An idea for Mackie
ABU0...@unccvm.uncc.edu
All the features of the MS1202 except also include ...
1) Long throw faders (high quality)Say "Goodbye to pots!".
2) Sweep mids EQ, PFL (with metering), 4 internal buses down to a steroe main
and a stereo monitor out
3) A very flexible arrangement for 4 track recordingthat will not be a major
compromise.
Not too major a rework of the MS1202, call it the 1402
Can you do it??????????
_______________________________________________________________________________
ABU0...@unccvm.uncc.edu º Living large between jazz and metal
mpd...@unccsun.uncc.edu º No foot on the brake, both feet on the gas pedal.
a.k.a. Michael Davis º DOD# 7303 º 78 Yamaha650SE "Bob",94 Accord "daCage"
HOWARDJILL
writes:
>Not too major a rework of the MS1202, call it the 1402
While you are at it, how about an 8-10 channel stand alone mic preamp that
I can use with the 1202. Also, the battery power mod that I got from you
guys back in Feb, Worked GREAT. Maybe that mic preamp could be batteried
itself, hmmmmmmm?
Scott Dorsey
>Come out with a new $499 price point mixer with
>
>All the features of the MS1202 except also include ...
>
>1) Long throw faders (high quality)Say "Goodbye to pots!".
This will really increase the price point. The faders are often much of
the cost of the mixer. Personally, I much prefer pots to faders, too, but
the Mackie controls are much too small. I want a knob that is at least 2"
in diameter for the level control (and my good studio mixer has 5" level
controls).
>2) Sweep mids EQ, PFL (with metering), 4 internal buses down to a steroe main
> and a stereo monitor out
That really hikes the cost up. I'd say keep the electronics the same, just
add a few more channels of mike preamps.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
rick.v...@mackie.wa.com
>Come out with a new $499 price point mixer with
>
>All the features of the MS1202 except also include ...
>
>1) Long throw faders (high quality)Say "Goodbye to pots!".
>2) Sweep mids EQ, PFL (with metering), 4 internal buses down to a steroe
> main
> and a stereo monitor out
>3) A very flexible arrangement for 4 track recordingthat will not be a
> major
> compromise.
>Not too major a rework of the MS1202, call it the 1402
>
>Can you do it??????????
Hi Michael,
Thanks for the suggestions !
We are planning a bunch of cool new products that I can't talk about yet !
{G} remember what I said about our new policy ? " Oops that was in
another forum....Basically after the 8-Bus and 3204 fiasco...We are no
longer allowed to discuss new product releases until they are ready to ship.
The only exception to this new rule is the 8-Bus automation system since we
were already discussing that prior to this new policy.
All I can say is that we welcome suggestions like yours and don't be
surprised if you recognize some of these ideas in Mackie products of the
future !
[]Rick Vartian
Mackie Designs Inc.
rick.v...@mackie.wa.com
working the bugs out of our Newsgroup feed.
>In article of 8:34 AM 8/30/94, ABU0...@unccvm.uncc.edu writes:
>
>>Come out with a new $499 price point mixer with
>>
>>All the features of the MS1202 except also include ...
>>
>>1) Long throw faders (high quality)Say "Goodbye to pots!".
>>2) Sweep mids EQ, PFL (with metering), 4 internal buses down to a
> steroe
>> main
>> and a stereo monitor out
>>3) A very flexible arrangement for 4 track recordingthat will not be a
>> major
>> compromise.
>>Not too major a rework of the MS1202, call it the 1402
>>
>>Can you do it??????????
>
>
Ross Morley
Add a PFL switch to each channel as well as the solo switch.
Only takes one extra switch per channel.
Basically, I think it was a mistake to sacrifice PFL to get the
nice stereo solo. Stereo solo is nice to have. PFL is essential,
especially for live work). My board has the PFL mod, but what a
lot of work the authorized tech. had to do (cut tracks! ever heard
of jumpers?). And because it is now non-standard, I might have trouble
selling it (not that I plan to soon).
Oh, and how about PFL/solo on the tape returns?
Gripes aside, I'm very pleased with my board (24x8). And congrats
to you guys for being on the internet!
Ross
--
Ross Morley .---. email: ro...@prpa.philips.com
Philips Research Palo Alto | ___\ Phone: (415) 354 0325
4005 Miranda Ave, Suite 175 \/ \ FAX: (415) 354 0309
Palo Alto, CA 94304, USA ^^
Jeff Ulmer
Could you ask Kieth to forward the schematics for the split EQ to solo mod that
we discussed? It would be cool to be able to solo the mix B bus, I keep losing
stuff up there.
Had my 32¥8 for 6 months now (as long as I waited for it-dig, dig) and love it.
rick.v...@mackie.wa.com
>OK, here's a simple idea to improve the 8-Buss.
>
>Add a PFL switch to each channel as well as the solo switch.
>Only takes one extra switch per channel.
>Oh, and how about PFL/solo on the tape returns?
>
>Gripes aside, I'm very pleased with my board (24x8). And congrats
>to you guys for being on the internet!
>
Thanks for your input Ross ! I will leave the replies to your comments to
our tech support guys who are also reading these.
We are glad to be here on the Internet and we appreciate those who
appreciate us doing it ! {G}
[]Rick.
Casimir J Palowitch
<rick.v...@mackie.wa.com> wrote:
>Thanks for your input Ross ! I will leave the replies to your comments to
>our tech support guys who are also reading these.
>
>We are glad to be here on the Internet and we appreciate those who
>appreciate us doing it ! {G}
Well, all hail and rejoice. Now, after just reading about five of these
congratulatory posts, could you utilize the 'reply-by-mail' option on your
newsreader rather than the 'followup' for these messages directed at
individuals?
Much appreciated.
--
*+* Casey Palowitch, Advanced Technology Projects Librarian
+*+ University of Pittsburgh Library Systems
+*+ cj...@pitt.edu NeXTmail: c...@acid.library.pitt.edu
*+* Web: http://www.pitt.edu/~cjp/
Jeff Martini
> , rick.v...@mackie.wa.com writes:
>We are planning a bunch of cool new products that I can't talk about yet
!
Rick, can you tell me, are the 16channel-8 bus mixers shipping? I've
only seen the 24 X 8 around.
I second the previous post on developing a new, say, 12 X 2 mixer with
some of the whizz bang faders and EQ from the 8-bus. Maybe even the
talkback too?! Keep us informed
Jeff
Andrew Wing
How about doing the same thing to mixers that was done with PCs?
Offer some sort of 'open architecture' system:
1) A basic backplane and power supply buss.
2) 'all in one' budget channel cards.
3) more specialized slot cards, balanced or unbalanced front ends
4) simple three band eqs or full four band parametrics
5) sub group cards
6) matrix cards with or without headphone amps
7) effect cards with reverb, eq, compression, etc...
8) mini patch bays
You could offer backplanes in a basic 8 channel configuration
and make them connectable for 16/24/32/etc channel setups.
With such a variety of plug and play modules, an engineer could
create their own custom console suited for studio use, live sound or
TV/radio production/broadcast.
<<fantasy mode on>>
Wouldn't it be nice if *all* manufacturers came up with standard
audio buss structures like ISA/EISA/PCI for PCs. By having a standard
buss to plug into, the manufacturers could concentrate on making the
plugin modules as good as possible, that way when newer technology
comes out, the engineer would only have to replace the cards, not the
whole chassis as well.
<<fantasy mode off>>
--
A fool and his net access soon go their separate ways.
"Any disclaimer issued by me is subject to change without notice"
Andy Wing Temple U. Computer Services agw...@astro.ocis.temple.edu
Mike Rivers
> How about doing the same thing to mixers that was done with PCs?
> Offer some sort of 'open architecture' system:
> You could offer backplanes in a basic 8 channel configuration
> and make them connectable for 16/24/32/etc channel setups.
One of the most costly (both in dollars and reliability) thing you can
put into a mixer is connectors. Such an architecture would push the
cost up (or the reliability down) so that it wouldn't be a good
marketing decision.
> Wouldn't it be nice if *all* manufacturers came up with standard
> audio buss structures like ISA/EISA/PCI for PCs. By having a standard
> buss to plug into, the manufacturers could concentrate on making the
> plugin modules as good as possible, that way when newer technology
> comes out, the engineer would only have to replace the cards, not the
> whole chassis as well.
There are a few consoles (and third party manufacturers) who are
building like that, but the price is out of Mackie's class. For
instance, you can get API mic preamps and EQ's, and GML automated
fader modules to go into a Sony MXP-3000 series console, but that's a
$75,000+ console.
> <<fantasy mode off>>
It's nice to fantasize, but more practical to look at what the
fantsies cost.
--
I'm really Mike Rivers (mri...@d-and-d.com)
Mike Porter
>Well, all hail and rejoice. Now, after just reading about five of these
>congratulatory posts, could you utilize the 'reply-by-mail' option on your
>newsreader rather than the 'followup' for these messages directed at
>individuals?
I tried; my mail to them gets bounced. I VERY highly second the recommendation
that they expand the EQ on the 1604, but I can't tell them that without going
over the 'net, at least so far.
---Michael...
Bryan Levin
> Offer some sort of 'open architecture' system:
> 1) A basic backplane and power supply buss.
[...]
> You could offer backplanes in a basic 8 channel configuration
> and make them connectable for 16/24/32/etc channel setups.
> With such a variety of plug and play modules, an engineer could
> create their own custom console suited for studio use, live sound or
> TV/radio production/broadcast.
This is EXACTLY what is being done in the network "hub" (physical backplane)
market. Each vendor has released a hub, and you buy the modules that apply to
you (repeaters, comm servers, bridges, etc). When I was back at DEC, our hub
costed under $2k, with typical module prices of $1000. Note, this was
high-speed LAN stuff; IMHO, more complicated and technical than 'regular audio
stuff' (gee, I'll get flamed for that, I'm sure!)
When you need different port capabilities, you buy new plug-in's.
Unfortunately, there is almost no cross-vendor development (other than special
'partner' relationships). This is actually a plus for manufacturers - can you
say 'account control'? ;-)
I see no reason at all why this could not be done for audio use, at even
"mackie" prices. Market share, product recognition, corporate direction: all
these would affect a decision to enter this realm, but believe me, it CAN be
done for much less than you think. (At DEC, we even had our modules built in
the USA - and the typical cost was under $400 per board!)
I've often though of doing this myself. Making a backplane that accepts
various filter modules - sort of a user-expandable parametric. You need to
cut this freq? Fine, plug in a board to notch it out and adjust. Need some
highpass, plug it in. You just insert what you need, and nothing more.
With LAN hubs, you even have multiple 'virtual backplanes'. You can choose
which bus (out of 64, say), the INs and OUTs are connected to. Yup, a patch
panel, but done with software (firmware, actually).
Sigh: I wish this stuff was commercially available, and at 'reasonable' prices
(ie, not a huge markup due purely to market demand). I guess you have to
CREATE a demand first, either by producing/pushing some pilot products, or
selling the experts, mags, etc on this idea and following up with product.
Given enough demand, the circuits could ultimately be done on custom VLSI, to
lower cost. Most hub vendors have an initial market offering first; a
low-tech version, if you will, then a cost-reduced version to follow (once
they see FCS revenue, their concept is now 'proven').
--
.bl
#
# web: http://www.hal.com/services/juggle/home/ble...@netcom.com
# ftp://ftp.netcom.com/pub/blevin/index.html
Technically Sweet
This is only feasible in an all-dig mixer. At some point it
will be standard for $5000+ mixers, I'd guess. It would need
mono, stereo, stereo+smpte connector options I'm guessing?
--
Lance Norskog
thi...@netcom.com
Artisputtingtogether. Art s th ow n aw y.
Chris Christensen
>[ modular computer-style mixer bus idea ]
>This is only feasible in an all-dig mixer. At some point it
>will be standard for $5000+ mixers, I'd guess. It would need
>mono, stereo, stereo+smpte connector options I'm guessing?
I have been dreaming about this since 1988. I ahve even designed most of
the user interface.... But I'm and audio engineer stuck in a video
design and manufacturing company.....
--
D.R. "Chris" Christensen Grass Valley Group (the day job)
chr...@fuggles.gvg.tek.com P.O. Box 1114 mail Stop N32B
916-478-3419 FAX 916-478-4195 Grass Valley, CA 95945
Neither I nor my employer is responsible for anything I say or do.
rick.v...@mackie.wa.com
>[ modular computer-style mixer bus idea ]
>
>This is only feasible in an all-dig mixer. At some point it
>will be standard for $5000+ mixers, I'd guess. It would need
>mono, stereo, stereo+smpte connector options I'm guessing?
>
One thing to keep in mind here. Although a fully digital mixer can provide
a lot of added automation control over EQ etc., until 24 BIT AD/DA becomes
affordable the dynamic range and overall performance of a digital board will
not equal that of a good analog mixer.
[]Rick.
Richard D Pierce
<rick.v...@mackie.wa.com> wrote:
>
>One thing to keep in mind here. Although a fully digital mixer can provide
>a lot of added automation control over EQ etc., until 24 BIT AD/DA becomes
>affordable the dynamic range and overall performance of a digital board will
>not equal that of a good analog mixer.
Are you suggesting, Rick, that a "good" analog mixer (I presume yours are
"great" :-) has 144 dB of dynamic range? For, say, a 10 volt RMS maximum
output capability, its unweighted noise floor is 144 dB lower than that,
meaning that the total broadband noise floor on output is 0.5 microvolts?
I doubt it, VERY seriously. I have NEVER seen this kind of performance
out of a "good" mixer. In my lab, I have laboratory microphone preamplifiers
that are MUCH better than the best mixer (yes, even a Mackie, which is a
pretty good mixer), that barely makes it to a broadband dynamic range of
130 dB, and those are at least 20 dB BETTER than the very best analog
mixers I have seen.
A little hyperbole? Or maybe some confusion about the number of bits
needed to fully represent a given dynamic range? Hmmm...
--
| Dick Pierce |
| Loudspeaker and Software Consulting |
| 17 Sartelle Street Pepperell, MA 01463 |
| (508) 433-9183 (Voice and FAX) |
rick.v...@mackie.wa.com
>
>A little hyperbole? Or maybe some confusion about the number of bits
>needed to fully represent a given dynamic range? Hmmm...
Dick,
The dynamic range is not simply the number of bits x 6.
Once analog is above the noise floor it has infinite resolution. 16 bit
digital systems are operated with about 20dB of headroom. Which makes
nominal zero at 20dB below maximum output, leaving around 72dB.
Our feeling is that 24 bits would give us a similar performance to analog.
Most A/D D/A converter manufacturers are producing 20 bit parts at this
time. These parts are quite expensive considering the sheer number required
in a typical high end console.
<<<For, say, a 10 volt RMS maximum
output capability, its unweighted noise floor is 144 dB lower than that,<<<
The converters on the market today do not produce levels anywhere near 10
volts RMS.
<<< I have NEVER seen this kind of performance
out of a "good" mixer. In my lab, I have laboratory microphone preamplifiers
that are MUCH better than the best mixer (yes, even a Mackie, which is a
pretty good mixer), that barely makes it to a broadband dynamic range of
130 dB, and those are at least 20 dB BETTER than the very best analog
mixers I have seen.<<<<
I am confused...Why are you comparing a Mic preamp to a mixer ? I don't see
what this has to do with the subject at hand.
We are very involved in digital technology and we *DO* intend to have fully
digital mixer when the time is right. This forward motion is the
appropriate action for us to take in our marketplace.
Thanks for your input.
[]Rick Vartian
Automation Projects Manager
Mackie Designs Inc.
Technically Sweet
Yes and no. The AES/BEU standard is 24 bits deep; assuming that quality
effects boxes sample at 18-20 bits and send 24 bits. The next effects
box accepts and sends 24 bits to the mixer. The mixer doesn't need
all analog inputs...
Richard D Pierce
<rick.v...@mackie.wa.com> wrote:
>In article of 10:12 PM 9/15/94, DPi...@world.std.com (Richard D writes:
>>
>>A little hyperbole? Or maybe some confusion about the number of bits
>>needed to fully represent a given dynamic range? Hmmm...
>
>
>Once analog is above the noise floor it has infinite resolution. 16 bit
>digital systems are operated with about 20dB of headroom. Which makes
>nominal zero at 20dB below maximum output, leaving around 72dB.
Rick, this a common myth: the resolution of ANY system (analog or digital)
can only be infinite if the dynamic range is infinite and/or the bandwidth
is infinite. A limited bandwidth, limited dynamic range system CANNOT
HAVE infinite resolution, as Shannon demonstrated 50 years ago.
><<<For, say, a 10 volt RMS maximum
>output capability, its unweighted noise floor is 144 dB lower than that,<<<
>
>The converters on the market today do not produce levels anywhere near 10
>volts RMS.
Nonsense, the converters in the analog section of the DSE 7000 produce 10
volts RMS with full 16 bit resolution and have an unweighted broadband
(measured, not spec'ed) noise floor of about 0.2 mV. Further, the sonic
performance due to the proprietary noise shaping and dither that we do is
damned near indistiguishable from analog noise at levels below 20 dB above
the quantization floor.
Regardless, say it's 5 volts or 1 volt, you're still in the realm of
physically unrealizable numbers.
