1 (edited by SonicExplorer 2019-10-13 18:49:06)

Topic: Recording Tracks don't NULL properly, anyone ?

Hi,

Guys, could really use some help on this one, I'm not quite as sharp in the digital realm as some of you younger guys out there...

Using FF400 with Sonar (I know, a bit old) and a Kemper Profiler.  Running SPDIF to Kemper (as master) I reamp a direct (dry) guitar track twice and it results in a pair of tracks that usually don't NULL when one is phase reversed.  Maybe 1 out of 5 times it will work, the rest fail. Does anyone have any idea why this might be?  How might I narrow down where the problem is (Sonar, RME or Kemper) ?  Can I simply connect a SPDIF cable from output back to input in order to remove the Kemper from the testing equation?  Or, if anyone has a Kemper (regardless of DAW or interface) and could try this test it would be helpful in maybe eliminating that variable.  Just don't forget to turn Constant Latency On.

Sonic

Re: Recording Tracks don't NULL properly, anyone ?

'Don't null, I guess I'd ask, by how much?
Presuming reamping means miced re-recording a direct track through a speaker -twice, and nothing else changed [mic position for one], I'd zoom into the resulting alignment of the two tracks? Even the slightest shift in the round trip would through it off. (clue.. Hard pan the two a equal volume. If the image shifts from center, that's time alignment.

3 (edited by SonicExplorer 2019-10-13 23:44:53)

Re: Recording Tracks don't NULL properly, anyone ?

mixsit wrote:

'Don't null, I guess I'd ask, by how much?
Presuming reamping means miced re-recording a direct track through a speaker -twice, and nothing else changed [mic position for one], I'd zoom into the resulting alignment of the two tracks? Even the slightest shift in the round trip would through it off. (clue.. Hard pan the two a equal volume. If the image shifts from center, that's time alignment.

No mics involved, reamping in this instance refers to reamping through the Kemper (which is a sophisticated amp modeler so whatever goes in should be consistently coming back out).

Anyone know if it is safe to connect a SPDIF cable from output back to input on a FF400 for a loop back, or will this toast the interface?   I can't imagine what would happen if the RME remain in auto-slave mode as it would, in this scenario, be detecting itself I assume?   My instinct says it isn't safe to connect a cable in this way, but I'd like to hear from the forum....

Sonic

Re: Recording Tracks don't NULL properly, anyone ?

Hi there!

I think that this is standard behaviour, as it is a modeller. It might be digital but it models the behaviour of analog equipment. And analog equipment, like tube amps and the like, are not linear in their behaviour.
For example, if you reamped the same signal through a tube amp, twice, would then the two tracks null eachother? I think not.
And you 'd need that in a recording, because the sound becomes fuller.

I don't have a Kemper, so this is just based on my understanding of complex modelling gear.
Unless I didn't understand SonicExplorer's post quite well.

RME Gear: Digiface USB, HDSP 9632

Re: Recording Tracks don't NULL properly, anyone ?

MetalHeadKeys wrote:

Hi there!

I think that this is standard behaviour, as it is a modeller. It might be digital but it models the behaviour of analog equipment. And analog equipment, like tube amps and the like, are not linear in their behaviour.
For example, if you reamped the same signal through a tube amp, twice, would then the two tracks null eachother? I think not.
And you 'd need that in a recording, because the sound becomes fuller.

I don't have a Kemper, so this is just based on my understanding of complex modelling gear.
Unless I didn't understand SonicExplorer's post quite well.

Conceptually I agree - but in reality we are talking about digital devices connected via SPDIF. So if I understand correctly this should be a case of exact duplication more likely than not.  And the fact I am able to get tracks to NULL every few times reinforces that, otherwise they would probably never NULL.  Random variance would inevitably surface somewhere in every track, but that's not the case when I am able to get a pass that NULLs - the entire track NULLs.

Re: Recording Tracks don't NULL properly, anyone ?

Yes, you re right..! I 'll think some more about it, do some more research and try to figure it out!

RME Gear: Digiface USB, HDSP 9632

7 (edited by MetalHeadKeys 2019-10-14 02:32:55)

Re: Recording Tracks don't NULL properly, anyone ?

