Nasty Clipping on S/Pdif transfer.
Richard Kuschel
ambitious and red-Lined my DAT recorder. On playback -no audible
distortion.
I transferred the file through s/pdif into my Mastering program (which
will currently go unnamed as I am in discussions with their tech
people) and got audible distortion on the transfer in. I had
previously transferred this part in earlier on a previous version of
the program and got no distortion.
I then tried the s/pdif transfer via Pro Tools. Everything was fine.
I imported the files from the new program and the old program into
ProTools to make comparisons. the PT and Old Mastering program files
were identical. The files from the new program had the absolute
polarity inverted and where the signal hit the top rather than
flatlining, it produced an inverted wave that ran to the other rail
and lasted as long as the over and then back to the top. The length of
the over was seven samples.
Really nasty on speakers.
Same Equipment -digi 002 box, Panasonic 3900 DAT.
I tried a CD made directly from the DAT and got the same result.
I then tried an analog recording and still got the same result. The
old program and ProTools accepted the overs with no problem. The new
program produced the same inversion of absolute polarity and Nasty
distortion.
I was able to stop the problem by running the s/pdif signal through a
digital compressor before recording on the new program, but the signal
still had inversion of absolute polarity.
Anybody ever heard of this before?
Don Pearce
Oh yes. I've already posted this on another thread.
http://soundthoughts.co.uk/look/badclip.gif.
This was the result of some software pretending to be a DAC in order
to produce a .wav file from Midi.
d
Peter Larsen
> I recorded a concert in which the concert Bass Drum Player got really
> ambitious and red-Lined my DAT recorder. On playback -no audible
> distortion.
There shouldn't be, my SV3800 flatlines cleanly, I except the 3 head version
to do likewise, in case of an overload.
> I transferred the file through s/pdif into my Mastering program (which
> will currently go unnamed as I am in discussions with their tech
> people) and got audible distortion on the transfer in. I had
> previously transferred this part in earlier on a previous version of
> the program and got no distortion.
I would take convert the file to 32 bit format, unclip and either lower the
level or hard limit or a combination thereof, but that is just because I am
finicky.
> I then tried the s/pdif transfer via Pro Tools. Everything was fine.
> I imported the files from the new program and the old program into
> ProTools to make comparisons. the PT and Old Mastering program files
> were identical.
As it should be.
> The files from the new program had the absolute
> polarity inverted and where the signal hit the top rather than
> flatlining, it produced an inverted wave that ran to the other rail
> and lasted as long as the over and then back to the top. The length of
> the over was seven samples.
> Really nasty on speakers.
> Same Equipment -digi 002 box, Panasonic 3900 DAT.
> I tried a CD made directly from the DAT and got the same result.
> I then tried an analog recording and still got the same result. The
> old program and ProTools accepted the overs with no problem. The new
> program produced the same inversion of absolute polarity and Nasty
> distortion.
> I was able to stop the problem by running the s/pdif signal through a
> digital compressor before recording on the new program, but the signal
> still had inversion of absolute polarity.
> Anybody ever heard of this before?
Sounds like codewriting at 4 am or as the late night error I just found in a
job application after sending it ... reckon I'll send an update right away.
Kind regards
Peter Larsen
Scott Dorsey
>I then tried the s/pdif transfer via Pro Tools. Everything was fine.
>I imported the files from the new program and the old program into
>ProTools to make comparisons. the PT and Old Mastering program files
>were identical. The files from the new program had the absolute
>polarity inverted and where the signal hit the top rather than
>flatlining, it produced an inverted wave that ran to the other rail
>and lasted as long as the over and then back to the top. The length of
>the over was seven samples.
>Really nasty on speakers.
>
>Same Equipment -digi 002 box, Panasonic 3900 DAT.
>
>I tried a CD made directly from the DAT and got the same result.
>
>I then tried an analog recording and still got the same result. The
>old program and ProTools accepted the overs with no problem. The new
>program produced the same inversion of absolute polarity and Nasty
>distortion.
>
>I was able to stop the problem by running the s/pdif signal through a
>digital compressor before recording on the new program, but the signal
>still had inversion of absolute polarity.
>
>Anybody ever heard of this before?
I have seen converters that have done this before. What you need is a
bitscope to see if the actual data coming out of the DAT player is doing
it.
You might try a different DAT player if it is.
The thing is, nothing "accepts overs." An over is entirely in the mind
of the user and in the configuration of the metering. But some converters
do roll over when they hit full scale... I haven't seen anything that has
done it for more than a decade, but it happened.
The solution is... don't ever go over....
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Richard Crowley
> The thing is, nothing "accepts overs." An over is entirely in the mind
> of the user and in the configuration of the metering. But some converters
> do roll over when they hit full scale... I haven't seen anything that has
> done it for more than a decade, but it happened.
