Tascam DM 4800 and Apogee Rosetta 200

OK Dan, you got it!... Done. I hate the idea of having to go to MP3 for sample rate comparisons. Nonetheless, I did one sample at 88.2 through channel one in the DM 3200. The second file is done through the UH7000 at 192Khz. Both were combined into one file and converted to a 160kbps MP3.
I talk quite a bit on the first half and it sounds kind of phasey because I was bending over my guitar trying to speak into it while it was hanging upside down, so disregard the phasey sound of my voice.
It's a crying shame that I can't upload MP3s of my own stuff here, so I will need your email. PM me when you see this.
BTW, both are raw files only, but I did normalize both recordings for volume's sake.
 
Last edited:
I'm currently using external hardware analog preamps in to the DM convertors.
Would I benefit from using the external convertors that you are mentioning?
If so, where do I insert the higher quality signal to avoid the DM lower quality?

Also wondering about optimal tracking specs.....
Sample rate and bit rate.
I'm generally running at least 32 tracks of audio and a bunch of plugs and VIs.
Is all of your gear working well at the high sample rates?
Does it slow down your back up procedures?
Does it sound noticeably better?

Thanks
 
Problem would have been eliminated just by recording your mixdown back to Cubase. Ater this it's only drag-and-drop operation to import it to WaveLab.
I hear you Jarno. I think we discussed this before. I prefer to do my capturing in a separate application. I find it distracting to have both my mix and the mixdown in the same project. Pressing PLAY/REC on the external unit and not having to deal routing is very satisfying. Personally, I find this method "cleaner". I have my Cubase project folder and my WaveLab master folder. Anyway, that was just an added bonus to bump in sonics and wasn't the sole justification for purchasing the DA-3000.
 
Last edited:
Hi Waterstrum,
We've had these discussions almost every year, at least once or twice, so I will try to keep it short;

I'm currently using external hardware analog preamps in to the DM convertors.
Would I benefit from using the external convertors that you are mentioning?
If so, where do I insert the higher quality signal to avoid the DM lower quality?

You might benefit from it's use. Let me again put one thing out there. There is nothing wrong with the DM converters. They do the job quite well and should not be thought of as toys or trash that must be bypassed at all costs. You can make perfectly professional albums using your DM converters as many have. For me, the UH7000 offers a different sound that I like with built-in signal and clock circuitry that, I feel, does exceed the signal path of the DMs. Not to mention, the converters are new as of this year, and I think there might be something to that. Also, having 2 excellent preamps allows me to track directly into the unit and convert immediately, using it's aforementioned signal path.
For your routing, you could take the line outs of your preamps and hook them into the UH's line ins. The unit will then convert internally. Then, hook the digital AES/EBU outputs to Digital in 1 or 2 in the DM. Now you can send that DIN any where you want. It will remain unaffected, as long as it doesn't convert back to analog.

Also wondering about optimal tracking specs.....
Sample rate and bit rate.
I'm generally running at least 32 tracks of audio and a bunch of plugs and VIs.
Is all of your gear working well at the high sample rates?
Does it slow down your back up procedures?
Does it sound noticeably better?
Thanks

All good objective questions that, unfortunately, require subjective answers. My likes may be different than yours.

I will say that recording at the highest bit rate possible is extremely important. I did NOT say Sample Rate. I mean 24bits or 32bits or 64bits. Really, anything over 16bit should really be a goal and I think 24 is recommended. Please don't ask me for the math behind that.
As for Sample Rate, I subscribe to the to the school of "record/edit/mix/master at a higher sample rate, for better A/D waveform reproduction, then down-sample to 44.1/16 for CD as the last possible conversion". I have been very happy working at 88.2. My DAW is set to record at 24bit and uses 64bit internal processing throughout. It offers many more bit depth options, but I haven't done a comparative analysis for any of the other audio bit depths. BTW, this has nothing to do with 32/64 bit versions of windows. I have a 32bit windows7 quad core computer that has had no problems with working at 88.2. I have had projects up to 50 tracks, although I typically don't use VIs. If I do, I will usually get them where I want them, then render a stem, so I am not constantly trying to struggle with loading the VI and having take up all my ram and processing power. I do use VSTs though, almost always on every track.
For backup, I will typically just copy my project folder, and all its audio files to a different hard drive. The speed of this is dependent on the hard drive type and interface. If I could afford large, Terabyte SSDs, I would use them. Right now, I have 4 WD Sata3 hard drives, all 500GBs, and they do just fine.