Well, after reading this:

https://www.kemper-amps.com/forum/index … p;pageNo=1

and the official Kemper page, we can be sure that nobody, actually, knows how the profiler works, except mr Kemper himself! smile

Anyway, if it's not much of a hassle, you could try reamping by having everything turned off, on the kemper patch(amp model, effects and speaker emulation). It would be like reamping the DI through a DI, just so to see that there is no problem with your SPDIF connections. That should null, I think! And we 'll take it from there!
And other members with kempers(or no kempers) could help as well !

RME Gear: Digiface USB, HDSP 9632

Re: Recording Tracks don't NULL properly, anyone ?

Probably the kempler has some random behaviour to simulate analogue circuits?..

www.yila.co.uk

9 (edited by SonicExplorer 2019-10-14 18:55:16)

Re: Recording Tracks don't NULL properly, anyone ?

I'm sort of leaning toward not being randomness because if there was randomness, at least in the conventional sense, it would always be present at some point in every track. But that's not what is happening, I either get a near NULL (audibly anyway) result across the whole track or else it is a mis-match across the entire track.

I did more testing and the results seem to indicate it is the amp section in the Kemper that is the main contributor. But that might simply be perception because of the very high frequencies tube amp distortion generates, which makes it a lot easier to hear when there's a mismatch. That said, I did quite a few passes turning off the various sections in the Kemper and also tried dialing the gain all the way back on the amp section while it was still engaged. In all instances it seemed to audibly NULL for the most part. On the surface it seems it is attributed to the gain logic in the amp section, but please take all this with a grain of salt at the moment. Not even 100% certain it is the Kemper that is the cause.

It sure would be helpful if somebody else with a Kemper could try this reamp test. Just pick any reasonably high gain rig, set constant latency to ON, turn off stomp and FX sections just to be safe, and reamp a few passes of the same direct track through the Kemper. Then invert phase on one of the tracks and see if you hear silence or rather a bunch or low-volume fizz.

Sonic

10 (edited by MetalHeadKeys 2019-10-15 00:55:41)

Re: Recording Tracks don't NULL properly, anyone ?

Some more thoughts:

What is your exact signal chain? You 're going Kemper SPDIF Out to FF400 SPDIF In(obviously), but how is the DI guitar track going into kemper? Is it through SPDIF as well(FF400 Out to Kemper In)? Or through instrument cable?

Only reason I 'm asking is because of the quality of the shielding of the cables. Co-axial SPDIF cables are very susceptible to electromagnetic interference, and guitar cables are as well(but there are many with good shielding, so I assume you have a good one for recordings); my point being that, maybe, the difference you 're getting in the null tests, is the interference(apart from the modelling aspect)?

Perhaps, the gentlemen from RME could verify the SPDIF loopback, for you to test? Or someone else who has tried it?

RME Gear: Digiface USB, HDSP 9632

11 (edited by SonicExplorer 2019-10-15 01:30:42)

Re: Recording Tracks don't NULL properly, anyone ?

MetalHeadKeys wrote:

What is your exact signal chain? You 're going Kemper SPDIF Out to FF400 SPDIF In(obviously), but how is the DI guitar track going into kemper? Is it through SPDIF as well(FF400 Out to Kemper In)? Or through instrument cable?

FF400 direct guitar signal out over SPDIF to Kemper then back to FF400 over SPDIF.  Everything is digital path, no analog.

Re: Recording Tracks don't NULL properly, anyone ?

I found this:

https://www.gearspace.com/board/music-c … tions.html

So we can assume it is ok to have an SPDIF loopback. But I can't check if it is possible to route it, on the TotalMix side of things, as I have the Digiface USB and it doesn't have co-axial SPDIF.

As for the auto-slaving, that, I believe, won't happen if you set clock source to 'internal', as it won't be looking for a clock source on the input. Frame of mind being: "sending from SPDIF Out to SPDIF In".

Please report back any results, if you try it ! If you don't, we can continue the discussion, as it is an interesting topic!

