Actually, going from unity gain to 30 dB noise gain (with 30 inserts) lowers the closed loop bandwidth from 10Mhz to about 250kHz. It's a tribute to your and your wive's ears that you can clearly hear this difference!
Jan
JD,
Would you expect distortion to rise, esp at high freqs? I would. When you add more test tones and more level, distortion rises. 30 more and that could be audible IMO.
THx-RNMarsh
Which is preferrable IYO when used with DAW for creating music files?
THx-RNMarsh
Ok Fair comment on my questions. If I want theory I can read a book or go ask the Wizard of OZ (Google). BTW - I accept the validity of dither... always have. Its the experience in application for making files of HiRez vs CD that is the question for me.... keep that in focus. And, if You are too busy to answer in this DIY thread, then I certainly would not want to distract you from your important work. I am retired however and have plenty of time to ask questions from those prof who do have the time to give their worldly experience in application of theory.
THx-RNMarsh
Dick, see my follow-up post to that with the data sheet stuff. At this point in time, having no further detailed info, I do not think he nor his wife heard it.
Jan
Last edited:
The statement was about 'dull sound'. As in drooping freq response. Which, in an intuitive way, can look entirely reasonable. Until you look at the facts.
I do not believe that a decrease in closed loop bandwidth from 10 MHz to 250 kHz is audible as 'dull sound'.
Jan
What are the mixing desk you know, and what are the OPA used in the bus mix ?Actually, going from unity gain to 30 dB noise gain (with 30 inserts) lowers the closed loop bandwidth from 10Mhz to about 250kHz. It's a tribute to your and your wive's ears that you can clearly hear this difference!
NE5532 was not early on the scene, and expensive at its time. A lot of µA741 were used.
And, by the way, just try on a bench, and make your opinion yourself, even with a NE5532 to figure out if *you* cannot hear the difference.
We listen with our ears, not theory.
Too, about theory, what about the phases at 20KHz with 250KHz of CF ?
Last edited:
Suck it and see. You can dither using plugins with cubase, other DAWs use proprietary algorithms, if you can figure out what they're doing. You've displayed considerable hostility to theory and properly substantiated evidence, you obviously don't need my opinion.
SOX seems to be the preferred tool in some quarters. Open source? You can see exactly what is being done?
If you think about writing a program to toggle LSBs so that the net error has a gaussian distribution, and then the same for a triangular distribution, then the effort, on the face of it has, has to be less in the case of the triangular, unless you've got some trick algorithm. I dunno, have you?
Dick here's another one guaranteed to make your head hurt.
http://www.robertwannamaker.com/writings/rw_phd.pdf
Much of this discussion is moot for staying in the 24 bit domain, producing 44.1/16 CD's from higher is where there is the most concern.
http://www.robertwannamaker.com/writings/rw_phd.pdf
Much of this discussion is moot for staying in the 24 bit domain, producing 44.1/16 CD's from higher is where there is the most concern.
I just did the listening test - no difference. Just like theory predicted.
What about it? Does that cause your 'dull sound'? Facts?
But I will grant you the last word as I do not think anything useful will come out of continuing this discussion.
Jan
In most summing amp designs in real world consoles I am familiar with (quite a few) there is no change in bandwidth as more or less channels are used. If a channel is muted or disconnected the summing resistor is taken to ground to keep the buss's impedance the same, so level, or any other quality, never shifts. That would be "a very bad thing". Muting is a standard part of the workflow.
Boards with voltage summing are especially careful about this and keep things constant (think old Neve's, pre inverting summing node summing adoption) as noise is very hard to keep acceptable in that design type.
If the mastering is the same for Hi-rez vs cd is a constantly changing reality and may be specific to the project. It is becoming more common on higher profile (budget) projects that the delivery spec now includes cd, hi-rez and mpeg versions. They MAY be done from the same hi-rez file, or they may not. More people are doing their mpeg creation from the hi-rez version as that is now accepted as making a "better" mpeg than the 16/44 file. This will lead you to the often correct assumption that the compression as such is about he same on the hi-rez and the cd version.
Cheers
Alan
Boards with voltage summing are especially careful about this and keep things constant (think old Neve's, pre inverting summing node summing adoption) as noise is very hard to keep acceptable in that design type.
If the mastering is the same for Hi-rez vs cd is a constantly changing reality and may be specific to the project. It is becoming more common on higher profile (budget) projects that the delivery spec now includes cd, hi-rez and mpeg versions. They MAY be done from the same hi-rez file, or they may not. More people are doing their mpeg creation from the hi-rez version as that is now accepted as making a "better" mpeg than the 16/44 file. This will lead you to the often correct assumption that the compression as such is about he same on the hi-rez and the cd version.
Cheers
Alan
Seen a lot, specialty Trident ;-(, at their begining; Rapidly changed for TL07X, yes.
