|
Post by Johnkenn on Jun 2, 2015 18:14:33 GMT -6
I am svarticus.
|
|
|
Post by Martin John Butler on Jun 2, 2015 18:50:51 GMT -6
Dang, I liked B, the svartbox is a contender !
A was thin in comparison to B, D was similar to B, but less three dimensional, B takes the gold.
|
|
|
Post by wiz on Jun 2, 2015 18:52:22 GMT -6
Listen to the word , crowd, when the bloke first sings "I'm in with the in crowd" in the original recording, there is a cool little grate sound on his voice, then listen to the other converters. All of them except the RMA catch that. I like listening to the RMA the best. Makes the reverb around the voice and that sound cool.. nice to listen to... but it missed that aspect of the recording. What does that mean to you? dunno ... means something for me. 8) Nice work svartcheers Wiz
|
|
|
Post by Martin John Butler on Jun 2, 2015 18:57:35 GMT -6
To me, with B, I felt the pace and rhythm, it felt more live. I found B's transparency gave clearer low level detail.
|
|
|
Post by tonycamphd on Jun 2, 2015 19:29:20 GMT -6
Dang, I liked B, the svartbox is a contender ! A was thin in comparison to B, D was similar to B, but less three dimensional, B takes the gold. B is the old motu converter clocked off the svartacus
|
|
kcatthedog
Temp
Super Helpful Dude
Posts: 16,042
Member is Online
|
Post by kcatthedog on Jun 2, 2015 19:33:46 GMT -6
in terms of clock, I thought the svatacus always gets clock from the spdif signal it is passing as it has no wordclcok i/o ?
|
|
|
Post by svart on Jun 2, 2015 19:51:54 GMT -6
in terms of clock, I thought the svatacus always gets clock from the spdif signal it is passing as it has no wordclcok i/o ? The ADC portion is clocked by it's internal clock always. It should always be the master clock to whatever the receiver is too.
|
|
|
Post by NoFilterChuck on Jun 2, 2015 20:05:34 GMT -6
you guys should be listening for how the converters change the sound from the original audio file, not how they sound individually.
|
|
kcatthedog
Temp
Super Helpful Dude
Posts: 16,042
Member is Online
|
Post by kcatthedog on Jun 2, 2015 20:06:55 GMT -6
Right if you are using the svartacus's AD (money tracks) it will be the master clock as it is doing the converting right, send signal to your daw channel on spdif as input to recording channel ?
|
|
|
Post by tonycamphd on Jun 2, 2015 20:48:16 GMT -6
OK, First, congrats to Svart on this thing, it really sounds great!(looking damn cool so far also 8) I just level matched these and did a good long listen to the Svartacus, RMA, and the symphony, what I find seems stupid to me, or like i'm confirmation biasing due to the fact that i'm getting all that RM conversion? But i really don't think i am. I compared everything against the original, not against each other, my favs go in this order... don't laugh, i'm totally serious! RMartin, the original track, the svart box, and the symphony lol. I picked the RM over the original track? wtf? really? I'm sitting here asking myself how is it possible that i'm hearing it this way? but i am, here is what i heard. 1. RM placed the vocal forward in the mix and imaged equal to the original with about the same freq representation, but it has a smoothness/softness to it that seems to add depth to the rearward content, that the original doesn't have... Before you get upset about my possible confirmation biasing, i have to say that its a bad observation as far as accuracy right? so it's not really a compliment, and i'm not sure how to quantify my feelings about it? But no doubt, the RM sounds the most pleasing and listenable to me overall. 2. The original track, same image placement, freq rep as i hear on the RM, but slightly more of a grain like texture to the sound overall. 3. The Svartacus, had about the same freq representation, but depth closed up front to back a bit compared to the original, it also seemed a bit less dynamic as compared, no bass problems to my ears, this box sounds really nice! 4. The Symphony, closed depth up front to back about the same as the Svartacus, slight low/mid bass bloat(but not as bad as I remembered) rolled off a tad in the highs, and most surprisingly a bit pinched/congested in the stereo field by comparison, the symphony still sounds really good also. i didn't listen to the old motu files at all lol I will listen to these again in the AM with fresh ears and a new perspective, maybe i'll hear it differently?
