres gear around that makes the signal louder or lower

usually adda conversion takes xtra time . only system ive heard that was able to compensate it was digis HD using their converters .

cheap converters ( prolly all converters , just the amount differs ) degenerade the sound . what you put in will not come out .

well , anyway ... hardware and itb is a cool thing . better converters make your outboard sound better of course . and its less noisy . only problem as usual ...

they are hell xpensive

[u b k]

are your channels at unity in the box? if not, you need to set the L channel at unity while spitting the signal out and pull the fader back to match R after it's recorded.

the only other thing i can think of is balanced/unbalanced issues. is the comp balanced? is the ad/da?

wait, there is one other thing i can think of: are there calibration pots on the a/d's?


gregoire
del ubik

[spacewars]

Originally Posted by eirikur

Hi guys!
In a mixerless setup, I will have to route signals out through my AD/DA to hardware units; fx/comp etc, and then back in through the same AD/DA.

What worries me is an issue probably related to my AD/DA, but I am curious how people have solved this... I do the following with a stereo track (All inputs/outputs of all equipment unbalanced):

1. Route a signal (L) out through AD/DA output channel 1, into a compressor, set compressor to bypass, signal from compressor into AD/DA input channel 1, summed together with another signal (R).

2. Right channel is left ITB, i.e. no outboard fx/comp applied, summed directly together with L and output to monitors.

The "issue" is that the signal, although the hardware unit is in bypass, becomes audibly quieter, and also appears to loose quite a bit of information in the upper mids/highs.

Why would I want to do this test, you might ask? Well, it was simply to test if there was a notable amount of latency when including hardware units into my digital setup. And the latency was just fine, but the other artifacts I discovered put me a bit off. Of course, there will be degradation when AD/DA'ing a signal multiple times, but I didn't think the degradation would be that apparent. Also, that still doesn't explain why the L signal became quieter than the R signal?

I am using a creamware card with a 2496 AD/DA... Could the level difference be due to differences in the AD/DA input/output level capabilities? And am I supposed to hear the degradation in signal quality this apparent?

Thanks for any help/thoughts!

The Creamware stuff doesn't have delay compensation for I/O like you're running.
Any system in which you're routing via DA to external gear then back to AD will
introduce some latency and requires delay compensation.
I have this problem on a much larger scale with PT Mix. The only real solution is
PT HD, which does have ADC. Otherwise it won't work correctly when you're trying to insert analog or digital external gear on I/Os.

PT HD = no more phase issues with external I/O inserts.

Time to spend money.

-spacewars

[cold c]

Originally Posted by spacewars

The Creamware stuff doesn't have delay compensation for I/O like you're running.
Any system in which you're routing via DA to external gear then back to AD will
introduce some latency and requires delay compensation.
I have this problem on a much larger scale with PT Mix. The only real solution is
PT HD

There is no need for the delay compensation to be carried out in the creamware audio hardware or in the ASIO drivers, as sequencer applications such as SL3/SX3/Nuendo3/etc already carry out this function. Furthermore, any sequencer that has automatic plugin delay compensation can extend this functionality to an external effect channel (as it 'simply' delays playback of non-delayed tracks to sync with delayed tracks, no time travel or PT HD required etc).

[spacewars]

Originally Posted by cold c

There is no need for the delay compensation to be carried out in the creamware audio hardware or in the ASIO drivers, as sequencer applications such as SL3/SX3/Nuendo3/etc already carry out this function. Furthermore, any sequencer that has automatic plugin delay compensation can extend this functionality to an external effect channel (as it 'simply' delays playback of non-delayed tracks to sync with delayed tracks, no time travel or PT HD required etc).

Yes, but in order for this to be the case, the software has to know the precise number of samples of delay introduced by the hardware that's doing the DA and AD conversion for the I/O insert. I doubt that SX and Nuendo "know" this information when it comes to I/O, because that would involve them already knowing the specs of any interface that was connected. PT HD only works with its own interfaces, so these delay figures have been precalculated and thus there is no guesswork involved. HD also calculates the delay for all TDM plugins, of course. I don't use SX or Nuendo, so I can't speak from practical experience here, but I'd be a bit surprised if this issue is handled seamlessly by those apps as it is in HD. Anyone with experience with those programs in this scenario, feel free to jump in and correct me, but the hard reality of the problem is what it is- it's just arithmetic and physics, and it's been around as long as digital mixers have existed. In fact, I believe that on the early Neve Capricorns, there was no delay compensation so this reared its ugly head even then. Imagine spending six figures on a high-end digital desk and getting phase problems when you plug in an outboard compressor and return it on a fader!

[eirikur]

Hi again, and thanks to everyone for their replies.
I have already tried sending the signal directly back to the AD/DA's input, without the hardware, to make sure that the hardware unit itself wasn't the reason for the lesser volume and general signal degredation. Also, the hardware unit I tested with initially (a RNC) is unbalanced, so all I/Os in the chain are unbalanced.

Also, the drop in volume is real, it's not simply perceived or a phase thing, since I can see the level drop in my scope mixer (appears to be 2-3 dB on the meter).

Ubik, the unit doesn't have any pots for calibration, so I am unsure how to calibrate the box. Any ideas?

I was originally trying to test the latency of the audio card to see if it would work perfectly to insert hardware compressors etc. In my setup, with my settings, the latency of I/Os should be 3ms, according to creamware's software. Given this, 6ms should be the latency difference applied to a track sent out into hardware and back. (Compared to a track sent directly out.)

Spacewars, I am not using delay compensation-able software to do this test, but I am pretty sure delay compensation works in the same way for both PT, Logic (finally ), and any other daw that has delay compensation. I don't know how they calculate the latency, but it shouldn't be too hard, given the fact that they are able to do digital mixdowns and bounces of stuff. Also, software has to be able to to this regardless of audio interface connected. So I am not too worried about that for now. (As long as one's using delay compensation enabled software, that is) Any other delay will be the same as in any analogue setup, and not noticeable, at least not to me..