So much for keeping it short.
 
Last edited:
Have you ever played a triangle waveform on your synth? Sounds kind of nasty doesn't it? You have just drawn a visualization of how 44.1khz sample rate reproduces a 22.05khz audio wave. Now, double the number of dots, equal spacing, and redraw. It looks a little better right? This is the same audio wave reproduced at 88.2khz.

Several years ago, on another audio forum, a learned engineer posited a very similar example of why 96kz bests lower sample rates. It made sense to several people including myself. However, the forum's Moderator - a mastering engineer with serious attitude - took the poster to task for his logic. He insisted that comparing digital audio to 'connect-the-dots' wave form resolution is apples-to-oranges, and the ONLY thing 96kz offers is reproduction of frequencies above the human audible range. Nothing more.

Long story short, a debate ensued resulting in flames and shames. The poster was ultimately banned from the forum for sticking to his guns. That was my first experience with the Sample Rate Nazis, and it wasn't pretty. :)

CaptDan
 
I hear you Jarno. I think we discussed this before.
I think too.

I find it distracting to have both my mix and the mixdown in the same project. Pressing PLAY/REC on the external unit and not having to deal routing is very satisfying.
Well ... I feel quite opposite. I like to just set my locators, enable punch in/out and hit PLAY and let Cubase to record my mixdown. And the most inportant thing on having mixdown on the same project is easy A/B comparison: I can try another mixdown and compare against the recorded one on-the-fly just by pushing DM's MON SEL buttons.
 
Last edited:
He insisted that comparing digital audio to 'connect-the-dots' wave form resolution is apples-to-oranges
INDEED! Connect-the-dots or strairsteps representation of ditial audio is just completely WRONG! Please people, try to understand how digital audio works before making foolish statements:
ONLY thing 96kz offers is reproduction of frequencies above the human audible range.
Well ... that's only true in ideal world. In true world we may have poorly implemented A/D converters and lousy DSP algorithms creating artifacts which using higher sample rates can be moved out of hearing range and then filtered out by good D/A conversion (or high-quality sample rate conversion). If this is a real problem ... I don't know.
 
Last edited:
If this is a real problem ... I don't know.

I don't think it's a problem but more a preference, these days. What I'm unsure of is whether the quality difference is perceivable in all musical examples and performances. I tend to think it's not.

CaptDan
 
WOW! That's the best video on the subject I have ever seen. Thanks Jarno.

My example was based on Nyquist's theory that the highest frequency producible by a sample rate requires twice that sample rate. The way it was explained to me was the same as how I explained it in my "connect the dots" explanation. And, I think that it is still correct when talking about a digitized signal with a continuous time function. But, I didn't know that the time between samples, when converting back to analog has no definable value at all, allowing the digital sample to be converted back to analog as a perfect continuous wave. This little fact pretty much trashes any theory that says a higher sample rate is a better reproduction of the waveform, a theory I have held for a long time. Thanks for ruining it all for me! Just kidding. Always learning...
Well, my apologies for incorrectly explaining it that way. My ass is right where it's always been, firmly planted in my chair of modesty!
PS, I still like my UH7000! even at 88.2!
 
Now that I think about it, music, of course, doesn't live in a 1Khz sine wave. Or any sine wave, for that matter. Unless it was originally a sine wave, of course. Music lives in complex waves. What happens to the freq harmonic that originally existed in the analog domain, but doesn't get represented by a sample. Or, at least not correctly represented by a the lower sample rate. Wouldn't that be a reasonable argument for higher sample rates? Can't more samples better represent a harmonically rich complex waveform? There are only two samples available in a 22.05Khz wave form when sampling at 44.1. What if there are more points in there that need to be represented? Aside from the perfect drawing of a sine wave, I am floating back to my original theory.
 
What happens to the freq harmonic that originally existed in the analog domain, but doesn't get represented by a sample. Or, at least not correctly represented by a the lower sample rate. Wouldn't that be a reasonable argument for higher sample rates? Can't more samples better represent a harmonically rich complex waveform?
Didn't you watch the video all the way to the end?