|
|
|
Post by svart on Jun 2, 2015 20:57:33 GMT -6
Ok, I have some results for the modding testing, and I'll let the graph speak for itself. I used nulling tests to see what signals were left when a channel was modified and then compared to a signal from an unmodified channel. Modification was done by adding 330uF ultra-low ESR electrolytic caps from panasonic to each opamp power rail, 6 caps total on 3 opamps. Methodology: 1. Do stereo 5hz-22Khz sine sweep over 1 minute after letting unit warm up for 5 minutes. 2. Reverse polarity on left channel in DAW. 3. Adjust level at 1khz for optimum nulling. 4. Render null tracks and peak normalize to see residual signal. (Peak normalizing is used to grossly exaggerate the residual signal so that it's visible on a graph) 5. Modify Left channel of unit. 6. Perform 1-4 again and compare. 7. Do 1-6 again on another unit. The level adjustment for optimum nulling at 1Khz was -0.12dB for the first unit, +0.11dB for the second unit to get better than -85dB null at 1KHz. Peak normalizing raised the residual signals 75dB on the first unit, so that the highest peak was at 0dB. The reason for residual signal was gain mismatch across frequency, < 0.2dB max below 100hz. The second unit needed 74.5dB of gain. The second unit showed extremely similar results, confirming that the results are not anomalous. These are the residual signals below. The one on the top is what is left when a modded and unmodded channel are level matched at 1khz, nulled and peak normalized (75dB of gain). The bottom graph is the result of two unmodded channels level matched at 1khz, nulled and then gained up (75dB) to match the amount of gain needed to peak normalize the modded+unmodded test, ensuring the same graph scale between the two. I ran the test 2 times on each unit, with results within a couple tenths of a dB of each test.
|
|
|
Post by svart on Jun 2, 2015 21:03:56 GMT -6
And another glamour shot:
|
|
|
Post by tonycamphd on Jun 2, 2015 21:14:49 GMT -6
can that test be done with a maxed voltage, square wave pulsing sweep?(especially down low) I think this is when the state of power would affect dynamic range? does your measuring device have any functions that can represent the aspects of the image that we hear? serious question, how would you measure depth of field with an analyzer? It's pretty clear that 2 converters can measure identical, and not sound the same... right? I'm wondering out loud here, teach me please 8)
|
|
|
Post by tonycamphd on Jun 2, 2015 21:15:30 GMT -6
And another glamour shot: Dude, enough already,... whats the name gonna be?
|
|
|
Post by Martin John Butler on Jun 2, 2015 21:26:03 GMT -6
Thsnks Tony. I was rushing earlier, and I just saw svart... , and didn't notice it was just the clock on B. I listened again, and it was interesting. Listen to the piano on the left. It's almost missing on B, but E, the RMA puts the piano back on both sides, but it's a little dull, C, the svartacus, gives a more transparent, more realistic representation of the piano on both sides. I liked C, B, E in that order.
|
|
|
Post by svart on Jun 2, 2015 21:36:12 GMT -6
can that test be done with a maxed voltage, square wave pulsing sweep?(especially down low) I think this is when the state of power would affect dynamic range? does your measuring device have any functions that can represent the aspects of the image that we hear? serious question, how would you measure depth of field with an analyzer? It's pretty clear that 2 converters can measure identical, and not sound the same... right? I'm wondering out loud here, teach me please 8) I can do squares, and possibly sweep them. I'll check tomorrow if my software can do it. If not, I can borrow a lab function generator from work and see about doing it. I always think that at some level, no matter how small, a difference picked up by the ear can be seen in a graph. I figure that a change in low end should be seen at some point when doing long term sweeps and comparing them. Depth of field is more related to psychoacoustics, and thus could probably be defined by phase relationships, but not frequency sweeps. My testing above was strictly to see if the capacitor mods made a measurable difference in the low end response of my unit, since the mod helped the RM unit for Dan. I have a theory that the RM unit's traces were too thin for the current demands, and the opamps were starving a bit. I also posit that the RM unit has a much tighter I/V integrator bandwidth and/or LPF bandwidth that brings the high end down slightly, tilting the whole spectrum to be bass-heavy. With less than ideal bulk capacitance, the low end would suffer the greatest, and through some miracle the design probably ended up relatively flat. So when Dan modded it, it improved the low end and exposed the slight frequency tilt. I know it sounds like a crazy theory, but I've seen crazier things happen. Since my unit uses power planes within the 4 layer board, and I use high speed regulators, I don't have much power loss from the regulators to the opamps and can get away with smaller cap values.