><<< I have NEVER seen this kind of performance
>out of a "good" mixer. In my lab, I have laboratory microphone preamplifiers
>that are MUCH better than the best mixer (yes, even a Mackie, which is a
>pretty good mixer), that barely makes it to a broadband dynamic range of
>130 dB, and those are at least 20 dB BETTER than the very best analog
>mixers I have seen.<<<<
>
>I am confused...Why are you comparing a Mic preamp to a mixer ? I don't see
>what this has to do with the subject at hand.
Because the lab mic preamps I use are a LOT better than the best mixer I
have ever seen. And at 130 dB, they don't reach the 144 dB dynamic range
you suggest is necessary to meet the dynamic range performance of your
mixers. Thus, I suggest there's a bit of hyperbole in your statement.
>We are very involved in digital technology and we *DO* intend to have fully
>digital mixer when the time is right. This forward motion is the
>appropriate action for us to take in our marketplace.
That's good, but let's understand exactly what is meant by dynamic range
and resolution, because the two are intricately and (most importantly)
inseparably tied together (regardless of the encoding, storage or
transmission methodology, analog, digital or whatever notwithstanding),
despite the oft-held notion that they are not.
William K. McFadden
Richard D Pierce <DPi...@world.std.com> wrote:
>In article <4.47707593.M...@mackie.wa.com>,
> <rick.v...@mackie.wa.com> wrote:
>>
>>until 24 BIT AD/DA becomes
>>affordable the dynamic range and overall performance of a digital board will
>>not equal that of a good analog mixer.
>
>Are you suggesting, Rick, that a "good" analog mixer (I presume yours are
>"great" :-) has 144 dB of dynamic range?
Maybe he meant that an A/D with 24 bits of resolution (which would probably
have much less than 24 bits of spurious-free dynamic range) would be required
to render it inaudible. Since nobody's made one yet (as far as I know), who
knows?
--
Bill McFadden Tektronix, Inc. P.O. Box 500 MS 58-639 Beaverton, OR 97077
bi...@tv.tv.tek.com, ...!tektronix!tv.tv.tek.com!bill Phone: (503) 627-6920
How can I prove I am not crazy to people who are?
Ronald J Mann
flame just good old fashion disagreement:
I, and many others who have used it, will put my Yamaha DMC1000 up against
any analog board you care to name. I like my 1604... but seriously... and
anyway if your going from an analog console to a digital multitrack your
doin the AtoD anyway.
The point here is that there is a need to be filled. People buy digital
multitracks and they want to say in the digital domain through to mastering.
A product opportunity! By the way its not just consoles, its effects processors
as well... I find it perplexing that of all these digital effects units only
a few high priced guys have aes or spdif interfaces.
Perhaps digital isn't Mackies balliwick which it cool, but lets face it,
affordable digital gear is here and sooner or later someone (Yamaha seems
very close with the ProMix) is going to give that man (and maybe me too )
what he wants.
=Ron Mann=
Richard D Pierce
GO MACKIE <goma...@aol.com> wrote:
>In article <Cw96E...@world.std.com>, DPi...@world.std.com (Richard D
>Pierce) writes:
>
>Hi Dick,
>
>Well...It looks like we are going to have to agree to disagree on this
>one.
No, Rick, it looks like you are choosing not to disagree with me, but
with the entire, well established concepts of incormation capacity,
resolution, bandwidth and the myth of infinite resolution. This is an
area that has LOT of understanding to it.
><<<Nonsense, the converters in the analog section of the DSE 7000 produce
>10 volts RMS with full 16 bit<<<
>
>Hmmm.. Could I ask who makes this part for you ? How would one confirm
>these numbers ? I wonder why none of us here have ever seen this amount
>of output from a A/D or D/A converter before ?
Yes, you can ask, but I'm not going to tell. I can tell you that it's one
semi-custom part in an entirely proprietary design. But that doesn't mean
it's difficult. Getting the 10 volts RMS out was the easy part.
>24 Bit converters will be the first to acheive the same or better
>performance levels as analog. 20 Bit converters do not approach the 130
>dB levels of dynamic range or offer the resolution at lower audio levels
>that our mixers do.
And, again, you make this assertion, apparently either without proff or,
apparently, without understanding what the term "resolution" means.
>Some of your other comments and numbers also don't jive with my
>experience with digital technology ..But, I don't consider myself an
>expert either.
Then your reaction is entriely consistant with your claimed experience.
That's okay, as long as you make claims that you're able to support.
> So.. I also showed your first message to our engineering
>staff in New Products Development one of which is a very accomplished
>digital engineer and he along with the others were equally confused by
>your comments.
I would then suggest they condier studying Shannon and learning why
analog system (or ANY system for that matter )cannot have infinite
resolution, as you claim.
>Oh well...No point in arguing about it ..It's a big world out there, I
>guess theres room for two schools of thought on the realities of digital
>technology! {G}
Or so you think. There are NOT two schools of thought about the realities
of resolution and information capacity of limited dynamic range, limited
bandwidth system of ANY kind, despite your claims to the contrary. This is
not a matter of opinion, Rick, this is a matter both of theoretically
sound and experimentally verifiable fact.
rick.v...@mackie.wa.com
>Maybe he meant that an A/D with 24 bits of resolution (which would
> probably
>have much less than 24 bits of spurious-free dynamic range) would be
> required
>to render it inaudible. Since nobody's made one yet (as far as I know),
> who
>knows?
There are 24 Bit converters available now. They are incredibly expensive
even for use in the 100K mixer price range. This level of converter will
yield dynamic range and S/N in the same area as a good analog mixer 130 +
dBu & 95dBu
Anything less than these 24 bit converters do not provide an across the
board equivilant to an analog mixer circuit.
This information is available from any of the vendors of A/D D/A converters
if anyone would like to confirm what we have said here.
Richard D Pierce
GO MACKIE <goma...@aol.com> wrote:
>Some of your other comments and numbers also don't jive with my
>experience with digital technology ..But, I don't consider myself an
>staff in New Products Development one of which is a very accomplished
>digital engineer and he along with the others were equally confused by
>your comments.
The reason they don't jive with your experience, Rick, is because your
experience and intuition fail you in the actual realities of
band-limited, range limited systems.
I'm posting one section from an article I have written on the myths of
digital and analog audio systems. I've given this as a lecture to several
professional audio groups as well. I would suggest you read it, pass it
along to your technical staff as well. It's a good, non-mathematical
introduction to the debunking of the infinite resolution myth.
THE "MYTH" OF INFINITE RESOLUTION
A persistant myth these days is at the root of an intense and often
irrational debate between anaphiles and digiphiles (neither term is
meant as anything more than a label). That is that, somehow, an analog
representation, because it is continuous and not discrete, has more
inherent information in it than a discrete, sampled digital
representation. From one popular book:
"Digital master tapes, ..., are encoded in a sequence of
numerical bits. Because this is not the infinite sampling
rate of analog, but only a discrete sampling rate, there are
gaps between the bits." (Laura Dearborn, Good Sound: An
Uncomplicated Guide to Choosing And Using Audio Equipment,
New York: William Morrow and Company, 1987)
Shannon, in his landmark work on information theory almost half a
century ago ("A Mathematical Theory of Communication," Bell System
Technical Journal 27, 1948), advanced definitive proof that a signal
limited to a certain bandwidth and dynamic range can be faithfully
represented by a discrete, sampled system whose sample rate is twice
the maximum bandwidth with each sample represented by enough bits to
handle the maximum dynamic range.
Rather than go through a complex, formal proof of this theorem, let's
look at it by a gedanken (a thought experiment).
Let's take two systems: one is a pure analog system, with a bandwidth
of 20 to 20 kHz, with a dynamic range of 86 dB, the other is a
sampled, discrete system, with a bandwidth of 20 to 20 kHz (that
suggests a minimum sample rate of 40 kHz) and a dynamic range of 86 dB
(and that suggests 14 bit a sample length). Let's also say (for
simplicity) that the maximum output of each system is 10 volts. Which
system has "more resolution?"
Well, what do we mean by resolution? Let's settle on a definition of
resolution as the number of unambiguously knowable signal levels and
time increments in the system (or, what we might call, the number of
uniquely identifiable "states" the system can be in). More knowable
levels means more dynamic range. Smaller knowable time intervals means
more bandwidth. Higher resolution means more unambiguous states.
But wait. We've already stated that the bandwidth and dynamic range
are the same. What does that mean in reality?
Well, let's look at the practical results. In our analog system, 86 dB
of dynamic range below 10 volts means that there is a continuous
random noise level of about 1/2 millivolt. That means that at any
given instant, the signal voltage is whatever it is, plus or minus a
random value (that's the noise) equal to, on average, 1/2 millivolt.
Since this noise value is random and unpredictable, we can never know
whether the signal we read is the real one, or the real one with
additional noise (there is a very small but non-zero chance the noise
at that instant will be 0). So we can never know exactly what the
value was: there will always be that 1/2 millivolt uncertainty to any
signal level because of the noise. So 1/2 millivolt seems to be the
minimum absolute limit of our certainty about the signal, and that is
the limit of the resolution of the analog system.
In our digital system, we have a slightly different story. Here, any
change in level corresponding to less than the least significant bit
(which represents 1/2 millivolt) is lost. When we read a signal coming
out, we don't know where in any particular 1/2 millivolt interval the
original signal was. We can never know whether the signal we read was
the real one, or the real one quantized (there is a very small but
non-zero chance that the signal at that instant may be exactly what
the digital representation says it is). So, because of the
quantization, we can never know exactly what the signal was: there
will always be that 1/2 millivolt uncertainty to any signal level
because of the quantization. So 1/2 millivolt seems to be the minimum
absolute limit of our certainty about the signal, and that is the
limit of the resolution of the digital system.
Well, what's the difference? First of all, in the case of the analog
system, the uncertainty is random and thus the "distortion" products
generated by that uncertainty will have the characteristics of random
noise: uncertain, but fairly innocuous in their audible character. In
a digital system, the uncertainty is due to quantization of the
signal, thus any artifacts of that are highly signal correlated. That
means what you hear, unless other measures are taken, is a very harsh
distortion, almost square-wave like in nature when the signal levels
are very low. However, by introducting dither, which is a random noise
value corresponding to about 1/2 least significant bit or so, we now
make the artifacts behave and sound just like simple analog noise, and
we still maintain the 86 dB dynamic range. Now the digital value is
just as unpredictable as the analog value and for precisely the same
reasons.
The point is that many people equate analog with continuous, then make
a leap of logic (or faith) from continuous to infinitesimal changes
and infinite resolution. But the noise in analog systems or the
quantization in digital systems simply prevents changes below a
certain level from having any meaning.
The quote I cited above, about the "infinite sampling rate of analog,"
is sheer nonsense. To have an infinite sampling, any system, be it
continuous or discrete, must have an infinite bandwidth.
If the bandwidth of a system (for whatever reason) is limited to 20
kHz, while there may a "signal" measurable in the "interval" between
the information relevant to satisfying the constraints of that 20 kHz
bandwidth, it has NO relevance to that information. Another way of
looking at it is that if the bandwidth is limited to 20 kHz, there
WILL NOT be any information between the information of relevance. A
bandwidth of 20 kHz means that no significant change in the signal can
happen at a rate faster than once every 1/40,000 of a second. That's
that. It makes NO difference whether it's a purely analog system or a
purely discrete time and amplitude sampled digital system. The notion
of continuous does not mean infinite resolution. If there WAS some
significant change happening at a rate faster than 1/40,000 of a
second, then the bandwidth must be wider than 20 kHz.
These are two VERY important concepts:
* The resolution in time of a system is a function of the
system's bandwidth. The ability to distinguish without
ambiguity two unique events in time is limited by the
bandwidth. The wider that bandwidth, the more time
resolution the system has.
* The resolution in amplitude of a system is a function of the
system's dynamic range. The ability to distinguish without
ambiguity two unique amplitudes is limited by the dynamic
range of the system. The wider dynamic range, the more
amplitude resolution the system has.
One can then combine the two and say that, with certainty, the total
resolution of the system, its information carrying capacity, is
essentially the product of its bandwidth (20 kHz, say) and its dynamic
range. Either one or the other MUST be infinite for the resolution of
that system to be infinite. And I challenge anyone to find ANY system
that has infinite resolution, because that would require an infinite
bandwidth, and/or an infinite dynamic range. We all know that no such
system exists, be it a studio analog tape recorder, a DAT recorder, a
cassette, a mixer or amplifier, a piece of wire, a chunk of air, the
human ear.
That means, by default, that since NO system has either infinite
dynamic range or infinite bandwidth, that NO system can have infinite
resolution under ANY circumstances.
Copyright (c) 1994 by Dick Pierce.
GO MACKIE
Hi Dick,
Well...It looks like we are going to have to agree to disagree on this
one.
<<<Nonsense, the converters in the analog section of the DSE 7000 produce
10 volts RMS with full 16 bit<<<
Hmmm.. Could I ask who makes this part for you ? How would one confirm
these numbers ? I wonder why none of us here have ever seen this amount
of output from a A/D or D/A converter before ?
<<<And at 130 dB, they don't reach the 144 dB dynamic range
you suggest is necessary to meet the dynamic range performance of your
mixers. Thus, I suggest there's a bit of hyperbole in your statement.<<<
Hmmm..I don't see it as applied to my original point. But...If you want
to make an argument for arguments sake I guess you could make a point
there.. Again as applied to my original statement, the comments I made are
accurate.
24 Bit converters will be the first to acheive the same or better
performance levels as analog. 20 Bit converters do not approach the 130
dB levels of dynamic range or offer the resolution at lower audio levels
that our mixers do.
Some of your other comments and numbers also don't jive with my
experience with digital technology ..But, I don't consider myself an
expert either. So.. I also showed your first message to our engineering
staff in New Products Development one of which is a very accomplished
digital engineer and he along with the others were equally confused by
your comments.
Oh well...No point in arguing about it ..It's a big world out there, I
guess theres room for two schools of thought on the realities of digital
technology! {G}
Nice talking with you.
[]Rick Vartian
Mackie Designs
GO MACKIE
Mann) writes:
Hi Ronn,
Appreciate your comments, I have no problem discussing mixers with anyone
here ! I have just made myself a promise that I will not enter into any
future shouting matches with anyone over any subject. If a debate forms I
may try to stay in it....If it gets out of the realm of providing tech
support for Mackie I will most likely bail. When the personal attacks
start flying...I am grabbing the first modem outta here ! {G}
>>The point here is that there is a need to be filled. People buy digital
multitracks and they want to say in the digital domain through to
mastering.<<<<
But what about tracking ? You are not going to loose anything by
mastering from a digital recorder to a good analog console to DAT. You
*are* going to loose your original working dynamic range coming in to a 20
bit digital console for the first time. ( 16 and 20 bit systems ) like
I already said, 24 bit will not reduce your dynamic range beacuse it has a
dynamic range that meets or exceeds the best analog mixer.
Our point here is that the current crop of digital mixing products are not
the be all and end all of audio. None of the current digital mixers/DAW
mixers in the under 60k price range offer the same dynamic range and
performance of a comparably priced analog mixer. Tape recorders and
effects units are really not in the same catagory with a mixer, their
analog couterparts were far below the perfornace levels of the new digital
models but the same cannot be said for mixers. A high quality analog
console is *very* tough to beat ! Even with the miracle of digital. Why
would anyone want to reduce the working dynamic range of their sources
from their natural levels down to 119 or 90 db before they even hit a tape
machine/recorder ? Thats exactly what you will be doing using a 16 or 20
bit digital mixer. Just being able to process audio completely in the
digital domain does not automatically buy you superior performance to a
conventional mixer tape recorder combo.
Caution be to those who put all their eggs in the current digital mixer
basket. Although many may think they are working on the cutting edge of
technology, there are others out there using more versatile and less
expensive analog mixers that will be acheiving far more dyanamic and
realistic sounding productions.
<<<I, and many others who have used it, will put my Yamaha DMC1000 up
against any analog board you care to name. <<<
I don't doubt that but how much did your DMC1000 cost ? Does it not have
converters on every channel ? Even so...How many albums have been tracked
and mastered using the DMC1000 ? The board is not anywhere as popular
for music production as an SSL or NEVE "or even Mackie {G} is and I think
I know why. In addition to sound quality there is also the issues of
Interface. It is a real chore to produce a mixer interface that will be
acceptable by users as a better approach than a conventional layout. It
has to offer better ergonomics and a more efficient method of working
which I personally don't see on the DMC1000 and most of the other all
digital mixers today. That coupled with the fact that it really doesn't
buy you a better sounding production than a comparable analog mixer
completes my point.
No flame wars now ! {G} I respect your opinion...Please respect mine.
Thanks for your input Ronn !
Gabe Wiener
> Once analog is above the noise floor it has infinite resolution.
I'm sorry, but you're wrong. Resolution is a function of dynamic
range and bandwidth of ANY system, digital or analog. PLEASE don't
try to suggest that analog systems have infinite resolution, because
this was proven false in Shannon's most basic work on information
theory, five decades ago, copies of which can be found at your nearest
engineering library.
Let's leave this most basic concept of electrical engineering alone.
NO SYSTEM ON THE FACE OF THE EARTH has infinite resolutionn. This is
not a debatable point.
>16 bit
>digital systems are operated with about 20dB of headroom. Which makes
>nominal zero at 20dB below maximum output, leaving around 72dB.
