|
Hardware Synth Wierdness
|
View this Thread in Original format
| johno27 |
Hey all,
Having a really strange problem which is affecting my mixdown/master ability. Not sure if this is common place or what exactly to do about it.
Here is my attempt (length) explanation:
If we setup a software synth with a single oscillator, no fx, no eq and select a saw wave,
send this out to the mixer and record the result back into Logic you get a waveform pretty much
as you'd expect with a saw-wave, centered around the 0 line. If you now normalize this audio
file both peaks and dips reach the maximum amplitude correctly (IE: the wave form was at least 99%
balanced around the zero line).
From this we know that we’re not having cable/phase issues. All testing done with multiple sets of cables, either via mixer or directly into the audio interface.
We have several hardware synths, routed directly into our audio interface (motu 24i/o)
for tracking via Aux channels within Logic. (Nord Rack 3, Virus TI, JP8080).
When we perform the same test using any of these hardware synths, single oscillator saw wave, no fx, eq etc
and record this into Logic the wave is not balanced around the zero line. There are major spikes/dips
that cause the amplitude average to be offset negatively. It varies differently from synth to synth as the oscillator shapes are not all exactly the same.
We've checked the audio file for DC offset (none found), tried using a low and hi cut to ensure that
no sub harmonics or high overtones could cause these spikes and nothing seems to help.
The jp8080 seems to be the worst of the lot, and by increasing the high eq on the jp8080 itself seems to make the spikes even worse.
Trying different oscillators waves seems to result in different amounts of this problem, for example:
a sine wave on the nord 3 centered perfectly, a square wave was off but only by a small amount in the positive direction.
The jp8080 saw and supersaw oscillators definately seem to be the worst affected.
Once this is put into the mixdown, the whole thing is affected as such and it means that the track is almost
un-masterable. Any attempt to gain/limit the track causes it go into gain reduction before you're actually
getting any real gain reduction.
http://gomobo.com/img/untitled.jpg (zoom in of a mixdown showing the spikes in the lead sound).
http://gomobo.com/img/untitled2.jpg (overall view of the mixdown showing how it makes the whole track unbalanced
and almost useless for mastering).
Any ideas?
Thanks
John |
|
|
| Eric J |
Is it possible there is a problem with the MOTU 24IO? DO you have the latest drivers for the MOTU? I have my setup with a 24IO exactly the same way and I have never noticed these problems.
Do you have another audio interface you can plug the synths into to determine if it is a problem with the audio interface? |
|
|
| johno27 |
Hi,
Unfortunately not. Although sending the soft synth out and recording it back in doesn't exhibit any problems, it only seems to be when using the h/w synths. It's really puzzling me, in fact it's driving me insane! hehe
I was wondering if anyone with one of the same h/w synths could try it and see what happens. Like the jp8080, no delay, no lfo, 2nd oscillator at 100%, shape saw, filter max 12db lp, no filter env. shortish amp envelop with minimum attack, release, no sustain shortish decay to give a plucked sound, sample it and see what the wave looks like. |
|
|
| Eric J |
| Looks fine on this end. I just performed the test. |
|
|
| johno27 |
Damn thats odd.. and not cool at all. I've got the latest MOTU pcie424 driver installed, osx 10.4.9, logic 7.2.3.
I'm going to have to find another audio interface to test with..
my reasoning was that all signals are effective a sum of sinusoids (fft theory) therefor for every dip there should be an opposite and equal peak, so all signals should normalise equally -/+ amplitude over time.
On a side note, you're using the 24i/o as well, I find that if the inputs are set to +4dbU that the synths come in with very little level, but if I set the inputs to -10dbV they go into clipping on the inputs (I guess one could just lower the synth/patch level in this case) but I'm not sure which is the best option as one wants the best signal to noise ratio. Which setting do you use for you synth inputs on the 24i/o ?
Cheers
John |
|
|
| johno27 |
Found a 2408 which i'll connect up to the same pcie 424 card and try with that tonight.
Do you think it could be possible that a cable problem could induce some sort of phase shift that might cause this result?
Bearing in mind that they're neutrik plugs, somner cable balanced plugs on the audio interface side, unbalanced plugs on the synth side with the cold wire joined to the shield. |
|
|
| Eric J |
| quote: | Originally posted by johno27
Damn thats odd.. and not cool at all. I've got the latest MOTU pcie424 driver installed, osx 10.4.9, logic 7.2.3.