RME Gear: Digiface USB, HDSP 9632

13 (edited by SonicExplorer 2019-10-15 20:23:51)

Re: Recording Tracks don't NULL properly, anyone ?

MetalHeadKeys wrote:

I found this:

https://www.gearspace.com/board/music-c … tions.html

So we can assume it is ok to have an SPDIF loopback. But I can't check if it is possible to route it, on the TotalMix side of things, as I have the Digiface USB and it doesn't have co-axial SPDIF.

As for the auto-slaving, that, I believe, won't happen if you set clock source to 'internal', as it won't be looking for a clock source on the input. Frame of mind being: "sending from SPDIF Out to SPDIF In".

Please report back any results, if you try it ! If you don't, we can continue the discussion, as it is an interesting topic!

Thanks!  I also tried your prior helpful suggestion for testing and it seemed to work (disabling everything in the Kemper one by one).  It seems to NULL properly - UNLESS the Amp section is enabled.  So we have no reason to believe doing a loopback with SPDIF would offer any further helpful information, and may even possibly cause damage.  We can probably assume if tracks audibly NULL with Amp section disabled in Kemper then a loop-back test would also audibly NULL.

What we need at this point is for somebody with a Kemper to please duplicate the test.  Just reamp a short direct clip twice via SPDIF into a gain rig and then flip phase on one of the tracks and see what you hear when played back together..... 

Sonic

Re: Recording Tracks don't NULL properly, anyone ?

How extreme is the this non null?

If I bounce a file in ableton at 24bit vs a 32bit bounce it doesn’t null..

Is all your gain staging correct? Panning law? Does volume changes after recording null it? Latency? Is it sample accurate? Line up the recorded files? Does it actually matter? - if it sounds the same to the ear but the null test produces some artifices maybe it doesn’t matter?..

www.yila.co.uk

Re: Recording Tracks don't NULL properly, anyone ?

FYI it is 100% safe to do a spdif loopback (or adat or analogue). It is the best way to test for problems with the rme itself.

Vincent, Amsterdam
https://soundcloud.com/thesecretworld
Babyface pro fs, HDSP9652+ADI-8AE, HDSP9632

Re: Recording Tracks don't NULL properly, anyone ?

that will still have latency and will not null

www.yila.co.uk

17 (edited by SonicExplorer 2019-10-16 20:12:14)

Re: Recording Tracks don't NULL properly, anyone ?

We are talking about an exact replication of audio path test, no bouncing or exporting.  Just simply sending a signal out of DAW over SPDIF to an external device and back in.  Recording what is being returned.  A pair of passes should NULL when played back with one phase reversed, or at least audibly NULL. Unless something in is inconsistent, timing issue or has made changes to the audio signal itself.

I *AM* able to achieve audible NULL but only about 1 out of 5 tries - unless I disable the Amp section of the Kemper.  And BTW I did try nudging the audio tracks 1 sample forward or backward and that didn't work either - it creates a REALLY noticeable difference when I do that, so clearly this isn't a time alignment issue in any conventional sense.  Evidence thus far points at Kemper as most likely cause which is why we really need somebody else with a Kemper to try this quick test....

Sonic

Re: Recording Tracks don't NULL properly, anyone ?

Yeh but does it sound different?

I bet if I did five passes reamping through my valve amp it wouldn’t null either.. cause valves, and if your amp section is causing it then it’s replicating real analogue circuits with constant discrepancies?

www.yila.co.uk

Re: Recording Tracks don't NULL properly, anyone ?

That's what I thought at the beginning, as well(if you read my first post). And Sonic, too! But as he pointed out, null should never occur, not once. But through his tests, that happened every once out of five or more takes. So now we 're looking for the reason why.

Thanks!  I also tried your prior helpful suggestion for testing and it seemed to work

You 're welcome! Glad I could be of some help!

FYI it is 100% safe to do a spdif loopback (or adat or analogue). It is the best way to test for problems with the rme itself.

Thanks! That's good info to know!

Sonic, another test to try, could be if you sent a sine wave through your Kemper, instead of your guitar track.