Excuse me? You're mapping analog concepts into the digital domain
where they are no longer valid. Except in some recording
circumstances where one is recording instruments of very low dynamic
range (harpsichords, for instance), the proper way to record a source
in a digital system is to record as hot as one can without clipping.
As such, there is no concept of "headroom"....it's all a matter of
quantization of voltages, and either your analog gain structure fits
into the digital gain range or else it doesn't. If it overshoots, you
clip. If it doesn't reach 0 dBfs, you've under-recorded the signal.
Headroom is only a reasonable concept when there is a gain range in
the middle of which a component operates best.
Recording with a 20 dB safety margin is a drastic under-utilization of
the 16-bit format.
>Our feeling is that 24 bits would give us a similar performance to analog.
Eh? This doesn't follow. Kindly explain to us how 24 bits would give
performance that is "similar to analog" whereas, say, 20 bits wouldn't?
--
Gabe M. Wiener -- ga...@panix.com | "I am terrified at the thought that
N2GPZ -- PGP public key on request | so much hideous and bad music may be
Quintessential Sound, Inc. | put on records forever."
Recording / Mastering / Restoration | --Sir Arthur Sullivan
C.M. Hicks
>In article <35ffkq$j...@newsbf01.news.aol.com>,
>GO MACKIE <goma...@aol.com> wrote:
[...heavily munched...]
>>Some of your other comments and numbers also don't jive with my
>>experience with digital technology ..But, I don't consider myself an
>>expert either.
>Then your reaction is entriely consistant with your claimed experience.
>That's okay, as long as you make claims that you're able to support.
>> So.. I also showed your first message to our engineering
>>staff in New Products Development one of which is a very accomplished
>>digital engineer and he along with the others were equally confused by
>>your comments.
>I would then suggest they condier studying Shannon and learning why
>analog system (or ANY system for that matter )cannot have infinite
>resolution, as you claim.
This is a common problem. Electronic engineers seem to become
specialists too early - they become analogue or (more commonly)
digital electronics engineers. Very accomplished they may be, but it
requires an intimate understanding of *both* fields, and a whole heap
of information theory to be competent, let alone very accomplished, at
converting information between digital and analogue representations.
>>Oh well...No point in arguing about it ..It's a big world out there, I
>>guess theres room for two schools of thought on the realities of digital
>>technology! {G}
Absolutely not. When physical representations of information (such as
a CD or LP) are analysed it is found that the fundamental limitations
of bandwidth and signal to noise ratio apply equally to any system,
analogue or digital, with equal validity. It is simply the origin of
the noise floor and bandwidth limitation that differs between the two
systems.
In fact, our segregation of systems into two buckets labelled
"analogue" and "digital" is somewhat artificial. Everything is
analogue in some sense of the word - the real distinctions are between
continuous and quantised amplitude, and between discrete and
continuous time.
Read Shannon for further information (no pun intended :-)
Christopher
--
=====================================================
Christopher Hicks http://www.eng.cam.ac.uk/~cmh
c...@eng.cam.ac.uk Voice: (+44) 223 332767
=====================================================
Richard Wallingford
>>
>>A little hyperbole? Or maybe some confusion about the number of bits
>>needed to fully represent a given dynamic range? Hmmm...
>
Negative. It is precisely this noise floor that creates a so-called
ambiguity (or noise component) in any measurement of the analog signal
at *any* level. Therefore, your resolution is not infinite.
You might take your ideal DVM and measure 3.23012344 Volts but I'm here
to tell you that it is not the *true* value.
>Our feeling is that 24 bits would give us a similar performance to analog.
>Most A/D D/A converter manufacturers are producing 20 bit parts at this
>time. These parts are quite expensive considering the sheer number required
>in a typical high end console.
You'll have to produce some numbers to prove that 24 bits is necessary.
I can almost guarantee you that those last few bits of A/D will be
digitizing the noise component of your analog signal. It's easy
to figure out exactly how many bits will be useless (wasted) by knowing the
dynamic range (noise level and maximum value) of your input signal.
To put it another way, if you move your fader such that you change
the least significant bit on your D/A converter, the change in
voltage at the output of the D/A will be *WAY* below the noise component
in an analog board having a 130 dB dynamic range (if one exists).
So there you see the futility in using so many unneeded bits.
--
Dick Wallingford di...@iastate.edu
Center for Nondestructive Evaluation wd0anb
Iowa State University 515-294-9751
Ames, IA 50011 515-294-7771 (FAX)
Bob Mills
wrote:
>But what about tracking ? You are not going to loose anything by
>mastering from a digital recorder to a good analog console to DAT. You
>*are* going to loose your original working dynamic range coming in to a 20
>bit digital console for the first time. ( 16 and 20 bit systems ) like
>I already said, 24 bit will not reduce your dynamic range beacuse it has a
>dynamic range that meets or exceeds the best analog mixer.
Rick,
The first 2 sentences here don't agree with the rest of the paragraph and
your other statements. Let me point out right up front that I'd like it to
be true, since I run an ADAT into a Mackie 1604 into a DA-30 for mastering.
But it ain't.
If you're upset about ONE 20-bit conversion, why would you not be concerned
about MULTIPLE 16-bit conversions? It doesn't compute!
Bob Mills
NoiseToys Phone: 1-609-683-0234
9 Charlton Street Fax: 1-609-683-4068
Princeton, NJ 08540 USA e-mail: deci...@tigger.jvnc.net
ke...@infoserv.com
#
# One thing to keep in mind here. Although a fully digital mixer can provide
# a lot of added automation control over EQ etc., until 24 BIT AD/DA becomes
# affordable the dynamic range and overall performance of a digital board will
# not equal that of a good analog mixer.
On the contrary.
Good 18 or 20 biut A/D and D/A will do the job.
Your limitation is not in A/D or D/A but in the width of
the accumulator in your DSP. A 56K has an accumulator with
50+ bits. That gives you plenty of space for processing.
Even if you add a large number of signals that are 18 bits
on input it will take quite a while to over flow at 50 bits.
I would argue that you would clip on an analog console on the
bus simply doing straigt mixing before you could clip in a
digital console because with 50 bits you get 300dB of dynamicr
range. l
I have yet to see a console that could do that.
Let's say you use a DSP that has an accumulator width of
36 bits. This still gives you 216dB.
You will need some intelligent bit reduction for output
but with a 20 bit D/A that still gives you 120dB of dynamic
range.
If you factor in EQ and other processing you have the same basic
limitations you get on an analog console. That is, sooner or later
if you run your signals too hot you will clip on one of the internal
buses, in digital you overflow the accumulator.
Ask SSL, AT&T, Yamaha and Neve about digital consoles.
Kent
--
/* "There is no king who has not had a slave among his ancestors and */
/* no slave that has not had a king among his." ---- Helen Keller */
/* Kent L. Shephard ----- K. L. Shephard Consulting */
ke...@infoserv.com
<rick.v...@mackie.wa.com> wrote:
>In article of 10:12 PM 9/15/94, DPi...@world.std.com (Richard D writes:
>The dynamic range is not simply the number of bits x 6.
>
>Once analog is above the noise floor it has infinite resolution. 16 bit
>digital systems are operated with about 20dB of headroom. Which makes
>nominal zero at 20dB below maximum output, leaving around 72dB.
This is completely untrue and erroneous.
Dynamic range is simply number of bits X 6dB check a basic course
in signal processing. When I record I run my signals into the A/D
at an ave. of right at clipping. Depending on the type of music
I record the nominal might be above or below 72.
Once analog is above the noise floor you don't get infinite resolution
simply because you have theshold voltages and thermal noise.
There will be a value below which no change will be detected. If you
have an analog system with a theshold voltage/thermal noise at some
voltage, let's call it X. If you have a change in signal level of
.2X you will probably not detect because it will be hidden under
the thernal noise and/or threshold voltage.
Matt Mora
>The reason they don't jive with your experience, Rick, is because your
>experience and intuition fail you in the actual realities of
>band-limited, range limited systems.
Maybe Rick is thinking about the extra bits needed for processing
( ie Eq, mixing) where you might want the extra bits for over flow/under
flow?
Just a guess.
Xavier
--
___________________________________________________________
Matthew Xavier Mora Matt...@sri.com
SRI International mxm...@unix.sri.com
333 Ravenswood Ave Menlo Park, CA. 94025
Richard D Pierce
>In article <CwAvF...@world.std.com> DPi...@world.std.com (Richard D Pierce) writes:
>
>>The reason they don't jive with your experience, Rick, is because your
>>experience and intuition fail you in the actual realities of
>>band-limited, range limited systems.
>
>Maybe Rick is thinking about the extra bits needed for processing
>( ie Eq, mixing) where you might want the extra bits for over flow/under
>flow?
>
No, Rick specifically mentioned 24 bit A/D converters. The Orban DSE 7000
uses 16 bit converters for going in and out to the real world, but
internally mixes, gains, pans, etc., are all computed using 32 bits.
Also note his assertion about the "resolution of analog being infinite"
once you get 20 dB above the noise. This is simply wrong. If you're 20 dB
above the noise, your resolution is 10%, 40 dB above the noise, it's 1%,
and so on.
The assertion that the resolution of ANY system is infinite is at
complete odds with the fact that infinite resolution requires infinite
dynamic range or infinite bandwidth and, I don't care how hard or how
long the people at Mackie or anywhere else work, they don't make mixers
with those specs.
Richard D Pierce
Two nits, Kent. First, is the somewhat erroneous way you edited the
attributions, making it possible to infer that the comments you object to
(correctly) were written by me: they were not, but came from Rick at
Mackie.
>In article <2.47709176.M...@mackie.wa.com>,
> <rick.v...@mackie.wa.com> wrote:
>>In article of 10:12 PM 9/15/94, DPi...@world.std.com (Richard D writes:
>>The dynamic range is not simply the number of bits x 6.
>>
>>Once analog is above the noise floor it has infinite resolution. 16 bit
>>digital systems are operated with about 20dB of headroom. Which makes
>>nominal zero at 20dB below maximum output, leaving around 72dB.
>
>This is completely untrue and erroneous.
Agreed
The second nit is VERY minor:
>Once analog is above the noise floor you don't get infinite resolution
>simply because you have theshold voltages and thermal noise.
>There will be a value below which no change will be detected.
It's not that no change will be detected, it's that ANY change detected at
or below the noise of the system will be completely ambiguous, random and
unpredictable, indistiguishable from noise, and thus contains NO
information about the original signal.
Chris Christensen
the parameters to consider in a mixer design.
With the possiblity of a signal being routed through a mixer many times,
the ultimate bandwith is a concern. Many prople believe that the
audiobandwith afforded by a 44.1kHz sample rate is fine. Maybe so but the
use of a band limited mixer, even digital, would reduce that bandwith a great
deal.
Am I all wet when it comes to the digital domain (I suspect so)?
rick.v...@mackie.wa.com
>
>Or so you think. There are NOT two schools of thought about the
> realitiesof resolution and information capacity of limited dynamic range,
limited bandwidth system of ANY kind, despite your claims to the contrary.
This is not a matter of opinion, Rick, this is a matter both of
theoretically sound and experimentally verifiable fact.
So you are trying to me us that a 20-Bit system contains *all* of the
information that first came into the microphone from the original "analog"
source ?
[]Rick.
Richard D Pierce
Chris Christensen <chr...@fuggles.gvg.tek.com> wrote:
>In article <4.47712901.M...@mackie.wa.com> rick.v...@mackie.wa.com writes:
>>How can a digital signal...Which is a *sample* not the entire signal but a
>>portion of the signal have the same sonic purity as the original ?
>
>Well this is one of my pet topics in the world..... The problem is that bits
>don't tell the true story. When it comes to the reconstruction of a high
>frequency signal, say 20kHz there are something like two samples to
>represent that signal. Am I right Dick? Gabe?
>
>Personally I would like to have _more_ samples available to reconstruct the
>HF of a digital signal. Rick, this is really what I believe you mean.
But it can be demonstrated theoretically, mathematically, via measurement
AND listening that if the highest frequency of interest is x, then a
sampling frequency of greater than 2x is both the necessary AND sufficient
sampling rate to completely and umambiguously represent the signal at x.
If you are then going to argue by saying, "well, I have this complex
waveform at x and my 2x sampling system doesn;t represnt it," then you've
stacked the deck by using a complex waveform whose highest frequency is
NOT x, but maybe 10x or 50x!
>Bits represent dynamic range, sure. For my purposes bits represents
>granularity ans quantizing noise. Those are far worse, IMHO, than the
>difference between even 90 to 110 dB S+NR. If you apply Dick's noise
>shaping, etc. as is done in the DSE 7000 noise becomes a moot point.
There are "bits" in the amplitude domain and there are "bits" in the
frequency domain. Turns out the SAME rules essentially apply to both.
>Let's move on to _my_ problem with digital, beaning sampling frequency. Mr.
>Hicks? Dick?
>The 20 bit digital mixer would sound just as good as the analog console if
>it weren't for the bandwith limitations.
Sure, could well be, but the system can only represent what's fed it. And
that is the ultimate bandwidth limiter.
>I was doing a location "studio" recording a coupla months ago. I was using
>my "trusty" Mackie 1202 and my sublime Denon DTR-2000 DAT Deck. Not this
>comparison _may_ be invalid but when I plugged my 'phones into the Mackie
>'phone fack vs. the Denon 'phone jack there was an audible difference in the
>HF response.... The reason I say the comparison may be invalid is that I
>haven't looked at the schematics to rule out an obvious electrical
>difference.......
The comparison may be totally invalid. I would find it not at all hard to
believe that our friends at Mackie are capable of making a MUCH better
headphone amplifier than is probably sitting in your Denon DAT machine.
For example, let's say the output impedance of the Mackie headphone is,
oh, 2 ohms, and that of the Denon is, maybe, 100 ohms. Now, let's say the
impedance of your headphones varies between 8 and 150 ohms. It's very
easy, given this kind of difference to explain ALL the differences you
heard and never once worry about digital or analog representation.
>I agree and disagree with the above statements. The performance of a high
>end analog console is best complimented by a good analog deck using Dolby S
>noise reduction. In that mode you get low noise and wide frequency
>response.
But why limit the output of this alledged good console by an analog deck.
Ultimately, the dynamic range of such a deck is limited compared to
digital, even with noise reduction schemes. Remember that it is the
device with the narrowest dynamic range that limits resolution, and
that's the tape. The noise reduction scheme may increase the dynamic
range, it will not increase the resolution, though (think about what
happens to the tape noise when the amplitudes are high, for example).
>And lastly, a good digital console could be as little as 2-3 years away
>given the proper resources, my design and my skils as a product development
>engineer..... :-)
Or it could have been on the market already for 4 years with a free 8
track digital deck, an all-digital editor, a 2 track mixdown deck, and a
real user interface thrown in for good measure, given mine! :-)
Brandon Mathew
>In article <4.47712901.M...@mackie.wa.com> rick.v...@mackie.wa.com writes:
>
>>>Eh? This doesn't follow. Kindly explain to us how 24 bits would give
>>>performance that is "similar to analog" whereas, say, 20 bits wouldn't?
>
>>source.
>
>>How can a digital signal...Which is a *sample* not the entire signal but a
>>portion of the signal have the same sonic purity as the original ?
>
>Well this is one of my pet topics in the world..... The problem is that bits
>don't tell the true story. When it comes to the reconstruction of a high
>frequency signal, say 20kHz there are something like two samples to
>represent that signal. Am I right Dick? Gabe?
I think that in a system that is band limited to 20Khz, it does tell the whole
story. If you want to see more wiggles in that 20Khz signal, you have to
have a system that has more than 20khz of bandwidth (which is what you
alluded to in the rest of your posting).
While I'm at it, I would like to thank Dick for the explanation of analog/noise
and digital/quantization. It has made some light bulbs go on in my cob web
infested brain.
--
Brandon Mathew - bra...@core.rose.hp.com
Chris Christensen
>In article of 5:37 PM 9/17/94, ga...@panix.com (Gabe Wiener) writes:
>>Eh? This doesn't follow. Kindly explain to us how 24 bits would give
>>performance that is "similar to analog" whereas, say, 20 bits wouldn't?
>More dynamic range and more of the sonic data from the original analog
>source.
>How can a digital signal...Which is a *sample* not the entire signal but a
>portion of the signal have the same sonic purity as the original ?
Well this is one of my pet topics in the world..... The problem is that bits
don't tell the true story. When it comes to the reconstruction of a high
frequency signal, say 20kHz there are something like two samples to
represent that signal. Am I right Dick? Gabe?
Personally I would like to have _more_ samples available to reconstruct the
HF of a digital signal. Rick, this is really what I believe you mean.
Bits represent dynamic range, sure. For my purposes bits represents
granularity ans quantizing noise. Those are far worse, IMHO, than the
difference between even 90 to 110 dB S+NR. If you apply Dick's noise
shaping, etc. as is done in the DSE 7000 noise becomes a moot point.
Let's move on to _my_ problem with digital, beaning sampling frequency. Mr.
Hicks? Dick?
>How can a 20 Bit digital mixer which can only acheive 119db of dynamic range
>offer the same dynamic response internally as an analog mixer that can have
>internal operating ranges of above 135db ? Also an alalog console doesn't
>suffer from that grainey mush at low levels.
The 20 bit digital mixer would sound just as good as the analog console if
it weren't for the bandwith limitations.