I'm going to have to find another audio interface to test with..
my reasoning was that all signals are effective a sum of sinusoids (fft theory) therefor for every dip there should be an opposite and equal peak, so all signals should normalise equally -/+ amplitude over time.
On a side note, you're using the 24i/o as well, I find that if the inputs are set to +4dbU that the synths come in with very little level, but if I set the inputs to -10dbV they go into clipping on the inputs (I guess one could just lower the synth/patch level in this case) but I'm not sure which is the best option as one wants the best signal to noise ratio. Which setting do you use for you synth inputs on the 24i/o ?
Cheers
John |
It depends on what you are referring to. I am using the CueMix software to set the gain levels on each channel, but I didn't see anywhere in the interface the be able to set the inputs to -10 or +4. This was odd to me because even my lowly Delta 66 card had this option. Generally I have the gain cranked all the way up on my 8080 and my JV1080 as these synths seem to have very low output, especially the 8080. Where are you see that option?
| quote: |
Found a 2408 which i'll connect up to the same pcie 424 card and try with that tonight.
Do you think it could be possible that a cable problem could induce some sort of phase shift that might cause this result?
Bearing in mind that they're neutrik plugs, somner cable balanced plugs on the audio interface side, unbalanced plugs on the synth side with the cold wire joined to the shield.
|
Well, you said that you had tested it with different cables, so I would think that rules out a cable problem.
My hunch is that if there is indeed a problem with the MOTU, its something physical with the TR inputs in the breakout box. If you are using soft synths then it never touches the breakout box when its being recorded. It would only go through the hardware on the PCI 424 card.
There is a slight difference between your setup and mine, as you are on OSX and Logic, where as I'm on a PC with Cubase. I can't see a situation where OS or sequencer would make any difference, but just something to keep in mind. We're using different drivers. Maybe also try recoding into BIAS Peak or similar audio application, to rule out some weirdness with Logic.
Keep me posted, I'm curious as to what might be going on. |
|
|
| johno27 |
Hey,
When I go into the PCI424 setup under OSX there is an advanced tab, in there you can switch the inputs between +4 and -10 in groups of 8.
so there's a (1-8) +4 / -10
(9-16) +4 / -10 ... etc
If you use them at +4, and change the gain in cuemix, does it remember this everytime you restart the machine, or do you have to adjust this input channel gain everytime?
I'll try recording into some other audio editors and test that too.
With the softsynth I send it out to an analog mixer and then back into the TRS input jacks on the interface, so it does go through the same cable outboard side of things as do the hardware synths. I've tried sending the h/w synths (jp8080 specifically) to the mixer first then into the soundcard, result was the same.
I'll keep you posted on what I find as I'm just getting more and more confuzzle about it :)
Cheers |
|
|
| Eric J |
| quote: | Originally posted by johno27
Hey,
When I go into the PCI424 setup under OSX there is an advanced tab, in there you can switch the inputs between +4 and -10 in groups of 8.
so there's a (1-8) +4 / -10
(9-16) +4 / -10 ... etc
If you use them at +4, and change the gain in cuemix, does it remember this everytime you restart the machine, or do you have to adjust this input channel gain everytime?
I'll try recording into some other audio editors and test that too.
With the softsynth I send it out to an analog mixer and then back into the TRS input jacks on the interface, so it does go through the same cable outboard side of things as do the hardware synths. I've tried sending the h/w synths (jp8080 specifically) to the mixer first then into the soundcard, result was the same.
I'll keep you posted on what I find as I'm just getting more and more confuzzle about it :)
Cheers |
Yeah, I dont think I have that option in CueMix, but I'll look again. Were using different versions of CueMix, as I have the PC version, but it wouldn't suprise me if they left it off the PC version. CueMix does remember all my settings, though, even after a restart. |
|
|
| johno27 |
It's a tab called "Interface Options" for me under the PCI setup application, not cuemix.
I wasn't able to get the other sound card in last night to test with, so as soon as I've tried that I'll let you know. Other than that I've run out of ideas on this one. :) |
|
|
| camsr |
| waves CAN be offset in one direction enormously, without DC offset. Take a regular sine wave and rectify it, and dc filter it. the spikes where the sine was inverted will reach a higher voltage than the rounded wave crests. Now try bumping up the high EQ and the spikes get even hairier. It's just the nature of the wave, and where the steep tangent lines reside. If you want to try to control this behavior, i suggest a multiband compressor on brickwall settings, Or you can try an ALLPASS FILTER. tweak the freq till the wave looks good. |
|
|
|
|