|
|
|
Post by wiz on Jun 2, 2015 21:49:55 GMT -6
and to put a little slightly different consideration on null test results...
take a mix, copy it, put it 180 degrees out of phase with the mix, zero right? Of course.
Now adjust one of those tracks volumes up or down, by 0.1dB and see what your left with in terms of amplitude.... 8)
It will be something around -40dB below the signal of the original track.. with only a 0.1dB difference between the two.
Null tests are interesting things.
cheers
Wiz
|
|
|
Post by dandeurloo on Jun 2, 2015 21:52:14 GMT -6
OK, First, congrats to Svart on this thing, it really sounds great!(looking damn cool so far also 8) I just level matched these and did a good long listen to the Svartacus, RMA, and the symphony, what I find seems stupid to me, or like i'm confirmation biasing due to the fact that i'm getting all that RM conversion? But i really don't think i am. I compared everything against the original, not against each other, my favs go in this order... don't laugh, i'm totally serious! RMartin, the original track, the svart box, and the symphony lol. I picked the RM over the original track? wtf? really? I'm sitting here asking myself how is it possible that i'm hearing it this way? but i am, here is what i heard. 1. RM placed the vocal forward in the mix and imaged equal to the original with about the same freq representation, but it has a smoothness/softness to it that seems to add depth to the rearward content, that the original doesn't have... Before you get upset about my possible confirmation biasing, i have to say that its a bad observation as far as accuracy right? so it's not really a compliment, and i'm not sure how to quantify my feelings about it? But no doubt, the RM sounds the most pleasing and listenable to me overall. 2. The original track, same image placement, freq rep as i hear on the RM, but slightly more of a grain like texture to the sound overall. 3. The Svartacus, had about the same freq representation, but depth closed up front to back a bit compared to the original, it also seemed a bit less dynamic as compared, no bass problems to my ears, this box sounds really nice! 4. The Symphony, closed depth up front to back about the same as the Svartacus, slight low/mid bass bloat(but not as bad as I remembered) rolled off a tad in the highs, and most surprisingly a bit pinched/congested in the stereo field by comparison, the symphony still sounds really good also. i didn't listen to the old motu files at all lol I will listen to these again in the AM with fresh ears and a new perspective, maybe i'll hear it differently? I just finished a long session and I am fried, so I haven't listened to all these files. But I also preferred the mix after I processed it compared to the original mix. That is a good and bad thing. It's good that I think the alterations to the mix where musical but, bad that it it doesn't sound like the original file. In fact, none of these converters (and I would bet none) are actually transparent like we all want. So, with that being the case I'll pick the one that changes the music in the most musically pleasing way. Hopefully, I can listen to these tomorrow before my other session. I am really curious to hear the symphony.
|
|
|
Post by svart on Jun 2, 2015 21:52:10 GMT -6
and to put a little slightly different consideration on null test results... take a mix, copy it, put it 180 degrees out of phase with the mix, zero right? Of course. Now adjust one of those tracks volumes up or down, by 0.1dB and see what your left with in terms of amplitude.... 8) It will be something around -40dB below the signal of the original track.. with only a 0.1dB difference between the two. Null tests are interesting things. cheers Wiz Yes they are! That's why you'll see that to get maximum nulling, I had to adjust by like -0.1dB and some change.
|
|
|
Post by dandeurloo on Jun 2, 2015 22:31:54 GMT -6
When you guys are listening to those files be sure to volume match them as well.
|
|
|
Post by tonycamphd on Jun 2, 2015 23:21:42 GMT -6
I level matched my files, I can probably re bounce the files and post them if yall want?