I was doing a location "studio" recording a coupla months ago. I was using
my "trusty" Mackie 1202 and my sublime Denon DTR-2000 DAT Deck. Not this
comparison _may_ be invalid but when I plugged my 'phones into the Mackie
'phone fack vs. the Denon 'phone jack there was an audible difference in the
HF response.... The reason I say the comparison may be invalid is that I
haven't looked at the schematics to rule out an obvious electrical
difference.......
For the sake of arguement, that all is OK as far as the headphone circuits.
Why should I be able to hear the difference between the two equipment? Is
it only the E-E bandwidth reduction of the DAT machine????
>I totally buy digital for use as a recording medium, mainly cause analog
>tape recorders were so much less capable than Digital ones.
The "features" of DAT is what originally brought me to them....
>Mixers on the other hand are a whole different ball game. There is very
>little sense in tracking through any of the current digital mixing boards.
>Most mid-priced analog consoles are far better at handling the signal and
>maintaining the sonic quality of those signals. All the producers that I
>know are still waiting for digital mixers to out perform analog. I would
>say that it is at least 7-10 years away.
I agree and disagree with the above statements. The performance of a high
end analog console is best complimented by a good analog deck using Dolby S
noise reduction. In that mode you get low noise and wide frequency
response.
Secondly I would agree that the people are waiting for good digital consoles
but isn't the problem that there is but one "real" studio console on the
market and it is very expensive? If there was a 64 input digital console in
the $200K range, wouldn't it sell like hot cakes??????
And lastly, a good digital console could be as little as 2-3 years away
given the proper resources, my design and my skils as a product development
engineer..... :-)
--
rick.v...@mackie.wa.com
>The reason they don't jive with your experience, Rick, is because your
>experience and intuition fail you in the actual realities of
>band-limited, range limited systems.
O.K....Let me ask you this..
Our 8-Bus mixer has a bandwidth of 20Hz to 60Khz internal from Input to
Output. How would I get a digital converter with a bandwidth of 20Hz to
20Khz to give me the *same* output ?
Mixers and tape recorders are not the same thing. Many people who are
actively involved in digital recording fail to realize this fact.
Digital came about to improve recording and editing. Mixers don't
necessarily benefit from being digital. Especially if they cost 10 times
more money and yield less audio performance than their analog counterpart.
If we could find a converter that offered us the same bandwidth as our
analog circuitry and could be purchased for the same price as an Op-Amp then
we would be talking. Even then this would be more of a marketing decision
than one driven by necessity.
rick.v...@mackie.wa.com
>Eh? This doesn't follow. Kindly explain to us how 24 bits would give
>performance that is "similar to analog" whereas, say, 20 bits wouldn't?
More dynamic range and more of the sonic data from the original analog
source.
How can a digital signal...Which is a *sample* not the entire signal but a
portion of the signal have the same sonic purity as the original ?
How can a 20 Bit digital mixer which can only acheive 119db of dynamic range
offer the same dynamic response internally as an analog mixer that can have
internal operating ranges of above 135db ? Also an alalog console doesn't
suffer from that grainey mush at low levels.
I totally buy digital for use as a recording medium, mainly cause analog
tape recorders were so much less capable than Digital ones.
Mixers on the other hand are a whole different ball game. There is very
little sense in tracking through any of the current digital mixing boards.
Most mid-priced analog consoles are far better at handling the signal and
maintaining the sonic quality of those signals. All the producers that I
know are still waiting for digital mixers to out perform analog. I would
say that it is at least 7-10 years away.
[]Rick.
Richard D Pierce
>>
>>No, Rick specifically mentioned 24 bit A/D converters. The Orban DSE 7000
>>uses 16 bit converters for going in and out to the real world, but
>>internally mixes, gains, pans, etc., are all computed using 32 bits.
>
>
>By the way, thanks for the article. I too was under the myth of infinte
>resolution of analog. Wow, I learned something from usenet. :-)
>
>I have a question though. In your text you stated a 86db headroom or
>something like that. 16bit systems are about 90+db right? What does
>20 bits buy you? Aren't analog boards around 130db of headroom?
I weas using 86 dB merely as an example. That would correspond roughly to
a 14 bit system, or the very best analog tape recorder.
16 bit system are theoretically capable of 96 dB of full dynamic range,
20 dB, theoretically, 120 dB and so on.
>If I'm not making sence, please excuse me, I just trying to get a handle
>on all this stuff. While your at it, (and if any of you have the time)
>can you explain the difference in S/N ratio, Dynamic headroom, db, dbv?
>Also why are some spec in negative values while others are in postitive
>values? (ie -129 db, S/N 86 )
Signal to noise:
Simply, the ratio between the signal level and the amount of noise in
a system
Dynamic headroom:
us, who knows? It's often an advertising term, and can mean damned
near anything, or nothing.
Dynamic range:
The difference between the smallest unambiguously knowable piece of
information that can be represented by a system and the largest that
can be represented without saturation and distortion.
dB, dBv, etc.
A dB is a deciBel. or one tenth of a Bel. It is a logartihmic
representation of power ratios. A Bel represents a factor of 10
change in power, as does 10 decibels of 10 dB.
A dBv is a ratio using 1 volt as a reference. A dBm uses 1 milliwatt,
a dBf 1 femtowatt and so on.
The number will be positive if the ratio is greater than 1. For example,
10 watts is 10 dB (+10 dB) greater than 1 watt. 1 watt is 10 dB less
(-10 dB) than 10 watts.
Richard D Pierce
Absolutely the case. If you have two systems, BOTH have the same
bandwidth and BOTH have the same dynamic range (say, 120 dB), BOTH will
pass the same amount of information, REGARDLESS of whether one is digital
or one is analog. Period. It's a matter of definition, not of opinion or
religion.
Now, that erroneously presupposes that microphones put out enough
"information" to tax such a system. Unless someone is using B&K or similar
lab mics that really do have 120 dB of dynamic raneg AND bandwidths wider
than 20 kHz, then the output of the microphone sets the limit on
resolution, and ANY system with resolutiuon greater than that is good
enough.
Matt Mora
>>>The reason they don't jive with your experience, Rick, is because your
>>>experience and intuition fail you in the actual realities of
>>>band-limited, range limited systems.
>>
>>Maybe Rick is thinking about the extra bits needed for processing
>>( ie Eq, mixing) where you might want the extra bits for over flow/under
>>flow?
>>
>
>No, Rick specifically mentioned 24 bit A/D converters. The Orban DSE 7000
>uses 16 bit converters for going in and out to the real world, but
>internally mixes, gains, pans, etc., are all computed using 32 bits.
Ah yes, I remember now. :-)
By the way, thanks for the article. I too was under the myth of infinte
resolution of analog. Wow, I learned something from usenet. :-)
I have a question though. In your text you stated a 86db headroom or
something like that. 16bit systems are about 90+db right? What does
20 bits buy you? Aren't analog boards around 130db of headroom?
If I'm not making sence, please excuse me, I just trying to get a handle
on all this stuff. While your at it, (and if any of you have the time)
can you explain the difference in S/N ratio, Dynamic headroom, db, dbv?
Also why are some spec in negative values while others are in postitive
values? (ie -129 db, S/N 86 )
Thanks
Richard D Pierce
<rick.v...@mackie.wa.com> wrote:
>In article of 5:37 PM 9/17/94, ga...@panix.com (Gabe Wiener) writes:
>
>>Eh? This doesn't follow. Kindly explain to us how 24 bits would give
>>performance that is "similar to analog" whereas, say, 20 bits wouldn't?
>
>More dynamic range and more of the sonic data from the original analog
>source.
That would be true IF AND ONLY IF the microphone itself had a wider
dynamic range AND wider bandwidth than the subsequent systems. No
performance or recording microphone I've come across meets both these
criteria simultaneously, and few lab grade mics do.
>How can a digital signal...Which is a *sample* not the entire signal but a
>portion of the signal have the same sonic purity as the original ?
Rick, this is the whole point of the "analog" myth. IT DOES NOT MAKE ANY
DIFFERENCE IF THE SYSTEM'S DYNAMIC RANGE IS LIMITED BY NOISE OR IF ITS
LIMITED BY QUANTIZATION! THE LEVEL OF KNOWABLE INFORMATION IS THE SAME
WHEN THE DYNAMIC RANGE IS THE SAME REGARDLESS OF THE METHOD USED IN
TRANSMITTAL OR STORAGE.
Don't believe us? go look it up for your self
>How can a 20 Bit digital mixer which can only acheive 119db of dynamic range
>offer the same dynamic response internally as an analog mixer that can have
>internal operating ranges of above 135db ? Also an alalog console doesn't
>suffer from that grainey mush at low levels.
Neither does a properly dithered digital system. In fact, the signal
characteristics of the two at the smae low levels look the same, save the
noise spectrum in the dithered digital case can be specifically tailored
whereas the random thermal noise in the analog cannot be.
>I totally buy digital for use as a recording medium, mainly cause analog
>tape recorders were so much less capable than Digital ones.
And if you buy that ANY recording medium and the microphone both are more
band and dynamic-range limited than the console, what's the point.
I have no particular issue with your claims of the eprformance of MAckie
mixers: they're not particularly outrageous nor do I care. However, your
pronouncements about the "analog" and "sonic purity" and "grainey mush"
suggest a fundamental misunderstanding of the principles of information
system and the limitations and characteristics of both analog and
digital systems. And while it may seem "intuitively" correct that analog
has "infinite" resolution, physical reality pays no heed to your
intuition, mine or anyone else's.
rick.v...@mackie.wa.com
Hi Chris and company,
I am going to try to get Carl Malone to respond to your comments. We are
very busy at this time and it is real hard for us to get into these three
page message debates when we have so many new products to turn out before
the end of the year.
I am having to reply to these messages late at night after working 12 hour
days... It gets real tough after awhile.
Carl is giving a talk on the Digital Control of analog for this years AES
show in San Francisco. He is also our Senior Digital Engineer for our New
Products department
If he can find the time I am sure that it will make for some very
enlightening messages.
Thanks.
[]Rick Vartian
Mackie Designs Inc
Jim T. Rusby
>Chris Christensen <chr...@fuggles.gvg.tek.com> wrote:
>>And lastly, a good digital console could be as little as 2-3 years away
>>given the proper resources, my design and my skils as a product development
>>engineer..... :-)
Richard D Pierce <DPi...@world.std.com> wrote:
>
>Or it could have been on the market already for 4 years with a free 8
>track digital deck, an all-digital editor, a 2 track mixdown deck, and a
>real user interface thrown in for good measure, given mine! :-)
>
And meanwhile- somewhere right now 5 stagehands are wrestling a Yamaha
PM3000 or heavier console out to the front of house position, and
grumbling about how some day "they" will invent a lightweigh mixer
control surface one guy could carry- keeping all the sig processing and
preamping on stage.
At this point the janitor looks up from his broom and mentions the
benefits of using the same system in the studio, just change a few ROMS
and use software to configure the setup for recording. With the right
software, the rig could also control machines and perform lab grade
studio maintenance tests.
Ain't dreamin fun?jr
--
Jim Rusby Phone: 503-346-5659
Recording engineer UO School of Music email:jru...@darkwing.uoregon.edu
Tech Director- Dept of Dance Two Tracks- 8 chickens and a dog.
Nigel Orr
>Well this is one of my pet topics in the world..... The problem is that bits
>don't tell the true story. When it comes to the reconstruction of a high
>frequency signal, say 20kHz there are something like two samples to
>represent that signal. Am I right Dick? Gabe?
Well Nyquist would suggest that you are right- I guess I'd tend to
believe him since he seems to know what he was saying and didn't post
to usenet :-)
However, I do have this one little niggling doubt, which perhaps those
more in the know can explain away...
You have a 20kHz sine wave, and are sampling at 44.1kHz- OK so far?
Now, this means that you will have slightly more than 2 samples per
period of the wave- right?
Doesn't this mean that the sampling points will be at different points
on the waveform at each sampling interval?
If so, surely the resultant output waveform would be modulated at some
other frequency (I would assume 2.05 or 4.1kHz, but I haven't really
thought about it)- for example, some consecutive samples will be at
the zero points of the waveform, and some at the peaks.
Does anyone follow what I'm getting at here- am I missing some vital
consideration?
I'd appreciate an arm-waving or mathematical explanation of why this
doesn't make 2 samples per period inadequate to accurately code the
signal.
--
Nigel o...@tks.oulu.fi
C.M. Hicks
[...munch...]
>Say we have a 20kHz sine wave, as I was talking about earlier, being
>sampled at 44.1kHz.
>Let's also say that the first sample is taken at time t=0
>Right, the waveform of the 20kHz sine wave in time will be given by
>y=sin (2 x 2E4 x pi x t) [I hope I got that right?]
Yep.
>so, sampling at 44.1kHz will produce a table of values starting [FX-
>gentle clicking as Nigel tries to figure out how to program his
>calculator...]:-
>sample no. sample value (sin (4E4 x pi x sample no. / 44100)
>0 (t=0) 0
>1 (t=1/44100) 0.28794
>2 -0.55149
>3 0.76833
>4 -0.92009
>5 0.99391
>So, assuming I have got the equation right, which I'm pretty sure I
>have, do you see what I mean? Can you resurrect a 20kHz sine wave of
>constant amplitude from the above values? What am I missing out here?
You resurrect the original sinewave by low-pass filtering at half
the sample rate.
>Try it on paper- draw a sine wave, and draw sample points at intervals
>of just under half the period, and see what the smoothest curve is
>that you can make through those points, which is presumable what will
>be regenerated?
The smooth curve you are drawing by eye contains components above half
the sample rate, most notably a very strong one at 24.1kHz (= 44.1 - 20)
which would be filtered out by a brickwall reconstruction filter whose
cutoff lies somewhere between 20 and 22.05 kHz.
Try plotting a graph of:
y = 0.5*( sin(2*pi*2e4*t) - sin(2*pi*2.41e4*t) )
Notice how this looks like the smooth curve you tried to draw through
the sample points.
Christopher
PS Nigel, you can plot these graphs very easily with Matlab.
Bill Vermillion
>Chris Christensen <chr...@fuggles.gvg.tek.com> wrote:
>>>It has often been my fantasy that because wide bandwidths may result in
>>>more linear phase across the audible part of the spectrum, we frequently
>>>like the "wide bandwith design" not for the stuff it does at 70KHz, but
>>>rather (as you point out) for what it doesn't do at 10KHz.
>>I have no proof of these assertions but I believe them.
>>This is my fundamental gripe with digital. It is my firm belief that because
>>of the restricted bandwith of the digital recording process, high frequencys
>>that, albeit minimal, are being filtered from the live sources. Music rich
>>in harmonics are being sterilized but the digital process. The convoluted
>>high frequency harmonic of the live source is now represented by a perfectly
>>reconstructed pure sine wave.....
>Now, look at the frequency and phase response of the BEST analog
>recording machines. Look at the frequency and phase response of the
>AVERAGE semi-pro DAT machine. Which is wider band, lower phas error,
>flatter frequency response? No, I mean REALLY!
I suspect many will fall into the higher maximum frequency vs
bandwidth trap.
On the BEST analog machines I had no problems getting them
'relatively' flat out to over 30KHz - at 30 ips.
The sacrafice, of course, is made at the lower end, as analog
bandwidth is awfully hard to get any wider than 10 octaves.
Even though the DAT top-end is about 20KHz, it has a 12 octave
bandwidth. So while analog has a higher maximum frequency it
still has a smaller bandwidth.
--
Bill Vermillion - bi...@bilver.oau.org | bill.ve...@oau.org
rick.v...@mackie.wa.com
>
>This is my fundamental gripe with digital. It is my firm belief that
> because
>of the restricted bandwith of the digital recording process, high
> frequencys
>that, albeit minimal, are being filtered from the live sources. Music
> rich
>in harmonics are being sterilized but the digital process. The
> convoluted
>high frequency harmonic of the live source is now represented by a
> perfectly
>reconstructed pure sine wave.....
This is hitting on some of the points I was trying to make early on in this
thread. WE DO HEAR A DIFFERENCE...It is enough to convince ourselves that
the current digital resources are not worth pursuing at this time. We see
no reason in offering a all digital mixer at 4-10 times the cost that in our
opinion will not sound as true to life as an analog console.
As far as myself and Mackie are concerned...This concludes our remarks on
this subject.
We are working on some new products that we feel will address this market in
a far more efficient manner and offer performance and internfaces that rival
anything that can be done in an all digital console.
[]Rick Vartian
Mackie Designs Inc.
[]Rick.
C.M. Hicks
[...Nyquist theorem...]
>Well Nyquist would suggest that you are right- I guess I'd tend to
>believe him since he seems to know what he was saying and didn't post
>to usenet :-)
>However, I do have this one little niggling doubt, which perhaps those
>more in the know can explain away...
>You have a 20kHz sine wave, and are sampling at 44.1kHz- OK so far?
>Now, this means that you will have slightly more than 2 samples per
>period of the wave- right?
>Doesn't this mean that the sampling points will be at different points
>on the waveform at each sampling interval?
>If so, surely the resultant output waveform would be modulated at some
>other frequency (I would assume 2.05 or 4.1kHz, but I haven't really
>thought about it)- for example, some consecutive samples will be at
>the zero points of the waveform, and some at the peaks.