Last edited:
Of course the indisputable effect of hanging a bunch more resistive inputs on the summing node is the increase in noise, unless the feedback resistor is reduced along the way. For the nominal 5nV/sq rt Hz 5532, with 30 inputs with R's the same as the feedback R, we have a noise gain of times 31 and in a 22kHz bandwidth an output noise of at least 23uV rms, and this neglects the resistors' thermal noise. If Rf is 1k, the thermal noise will be mostly from the paralleled input resistors and if they are 1k each, we wind up with another 3.4uV rms, so relatively small compared to the op amp voltage noise (the op amp current noise effect will be still smaller). If we swing to 20V peak at the output (running at abs max rails, risky) or 14.1V rms, the signal-to-noise ratio at best will be about 116dB, referenced to 1V rms out about 93dB. A 5532 could just manage the output current. Not too bad, and likely the noise in the sources will dominate anyway.
Thanks, Scott. A real brain buster.
I think I found what I am looking for already. Some late comers to the party here think I dont understand dither. I think my inquiry about dither as applied by DAW music creation software has been identified as potential trouble. Thats as far as i need to go for now. It might not explain well the total differences between CD and hi-rez download master files. It just has the potential for abuse and making music sound less real, if over-used, and more canned/hi-fi-ish..... which we have plenty already.
THx-RNMarsh
Last edited:
Hi Richard,
It is pretty late to respond to your question on CD player transports much earlier, but it looks like it may help. Here goes:
Computer drives read data at higher rates, but the audio error correction is not the same as a data disk. Bit perfect audio data retrieval is a myth. Now, for actual computer data you obviously do need perfect retrieval, and the data disc error correction makes that possible even with the high BLER rates that all CDs and DVDs have.
As for which is better, I have to say that for audio use, the audio transport is far superior if you look at the error rate in the retrieved data. A C2 error count would arbitrate this question, wouldn't it? I guess the question you pose might be considered on a convenience view point for the consumer. The public is both unaware and more often will accept badly flawed performance from any or all of their equipment. I make this statement after over 30 years of observing what is brought in for repair and the condition we find it in. Even new equipment is typically not adjusted properly. The buying consumer hasn't got a chance when evaluating products during the sales process. The added confusion injected by the high end audio market has made the situation even worse due to the flawed information the market is flooded with.
Look at this another way. No matter what the medium is where stuff is moving, the more consistent the reading device is passing the information to read it, the better your information is going to be. For records that means a more constant angular velocity. For CD / DVD / tape that means a constant linear velocity. One thing everyone should clearly see is that reading a CD is different than reading a record disc. So why do some manufacturers add "stable platters", or even drive the disc table with a belt for some higher end CD players??? They crippled the transport, and CEC ought to have known better. It is all marketing folks, quality be damned.
-Chris
It is pretty late to respond to your question on CD player transports much earlier, but it looks like it may help. Here goes:
The two drives are completely different. The audio version normally runs through the track one time only. The computer based one will take several stabs at retrieving the data as they store the information in a much larger buffer than do most audio machines. But they spin the disc faster which amplifies the inertia of the spinning disc. The poor little disc motor can't speed up or slow down quickly enough to read through errors due to off center or warped discs. The quality of the data is far better with a good audio transport, the NEC made for Nakamichi OMS 5/7 being the very best. This transport was used by a few other manufacturers as well. The eye pattern from this transport was just about perfect in every aspect. I have rebuilt many of these and many other types of audio players.
Computer drives read data at higher rates, but the audio error correction is not the same as a data disk. Bit perfect audio data retrieval is a myth. Now, for actual computer data you obviously do need perfect retrieval, and the data disc error correction makes that possible even with the high BLER rates that all CDs and DVDs have.
As for which is better, I have to say that for audio use, the audio transport is far superior if you look at the error rate in the retrieved data. A C2 error count would arbitrate this question, wouldn't it? I guess the question you pose might be considered on a convenience view point for the consumer. The public is both unaware and more often will accept badly flawed performance from any or all of their equipment. I make this statement after over 30 years of observing what is brought in for repair and the condition we find it in. Even new equipment is typically not adjusted properly. The buying consumer hasn't got a chance when evaluating products during the sales process. The added confusion injected by the high end audio market has made the situation even worse due to the flawed information the market is flooded with.
Look at this another way. No matter what the medium is where stuff is moving, the more consistent the reading device is passing the information to read it, the better your information is going to be. For records that means a more constant angular velocity. For CD / DVD / tape that means a constant linear velocity. One thing everyone should clearly see is that reading a CD is different than reading a record disc. So why do some manufacturers add "stable platters", or even drive the disc table with a belt
for some higher end CD players??? They crippled the transport, and CEC ought to have known better. It is all marketing folks, quality be damned.-Chris
Thank you SyncTronX for this.
If anyone is interested on that, more details can be found in these two patents:
US4032855
US4117412
For commercial products, see the RIAA preamp of the Advent Model 300 receiver.
Behaviour is predictable, thus straightforward. But why shoot the messenger?
Esperado should only have mentioned from the start that his expressed experience was from around 1970, the same years that the attachment was printed, yes 741s
George
>Edit. For a unity gain mixer, the problem does not really exist
Attachments
Last edited:
actually, we can now go to a discrete 741! In a large format package!
XL741 Discrete Op-Amp Kit
have fun with that.
Alan
- Status
- Not open for further replies.