As far as analyzer vs ears goes, the ship of doubting I can hear what analyzers can't sailed a long time ago. To be clear, I'm not being braggart, I just know what I'm listening for, and that's more than 1/2 the battle. If u listen for the depth of field front to back, it's totally there, is it phase diffs or accurate rep of a mixers placement? I say the later because I hear it in the original.
svart, that would be awesome if u can do the squares test. I also don't think Ross got lucky, he's a pretty talented cat
|
|
|
Post by NoFilterChuck on Jun 2, 2015 23:36:05 GMT -6
In the interest of honesty, i'll re-do the files, as I had to bump the svart file up 5db, and the motu files up 2db. svart, you still haven't addressed the volume difference of your unit. We feed it a -12dbFS sine wave, and the computer sees a -17dbFS sine wave come back in. That doesn't happen with the other interfaces.. edit: updated the files in the dropbox. They are exactly as they came back into the interface used, no bumping up of signals or anything. Offhand, you'll need to raise them as follows: Svart: ~5db Motu: ~2db Symph: ~0.7db for the mic In file.
|
|
|
Post by svart on Jun 3, 2015 8:33:46 GMT -6
In the interest of honesty, i'll re-do the files, as I had to bump the svart file up 5db, and the motu files up 2db. svart, you still haven't addressed the volume difference of your unit. We feed it a -12dbFS sine wave, and the computer sees a -17dbFS sine wave come back in. That doesn't happen with the other interfaces.. edit: updated the files in the dropbox. They are exactly as they came back into the interface used, no bumping up of signals or anything. Offhand, you'll need to raise them as follows: Svart: ~5db Motu: ~2db Symph: ~0.7db for the mic In file. I'll look at the level more closely although I'm measuring a ~2dB difference here. I also have some extra series resistance on the output of the DAC to protect the opamps from stray loads/grounding/capacitance which will decrease signal a little. It's a small tradeoff for DC coupling I would think. I also have extra series resistance on the input of the ADC for some protection of the ADC chip, to decouple the input/feedback capacitance, and to match headroom on the opamps. The ADC chip can see a max swing of 5V on the inputs, centered around 2.5V. The opamps are running a 30V swing, so I've traded a lot of opamp headroom for a lower signal. It should help keep the opamps away from the rails and out of non-linearity. I can gain up both a couple dB, but we'll lose what I think is some extra protection. Why not just turn your faders up a bit?
|
|
|
Post by svart on Jun 3, 2015 8:35:50 GMT -6
So I was going for a glossy white look on the front panel. I think it looks good, but I'm going to try a matte white and a semi-gloss too. Matte might show more dirt, so I'm thinking semi-gloss should look great.
Thoughts?
|
|
|
Post by tonycamphd on Jun 3, 2015 8:52:18 GMT -6
In the interest of honesty, i'll re-do the files, as I had to bump the svart file up 5db, and the motu files up 2db. svart, you still haven't addressed the volume difference of your unit. We feed it a -12dbFS sine wave, and the computer sees a -17dbFS sine wave come back in. That doesn't happen with the other interfaces.. edit: updated the files in the dropbox. They are exactly as they came back into the interface used, no bumping up of signals or anything. Offhand, you'll need to raise them as follows: Svart: ~5db Motu: ~2db Symph: ~0.7db for the mic In file. I'll look at the level more closely although I'm measuring a ~2dB difference here. I also have some extra series resistance on the output of the DAC to protect the opamps from stray loads/grounding/capacitance which will decrease signal a little. It's a small tradeoff for DC coupling I would think. I also have extra series resistance on the input of the ADC for some protection of the ADC chip, to decouple the input/feedback capacitance, and to match headroom on the opamps. The ADC chip can see a max swing of 5V on the inputs, centered around 2.5V. The opamps are running a 30V swing, so I've traded a lot of opamp headroom for a lower signal. It should help keep the opamps away from the rails and out of non-linearity. I can gain up both a couple dB, but we'll lose what I think is some extra protection. Why not just turn your faders up a bit? I've read this out and about, so don't hold me to it, it's been said the hotter you can run the V on the 4898's, the more opened up the signal they pass becomes?
|
|