(For those of you who know about these things be aware that I am being
deliberately vague in the following explanations about the distinction
between sinusdoids and complex exponentials, and also about the phase
relationships of the various components!)
You're roughly right. The point is that the sampled signal is composed
of not just one sinewave at 20kHz, but an infinite number of sinewaves
at (n*fs +/- 20) kHz (where n is an integer and fs is the sample rate)
all added up together. Now, when converting back to analogue we use an
anti-image filter which removes all the extra sinewaves we don't want
and leaves behind just the two corresponding to n=0 (ie the original
signal).
You also mention the fact that the signal looks modulated - perhaps
the following will explain something of why... Take the four signal
components at +/-20kHz and +/-(fs-20)kHz (corresponding to n=0 and n=1
respectively). On their own this this little lot looks to all intents
and purposes like a 22.05kHz (fs/2) carrier, amplitude modulated to a
depth of 200% by a 2.05kHz signal (this is known as suppressed carrier
modulation, since there is no component actually at the carrier
frequency). An ambiguity arises if you don't know which 44.1kHz of
bandwidth the signal started its life in.
In fact, this can be exploited to do demodulation of AM signals. Take
an AM signal of 10kHz bandwidth on a carrier of 1MHz. Bandlimit this to
the range 1MHz-10kHz to 1MHz+10kHz. Now take a clock, phase locked to
the carrier, at say 20kHz and use it to sample the AM signal.
Because one of the harmonics of the 20kHz sample clock is guaranteed
to be at precisely the same frequency as the carrier (they are phase
locked, remember) the resulting data stream contains copies of the
audio modulation signal centred around all integer multiples of 20kHz
including n=0. In the discrete time domain it cannot be distinguished
from the original audio, sampled at 20kHz, and can therefore be
treated in exactly the same way as any other baseband digital audio
signal, including conversion to analogue if desired.
Neat.
>Does anyone follow what I'm getting at here- am I missing some vital
>consideration?
Yes and no, respectively!
Does anyone follow what *I'm* getting at?
>I'd appreciate an arm-waving or mathematical explanation of why this
>doesn't make 2 samples per period inadequate to accurately code the
>signal.
I promise I am waving my arms around! Oh for a blackboard and some
pictures - I could explain this clearly then!
Christopher
Chris Christensen
writes:
>chr...@fuggles.gvg.tek.com (Chris Christensen) writes:
>>In article <4.47712901.M...@mackie.wa.com> rick.v...@mackie.wa.com writes:
>[...munch...]
>>>How can a digital signal...Which is a *sample* not the entire signal but a
>>>portion of the signal have the same sonic purity as the original ?
I'll add the typical artistic analogy here:
Digital audio is sampled and reconstructed.
Television is samples and reconstructed.
Analog audio recording in not sampled.
A photogtaph, while a sample in time, is more like analog audio...
>>Well this is one of my pet topics in the world..... The problem is that bits
>>don't tell the true story. When it comes to the reconstruction of a high
>>frequency signal, say 20kHz there are something like two samples to
>>represent that signal. Am I right Dick? Gabe?
>I'm not Dick or Gabe, but Chris! Yeah, that's kind of OK.
>>Personally I would like to have _more_ samples available to reconstruct the
>>HF of a digital signal. Rick, this is really what I believe you mean.
>>Bits represent dynamic range, sure. For my purposes bits represents
>>granularity and quantizing noise. Those are far worse, IMHO, than the
>>difference between even 90 to 110 dB S+NR. If you apply Dick's noise
>>shaping, etc. as is done in the DSE 7000 noise becomes a moot point.
>Yep. This is precisely the point of dither. Noise-shaping is a scheme
>for reducing the audibility (not the power) of quantisation and dither
>noise that can sometimes be used to great effect, and can often be used
>to no beneficial effect whatever.
I have no problems with processes that mask the negative artifacts of
digital conversion...!
>But, yes. On the whole dither+quantisation represents the addition of
>low level white noise to the signal.
If I read this correctly, dither reduces quantization and adds a small
ammount of noise? I can live with that...
>>Let's move on to _my_ problem with digital, beaning sampling frequency. Mr.
>>Hicks? Dick?
>Ah, ME! :-)
>[...a rather shaky conjecture deleted...]
It;s wnat I am famous for...
<Location recording story deleated>
>>For the sake of arguement, that all is OK as far as the headphone circuits.
>>Why should I be able to hear the difference between the two equipment? Is
>>it only the E-E bandwidth reduction of the DAT machine????
>As you pointed out this test is fatally flawed and I don't believe we
>can draw any conclusion from it.
I have looked up the circuitry of rhe two devices. Here are the results:
The Denon uses a 120 buildout resistor.
The Mackie uses two OP Amps in parallel with 120 Ohm bulitouts on each OP
Amp.
The source impedance difference is a factor of 2......
>The anecdote often wheeled out by digiphobes in situations like this
>is the one attributed to Rupert Neve and the mixer channel that was
>3dB down at 40kHz when it should have been 80kHz or something. Nobody
>that has quoted this to me has been able to answer either of the
>following questions with any certainty:
> 1) Is the story true?
yes!
> 2) Assuming it *is* true, what was the fault?
I don't rember exactally (I read the article some time ago) but there was a
misstffed cap that reduced the bandwith of the channel.....
>See, the bandwidth reduction could be only one of many symptoms of an
>oscillating transistor, a leaky capacitor, an incorrect resistor or
>any other electronic fault. Other effects could easily include nasty
>distortion-inducing non-linearities - potentially highly audible. The
>fact that when a sinewave was fed in the voltmeter on the output read
>lower than it should is largely irrelevant.
I having trouble shot a few thousand circuits I would agree completely with
your assumption. I really should look up that article.....
As I noted in a previous posting, the direct comparison possiblities in
something like a studio console situation makes this type of revelations
possible... (?).
>Having said all that your basic logic is sound; if you can reliably
>hear differences between the outputs of your Mackie and your Denon
>then it may be due to the bandwidth limitation. I personally doubt it,
>but if true then I suggest that is equally likely that the sampling
>process has actually *prevented* distortions from occuring in the
>headphones, which are unlikely to be linear at very high frequencies.
With the 'phone amp information I have provided and the next revelation:
The phones used were Sony MD-V6's....... Do you believe that the difference
is electrical between the devices or bandwith limitations....?
>But perhaps you just like the way the Mackie sounds. That's fine
>by me - I can't argue that you *must* prefer the digital system :-)
I should state here for the record, because I am sure that I sound like
andigiphobe, I'm NOT!!!!! I record and edit digitally! I love it, I
really do. But not because of how it sounds. I really do wish thet the
sample frequency was 88.2kHz instead of 44.1kHz.......:-)
>There are lots of other likely explanations too, but if you are
>saying, as I think you are, that you would prefer a higher sampling
>rate because you want a higher system bandwidth then I can't argue!
See! You agree!
>I do to some extent challenge your motives for wanting such a system.
Please do, it won' hurt my feelings.
>I am not a psychoaccoustics expert but everything I have read (for
>example: Moore, Introduction to the Psychology of Hearing) seems to
>suggest that the Fourier basis is good for analysing hearing, and that
>in this basis the ear's sensitivity varies with frequency and
>amplitude in a fairly predictable manner.
I don't disagree here at all. I also agree that the usual human hearing is
bandwith limited, as well as speakers, microphones, etc...... I believe that
I have experienced the "artifacts" of bandwith limiting. The engineer in me
wants to explore the posiblities to satisfy the artistic demands! My
problem is that my daytime employer doesn't share my quest for this specific
answer..... Then my evenings and weekends are "shared" with my artistic
quests....
>In particular its sensitivity to both sinusoidal tones and broadband
>noise decreases rapidly above (typically) 16kHz, and that most people
>(other than some young children) are effectively completely deaf above
>19 to 20kHz. Given this premise, a sampled data system operating at a
>sample rate of 44.1kHz is theoretically capable of perfectly
>describing any sound to which the ear/brain will respond.
I have said this before, I believe that the filtering of the harmonics of
music above 20kHz, while outside the realm of scientific possiblity (which I
embrace, science that is... :-) Is audible...... Simple ABX testing would
prove this, I haven't set this up and tested it though...... Maybe Dick has?
>Research into other physiological effects of airborne vibration is
>sketchy, to say the least. However there is evidence to suggest that
>ultrasonic cleaners etc can cause headaches, but the pressure levels
>have to be enormous. At the other end of the spectrum it has been
>reported that subjecting people to high intensity sound at 3 or 4 Hz
>can be very messy...
I haven't read details of the airborne effects you cite, but I have
experience ultrasonic annoyances.
Thanks for the discussion
Would anyone like to set up a test at AES?
Richard D Pierce
<rick.v...@mackie.wa.com> wrote:
>
>This is hitting on some of the points I was trying to make early on in this
>thread. WE DO HEAR A DIFFERENCE...It is enough to convince ourselves that
>the current digital resources are not worth pursuing at this time. We see
>no reason in offering a all digital mixer at 4-10 times the cost that in our
>opinion will not sound as true to life as an analog console.
>
>As far as myself and Mackie are concerned...This concludes our remarks on
>this subject.
Well, Rick, I guess that about does it, eh?
So far, you've simply stated, as almost a matter of universal TRVTH(tm),
the digital is Bad. Yet, other than hyperbole, strange, somewhat less
than technical terminology akin to "yucky", and some way off-base
pronouncements (that turned out to be quite incorrect) about the
"infinite resolution of analog", you have yet to come up with ANY
justification for such a statement.
Gee, that's okay, but lest the unwary take such pontifications as
absolute gospel, which I am sure some will, regrettably, I think, to be
fair, you have a pile of challenges to some of your stranger assertions
that you have left totally unanswered.
And that's okay, too, I suppose. But, do us a favor, okay? Leave the ad
hominem attacks at home, alright? If you make an assertion about stuff
like "analog's infinite resolution," even of you made it in an honest
misubnderstanding about the principles involved (and it ain't necessarily
an easy topic, agreed), then let it go at that.
Richard D Pierce
Chris Christensen <chr...@fuggles.gvg.tek.com> wrote:
>>>In article <4.47712901.M...@mackie.wa.com> rick.v...@mackie.wa.com writes:
>>>>How can a digital signal...Which is a *sample* not the entire signal but a
>>>>portion of the signal have the same sonic purity as the original ?
>
>I'll add the typical artistic analogy here:
>
>Digital audio is sampled and reconstructed.
>
>Television is samples and reconstructed.
>
>Analog audio recording in not sampled.
>
>A photogtaph, while a sample in time, is more like analog audio...
Sorry, Chris, they are ALL sampled and ALL reconstructed. The methodology
of sampling and reconstructiod differs, but they are all sampled and
reconstructed nonetheless.
The representation can be either continuous or discrete, in both the
amplitude domain and the time domain. There exist continuous time and
amplitude representations (what you are refering to as "analog"),
discrete time and continuous amplitude representation (CCD bucket
brigade, switched capacitor filters, etc.), continuous time and discrete
amplitude representations (? don't remember any off the top of my head)
and discrete time, discrete amplitude representations (so-called "digital
audio"). ALL of them are sampled systems, its just the nature and details
of the sampling differ. The fact that some of them are continuous in
either domain does NOT make them inherently more accurate or better.
Your photograph analogy, being a continous representation in both the
spacial and intensity domain, is a good example of the problems. It's
bandwidth limited ( due to the spectral sensitivity of the materials),
its spacial resolution is seriously limited (because of the inability to
focus infinitely small points to infinitely small points due to
diffraction, the inability of the media to resolve infinitely small
points due to silver grains, and on and on and on). The very best
photograph suffers from a tremendous loss of information from the
original scene.
[ about dither, etc... ]
>I have no problems with processes that mask the negative artifacts of
>digital conversion...!
No, they don't mask the negative artifacts of digital conversion, they
make those artifacts behave like the negative artifacts of analog:
uncorrelated noise!
>>But, yes. On the whole dither+quantisation represents the addition of
>>low level white noise to the signal.
>
>If I read this correctly, dither reduces quantization and adds a small
>ammount of noise? I can live with that...
No, dither decorrelates the noise from the quantization process.
>The Denon uses a 120 buildout resistor.
>
>The Mackie uses two OP Amps in parallel with 120 Ohm bulitouts on each OP
>Amp.
>
>The source impedance difference is a factor of 2......
And that could well be more than enough, given that the output impedance
is proximal to the load impedance, to have pretty serious effects on the
overall frequency response, not to mention the fact that you mentioned no
attempts at accurate levbel matching and so on.
rick.v...@mackie.wa.com
:-)
>
>>There are lots of other likely explanations too, but if you are
>>saying, as I think you are, that you would prefer a higher sampling
>>rate because you want a higher system bandwidth then I can't argue!
>
>See! You agree!
>
So do we ! Part of our argument agaist a digital mixer is bandwidth
reduction, dynamic range reduction and first and foremost...COST !
We started with converters...Never did get to sampling rates.
Nigel Orr
of audio.tech- Followups set accordingly...
In article <35s18b$1...@lyra.csx.cam.ac.uk>,
C.M. Hicks <c...@eng.cam.ac.uk> wrote:
>In fact, this can be exploited to do demodulation of AM signals. Take
>an AM signal of 10kHz bandwidth on a carrier of 1MHz. Bandlimit this to
>the range 1MHz-10kHz to 1MHz+10kHz. Now take a clock, phase locked to
>the carrier, at say 20kHz and use it to sample the AM signal.
Or 18kHz for US readers, I guess?
>Does anyone follow what *I'm* getting at?
Yes, I followed this discussion in sci.electronics a while back with
some intrigue... only snag is that it apparently demands higher
accuracy from the sample clock (by a factor of
f_transmission/f_sample, I would expect from thinking about lots of
reflected sampled signals... but then again, my visualization methods
haven't exactly been very successful in the past :-) )
>I promise I am waving my arms around! Oh for a blackboard and some
>pictures - I could explain this clearly then!
Indeed, it's awfully hard to put these things into ASCII- probably
easiest to just "draw by formula", assuming everyone can follow the
maths, complex exponentials and the like... I mean surely all
pro-audio folks can manage that? Or is this going back to a Full-Sail
slagging match again?
--
Nigel o...@tks.oulu.fi
rick.v...@mackie.wa.com
>I'm posting one section from an article I have written on the myths of
>digital and analog audio systems. I've given this as a lecture to
> several
>professional audio groups as well. I would suggest you read it, pass it
>along to your technical staff as well. It's a good, non-mathematical
>inFINITE RESOLUTION
>
>A persistant myth these days is at the root of an intense and often
>irrational debate between anaphiles and digiphiles (neither term is
>meant as anything more than a label). That is that, somehow, an analog
>representation, because it is continuous and not discrete, has more
>inherent information in it than a discrete, sampled digital
>representation. From one popular book:
Thanks for the article Dick,
I have circulated it to all of the engineers including Greg Mackie.
If we have any comments I will post them.
Chris Christensen
>Chris Christensen <chr...@fuggles.gvg.tek.com> wrote:
>>I'll add the typical artistic analogy here:
>>Digital audio is sampled and reconstructed.
>>Television is samples and reconstructed.
>>Analog audio recording in not sampled.
>>A photogtaph, while a sample in time, is more like analog audio...
>Sorry, Chris, they are ALL sampled and ALL reconstructed. The methodology
>of sampling and reconstructiod differs, but they are all sampled and
>reconstructed nonetheless.
I "knew" that, I just wanted to set you off on the folowing useful
discussion.
<really informative discussion deleted>
I wasn't clear in my analogy...... How about the difference between a nice
pin registered 35MM motion Picture Camers and a really nice Television
Camera.
Both make wonderful pictures.
Both record and reproduce images in a similar manner, frame samples
exposure, but there is a clear visible difference in the picture
quality......
I did a test last night. I have the same artists album on both CD and LP.
The LP has only been played on a inline tracking turntable just a few times.
It hasn't been deep cleaned.
I set up the two sources on a simple bandwith limited mixer that has a
control that fades between two sources. The mixer also has level controls
for each channel. Here is where you have to trust my experience, at least
for this simple test...... I adjusted the levels of the two sources by ear.
This test _IS_ flawed but I wanted to report the results anyway......
This mixer was connected to my bandwith limited studio console, potted up to
my bandwith limited monitor amplifier and finally fed to my band limited (and
usually maligned) monitor speakers!
I hope you will bear with me on this test when I say what I have to say but
in balancing the level I was listening to midrange sounds so I could be
surprised by what I was expecting to hear.... :-)
So I run both playbacks badk to the stary and drop the needle. I started on
the LP side, listened then switched to the CD side. I was quite surprised
that I noticed just about what I expected to hear. The LP had more top end.
The LP had more top end on midrange sounds. What I heard was what I would
have expected to hear if I was mixing either live or studio and wanged the
shelving HF control up a bit.......
I don't for a minute expect anyone to accept this test on anything other
than face value, there was no precision accuracy in level matching, the
RIAA curve of the phonl preamp wasn't tweeked for the precise transfer
curve, and the whole test was done using direct comparison by someone who
is addmitedely predisposed to wanting to hear a difference....
