Totally wierd phase phenomenon. . . try it yourself

[honkyjonk]

In an effort to track down what I thought was reversed phase somewhere, (well I did a bunch of things, but I don't need to mention them all)
but in my HD24, I took a single ac. guitar track (recorded w/ 1 mic)(on track 1) , straight from it's HD24 output, into
track 2 of the HD24.

Now that I had two identical tracks (outside of the DA/AD conversion)
I sent these two tracks to the board. I panned them both to the center, made sure all the EQ was the same, everything the same, then I pulled one fader up while simultaneousley pushing the other down. and then went back the other way.

The tracks sound okay with one another while one is significantly louder than the other, but when they start to reach the same exact volume (i.e. the two faders are level) there is a bit of a phasey sound, or maybe you'd call it an out of phasey sound, not that they start to cancel eachother out, but they're working toward wanting to cancel each other a bit. I should, if both tracks are in phase, ideally be able to fade 1 in and the other out without noticing anything right? Hmmmm.

I at first thought, oh, well maybe I have a cable that I screwed up when I made it, and I've never noticed until now, so I flipped the phase on one channel on the board, and it was much worse, and both tracks were very much canceling each other out.

Okay, so as an experiment I digitally cut and pasted the original ac. guitar from track 1 to track 2.
So now they are truly identical right?
Well, I did the same thing, using the same channels on my board, same cables to the board, and there's not the same phasey sound when the two tracks reach the same volume.

Wierd.

What's going on here? It can't be:
1.the board
2. the cables
3. one of the trs connections on the HD24 mistakenly wired backwards

???????

The only pheasable answer is that the patch cable from track 1 output to track 2 input was wired backwards, but I'm sure this is not the case because I tried several cables. And wouldn't it have been cured anyway when I flipped the phase on the board?

???????????

Has anybody ever encountered this before?

[scarygroover]

i believe it might be a latency issue. Computers have latency. when you go out of them and back into them there will be a slight delay causing phasing.

[slowblue]

I"m guessing it's the latency with your a/d d/a conversion. The tracks, while identical, are not aligned due to the time it takes the converters to convert the signal, so you're getting phase cancellation between the two tracks. This is a good opportunity to see what the latency on your system is, for my 001 it's around 40 samples, so after I overdub to a track I always nudge it back 40 samples so that what I played is actually where I intended it to be played, rather than where the computer put it after conversion.

[honkyjonk]

Holy Shiza!!!

That has to be what it is.
I've never dealt w/ it w/ computers because I've never recorded w/ a computer before, and I'm pretty new to digital anyway.

I wonder if I can nudge the track a bit on the HD 24. . . . . . .

[rainsinvelvet]

I would also agree on the latency of the converters. There would be a delay going out from track one into track two. Even if its not much of a delay the two signals will be slightly out of phase with each other.

ERic

[Roboburger]

don't forget- you got two different d/a converters working there, each with thier own gain path chock full of reisistors, chips and caps. there might be a decent enough bit of dicrepency between each component (we are talking Alesis here) to make the post digital world have a slightly different thang happenin'. At least that's my guess.

Edit: But then again, phase problems are usually time based, right? hmmm.

[Piotr]

Plus the cable run from channel one to channel two

[wardshorsehead]

i'm sure its a phase issue as a result of latency. if you are going out of one output into another input, that would be two conversions...da and ad. more that enough to cause this.

frank

[rhythm ranch]

i'm sure its a phase issue as a result of latency. if you are going out of one output into another input, that would be two conversions...da and ad. more than enough to cause this.

I concur.

[honkyjonk]

Kinda sucks for a bounce though, if I want to add some compression and free up some extra tracks at the same time (which I'm very used to doing in 1/2" 8 track world) and then have a compressed track to bring up w/ the original.

I guess I can do this, but I'm not sure if I can align the track right afterwards. I was messing around, and the HD24 let's you slide the track, but in milliseconds, so I don't know yet how to get it closest. No millisecond measurement has sounded quite right yet.

[Mark Alan Miller]

My suspicion is to overcome that latency issue (which is exactly what it is - the cumulative delay of the D/A out and the A/D back in) millisecond resolution is going to be far too coarse... teh proper time is going to be somewhere between the closests milliseconds.
You'll have to import both offending tracks into a DAW, align them visually at the sample-level, render them back out as contiguous files starting at the exact same time, and push the files back onto the HD24.
Alesis makes that hard drive dock specifically for tasks like this...

[Roboburger]

Oh! I thought the track copy was a digital one...

I don't think the latency would be enough to make the instruments feel off, It would just be enough so you can't use two sources of the same track...

[Huntlabs]

You should be able to slide or nudge the track so it will exactly align.

I have a Digi 001 and have been experimenting with this issue this past week.

My session set up was 44.1K / 24. I recorded a sharp click to a track and then recorded that output back to another track. There was about 55 samples of delay, from the headphones using the 001 mic input. (Not a good pre amp by the way.)

The same test using my Octopre with lightpipe showed about 90 samples of delay.

Do the same experiment on your gear. Zoom in until you can see the actual wave and then figure out how to slide or nudge until they are aligned. Should be in the manual.

Samples = Milliseconds x sample rate
OR
Milliseconds =Samples / Sample Rate

Good luck.

John

[heylow]

All converters take a certain amount of time to pass the signal....that's just the way it is. Going out of the HD and back in is 2 convesions. I can't remember how much it is...something like 3 to 6 ms.

You can, in fact, nudge (shift) on the HD but if you do, the machine won't record for whatever reason on that song.....basicly that is to say that a person would have to nudge at mix time (or after the song is completely finished) rather than nudge and save like in a DAW.

Fun, fun , fun.



heylow

[joel hamilton]

this is why overt latency is unacceptable in a professional recording situation.
Recording someone whose "feel" costs a lot of money means you cant just go in and slide it around! Imagine someone slipping the drum tracks up on "when the levee breaks" so they didnt "lag" at all!

All the manufacturers of prosumer converters (host based) TOTALLY downplay this fact. Latency= Phase problems. Period. If you track a pair of room mics split across the odd even pairs (meaning using inputs 4-5 or 6-7 rather than 3-4 or 5-6) you will not be getting phase accurate recordings.

It pisses me off that companies pass this off as acceptable. We should all be writing letters to digidesign et al. I know that feels like writing a letter to Bush and asking him to get smarter, but.... Ya know? I would love to have a small format digital multitrack that was acceptable for professional recording (other than ADAT, oddly enough!).