If there's complexity in the wayform which cannot be recreated from samples, the original signal contains frequencies above Nyqvist frequency. That's the beauty of the sampling theorem: if you put properly bandwidth limited signal in, you'll get the same signal out. And guess what: our hearing is bandwidth-limited, which makes it no-brainer to bandwith-limit audio before storing (and sampling) it.

You'll have to remember: complex waves are sums of sine waves. This means we have to only sample a "sum of sine waves" up to about 20k. Now, Monty demonstrated on the video he can sample and recreate 20k sine wave "perfectly". And I think you also believe addition is well-defined mathematical operation. These two things together leads us to conclusion: digital audio works (even at 44.1 sample rate).
 
"digital audio works (even at 44.1 sample rate)."

Yes it does. And that's why a lot of stalwart, previous 96kz adherents have gone back to riding the olde 44.1 horse. Many claim they do not hear any difference, and don't find larger files worth the hassle. And they're right - for THEM.

What these demos fail to present scientifically are the relative behaviors of various converters and how they function at different sample rates. The question is, would - say - a DM 96kz-capable AKM chip/converter work best at its maximum sample rate, and by extension, would the D/As on the mixer present playback in its truest form accordingly? Simply put, if a converter is a 96kz example, does it function 'best' at the highest rate? And if it does, will the engineer's job be made easier by virtue of more accurate reproduction, affecting mix decisions all along the way?

That's the question I've had to answer for myself - unscientifically, mixed liberally with 'ASSumption' and qualities I believe my ears are conveying.

YMMV.

CaptDan
 
Not to mention the other infinite variables that will effect the sound. For me, namely, the circuitry involved in a dedicated unit. Its not like we are taking a single wire and jamming it into a converter chip, so much more is going on. I often think of the Burl B2 and how the analog circuit was designed prior to conversion. This is one of the reasons I like the 7000.

Jarno, I did watch the video all the way to the end. Monty speaks like an electrical engineer. He makes excellent points that come off as obviously simple, but often says small phrases that require me to take some time to mentally digest them. When he said that any additional complexity in the waveform would be beyond the Nyquist frequency, I had to think about that one. It dawned on me in the shower that he was referring to the frequency at half the sample rate. So, if the sample points at 44.1 can't perfectly reproduce a complex waveform, it's because the frequency involved is above 22.05. I realized that after I wrote my last reply. Also, I still can't wrap my mind around the undefined time between sample points. If I am sampling at a regulated 44.1Khz, and assuming that each sample taken happens at the speed of light, or "c", then the time between samples would be 1/44.1Khz -"c". A defined amount of time.
 
I still can't wrap my mind around the undefined time between sample points. If I am sampling at a regulated 44.1Khz, and assuming that each sample taken happens at the speed of light, or "c", then the time between samples would be 1/44.1Khz -"c". A defined amount of time.
Yes, 44,100 times per second a voltage is measured and translated into a digital word consisting of X bits, so each sample including time to the next is 1/44,100 = 22.67 microseconds. So, you're correct, there is 'unaccounted for' time - but it's irrelevant. As can be seen in the video, the reconstruction process in the DA-converter smooths everything out again to recreate the perfect copy of the input. The reverse translation from digital to a voltage from one sample to the next goes without 'stepping' to the next level.
 
So, if the sample points at 44.1 can't perfectly reproduce a complex waveform, it's because the frequency involved is above 22.05.
Exactly!

Also, I still can't wrap my mind around the undefined time between sample points.
Don't waste too much on that. While mathematically it's true (in digital domain), I would rather use word "predictable" ie. can be perfectly reconstructed (into analog domain). And this predictability is direct consequence of requiring sampled signal not containing frequencies at or above 1/2 of sampling frequency.
 
Reviving an old thread. I've always regretted selling my Rosetta 200 as it was an impulsive reaction to having just spent $700 on the DA-3000. I always loved its "finished" sound and how it improved the stereo imaging and separation of the instruments. Even though it is as old as the DM series, it does sound better. I originally thought the DA-3000 converters would replace it but it ended up being used solely as a mixdown recorder capturing the DM's digital output. I decided to end my 3 year regret and bought another Rosetta 200. This time I thought about how I could use it to improve what I think is the DM's weakest audio path (D/A conversion) as well as capturing that improvement in the mixdown and monitoring processes. Here's what I did:

First I had to assign the Stereo L/R to output on DIN 1 L and R. I am then going out digitally from the DM (DIN 1 Out) to the Rosetta (S/PDIF In) then going Analog Out on the Rosetta to the Analog In on the DA-3000 to record the Rosetta's D/A conversion when mixing down. I'm then monitoring this signal by connecting the DA-3000's Analog Outputs to a Mackie Big Knob Passive Studio Controller to provide volume control to my monitors. I connected it this way to maintain sonic consistency between DAW playback and the DA-3ooo's recording & playback.