Will test the Rightmark later today, to see what results I get, and post them here.

Thanks guys!

[max cooper]

Originally Posted by cold c

One further possible cause is the latency of the sound as it is converted to and from an analogue signal. This could mean that the returned signal is 'offset' by some samples which could cause a perceived drop in volume particularly if the tracks being summed are left/right of the same source. You can test if there is an audio offset issue by using a loopback again and simply playing a track on one channel out through one output and recording it back on another track in you sequencer, then look at where the recorded samples are in relation to the original.

I think there's almost no way this is NOT the problem.


The way to check is to run both L and R outs into the compressor set to 'bypass' then run them back into the A/D. If latency is indeed the problem, you should now have equal latency on both channels and any cancellation should disappear.

[Lynn Fuston]

A) You must figure out a way to make sure the level that is coming out of your card is identical to what's coming back in. Use a tone to compare the output to the input. Until you do that accurately, any sonic comparisons are absolutely meaningless. You'll only confuse people by posting your impressions.

B) Don't process only one side of a stereo signal. While it's an interesting exercise, I can't think of any reasons to ever do that in real life.

[eirikur]

Originally Posted by Lynn Fuston

A) You must figure out a way to make sure the level that is coming out of your card is identical to what's coming back in. Use a tone to compare the output to the input. Until you do that accurately, any sonic comparisons are absolutely meaningless. You'll only confuse people by posting your impressions.

B) Don't process only one side of a stereo signal. While it's an interesting exercise, I can't think of any reasons to ever do that in real life.

A) Any ideas as to how I can do this, when there is no pots or anything on the converter to level up input and output?

B) True, but as I stated earlier, I was only trying to figure out an easy way to check if the latency imposed by implementing hardware devices in my signal chain was audible to me, and the artifacts I discovered were merely a biproduct of that exercise.

[Lynn Fuston]

Originally Posted by eirikur

A) Any ideas as to how I can do this, when there is no pots or anything on the converter to level up input and output?

B) True, but as I stated earlier, I was only trying to figure out an easy way to check if the latency imposed by implementing hardware devices in my signal chain was audible to me, and the artifacts I discovered were merely a biproduct of that exercise.

There may be calibration in the software somewhere. Short of that, I have no idea.

The way to check for absolute latency is to route the output directly back to the input and then play back a single track which has an easily discernable, very fast leading edge. A click track works well. Re-record the click onto an adjacent track and then compare the two waveforms. Set the counter to SAMPLES and measure the difference between the two leading edges of the waveform. That will tell you the latency of the DA/AD path.

[cold c]

Originally Posted by spacewars

Yes, but in order for this to be the case, the software has to know the precise number of samples of delay introduced by the hardware that's doing the DA and AD conversion for the I/O insert. I doubt that SX and Nuendo "know" this information when it comes to I/O, because that would involve them already knowing the specs of any interface that was connected.

Yes any ASIO driver must report its input/output latency so the sequencer applications already know this. (ASIO applications often add the hardware latency to the ASIO buffer latency to calculate/display the total latency to the user.)

Originally Posted by spacewars

HD also calculates the delay for all TDM plugins, of course. I don't use SX or Nuendo, so I can't speak from practical experience here, but I'd be a bit surprised if this issue is handled seamlessly by those apps as it is in HD. Anyone with experience with those programs in this scenario, feel free to jump in and correct me,

It's relatively seamless in SL3/SX3/Nuendo3, the external fx channel can be set up in VST connections, the delay to be compensated can be set along with the gain compensation. The effect can then be used anywhere in any channel (such as in a chain of software and hardware fx), there is also a 'ping' button to automatically detect the total latency of the hardware chain. (see image below.)

As I understand it, it will be integrated further in a future version (further integration of the device panel for automating MIDI controlable hardware etc).

Originally Posted by spacewars

In fact, I believe that on the early Neve Capricorns, there was no delay compensation so this reared its ugly head even then. Imagine spending six figures on a high-end digital desk and getting phase problems when you plug in an outboard compressor and return it on a fader!

Ouch.

[eirikur]

Sent a 20ms sine test tone, level set to 100% out just now, and the latency is exactly 3ms, just like the creamware software said it would be. However, the AD/DA is able to output the tone just fine without distorting its output, but it distorts on the input... seems like a bad design to me, creating a AD/DA that is able to output more than it is able to receive at its input, unless there is some way to calibrate input/output. There is certainly no such option on the AD/DA box itself, and I know the creamware scope software pretty well, and haven't seen such an option.... I will try to figure out if it is at all possible to calibrate input/output... in the meantime, if anyone here is a 2496 user, or has any other idea that I might try, please let me know....

Will also try to borrow another converter from another studio to see if the signal degradation is simply because the 2496 doesn't have good enough converters to implement hardware devices in the signal chain...

[jwh1192]

hi, since you are wishing to add analog devices to a digital world ... can you try using a more analog friendly level on the way out of your hardware ... say -20 or -18 on the digital scale ... this will be more in line with what your analog gear will like to see ... full scale digital and analog devices do not play in the sandbox very well ... too much gain for an analog circuit will make it work at it mamimum all the time ... yes, it can be cool to overload some analog devices but it general good gain practices will make your gear and your final product much happier ... just my .02

peace and happy outboarding ....

also: what is the level difference at this -20 level when you do not use outboard gear ... this may be easier to read as it will not be overloading the input ... hopefully it will only be a DB or 2 of difference ... just a note: my Avalon 2055 EQ adds close to 3 DB in gain between bypass and on ...

john