I really want to set up a live test using spectrum analyzers, and really
wideband flat microphones and a fine string ensamble.... Maybe next summer
when I do The Festival again..... Anyone wanna be a judge to this test the
last coupla weeks of June 1995? Co-author to the paper?
rick.v...@mackie.wa.com
>So far, you've simply stated, as almost a matter of universal TRVTH(tm),
>
>the digital is Bad. Yet, other than hyperbole, strange, somewhat less
>than technical terminology akin to "yucky", and some way off-base
>pronouncements (that turned out to be quite incorrect) about the
>"infinite resolution of analog", you have yet to come up with ANY
>justification for such a statement.
We are not saying that digital is bad. We are saying that it is a fabulous
alternative for analog recording and editing of audio and video. When it
comes to mixers however, it is a far more complex method of arriving at
results that in many ways do not provide the same reproduction as current
quality analog mixer designs. Until now I havn't really talked to anyone
about this that disagreed with that assesment.
>
>Gee, that's okay, but lest the unwary take such pontifications as
>absolute gospel, which I am sure some will, regrettably, I think, to be
>fair, you have a pile of challenges to some of your stranger assertions
>that you have left totally unanswered.
I think there could be considered an equal amount of disservice done to the
same group by asserting that a digital mixer is superior to an analog mixer
simply because it's digital. This is what we are dealing with more and more
every day and it just ain't so..
>Yet, other than hyperbole, strange, somewhat less
than technical terminology akin to "yucky", and some way off-base
pronouncements (that turned out to be quite incorrect) about the
"infinite resolution of analog", you have yet to come up with ANY
justification for such a statement. <
You may want to take a trip to the AES convention in Texas this November.
Carl Malone will be giving a lecture on the digital control of analog and
it's advantages over pure digital mixer technology. Since he provided me
with much of the information I have on posted here on this subject, perhaps
you would like to challenge his findings in a public setting surrounded by
some of the other prominent players in digital and analog audio today. I am
confident that he may be able to clearify his statements for you. I assume
that you are an active AES member ?
We will also be showing at the AES show in San Fransisco this November 10th.
I and most of the New Products group will be there. Please stop by our
booth, I would love to see you discuss this topic with our digital guys
face to face.
[]Rick.
A Bellware
(Chris Christensen) writes:
>So I run both playbacks badk to the stary and drop the needle. I started
on
>the LP side, listened then switched to the CD side. I was quite
surprised
>that I noticed just about what I expected to hear. The LP had more top
end.
>The LP had more top end on midrange sounds. What I heard was what I
would
>have expected to hear if I was mixing either live or studio and wanged
the
>shelving HF control up a bit.......
It may very well have been that when the LP was mastered, someone did
indeed wang the shelving HF control up a bit.
I haven't done too many vinyl and CD projects, but in all of them the
vinyl mastering was done separately, with a different mastering engineer,
with different aesthetic and technical needs. In CD mastering, technical
concerns about stereo phase, overloading the low end and mechanical
questions about tracking and the angle of the lathe are not as great as
with vinyl mastering. One would have to master an LP and a CD at the same
time (and presumably without using a Neve DTC which would band-limit the
LP the same way as the CD) to obtain a real scientific result.
Of course, if the CD or LP had a greater amount of HF noise, test subjects
might interperate the recordings as being "brighter."
I'd be interested in seeing the results of a double-blind study.
-Andrew Bellware
Gabe Wiener
Bill Vermillion <bi...@bilver.oau.org> wrote:
>On the BEST analog machines I had no problems getting them
>'relatively' flat out to over 30KHz - at 30 ips.
But Bill, the problem is that there are almost no electronic sources
that produce correlated information up that high that is worthy of
being recorded.
--
Gabe M. Wiener -- ga...@panix.com | "I am terrified at the thought that
N2GPZ -- PGP public key on request | so much hideous and bad music may be
Quintessential Sound, Inc. | put on records forever."
Recording / Mastering / Restoration | --Sir Arthur Sullivan
Gabe Wiener
<rick.v...@mackie.wa.com> wrote:
>
>Our 8-Bus mixer has a bandwidth of 20Hz to 60Khz internal from Input to
>Output. How would I get a digital converter with a bandwidth of 20Hz to
>20Khz to give me the *same* output ?
Why would you want to? Why do you wish to pass up to 60 kHz when nearly
all recording devices used today are constrained to the audio bandwidth?
>Mixers and tape recorders are not the same thing. Many people who are
>actively involved in digital recording fail to realize this fact.
No, but both of them need only pass what we are capable of hearing.
>Digital came about to improve recording and editing.
Actually, digital came about to improve telephony.
>Mixers don't
>necessarily benefit from being digital. Especially if they cost 10 times
>more money and yield less audio performance than their analog counterpart.
While the price argument may be a good one, you have yet to show us in
any convincing scientific way how they "yield less audio performance
than their analog counterparts."
>If we could find a converter that offered us the same bandwidth as our
>analog circuitry and could be purchased for the same price as an Op-Amp then
>we would be talking. Even then this would be more of a marketing decision
>than one driven by necessity.
You seem to fail to grasp the the concept that your bandwidth is not limited
by the mixer, but by what you're putting into it.
bi...@mix.com
<rick.v...@mackie.wa.com> writes:
> You may want to take a trip to the AES convention in Texas this November.
> Carl Malone will be giving a lecture on the digital control of analog and
> it's advantages over pure digital mixer technology. Since he provided me
> with much of the information I have on posted here on this subject, perhaps
> you would like to challenge his findings in a public setting surrounded by
> some of the other prominent players in digital and analog audio today. I am
> confident that he may be able to clearify his statements for you.
But this is a public setting - in fact it doesn't get more "public" than
this - and I'm sure there are some "other prominent players" present too.
How about Carl comimg here? Then all the rest of won't be left hanging..
Billy Y..
RobtLee
truth.
The real need for 24-bit audio comes in the DSP part of a digital mixer.
Equalization in particular requires many mathematical iterations, and 24
bits gives you the mathematical headroom to do complex equalization with
little accumulated error. Each iteration carries with it some quantization
error (depending on the function) because you can't get resolution finer
than the least significant bit; any remainder below the LSB goes into the
bit bucket. On successive iterations this error can grow and grow if the
DSP designer isn't careful. 24 bits gives you 256 times the resolution of
16-bit audio, and thus more precision for DSP.
-Bob Lee
| "If it sounds good, it is good." -Duke Ellington |
C.M. Hicks
>In article <35qd9l$d...@lyra.csx.cam.ac.uk> c...@eng.cam.ac.uk (C.M. Hicks)
>writes:
>>chr...@fuggles.gvg.tek.com (Chris Christensen) writes:
>>>In article <4.47712901.M...@mackie.wa.com> rick.v...@mackie.wa.com writes:
>>[...munch...]
>>>>How can a digital signal...Which is a *sample* not the entire signal but a
>>>>portion of the signal have the same sonic purity as the original ?
>I'll add the typical artistic analogy here:
>Digital audio is sampled and reconstructed.
>Television is samples and reconstructed.
>Analog audio recording in not sampled.
>A photogtaph, while a sample in time, is more like analog audio...
I am a photographer as well as an audio engineer! A photograph has
grain (in simple terms, the smallest speck of colour that can appear)
that is directly analogous to electronic noise in an analogue audio
system. Details in the photograph smaller than the grain size are lost
in just the same way as details in the music below the noise are lost.
>>Ah, ME! :-)
><Location recording story deleated>
[...munch...]
>The source impedance difference is a factor of 2......
[... Rupert Neve anecdote and discussion deleted...]
[...more deleted...]
>problem is that my daytime employer doesn't share my quest for this specific
>answer..... Then my evenings and weekends are "shared" with my artistic
>quests....
I know the feeling!
[...more munching...]
>I have said this before, I believe that the filtering of the harmonics of
>music above 20kHz, while outside the realm of scientific possiblity (which I
>embrace, science that is... :-) Is audible...... Simple ABX testing would
>prove this, I haven't set this up and tested it though...... Maybe Dick has?
Nor me, though it is a very difficult test to set up. An apparently
simple question to ask would be (for example) "Does a 12kHz sinewave
sound different from a 12kHz squarewave?" On the face of it this is a
simple thing to test: put headphones on a load of people and play
tones to them and ask. However I fear it is fraught with difficulties,
not least the requirement for a linear transducer with sufficient
(linear) bandwidth. I would be very interested in hearing of such
tests if anybody knows of any case histories.
[...getting really full now! ]
>Thanks for the discussion
Likewise!
>Would anyone like to set up a test at AES?
Would that a poor UK student such as I could afford to come!
Monte P McGuire
>In article <4.47713235.M...@mackie.wa.com>,
> <rick.v...@mackie.wa.com> wrote:
>>
>>Our 8-Bus mixer has a bandwidth of 20Hz to 60Khz internal from Input to
>>Output. How would I get a digital converter with a bandwidth of 20Hz to
>>20Khz to give me the *same* output ?
>
>Why would you want to? Why do you wish to pass up to 60 kHz when nearly
>all recording devices used today are constrained to the audio bandwidth?
Pardon my stomping in here, but the answer is that to pass ??-20KHz in
the most clean manner, it is sometimes simpler to also pass 60KHz than
to agressively remove everything above 20KHz.
The simplest possible lowpass filter is the first order filter. It
has very good time response (i.e. no ringing or other nasties) and it
starts rolling off early and takes a long time to do its job. It
'wins' in the transparency category since it is simple to implement;
the simplest form is a passive RC filter, which we can make very
cleanly these days.
However, to keep 20KHz relatively untouched by a first order filter,
you must move the turnover point a few octaves away from 20KHz. That
is why I too feel the need for wide bandwidths.
>You seem to fail to grasp the the concept that your bandwidth is not limited
>by the mixer, but by what you're putting into it.
True, but if you look at the problem as a cascade of two filters; the
source and the mixer, you see that if you want to have the source
dominate the overall system response, the mixer's bandwidth must
exceed that of the source. It's not like we're going to find new
signals at ultrasonic regions just because we can pass them; the point
is to make the source dominate the overall response of the system, not
the mixer. That is what tranparency is about.
Regards,
Monte McGuire
mcg...@world.std.com
Steven Graham
explanatory post, wrote:
> A dBv is a ratio using 1 volt as a reference. A dBm uses 1 milliwatt,
dBV (Capital V) is referenced to 1 volt. If I'm not mistaken, dBv (small
v), like dBu and dBs are referenced to .775 volts. (If there is a
difference between these three terms, what is it?)
--
-------------------------------
Steve Graham: sgr...@umich.edu
Nigel Orr
Steven Graham <sgr...@umich.edu> wrote:
>Richard D Pierce (DPi...@world.std.com), in his otherwise excellent
>explanatory post, wrote:
>> A dBv is a ratio using 1 volt as a reference. A dBm uses 1 milliwatt,
>
>dBV (Capital V) is referenced to 1 volt. If I'm not mistaken, dBv (small
>v), like dBu and dBs are referenced to .775 volts. (If there is a
>difference between these three terms, what is it?)
I'll take a stab at this, hopefully someone will jump in if it's
wrong...
First off, the easy one- dB's are a relative measurement- on their
own, they are not referenced to anything.
The dB is one tenth (deci-) of a Bel, where the Bel is calculated from
the measured value and the reference value as
log(measured_value/reference_value)
Now, as I recall, the dBu is a voltage reference, which is referenced
to the voltage required for 1mW of power transfer on a 600ohm line, so
using simple ohms-law derived power formula,
P=V^2/R
V=sqrt(P*R)
=sqrt(1E-3*600)
=0.7746V
or approximately 0.775V. I think dBv is referenced to the same value,
but is not the standard use due to the possibility of confusion with
the dBV referenced to 1V, mentioned above.
So, for a 600ohm line, the dBm value will be the same as the dBu one.
--
Nigel o...@tks.oulu.fi
Nigel Orr
>Hey, let's stop bouncing on Rick about 24 bit digital audio. He speaks the
>truth.
>
>The real need for 24-bit audio comes in the DSP part of a digital mixer.
Well I would agree with this- however, I seem to remember that it was
in the ADC and DAC's that Rick was suggesting 24 bits was required.
The more bits in the DSP the better, I would expect, with the law of
diminishing returns setting in when you produce a more accurate signal
from the DSP than the DAC can successfully convert.
--
Nigel o...@tks.oulu.fi
Gabe Wiener
>v), like dBu and dBs are referenced to .775 volts. (If there is a
>difference between these three terms, what is it?)
Folks, there are four terms here:. dB, dBv, dBu, and dBV.
The dB is simply a logarithmic unit of ratio, figured as 10 log P/Pref.
I don't think this one needs any explaining.
In the early days of electronics, the folks at Bell Telephone decided
on the standard references for their system, and they had a jump on
all of us, since the telephone company was the first group of people
who were concerned with moving audio around. The paradigm is as follows:
0.775 volts across 600 ohms produces 1 milliwatt.
Thus, across 600 ohms, 0 dBv (0.775 V) = 0 dBm (1 milliwatt). dBm is
thus defined at 10 log P/1mw, and dBv is thus 20 log V/0.775, the
coefficient being 20 because P=V^2/R. By the way, this also means that
dBm=dBv-dBz (where dBz, not really dB per se, is 10 log Z/600). A useful
formula at times.
The Europeans, not having Bell Telephone to dictate gospel, decided to
use 1 volt as their reference, and that is notated as dBV. Since the
American standard dBv could easily be confused with their dBV, most
Europeans call the American standard dBu.
Thus:
dBv, dBu 0.775 V reference
dBV 1 V reference
dBm 1 mW reference.
Bill Vermillion
>In article <1994Sep22....@bilver.oau.org>,
>Bill Vermillion <bi...@bilver.oau.org> wrote:
>>On the BEST analog machines I had no problems getting them
>>'relatively' flat out to over 30KHz - at 30 ips.
>But Bill, the problem is that there are almost no electronic sources
>that produce correlated information up that high that is worthy of
>being recorded.
Oh yes, I know that. There are even a lot of sources with
much lower frequency response that aren't worth of being
recorded too :-) (But I guess that's for a producer to decide).
I was following up to Dick's query on what had the widest
bandwidth, flatest response, and lowest distortion.
My comment mentioned the 'trap' that many would fall into by
thinking a higher frequency response meant broader bandwidth,
but such is not the case in analog. Each time you move the
top end up, you typically will move the bottom end too.
Back to the first line of 'information that high that is worthy
of being recorded'. True - but to my ears when I was working
with a machine that was relatively flat out to 30KHz, the
signals we are really intersted in - those below 20Khz usually
sounded much better. Sort of like the oversampling approach
in digital - not really - but sort of :-)
That being said - the machines that went out that far easily
were Studer's and they typically sound much better than their
competition in their higher end models.
th
Nigel Orr
Bill Vermillion <bi...@bilver.oau.org> wrote:
>My comment mentioned the 'trap' that many would fall into by
>thinking a higher frequency response meant broader bandwidth,
>but such is not the case in analog. Each time you move the
>top end up, you typically will move the bottom end too.
Now, I notice that digital machines have quoted responses from 20Hz to
20kHz- does it really not extend below 20Hz, or is that just all that
it's felt is necessary to spec?
I would have thought they would have response down to DC, neglecting
any AC coupling on the inputs etc- so why the quoted 20Hz minimum?
Are 20Hz and 20K really the approximate 3dB points?
--
Nigel o...@tks.oulu.fi
Richard D Pierce
Rick from Mackie SPECIFICALLY asserted the need for 24 bits width in the
converter statges, and never once mentioned the issues you talk about
above, which are completely separable. No one argues that internal
caluclations may well require extended precision, but he specifically
stated that it as an issue of "infinite resolution" of analog "once you
gte above the noise floor", which is demonstrably wrong.
Richard D Pierce
<rick.v...@mackie.wa.com> wrote:
>In article of 1:46 AM 9/23/94, DPi...@world.std.com (Richard D writes:
>>
>>Gee, that's okay, but lest the unwary take such pontifications as
>>absolute gospel, which I am sure some will, regrettably, I think, to be
>>fair, you have a pile of challenges to some of your stranger assertions
>>that you have left totally unanswered.
>
>I think there could be considered an equal amount of disservice done to the
>same group by asserting that a digital mixer is superior to an analog mixer
>simply because it's digital. This is what we are dealing with more and more
>every day and it just ain't so..
If, Rick, you are suggesting that I am among the camp that suggested, a
priori, that digital is superior to analog, then you defame me by putting
words in my mouth of your own confabulation.
I specifically challenged your demonstrably incorrect assertions about
the "infinite resolution" of analog and such, an assertion the is
demonstrably false by any rational means by which you care to test it.
A challenge, by the way, you have for one reason or another, chosen to
avoid completely, instead taking a completely diversionary tack that does
not address the flaws of your original assertions.
>You may want to take a trip to the AES convention in Texas this November.
>Carl Malone will be giving a lecture on the digital control of analog and
>it's advantages over pure digital mixer technology. Since he provided me
>with much of the information I have on posted here on this subject, perhaps
>you would like to challenge his findings in a public setting surrounded by
>some of the other prominent players in digital and analog audio today.
Other than the silliness about the "infinite resolution of analog" and
"grainy mushiness of digital" and other stuff like that, you've failed to
convey any real information on the topic. WHat, specifically, does your
friend Carl have to say on the challenges I and many others made to those
assertions (which, given the lack of ANY information to the contrary,
seem to be YOUR assertions, not his)?
>I assume that you are an active AES member ?
Of what relevance is this?
>I would love to see you discuss this topic with our digital guys
>face to face.