That way the signal would be consistent to the Kemper. My reasoning for this is that, perhaps, the "randomizing" factor on the amp modelling has to do with how strong a signal is(because the kemper has a way of mimicking impedance switching for different behaviour of different pickups, or lowering your guitar volume and mimicking valve behaviour).

Could you, also, try the same tests(your guitar track and the sine wave one) with a different high gain patch as well?

RME Gear: Digiface USB, HDSP 9632

20 (edited by SonicExplorer 2019-10-17 22:13:18)

Re: Recording Tracks don't NULL properly, anyone ?

I will try some more tests and report back.  But to clarify something important: Part of this exercise has now become to determine whether or not the result should be random.  We cannot make that assumption.  I would actually assume the opposite.  A digital device being fed the same signal over SPDIF should, possible latency aside, generate the same result every time unless there is logic intentionally designed to cause differences. There is no rational reason engineers would go to extra lengths to implement an algorithm in such a way.  In fact, it would make designing and troubleshooting much more difficult for the manufacturer if they had that kind of logic in place.  Instead, it makes more sense on every practical level to have the algorithm react the same way every time to a consistent input.  Which we are seeing in some of my tests, but not in others.  If we were dealing with a random algorithm it would likely never NULL and would be variable throughout any given track. Neither is the case so far.

At this point we are trying to answer two basic questions:
1) Is the behavior I'm seeing normal, intended behavior for the KPA
2) If the answer to the above is no, then why am I experiencing such behavior?

Sonic

Re: Recording Tracks don't NULL properly, anyone ?

Okay, so to make a long story short I ran a crap load of various tests and the results are the same.  However I did manage to determine clean rigs always seem to NULL while dirt rigs tend to not NULL reliably.  So the issue seems to be more related to the gain logic in the Amp section of the Kemper.

This still however doesn't answer the two fundamental questions I posed above. 

Sonic

22 (edited by iamyila 2019-10-21 12:33:23)

Re: Recording Tracks don't NULL properly, anyone ?

Yes it because if analogue modelling, which will not return the same result, even if its digital.. distortion characteristic of a valve is always in flux and software designers emulate that.. like i said

What is the artifact that is left over from the null test? How does it sound?

www.yila.co.uk

Re: Recording Tracks don't NULL properly, anyone ?

iamyila wrote:

What is the artifact that is left over from the null test? How does it sound?

Mostly sounds like high end fizz, the harmonic portion of the amp distortion.  But is consistent across track, does not phase or swell.  This indicates the problem is most likely NOT randomization.

My theory is this:  The Kemper up-samples internally and under certain conditions the signal is getting shifted a few of those 'micro' samples.  So the resulting output is consistent, but is actually shifted in time, to a difference too granular to see in a typical DAW.

Sonic

24 (edited by MetalHeadKeys 2019-10-21 20:10:16)

Re: Recording Tracks don't NULL properly, anyone ?

Sonic,
I 've been digging some more, today and came up with this:

https://worldwide.espacenet.com/publica … ale=en_EP#

So according to the diagram, a sound generator(guitar) signal is going through an adjustable filter which is oscillated by a resulting mix(or difference because I think I see a Delta on the diagram) of a reference IR(which I think is stored internally on the Kemper) and a characteristic IR(user IR by profilling), and then to the output.

My guess is that there is some random morphing going on between the reference IR and the characteristic IR, in order to impart "virtual" nonlinearity to the system. Maybe there are only a few morphing equations(let's say five, for the sake of discussion), and the algorithm chooses randomly between them. And that's why you had one test out of five, nulling, but not every once out of five takes.

I 've also read most of this thread on the Kemper forum:

https://www.kemper-amps.com/forum/index … p;pageNo=1

and the OP there, was claiming that he could hear a certain amp character which was imparted in every profile, regardless of the amp that was profilled(be it a MESA, Soldano, ENGL etc). He was calling it the Kemper ampsim and thought it was similar to vst ampsims(but a very-very complex one). I think it is the reference IR of the diagram, he was reffering to(no pun intended smile )

So that is my theory, with the upsampling portion taken in of course!

Perhaps, you could, also, post at the Kemper forum as well?

RME Gear: Digiface USB, HDSP 9632