actually some point out that DAW SW allow literally pulling on individual sample points with mouse
so the DAW user could easily create artificial clips, introduce errors that would interpolate to above 0 dBFS with proper reconstruction filtering
a safer DAW editing process is having the DAW display a upsampled, interpolated, filtered Gibbs'ed "anlalog" output - and comparing that wavefrom to the 0 dBFS
Nyqusit-Shannon, et. al still works fine with Amplitude, Bandwidth Limited input - but standard pro audio processing uses nonlinear effects in the digital domain, can then cause harmonic and IMD products to be added in the digital processing that violate the conditions, require renormalizing amplitude, filtering the new nonlinear harmonic and IMD sum products before continuing
which is too cumbersome to redo at every effects step, so most just work with wordlength headroom, usually at 96k with most digital audio still being downsampled, released at lower sample rate too
so the DAW user could easily create artificial clips, introduce errors that would interpolate to above 0 dBFS with proper reconstruction filtering
a safer DAW editing process is having the DAW display a upsampled, interpolated, filtered Gibbs'ed "anlalog" output - and comparing that wavefrom to the 0 dBFS
Nyqusit-Shannon, et. al still works fine with Amplitude, Bandwidth Limited input - but standard pro audio processing uses nonlinear effects in the digital domain, can then cause harmonic and IMD products to be added in the digital processing that violate the conditions, require renormalizing amplitude, filtering the new nonlinear harmonic and IMD sum products before continuing
which is too cumbersome to redo at every effects step, so most just work with wordlength headroom, usually at 96k with most digital audio still being downsampled, released at lower sample rate too
Last edited:
That's still artificially created. I don't see how the full scale impulse between sampling points could get presented to the A/D or D/A process in normal recording of sound. Audacity presents the continuous sinc interpolated signal when CD data is ripped I don't notice a problem I would worry about, yet it's easy to to push samples around to create the examples shown.
EDIT - OK so heavy effects laded music can cause mastering problems but I have yet to see 44.1 CD data that has more than trivial "overs" unless it's simple hard clipping of music I don't usually listen too. It's easy to find examples with clipping almost continuously happening.
Last edited:
One audio power amplifier manufacturer prefers to use almost no power transistors in their output stages. In the past they have used parts well known in the industry above the manufacturer's ratings. So the part everyone else used with a 150 volt limit they ran at 200 volts. Everyone else used six or more in an output stage they used 4. They do have the manufacturer test their shipments to a different standard than the data sheet parts. One side effect is that to repair one of their amplifiers I have to buy their house branded part for $18.00 instead of the standard version for $2.00.
As to bit size the MIDI standard allows 7 bit for "Velocity" which corresponds to volume. So if you allow 1/2 bit for clipping margin, 7 to go from ppp to fff, that leaves 8 1/2 bits for playing a ppp (very very soft) passage from a 16 bit recording.
As to bit size the MIDI standard allows 7 bit for "Velocity" which corresponds to volume. So if you allow 1/2 bit for clipping margin, 7 to go from ppp to fff, that leaves 8 1/2 bits for playing a ppp (very very soft) passage from a 16 bit recording.
hi Scott, as usual I assume you know what your talking about, sorry if I missed a distinction you are making in my trying to catch back up,
as I added to the above, audio mixing/production does introduce hard clipping, other processes that can violate the conditions for well behaved sampled data
some claim Michael Jackson adopted digital hard clipping as a intentional effect - with no flitering/rate limiting allowance for Gibbs
as I added to the above, audio mixing/production does introduce hard clipping, other processes that can violate the conditions for well behaved sampled data
some claim Michael Jackson adopted digital hard clipping as a intentional effect - with no flitering/rate limiting allowance for Gibbs
maybe I misremember - but I thought I recalled a BAS discussion on overs and thought it was a MJ track being shown - with flat topped but otherwise pretty clean bass sinewave
it was pointed out that it strictly didn't hit 0 dB, was clamped to ~ -0.1 dB FS, but still "clipped" for maybe 20-30 % of the sine period
it was pointed out that it strictly didn't hit 0 dB, was clamped to ~ -0.1 dB FS, but still "clipped" for maybe 20-30 % of the sine period
Last edited:
-48 just doesn't sound as good! So do you compress the loud parts, boost the soft stuff or leave the dynamic range intact? Of course not an issue if you have an NC 35 room and listen at 75-80 dB levels.
OT (or is it?), I'm dying to see your 20pV/rtHz amplifier mentioned in the now closed thread, have mercy
.
I have an RCA/JVC 45RPM "super" disk with passages 45dB below max, they are perfectly audible and sound fine in context even LP to 24/96 with a USB sound card. 10yr. ago the A/D designers had not figured out how many corners could be cut before the masses could tell.
OT (or is it?), I'm dying to see your 20pV/rtHz amplifier mentioned in the now closed thread, have mercy
The mention of transformer and multiplier indicates more to the story. The details will be extracted by sessions of 20 questions.
With internal circuitry composed of current sources most everywhere and current limited/protected OPS et al it is conceivable that the circuitry will function very well at higher voltages. Even IF there would be some small degradation over time, it would be quite long for consumer use...
Maybe 'we' can do some short and long term tests on those popular opamps (VFA and CFA) for audio apps and see if we can get some working directly at +/- 24vdc.
THx-RNMarsh
Maybe 'we' can do some short and long term tests on those popular opamps (VFA and CFA) for audio apps and see if we can get some working directly at +/- 24vdc.
THx-RNMarsh
Last edited:
Thanks for that cool link - check out the Rush and Dream Theater levels - wow !
I was not denigrating modern SACD's ? (only way to clip my amp)
)
What's the point of that ? Does it make for better reproduction ? No emitter
R ? versus one with non-inductive R's - would it sound "better" ?
PS - marketing BS.
HUH ? All is intermingled .... digital starts (or ends) as analog somewhere - even my
fancy pants super sound card is/was "{soiled) by a OPA2134 or ne5532.
yeah , just want to hear lots of that Johnson noise.
OS
Oooh...... people like.... You?
Anyone. I have been away in Asia on more R&R from the last R&R but when I get back to the real world, I can put a +/- 24v supply on a few opamps powered in parallel and let it run for a long time and then remeasure. But who knows when I will do that.
I am going to have to live much longer than my insurance company is counting on if I keep thinking of things to do.
-RNM
Last edited:
I love defying the actuarial tables. I could have been drawing SS for six years now.
I think that over-voltaging IC's is a bad idea. Think what might happen if one broke down prematurely when attached to an amp and speaker. Rather pointless isn't it? I DO run up to +/- 18V sometimes, as that is what many of them are rated for, but for long term, I usually use +/-15V just to give a 'guardband' as I expect my designs to run a number of years without trouble. Prototypes? Who cares? For example, in a recent prototype, I accidentally put some 18V Zeners in, instead of 15V Zeners that were specified. I left them in. The prototype will never be commercially available to anyone else. IF it was a production unit, I would put 15V Zeners in, just for a safety margin, normally.
- Status
- Not open for further replies.