Rick, you made some pretty wild assertions about analog and digital
principles in THIS forum, how about answering the objections in THIS
forum?
Richard D Pierce
<rick.v...@mackie.wa.com> wrote:
>In article of 12:31 PM 9/25/94, ga...@panix.com (Gabe Wiener) writes:
>
>>While the price argument may be a good one, you have yet to show us in
>>any convincing scientific way how they "yield less audio performance
>>than their analog counterparts."
>
>suggestion. Lets resume this debate in 18 months. By then there will be
>far more evidence to help emphasize our point. Due to our policies
>surrounding new products, I am at a dissadvantage here by not being able to
>offer supporting information as it would possibly divulge our future plans.
Rick, stop the equivocation and answer the question. The question had
nothing to do with price, it had nothing to do with future development,
it had nothing to do with you you or me or anyone else may have in
development. The question was plain and simple: prvide us with your
SCIENTIFIC reasoning why digital does not provide the necessary ;level of
performance for apssing audio. There may well be good reasons: you have
not come up with any.
>>You seem to fail to grasp the the concept that your bandwidth is not
>> limited
>>by the mixer, but by what you're putting into it.
>
>No...The mixer can limit bandwidth if it fails to reporduce the levels being
>fed into it.
You are confusing level and bandwidth AND avoiding the specifi question
Gabe was asking, all at once?
>If you can produce excellent results using a cost effective and time tested
>design approach, why would you want to go the other direction in costs to
>acheive the same thing at 10 times the cost and complexity ?
That assumes that every one of your premises leading to this conclusion
are true, and in at least two cases, the "infinite resolution of analog"
and the "grainy mushiness of digital," they are not. Additionally, you've
picked this 10x cost factor from I don't know where. Thus, I would
suggest, simple logic holds that if your premises are flawed, how
dependable is the conclusion?
Richard D Pierce
Monte P McGuire <mcg...@world.std.com> wrote:
>In article <3648ke$1...@panix.com>, Gabe Wiener <ga...@panix.com> wrote:
>>In article <4.47713235.M...@mackie.wa.com>,
>> <rick.v...@mackie.wa.com> wrote:
>>>
>>>Our 8-Bus mixer has a bandwidth of 20Hz to 60Khz internal from Input to
>>>Output. How would I get a digital converter with a bandwidth of 20Hz to
>>>20Khz to give me the *same* output ?
>>
>>Why would you want to? Why do you wish to pass up to 60 kHz when nearly
>>all recording devices used today are constrained to the audio bandwidth?
>
>Pardon my stomping in here, but the answer is that to pass ??-20KHz in
>the most clean manner, it is sometimes simpler to also pass 60KHz than
>to agressively remove everything above 20KHz.
>
>The simplest possible lowpass filter is the first order filter. It
>has very good time response (i.e. no ringing or other nasties) and it
>starts rolling off early and takes a long time to do its job. It
>'wins' in the transparency category since it is simple to implement;
>the simplest form is a passive RC filter, which we can make very
>cleanly these days.
>
>However, to keep 20KHz relatively untouched by a first order filter,
>you must move the turnover point a few octaves away from 20KHz. That
>is why I too feel the need for wide bandwidths.
The effects you discuss are certainly true of the kinds of filters built
with simple linear analog components, first, second, etc. linear cascaded
filters. It is NOT necessarily true of filters implemented with other
topolgies nor is it necessarily true of filters implemented in the
digital domain.
Design a simple passive ladder network of nothing but R's and C's that
has the characteristic of flat frequency response at 20 kHz (+- .25 dB)
and has 80 dB rejection at 22 kHz and look at the phase response of that
filter below cutoff. Now, do the same thing in a 64x oversampled digital
filter and measure the phase response.
It's possible (and EASY), in the digital domain to have a bandwidth of 20
kHz with NO phase shift (say, less that 1-2 degrees TOTAL) in the
passband. It's very difficult to do it in analog AND do it right in
production. That's one reason why analog system require large bandwidths
to satisfy the constraints for a 20 kHz bandwidth.
>True, but if you look at the problem as a cascade of two filters; the
>source and the mixer, you see that if you want to have the source
>dominate the overall system response, the mixer's bandwidth must
>exceed that of the source. It's not like we're going to find new
>signals at ultrasonic regions just because we can pass them; the point
>is to make the source dominate the overall response of the system, not
>the mixer. That is what tranparency is about.
Look at the phase and frequency (and, thus, the transient) response of
the microphones used for recording: THAT's the source.
Richard D Pierce
That's exactly why the low end is limited: to keep DC out of the
converters.
I can quote the frequency response of the entire throughput chain of the
Orban DSE 7000 (because I've measured a bunch of them). They all fit
within a spec of 20 to 20 kHz, +- 0.25 dB at any input level at a sample
rate of 44.1 kHz, which is the spec that Orban will guarantee. They are 3
dB down in the low end at typically about 2-4 Hz because of the input
coupling cap. At the high end, -3 dB point is somewhere between 21 and 22
kHz. In reality, I have never seen one deviate beyond about +- 0.12 dB 20
to 20 kHz.
rick.v...@mackie.wa.com
>In article <4.47713235.M...@mackie.wa.com>,
> <rick.v...@mackie.wa.com> wrote:
>>
>>Our 8-Bus mixer has a bandwidth of 20Hz to 60Khz internal from Input to
>
>>Output. How would I get a digital converter with a bandwidth of 20Hz
> to
>>20Khz to give me the *same* output ?
>
>Why would you want to? Why do you wish to pass up to 60 kHz when nearly
>all recording devices used today are constrained to the audio bandwidth?
Hi Gabe,
Since so much of this discussion has been based around the differences or
lack of them as they appear on paper, let me ask you this..
If we came out with a new mixer that had a frequency response and bandwidth
spec only from 20Hz-20kHz how many mixers would we sell ?
>>Mixers and tape recorders are not the same thing. Many people who are
>>actively involved in digital recording fail to realize this fact.
>While the price argument may be a good one, you have yet to show us in
>any convincing scientific way how they "yield less audio performance
>than their analog counterparts."
Rather than bashing our heads against the wall over this I have a
suggestion. Lets resume this debate in 18 months. By then there will be
far more evidence to help emphasize our point. Due to our policies
surrounding new products, I am at a dissadvantage here by not being able to
offer supporting information as it would possibly divulge our future plans.
>
>>If we could find a converter that offered us the same bandwidth as our
>>analog circuitry and could be purchased for the same price as an Op-Amp
> then
>>we would be talking. Even then this would be more of a marketing
> decision
>>than one driven by necessity.
>
>You seem to fail to grasp the the concept that your bandwidth is not
> limited
>by the mixer, but by what you're putting into it.
No...The mixer can limit bandwidth if it fails to reporduce the levels being
fed into it.
If you can produce excellent results using a cost effective and time tested
design approach, why would you want to go the other direction in costs to
acheive the same thing at 10 times the cost and complexity ?
[]Rick.
rick.v...@mackie.wa.com
>But this is a public setting - in fact it doesn't get more "public" than
>this - and I'm sure there are some "other prominent players" present
> too.
>How about Carl comimg here? Then all the rest of won't be left
> hanging..
Well,
If Carl ever has a couple of hours a day free to participate in these very
lengthy discussions I will suggest that he get involved. Based on the work
load that we have in front of us over the next 18 months, I wouldn't hold
your breath. {g]
In fact...My time here is growing short as well. I am training our tech
support staff on the basics of Internet participation. My goal is to hand
the whole thing over to them on the first of October. I will still be able
to deal with automation questions through E-Mail for the time being. That
too will eventually be handled exclusively by our tech support staff.
The Fall is tradeshow hell for us and is also the time when new products are
readied for release. As one of the New Products Managers I am smack dab in
the middle of all the fun !
[]Rick Vartian
Automation Projects Manager
Mackie Designs Inc.
rick.v...@mackie.wa.com
>Hey, let's stop bouncing on Rick about 24 bit digital audio. He speaks
> the
>truth.
>bit bucket. On successive iterations this error can grow and grow if the
>DSP designer isn't careful. 24 bits gives you 256 times the resolution
> of
>16-bit audio, and thus more precision for DSP.
>
Thanks Rob,
Our original point was the ability to achieve the same performance that is
currently available with analog. When we are talking an all digital mixer
there are many aspects of audio processing that are taken for granted in
analog that can get really altered in a lower than 24bit digital process.
Thanks for helping to clearify our original point.
[]Rick.
Brandon Mathew
to mix down my multiple tracks of digital that reside on the ADAT/DA-88.
Now are you telling me that I want to go through another D->A and A->D
just so that I can mix this? I only want one A->D and one D->A in the
process. Once it has been converted to digital, I don't want it to be
converted to analog until just before the power amp at audiences site.
This device would not have any A->D's in the front end and would only
have D->A's in the back end as a convenience for monitoring the mix.
Am I the only one who feels this way? Is this need already being met by
the various multitrack DAW's?
--
Brandon Mathew - bra...@core.rose.hp.com
Chris Christensen
>In article of 12:31 PM 9/25/94, ga...@panix.com (Gabe Wiener) writes:
>
>>In article <4.47713235.M...@mackie.wa.com>,
>> <rick.v...@mackie.wa.com> wrote:
Rick sez:
>>>Our 8-Bus mixer has a bandwidth of 20Hz to 60Khz internal from Input...
Gabe responds:
>>Why would you want to? Why do you wish to pass up to 60 kHz when nearly
>>all recording devices used today are constrained to the audio bandwidth?
Rick responds:
>Since so much of this discussion has been based around the differences or
>lack of them as they appear on paper, let me ask you this..
>If we came out with a new mixer that had a frequency response and bandwidth
>spec only from 20Hz-20kHz how many mixers would we sell ?
Chris sez:
Ask Yamaha, their analog audio mixers are all bandlimited for noise
reasons.... I am NOT defending the practice, just answering the question...
Warren Harris
Excellent point. Ideally, once your material is in the digital domain,
it would be preferable to process it entirely as 1s and 0s until it needs
to be "heard" by someone. Needless to say, the fewer A/D-D/A conversions
that take place, the less "distortion" you have to accept.
The trick has been to do this in any affordable manner - to date.
---Warren
--
__ __ __
/ \_/ \_/ >__
\__/ \__/ \__> >_Ich habe keine idee!
/ \_/ \_/ >__> ====================
\__/ \__/ \__> Duck! (:<
RobtLee
Pierce) writes:
>In article <364tp2$f...@newsbf01.news.aol.com>, RobtLee <rob...@aol.com>
>wrote:
>>Hey, let's stop bouncing on Rick about 24 bit digital audio. He speaks
the
>>truth.
>>
>>The real need for 24-bit audio comes in the DSP part of a digital mixer.
>>Equalization in particular requires many mathematical iterations, and 24
>>bits gives you the mathematical headroom to do complex equalization with
>>little accumulated error. Each iteration carries with it some
quantization
>>error (depending on the function) because you can't get resolution finer
>>than the least significant bit; any remainder below the LSB goes into
the
>>bit bucket. On successive iterations this error can grow and grow if the
>>DSP designer isn't careful. 24 bits gives you 256 times the resolution
of
>>16-bit audio, and thus more precision for DSP.
>Rick from Mackie SPECIFICALLY asserted the need for 24 bits width in the
>converter statges, and never once mentioned the issues you talk about
>above, which are completely separable. No one argues that internal
>caluclations may well require extended precision, but he specifically
>stated that it as an issue of "infinite resolution" of analog "once you
>gte above the noise floor", which is demonstrably wrong.
############ AND ###########
In article <365tp6$3...@ousrvr.oulu.fi>, orr@tk1 (Nigel Orr) writes:
>In article <364tp2$f...@newsbf01.news.aol.com>, RobtLee <rob...@aol.com>
>wrote:
>>Hey, let's stop bouncing on Rick about 24 bit digital audio. He speaks
the
>>truth.
>>
>>The real need for 24-bit audio comes in the DSP part of a digital mixer.
>Well I would agree with this- however, I seem to remember that it was
>in the ADC and DAC's that Rick was suggesting 24 bits was required.
>The more bits in the DSP the better, I would expect, with the law of
>diminishing returns setting in when you produce a more accurate signal
>from the DSP than the DAC can successfully convert.
>--
>Nigel o...@tks.oulu.fi
I see what you guys are saying, but do you propose having 16- or 18-bit
ADCs and DACs, and 24-bit DSP in the middle? I think you might as well
start and finish with the precision you need. It's not completely
separable, especially if it's all in the same box.
I have nothing against 16-bit audio--in fact, it sounds wonderful coming
from my CDs--but if you need to do any signal processing and mixing, it's
better to have the extra bits of data.
I inferred from Rick that by infinite resolution of analog he meant the
lack of discrete steps, with any signal elements that are not obsured by
noise being reproducible. Sounds like a reasonable need for 24-bit ADCs
and DACs to me. The more bits, the more precision and less quantization
error (quantization noise) you get, and that's a good way to start your
signal path in a digital mixer, I would think.
In an analog mixer, even the inputs and outputs have filtering, not just
the EQ section. Analog filtering is based on resistive and reactive
components, along with any active circuitry appropos to the application.
Essentially you've got integrators and/or differentiators (for you
calculus freaks) and a few other configurations, with feedback loops. The
feedback signals are continuous analog signals, and in a filter circuit
it's not just the instantaneous voltage of the signal that matters, but
also its rate of change (or derivative--uh oh, calculus again). But it's
all continuous, analog signal.
In DSP filtering, though, typically you've got to look at the current
sample--and the one before--and the one before--and the one before--and so
on. You can get only as precise as the discrete, FINITE sampling intervals
and LSB values allow.
Let's give it a rest now and let the engineers design something for us.
Richard D Pierce
>> Dick Pierce said:
>>Rick from Mackie SPECIFICALLY asserted the need for 24 bits width in the
>>converter statges, and never once mentioned the issues you talk about
>>above, which are completely separable. No one argues that internal
>>caluclations may well require extended precision, but he specifically
>>stated that it as an issue of "infinite resolution" of analog "once you
>>gte above the noise floor", which is demonstrably wrong.
>############ AND ###########
>In article <365tp6$3...@ousrvr.oulu.fi>, orr@tk1 (Nigel Orr) writes:
>
>>The more bits in the DSP the better, I would expect, with the law of
>>diminishing returns setting in when you produce a more accurate signal
>>from the DSP than the DAC can successfully convert.
>
>ADCs and DACs, and 24-bit DSP in the middle?
No, in the case of the DSE 7000, we're talking 16 bits in, 32 bits in the
middle, and 16 bits out. The 32 bits is required because during the
mixing process, a 16 bit audio sample is multiplied by 16 bit
normalized gain/pan/send/whatever value, added together to form the mix,
noise shaped and re-stored in RAM or sent to the DACs.
>I think you might as well
>start and finish with the precision you need. It's not completely
>separable, especially if it's all in the same box.
If you scale gain and noise shape properly, it's TOTALLY separable.
>I have nothing against 16-bit audio--in fact, it sounds wonderful coming
>from my CDs--but if you need to do any signal processing and mixing, it's
>better to have the extra bits of data.
And most workstations do.
>I inferred from Rick that by infinite resolution of analog he meant the
>lack of discrete steps, with any signal elements that are not obsured by
>noise being reproducible. Sounds like a reasonable need for 24-bit ADCs
>and DACs to me.
Look, we'll say it again: CONTINUOUS DOES NOT MEAN INFINITE RESOLUTION IN
AMPLITUDE. That is the bottom line. The presence of finite noise and
finite levels means finite resolution, period. There is no way to get
around that. The noise completely irretrievably hides ANY and ALL real
changes in the signal below the noise.
>The more bits, the more precision and less quantization
>error (quantization noise) you get, and that's a good way to start your
>signal path in a digital mixer, I would think.
But argued against the absolutely irrefutable presence of real noise,
extra bits do not buy you any less quantization error below the noise.
Extra bits simply let you accurately quantize the analog noise, and of
what benefit is that?
Once the quantization error is the same as the analog noise level, NO
EXTRA BITS WILL BUY YOU ANY MORE RESOLUTION.
>In an analog mixer, even the inputs and outputs have filtering, not just
>the EQ section. Analog filtering is based on resistive and reactive
>components, along with any active circuitry appropos to the application.
>Essentially you've got integrators and/or differentiators (for you
>calculus freaks) and a few other configurations, with feedback loops. The
>feedback signals are continuous analog signals, and in a filter circuit
>it's not just the instantaneous voltage of the signal that matters, but
>also its rate of change (or derivative--uh oh, calculus again). But it's
>all continuous, analog signal.
>
>In DSP filtering, though, typically you've got to look at the current
>sample--and the one before--and the one before--and the one before--and so
>on. You can get only as precise as the discrete, FINITE sampling intervals
>and LSB values allow.
Absolutely not correct. First, you ignore the fact that in the analog
situation, the bandwidth of the system STRICTLY sets a maximum limit on
the slope of the change (the derivative, more accurately, the rise and
fall time) of the signal. Thus, there can be no more information
"between" two points that are separated by the reciprocal of twice the
bandwidth that have ANY significance to the signal. If there were, the
bandwidth would have to be wider. Once again, you miss the principle:
CONTINUOUS DOES NOT MEAN INFINITE RESOLUTION IN TIME, EITHER! The
resolution is a strict and unchangeable function of bandwidth.