Sounds like nothing more than ignorant and/or sloppy coding of
the driver. Wraparound or "roll-over" are mathematical phenomenon,
but not physical waveform phenomenon.
Scott Dorsey
I've seen it being a case of sloppy hardware design rather than software
design.
I know a lot of instrumentation converters do this also.. when I asked
why they didn't add extra hardware for checking for the condition I was
told that clipping so so inherently bad that it doesn't matter WHAT happens
when it clips.
mikea
But when you hit a "can't handle this" condition, the action the gear
takes can be very important to the user. Ideally, the equipment would
give you a configuration option: roll over, clip hard-line at the
extreme, or _something_. If nothing else, the specs for the gear should
include a description of its behavior in this particular situation.
Ideally, of course, the gear would never be presented with signal levels
past the design limits, too, but we live in the real world.
--
Mike Andrews, W5EGO
mi...@mikea.ath.cx
Tired old sysadmin
Richard Kuschel
It isn't in the hardware.
I get two results using the same hardware but different recording
programs.
This thread has pretty much confirmed what I suspected. it's the
software.
You shouldn't be able to hear 7 overs on a peak. Im did manage to get
the same result with three samples #1 and #3 were at full scale and #2
was at full negative scale. Extremely audible as a click.
At least I now know that it is called wraparound or rollover. That
doesn't account for the inversion of absolute polarity. That has to be
sloppy code writing.
I know i shouldn't hit the limits, but as Mikea said "We live in the
real world. "
It's gonna happen.
Thanks for the confirmation.
Mark
>
> - Show quoted text -
I have seen op amp circuits that will invert or wraparound when
overdriven too hard. And yes the sound this makes is MUCH MUCH worse
then simply flat limiting. I'm not sure in your situation if this
inversion on limit is happening in the digital domain or the analog
domain but your options are:
1) redesign the equipment / software
2) change the equipment / software
3) lower the level into the stage or process where the inversion
happens.
Mark
Les Cargill
> I recorded a concert in which the concert Bass Drum Player got really
> ambitious and red-Lined my DAT recorder. On playback -no audible
> distortion.
>
> I transferred the file through s/pdif into my Mastering program (which
> will currently go unnamed as I am in discussions with their tech
> people) and got audible distortion on the transfer in. I had
> previously transferred this part in earlier on a previous version of
> the program and got no distortion.
>
> I then tried the s/pdif transfer via Pro Tools. Everything was fine.
> I imported the files from the new program and the old program into
> ProTools to make comparisons. the PT and Old Mastering program files
> were identical. The files from the new program had the absolute
> polarity inverted and where the signal hit the top rather than
> flatlining, it produced an inverted wave that ran to the other rail
> and lasted as long as the over and then back to the top. The length of
> the over was seven samples.
> Really nasty on speakers.
>
Clearly, the new program is broken. Continue discussions with tech
support. Congratulations! You just donated a test vector.
> Same Equipment -digi 002 box, Panasonic 3900 DAT.
>
> I tried a CD made directly from the DAT and got the same result.
>
> I then tried an analog recording and still got the same result. The
> old program and ProTools accepted the overs with no problem. The new
> program produced the same inversion of absolute polarity and Nasty
> distortion.
>
> I was able to stop the problem by running the s/pdif signal through a
> digital compressor before recording on the new program, but the signal
> still had inversion of absolute polarity.
>
> Anybody ever heard of this before?
They can't do math, apparently.
--
Les Cargill
anahata
> I have seen op amp circuits that will invert or wraparound when
> overdriven too hard.
With opamps that do that, the amount of overdrive has to be excessive.
If I understand it correctly, the problem described here is wrap around
in the digital domain when the signal goes just 1 lsb unit over the top -
easy to understand how it happens but also easy to avoid and therefore
inexcusable in a working design.
--
Anahata
ana...@treewind.co.uk ==//== 01638 720444
http://www.treewind.co.uk ==//== http://www.myspace.com/maryanahata
anahata
> But when you hit a "can't handle this" condition, the action the gear
> takes can be very important to the user.
Indeed. Some folks believe that the reason why two amplifiers sound
different in use despite have the same measured performance has to do
with how they handle clipping, along with the notion that brief clipping
occurs quite frequently in real life.
Response to clipping events is certainly something that's not often
characterised in amplifier specs.
PStamler
an automatic sample-rate conversion when it opens or records a signal?
I've seen that a few times, even when the original sample rate was the
desired one. If so, it may be generating further overs, which manifest
themselves as wraparounds.
Peace,
Paul
Richard Kuschel
There's no automatic sample rate conversion in the software, and the
sample rate conversion it does takes forever,
The hardware is a digidesign 002 which has never shown problems in the
three years I have owned it. A/D D/A conversions have always been
decent and in absolute polarity.
Really certain it is the new software.
I haven't heard much from the manufacturer lately which means either
that they know they have a problem and don't want to admit it, or they
are ignoring me.