Hi Rob,
I can understand the methodology of comparison tests and they are
invaluable for noticing artifacts that would be difficult to
determine by evaluating numbers, such as an annoying audible tone
from a power supply. The trouble with a single EIN number is that it
is missing the spectrum graphical view. This is probably why the LS-
10 appeared at first to have a lower noise figure due to it's
decreased low frequency performance. I guess I was suggesting a more
common engineering test for devices. One that wouldn't rely on having
such good ears to hear detailed differences, and one that was
absolutely repeatable. I realize that kind of testing would require
specialized equipment like a spectrum analyzer and a frequency
generator. I found this link that could be interesting,
http://tinyurl.com/bd49vo, for future tests. One other thing I wanted
to suggest is to test the unit at different gain setting. Common, one
chip amps often have much worse noise performance at less than full
gain. I don't think we would normally use max gain in the field. I
tend to use 10-40dB gain.
I agree that the FR2-LE so far looks to be the modest budget choice.
I just wish they had a SPDIF input, which is why I had some hopes to
see how well the PMD661 performed. At least the PMD661 looks to be a
decent performer, and could be a second recorder for my applications.
Wow, none of these lower cost units tested have lower input noise
than microphone self noise of say a NT1-A? That is sad, but
understandable for equipment that is built for around $50.
Thanks for all the work that goes into the comparison tests. I
understand they are time consuming. I think we all appreciate anyone
who takes the time to share.
Bruce Rutkoski
--- In Rob Danielson <>
wrote:
>
> At 4:52 PM +0000 2/20/09, Bruce wrote:
> >Thanks for the file post Tom as I have been very interested in how
> >the PMD661 works. I worked on the file a little to try to normalize
> >the Tic Tocs and applied a rumble filter. With this done, the
middle
> >recording, the PMD661 is actually lower noise by 2-4dB. It seems
the
> >PMD661 has better low frequency response.
>
> Hi Bruce--
>
> As you infer, using normalization on the lowest frequencies or on
the
> clock ticks to match playback levels tends to assume that all of
the
> recorders' pres are "flat." The Hz response differences in Tom's
> samples are pretty dramatic, though we usually detect differences
> when the recording chains are used. I find it possible to match
> levels for these purposes by ear. I play them on several monitor
> systems and concentrate on matching overall "impact." I can hear
> subtleties in sound elements that come and go across the edits
> joining the test segments better than listening within the clips.
>
> As for the greater low-end response of the 661, we've observed that
> Hi-MD recorders have less low-end response (under 100Hz) compared
to
> many recorders. The LS-10 seems to also have lower response
compared
> to the 661. The ringing resonance around 250 Hz in the PMD661 and
> LS-10's samples is curious in that its not in the Hi-MD sample. Tom
> and I corresponded and it seems the he exercised excellent control
> over background sounds and took the best snippets from long takes.
> Its possible that the comparison would be a little easier to judge
if
> the clock was moved closer. The room is quite 'live." Term-wise, as
> per the subject of this discussion, recorders have "input noise"
and
> mics have "self-noise."
>
> >
> >I have a suggestion for future tests. If we want to test a new
unit's
> >noise level, wouldn't it be best to terminate the input with a 150
> >ohm resister with the gains normalized first with a reference
signal
> >at the input?
>
>
> Collecting reliable input noise measurements is very useful for
> seeing the "tiers" of mic pre performance that are available.
Here's
> an example:
>
> Let's say I've done my homework and have determined that my planned
> rig will have a pair of very low-noise mics for recording ambience
in
> quiet locations. With some guidance, I can tell by looking at
> Raimunds' chart http://www.avisoft.com/recordertests.htm, that
> there's a crop of recorders measured at -118 dBU and below that are
> not going to be well-suited for my purpose. Now I'm down to a more
> manageable list.
>
> When it comes to the next tier "up" of lower-cost recorders (-120
dBu
> to -126 dBu), its tempting to think I can rely on the numbers to
> determine if the noise added by the recorder's pre will be
acceptable
> or not. We have experiences that suggest this is probably not the
> case for my demanding gear requirements.