Secondly, your description of digital filtering ignores the fact that the
samples take place at the SAME intervals, given the same bandwidth, that
an analog system looks at in terms of resolution in time.
>Let's give it a rest now and let the engineers design something for us.
I and others would be more than happy to give it a rest if people would
get the underlying principles right and get over the assumption that
continuous means infinite. IT DOES NOT. It might seem intuitively correct,
but the intuition, in this case, is dead wrong.
And, oh by the way, I AM one of those engineers that design things for you,
and I HAVE to obey these underlying principles.
Those rules are VERY simple. They are also extremely rigid and
unbendable, despite the shrill protestations of many:
1. The resolution of amplitude is limited strictly by the
dynamic range of the system, regardless of whether the
system represents amplitude as continuous or discrete
values.
2. The resolution in time is limited strictly by the
bandwidth of the system, regardless of whether the
system represents time as continuous or sampled
intervals.
Some correlaries of the above rules that are important:
1. In order for a system to have infinite resolution, it
must have either a finite bandwidth and INFINITE dynamic
range, INFINITE bandwidth and finite dynamic range, or
both INFINITE bandwidth AND INFINITE dynamic range.
2. No currently extant system of ANY KIND has either infinite
bandwidth and/or infinite dynamic range, thus NO system has
infinite resolution.
And, lastly, for those that want to try:
3. Based on two other very solidly established principles, the
conservation of energy and the time/frequency uncertainty
principle, no system could ever have infinite dynamic range
(because it would require infinite energy to represent infinite
dynamic range), and no system could ever have infinite bandwidth
(since that would require the system AND the signal to exist for
infinite time). Thus IT IS FUNDAMENTALLY IMPOSSIBLE FOR ANY SYSTEM
TO EVER HAVE INFINITE RESOLUTION.
Q.E.D.
Chris Christensen
>In article <35nfs0$4...@gv-gate.gvg.tek.com> chr...@fuggles.gvg.tek.com (Chris
>Christensen) writes:
>#I was doing a location "studio" recording a coupla months ago. I was using
>#my "trusty" Mackie 1202 and my sublime Denon DTR-2000 DAT Deck. Not this
>#comparison _may_ be invalid but when I plugged my 'phones into the Mackie
>#'phone fack vs. the Denon 'phone jack there was an audible difference in the
>#HF response.... The reason I say the comparison may be invalid is that I
>#haven't looked at the schematics to rule out an obvious electrical
>#difference.......
>It is not valid because the headphone sections could and probably
>do vary a great deal.
Kent, Thank you for your observations but I had applied a disclaimer to the
posting that noted the in valid nature of the tests..... In a recent
follow-up I also reported the electrical differences that might have been
responsible for the differences as well as a plan to minimise the
differences for future comparisons....
It is also my intent to utalize a seperate headphone amp with a switchable
line input to test this theory further....
>#For the sake of arguement, that all is OK as far as the headphone circuits.
>That's a gross assumption.
I know.
>The analog sections of the DAT are different fron the analog signal path
>in the mixer and probably aren't as good so they should and would sound worse.
The analog path in the Denon is quite short and sounds very good. The only
tifferences that I heard were in thenHF region. Those differences can be
attribuued to the fact that the Denon HP circuit had a buildout resistor
twice the value of the Mackie.....
Ronald J Mann
No, and in fact I spent a wad of money, time and energy to do just that
(DMC1000 + Lexicon300 + Spectral Synthesis 16 track DAW +++)!
There, in my opinion, is a big opportunity for lower cost digital mixers
at the same price point as an ADAT or DA88. Digital io for so-called digital
effects units would be nice to!
Cheer up! despite all the ravings of the Mackie mavens low cost digital mixers
are on the horizon. I paid ~27000 for my DMC1000. Yamaha just dropped the price
about $8000. The ProMix demonstrates that it can be done for far less.
Companies stick there heads in the sand all the time. If theres a demand
some one will fill it. Given Mackies attitude toward digital my money is on
Yamaha to really shake things up (much the way Mackie did) in the next few years.
=Ron Mann=
rick.v...@mackie.wa.com
>If, Rick, you are suggesting that I am among the camp that suggested, a
>priori, that digital is superior to analog, then you defame me by putting
>
>words in my mouth of your own confabulation.
>I specifically challenged your demonstrably incorrect assertions about
>the "infinite resolution" of analog and such, an assertion the is
>demonstrably false by any rational means by which you care to test it.
I'll tell you what...I will try to set up a conference call with Carl myself
and you. Hows that ? I think we can resolve this a lot more efficiently
and after that you and I can post the results here. This is the only
solution I can see for this as we just can't take the time to address this
through E-Mail any longer. Let me know when we can call you voice.
>>I assume that you are an active AES member ?
>
>Of what relevance is this?
Ahhmmm,
You need to be an AES member to attend the lectures. Does this mean that
you are not a member of the AES ? That truly surprises me ! I would have
guessed you to be one of the administrators based on the amount of
information you have and your incredible desire to share it with others.
I could put you in touch with Les Tyler "President of That Corporation"
who is one of the active AES administrator types.
>>I would love to see you discuss this topic with our digital guys
>>face to face.
>
>Rick, you made some pretty wild assertions about analog and digital
>principles in THIS forum, how about answering the objections in THIS
>forum?
Like I said...I think at this point it would make more sense for those
engineers who have disagreed with what we have said here are welcome to talk
to us at the AES show in San Fransisco November 10th - 14th. I am sure that
we could clearify in a couple of minutes what looks like it will take months
of three page messages to resolve here.
[]Rick
Chris Christensen
Brandon, I am reasonable confident that you might already know my reply...:
>So one a somewhat related topic, lets assume that I want a digital mixer
>to mix down my multiple tracks of digital that reside on the ADAT/DA-88.
>Now are you telling me that I want to go through another D->A and A->D
>just so that I can mix this?
This is just exactally what happens in most pop production..... The project
may be tracked on a digital multi but it was put there through an analog
desk!
Then the signal takes another pass through the analog domain on it's way to
the two track.....
And then, depending on the house, it will make it's final trip through the
A-D/D-A debauchery in the mastering facility....
This is typican and there _are_ many exceptions (like Gabe's work).
>I only want one A->D and one D->A in the
>process. Once it has been converted to digital, I don't want it to be
>converted to analog until just before the power amp at audiences site.
This is a truely worthy goal. As soon as I sold my soul into digital I have
been progressing to this end, not there yet, but soon...
>This device would not have any A->D's in the front end and would only
>have D->A's in the back end as a convenience for monitoring the mix.
Yep, analog monitoring, that 'necessary' evil, that is until someone
develops the Serial Digital Spinal Tap..... :-)
>Am I the only one who feels this way? Is this need already being met by
>the various multitrack DAW's?
You are in good company. Some of the finest recordings in the world are
done this way! The quantity is low relative to the total music business
output.... :-(
A decent multi track DAW in the size necessary to do a large pop project is
very expensive indeed, although there are a couple of notable products on
the market.... The main limitation is (still) disk space. If one inmports
just the takes that one wants to use it still could be several hours of
material. The proces sounds very attractive though. The ability to cut and
splice from various takes, bits and snips........
Richard D Pierce
<rick.v...@mackie.wa.com> wrote:
>>>I assume that you are an active AES member ?
>>
>>Of what relevance is this?
>
>Ahhmmm,
>
>You need to be an AES member to attend the lectures.
No, I don't. I have both attended and given lectures before professional
organizations of which I was not a member.
> Does this mean that
>you are not a member of the AES ?
No, it doesn't mean anything of the sort.
>That truly surprises me ! I would have
>guessed you to be one of the administrators based on the amount of
>information you have and your incredible desire to share it with others.
> I could put you in touch with Les Tyler "President of That Corporation"
>who is one of the active AES administrator types.
Thanks, I met Les quite some years ago.
>>>I would love to see you discuss this topic with our digital guys
>>>face to face.
>>
>>Rick, you made some pretty wild assertions about analog and digital
>>principles in THIS forum, how about answering the objections in THIS
>>forum?
>
>Like I said...I think at this point it would make more sense for those
>engineers who have disagreed with what we have said here are welcome to talk
>to us at the AES show in San Fransisco November 10th - 14th. I am sure that
>we could clearify in a couple of minutes what looks like it will take months
>of three page messages to resolve here.
No, Rick, I have not seen a single one of your engineers disagree with
the points I have made. I've only seen you disagree. And you're not
disagreeing with me, you're disagreeing with Shannon, Nyquist, Blesser,
Pohlmann and an array of others who I will more than willingly defer to.
You also are on the verge of disagreeing with Newton and others, but
hopefully, we'll catch you before you fall of THAT precipice!
Michael Hyman
: >>Hey, let's stop bouncing on Rick about 24 bit digital audio. He speaks
: the truth.
The sad fact, unfortunately, is that he often doesn't.
: I see what you guys are saying, but do you propose having 16- or 18-bit
: ADCs and DACs, and 24-bit DSP in the middle? I think you might as well
: start and finish with the precision you need. It's not completely
: separable, especially if it's all in the same box.
: I have nothing against 16-bit audio--in fact, it sounds wonderful coming
: from my CDs--but if you need to do any signal processing and mixing, it's
: better to have the extra bits of data.
Audio DSPs often have internal 48-bit data paths. Are you suggesting
that 48-bit DACs are required?? The fact of the matter is that the very
best pro audio effect boxes use only 16- or 18-bit DACs on their input
stages. The *internal* data path's wider bit-width can accomidate any
lesser input signal by padding it with 0's in the lsb's. This is common
practice, and suffers from no sonic compromise due to the bit-width of the
original signal. I suppose your armchair comments can be excused since
you're from AOL. :-)
However...
From Rick's congratulatory comments on your original incorrect technical
argument, I can only assume that he's only interested in furthering his
*own* agenda, and could care less about "speaking the truth" (as you put
it).
--
- Mike (mi...@netaxs.com)
Gabe Wiener
>If we came out with a new mixer that had a frequency response and bandwidth
>spec only from 20Hz-20kHz how many mixers would we sell ?
Plenty, since that bandwidth would exceed the useful bandwidth of most
recorders found today in the oodles of semi-pro studios that compose an
ever-increasing market share.
>>While the price argument may be a good one, you have yet to show us in
>>any convincing scientific way how they "yield less audio performance
>>than their analog counterparts."
>
>Rather than bashing our heads against the wall over this I have a
>suggestion.
So do I. Rather than bashing our heads against the wall, I suggest that
you STAND BEHIND WHAT YOU SAY and tell us what "yield audio performance"
means, and tell us how analog has "infinite resolution" means, and that
you justify the wild, unscientific assertions you have made here.
My dear sir, I respect your company's fine products more than you'll
know, and I recommend them consistently in my writings, talks,
discussions, etc. But if the knowledge of electrical engineering that
you have seen fit to show us is even remotely prevalent at your firm,
then I dare say that your firm has come up with such a dazzling array
of fine products out of sheer luck.
And if your rejoinder will be that your job position does not require
you to have a full command of freshman electrical engineering, then
kindly refrain from trying to re-educate us on information theory that
was proven in 1928.
Sorry, but I get a little irate when people actually try to convince
others that a mixer can have infinite resolution, infinite bandwidth,
or infinite dynamic range. Any one of these assertions violates a few
small rules. Oh, nothing serious, just things like, say, conservation
of energy, or finite existence of the universe.
>Lets resume this debate in 18 months. By then there will be
>far more evidence to help emphasize our point.
No, let us have the discussion now, since you got us into it.
>Due to our policies
>surrounding new products, I am at a dissadvantage here by not being able to
>offer supporting information as it would possibly divulge our future plans.
The supporting information can be found in any undergraduate textbook on
information theory or signal processing. You needn't divulge a set of
plans in order to illustrate concepts like bandwidth, frequency, dynamic
range, discrete time sampling....
And believe me, while I'm sure your next product will be a zinger, I can
guarantee you and everyone else that you are _not_ about to re-define
electrical engineering with its introduction.
>No...The mixer can limit bandwidth if it fails to reporduce the levels being
>fed into it.
My good fellow, please spare us. ANY good digital mixer made today
can reproduce the levels produced by *ANY* sound source found in the
music industry today, and certainly anything that can fit on the
dynamic range of a compact disc. This is not a debatable point. In
any system of even marginal quality, the mixer is not the
bandwidth/dynamic-range bottleneck. The head end equipment nearly
always is. And if not the head-end electronics, then it's the
ambience.
Timothy E. Onders
>>all recording devices used today are constrained to the audio bandwidth?
>
>the most clean manner, it is sometimes simpler to also pass 60KHz than
>to agressively remove everything above 20KHz.
True. The whole reson behind over-engineering a mixer is to minimize
it's contribution to the degredation of the signal (the reason that
the mic pre should provide the dominant gain in the flow, while others
hover around unity).
>The simplest possible lowpass filter is the first order filter. ...
>... simplest form is a passive RC filter, which we can make very
>cleanly these days.
Quality higher-order active filters are also fairly easy to build
these days too. Filters are not usually the reason for the huge
bandwith, but the principle is correct. By extending the analog
bandwidth significantly beyond the audio range, you tend to have better
phase and amplitude characteristics in the range of interest.
These problems, however, do not directly translate to digital audio. In
the case of filters, properly designed digital filters are not restricted
by the physical properties of active components. This is part of the
magic of digital! A high-order filter can be designed with linear
phase response, or any other design parameter you wish to set. The
only real limit is the number of MACs that can be done in a single sample.
Similarly, it might be expected that the anti-imaging filter of the D/A
converter would cause significant distortion since it must have as close
to a brick-wall response as possible around 20 kHz. This is solved by
another trick, called oversampling. Oversampling effectively raises the
sampling frequency, often to as high as 400 kHz or higher. A low-order
filter can then be used to eliminate the images, without significant
distortion in the band of interest.
Digital audio may sometimes seem kind of hokey, but it really does all
work.
Tim Onders
Audio Design Engineer
ond...@netcom.com
ke...@infoserv.com
writes:
#In article of 5:37 PM 9/17/94, ga...@panix.com (Gabe Wiener) writes:
#
#
#>Eh? This doesn't follow. Kindly explain to us how 24 bits would give
#>performance that is "similar to analog" whereas, say, 20 bits wouldn't?
#
#
#More dynamic range and more of the sonic data from the original analog
#source.
#
#How can a digital signal...Which is a *sample* not the entire signal but a
#portion of the signal have the same sonic purity as the original ?
#
#How can a 20 Bit digital mixer which can only acheive 119db of dynamic range
#offer the same dynamic response internally as an analog mixer that can have
#internal operating ranges of above 135db ? Also an alalog console doesn't
#suffer from that grainey mush at low levels.
You answered his question but digital mixers have internal accumulators of
40-56+ bits which gives you several hundered dB of internal gain/headroom.
So the internal operating ranges of digital consoles is much better than
any analog.
#
#I totally buy digital for use as a recording medium, mainly cause analog
#tape recorders were so much less capable than Digital ones.
#
#Mixers on the other hand are a whole different ball game. There is very
#little sense in tracking through any of the current digital mixing boards.
#Most mid-priced analog consoles are far better at handling the signal and
#maintaining the sonic quality of those signals. All the producers that I
#know are still waiting for digital mixers to out perform analog. I would
#say that it is at least 7-10 years away.
Like I said in a previous post, you had better talk to Yamaha, Neve, SSL, and
AT&T because you are very wrong.
Kent
--
/* "There is no king who has not had a slave among his ancestors and */
/* no slave that has not had a king among his." ---- Helen Keller */
/* Kent L. Shephard ----- K. L. Shephard Consulting */
ke...@infoserv.com
Christensen) writes:
#
#I was doing a location "studio" recording a coupla months ago. I was using
#my "trusty" Mackie 1202 and my sublime Denon DTR-2000 DAT Deck. Not this
#comparison _may_ be invalid but when I plugged my 'phones into the Mackie
#'phone fack vs. the Denon 'phone jack there was an audible difference in the
#HF response.... The reason I say the comparison may be invalid is that I
#haven't looked at the schematics to rule out an obvious electrical
#difference.......
It is not valid because the headphone sections could and probably
do vary a great deal.
If you are looking at quality out of headphone output, boy do I have
news for you.
#
#For the sake of arguement, that all is OK as far as the headphone circuits.
That's a gross assumption.
The analog sections of the DAT are different fron the analog signal path
in the mixer and probably aren't as good so they should and would sound worse.
Kent
Nigel Orr
>
>I'll tell you what...I will try to set up a conference call with Carl myself
>and you. Hows that ? I think we can resolve this a lot more efficiently
[...]
>Like I said...I think at this point it would make more sense for those
>engineers who have disagreed with what we have said here are welcome to talk
>to us at the AES show in San Fransisco November 10th - 14th.
Unfortunately, you posted to a worldwide net- I for one won't be in
San Fran, and, as Dick said, you did post your assertions to this
forum originally.
>we could clearify in a couple of minutes what looks like it will take months
>of three page messages to resolve here.
We could- give us at least _one_ reference to a scientific paper
published in a peer-reviewed journal which backs what you say- surely
you can manage that? I'm assuming here that the design engineer's
opinions of which you speak are based on some experimentation, not
just on a personal theory? If so, tell us where we can read about it-
it would be _so_ much simpler.
--
Nigel o...@tks.oulu.fi