So to sum it up:
  • D/A conversion from the DM3200 is being done by the Rosetta & passed through the DA-3000 to my monitors via analog connections.
  • The Rosetta's D/A conversion of the mix is recorded on the DA-3000. There's also the DA-3000's A/D conversion when recorded.
  • A/D conversion is still being done by the DM-3200 using onboard or external preamps.
I'm now able to record anything from the mixer through the Rosetta directly to the the DA-3000 (straight to SD, no DAW required for live performances). Since I'm going Analog In on the DA-3000 for the first time, I'm now able to record to the DSD format which is impressive sounding to say the least. Lastly, I have the ability to connect any of my preamps to the Rosetta's analog inputs to make use of it's A/D conversion and route those signals to my DAW. I feel like I'm getting the best of several worlds while maintaining my original tracking, mixing and mastering workflows. This configuration should work with any AD/DA converter. No need for a Big Knob type of device if it's something like the Antelope Pure 2 which has a volume control built in. Hope all that made sense!!
 
  • Like
Reactions: Peter Batah
I’m really loving the Rosetta 200 as the DM’s D/A converters. If you want a little bump to the DM’s quality for about $500 then I recommend picking one of these up. Yes they are old but they still sound great. They are regularly selling on eBay for about $450 and the Mackie Passive Big Knob is about $60. In my estimation, this is the cheapest and most elegant way to upgrade your DM’s sound. Any better would possibly warrant replacing your DM depending on your needs (Burl B2 Bomber with same design goes for $2500). Be warned that it’s a subtle bump in quality upon first listen. You’ll probably say WTF Charlie! However once you start using it, you’ll find your older mixes sound flat and smeary when compared. For me the most noticeable differences are the focused and rounded bass, how crisp yet warm the high end is (especially cymbals) and how panned instruments are more clearly separated. Lastly, the Rosetta 200 was known for giving a sound that is characteristic to analog tape. The trick is capturing the D/A conversion so either an external deck or capturing the analog signal in your DAW will work.
 
  • Like
Reactions: Peter Batah
Happy New Year Charlie. I don't get by here that much but from time to time I like to check in. For some reason I stumbled onto your thread revive. I'm interested to know if you tried any other DAC's. I've always clocked my rig with a Mytek Studio Clock so I've looked at those on the 2nd hand market and that led me also to the Benchmark DAC1 (good prices for that unit). It seems these are all competitive with the Mytek being a bit more spendy. Currently I see less of the Rosetta. You may remember that my DM4800 monitor section died awhile back so I found a workaround using a split off of the L/R channels from my surround card that feeds a Mackie Big Knob. That seems to work fine but I like your tweak since I wouldn't mind a better stereo DA image and also putting my Surround needs back to normal. Again curious if you did any comparos.

THX,
Wilson
 
Keeping this thread alive...Here to report that I scrounged a Benchmark DAC1 off ebay for a good price and it was easy peasy to set it up with my DM4800. Not only does it improve the monitoring quality as Charlie mentioned but it solved my workaround monitoring fix where I was stealing the LR of my surround card to replace my dead 2 channel monitor section. I now have my surround setup back to its proper config...in my case the snake routes to a BLueSky Media desk Surround speaker array. The DAC1 is seeing one of the AES stereo outs and then converts it to analog into a Mackie Big Knob which I had been using as part of the workaround. For me this is a win win ie. proper monitoring that does indeed have tighter bass, and mids. The top end seems about the same but I'm very happy to get a tighter image overall. Once again the DM series survives for us true believers. Hell I haven't even had to raid my boxed/offline DM3200 for parts yet...tho my 4800 screen is starting to have the dreaded line condition....but I hardly use it so I don't notice too much. In my world my 3 LED displays from my MacPro/Pro Tools rig show all the visual info that I need.
 

New posts

New threads

Members online

No members online now.