>
> Within the last 12 months, I was one of the recordists who became
> encouraged by Raimund's input noise measurement of the LS-10 at
> -121dBu, only 3 dBu noisier than respected Hi-MD units. Some
> recordists bought LS-10's and favorable verbal accounts began to
> trickle in. Then some concerns began to creep in. Vicki Powys
> eventually conducted some comparison tests with low noise mics and
we
> learned, as Tom's test confirms, that the input noise of the LS-10
is
> quite warm and is considerably more audible than one would expect.
> With more side-by-side comparison testing, we discovered an
imbalance
> in the stereo channels in Vicki's LS-10 (which might be in Tom's
> recent sample as well?).
>
> The number of relevant comparison tests on the newest recorders is
> quite limited (most testers use noisy mcs and/or low pre gain). As
a
> result, we are forced to recommend the Fostex FR2-LE because its
> still the only, known, lower-cost recorder with pres quiet enough
for
> mics with the lowest self-noise ratings. Beginners on a tighter
> budget should be able to make a good first-investment. As they are
> usually making several quality compromises we can only guess about
> the expected performance of certain gear combinations without
actual
> performance comparisons to refer to. We've discovered both dogs
and
> diamonds in rough by not making assumptions.
>
>
> >The recording can then be graphed by frequency and
> >amplitude in a program like Adobe Audition. When a mic is added it
> >seems we are testing the microphone noise level more than the unit.
> >Any microphone should have much higher noise level than any descent
> >input amplifier. The noise of a variable test environment and
> >variable set-up sounds would greatly effect the test results from
> >person to person, and over time. Maybe I am missing the point of
> >these tic toc tests? I realize they have value and are helpful but
I
> >really want to know the noise performance of a unit in numbers and
> >graphs, rather than listening to different effects of mic noise,
when
> >making a purchasing decision.
>
> I don't think any of the recorders in Tom's test have pres that are
> "clean" enough to be completely inaudible behind NT1-A's self-
noise.
> In this case, and many, one is judging which recorder's input noise
> is more acceptable and eventually, "acceptable enough." Its great
> when one can include a more universally praised standard like a
744T
> in the same test, but one can also make this cross-comparison using
> other tests like, http://tinyurl.com/2ejuob
>
> A good way to make initial investments that we have often discussed
> is to do the research, narrow it down to a best guess _system_ and
> then buy the gear from dealer that will allow it to be returned if
it
> doesn't meet one's needs. Oryoki has made a list:
> http://tinyurl.com/b6985l This way, one will get tangible results
and
> gain a much better sense of other options to pursue.
>
> Tests and measurement charts will not help the person who has
bought
> a recorder before knowing which mics, powering, and stereo array
will
> be used. This practice used to be a tad more common it seems to me.
> Perhaps our discussions about not accepting the vague claims of
> salespeople and manufacturer literature is having some positive
> effect. Rob D.
>
> >
> >Bruce Rutkoski
> >www.natureguystudio.com
> >
> >--- In
> ><naturerecordists%
Message: 40yahoogroups.
Subject: com>
> >"tk7859" <g0sbw@> wrote:
> >>
> >> Hi Rob and the Group
> >>
> >> I ran the tests this morning and have sent you direct some WAV
> >files.
> >> I did two sets of tests; the first with the pre-amp gain turned
to
> >> maximum and a second series with the gain at 80% For a quick
> >> comparison I have put together a compilation of 10 second
snippets
> >> from the 100% gain recordings out of each of the three machines
The
> >> quietest (LS-10) is first and the loudest (MZ-NH1) last. Other
than
> >> being converted to mp3 the recordings are "as recorded"
> >>
> >> I don't think a lot can be deduced from this until the levels
have
> >> been equalised. However, the Marantz 661 seems to be good.
> >>
> >> The mp3 of the compilation is in the files section
> >>
> >> compLS66NH.mp3
> >>
> >> Cheers
> >>
> >> TomR
> >>
> >
> >
> >
>
>
> --
>
>
>
|