Topic: new web page lists recorder test results

[flintstone]

A German fellow named Raimund Specht has published
his measurements of the self-noise of the mic input
of many portable recorders.  His results are posted here
http://www.avisoft.com/recordertests.htm

My observations, from these test results:
The Sound Devices 722 delivers the goods.
The Tascam HD-P2 is right there with the 722.
The new Fostex FR-2LE is a pleasant surprise.
The Korg MR-1000's preamp is mediocre.
With other recorders, you pay less and get less.

Dr. Specht's numbers reflect some of the observations
made by members of this group.  The Edirol R-09's relative
insensitivity to clipping is there.  The Marantz PMD660's
need for a pad at high sound pressures is there.  And the
low noise level of the Hi-MD recorder compared to the
M-Audio Microtrack and Edirol R-09 is noted. So I'm pretty
confident in these test results.

These tests were made at each recorder's maximum gain setting.
That's a worst-case scenario. It's likely that many of the
recorders tested, particularly the less expensive models,
would perform better with the gain turned down.

Dr. Specht's company, Avisoft, sells PC software used in
bioacoustics research.  Avisoft makes an expensive preamp
that's designed to help capture very high frequency sound,
like the calls that bats make.  Read about Avisoft at
http://www.avisoft.com

Flintstone

[silence]

A German fellow named Raimund Specht has published
his measurements of the self-noise of the mic input
of many portable recorders.  His results are posted here
http://www.avisoft.com/recordertests.htm

My observations, from these test results:
The Sound Devices 722 delivers the goods.
The Tascam HD-P2 is right there with the 722.
The new Fostex FR-2LE is a pleasant surprise.
The Korg MR-1000's preamp is mediocre.
With other recorders, you pay less and get less.



Flintstone

Wow, the SONY HiMD MZ-RH really holds its own for a $ 300 recoder

[spyder9]

I'm stunned on the Tascam's dynamic range.   

[libfab]

I agree with flintstone on the fact that the Specht tests are biased against cheaper recorders. As both SunWizard and guysonic show, R-09 self-noise sharply increases between "low" and "high" recording levels, and at "high",  between #25 and #30. Recording with maximum gain as Specht does give disappointing results wrt lower levels. In most cases though, recording at "low" levels will be more than enough, at least at 24bits.
My conclusion is: don't run for an Hi-Md just for those tests : in many cases, it will clip when an MT/R-09 would not (see the thread on setting levels with wide dynamics). That's for silence.

[flintstone]

libfab's right, the Hi-MD recorder did clip at much lower
sound pressure level in these test results (this is the
one recorder that Specht did not test personally). 

So if you're recording a loud venue, then one of the other
recorders would be a better choice (or you'd need to pad
the mic input, just like you have to with the PMD660).

Flintstone

[gratefulphish]

I cannot seem to get this page to open. Is anyone else having problems?

[Kindguy]

I cannot seem to get this page to open. Is anyone else having problems?

yes wont open.

[firebaugh]

I cannot seem to get this page to open. Is anyone else having problems?

yes wont open.

ah, she's dead jim. 

[libfab]

Just in case you can't get to the Specht page, I've copied parts of it below:

""Noise levels at the maximum available gain settings

 Equivalent Input Noise (EIN)
impedance: 150 ohms

Input Clipping Level (2) Dynamic Range (3)  A-weighted unweighted

20Hz - 20kHz dBu mVrms A-weighted

SoundDevices 722 -130dBu (-130dBu) -128dBu (-128dBu) -46dBu 3.8mV 84dB
Tascam HD-P2 -127dBu -125dBu -40dBu 7.7mV 87dB
Fostex FR-2LE -129dBu -125dBu -45dBu 4.3mV 84dB
(85dB)
Marantz PMD671 -125dBu -123dBu -50dBu 2.4mV
(1.2mV) 75dB
(65dB)
Marantz PMD670 -120dBu -118dBu -52dBu 1.9mV
(1.2mV) 68dB
(65dB)
Marantz PMD660 -120dBu -118dBu -52dBu 1.9mV
(1.2mV) 68dB
(60dB)
KORG MR-1000 -117dBu -115dBu -46dBu 3.8mV 71dB
M-Audio MicroTrack 24/96 -115dBu -113dBu -45dBu 4.4mV 70dB
Edirol R-09 -106dBu -104dBu -41dBu 6.9mV 65dB
SONY HiMD MZ-RH1 (4) -124dBu -122dBu -64dBu 0.49mV 61dB
SONY DAT TCD-D3 -117dBu -115dBu -56dBu 1.3mV 61dB
INA163-based preamp -130dBu -128dBu   

Measurement conditions: Recording format: 44.1 KHz, 16 bit, mono (left channel). All parameters were measured at the maximum available gain setting. Limiter and noise cancel options were disabled. Note that a bit-depth of 24 bit would not improve the noise level figures at these high gain settings. When available, the figures in brackets indicate the manufacturer's specifications. 
 The errors of the measurements should be less than +-2dB
2: The input (clipping) level at the maximum available gain setting represents the maximum input sensitivity of the recorder. 0 dBu is equivalent to 775 mVrms. Note that the dB gain figures that are often specified by the manufacturers do not represent a meaningful characterization of the relevant input sensitivities that would be needed in practice.
3: The dynamic range at the maximum available gain setting alone should not be interpreted as an indicator the performance of a recorder because these numbers will depend on the actual input level (or effective overall gain).
4: The MZ-RH1 measurements were done by John Beale
 The calculation and filtering procedures were executed by using the Avisoft-SASLab Pro sound analysis software, version 4.39. ""

[flintstone]

This link to Dr Specht's test results works for me today.
Maybe the server was offline for a while.  Give it a try.

http://www.avisoft.com/recordertests.htm

Flintstone

[paullySC]

Would like to see it with modded units.

[meatling]

I'm stunned on the Tascam's dynamic range.   

I believe this would be a fallacy of sorts: As you can see, the dynamic range is simply the (arithmetic) difference between EIN and clipping level.

In this table, the clipping level is the lowest level sufficient to overload the input stage when set to full gain.  Hence, this test punishes recorders with powerful preamps, and the Tascam simply isn't one of those.

Nevertheless these data are very useful, if I'm not mistaken: Just subtract your mic's sensitivity (re 0.775V/Pascal) from EIN, and you've got a rough idea about the preamp noise in your recordings.

m

[DSatz]

Old thread, I know, but: Preamp noise measurements are very tricky. For one thing, as Meatling pointed out, different recorders have different gains when turned up all the way (which is how they were apparently all tested by this guy), so the numbers for different units weren't obtained on any consistent basis whatsoever.

The only noise level that matters to me is the noise level when the preamp is set for the amount of gain that I'm actually going to use--and there is simply no way to derive or predict that from knowing the noise level with any other gain setting. Turning up the gain by 10 dB doesn't usually increase the noise by 10 dB, though with some equipment in some ranges of gain it does; my point is that this behavior is highly individual with different circuits, and not "extrapolatable" from some other measurement.

Then there's the fact that this noise is at very low equivalent sound pressure levels where our hearing is attuned mainly to high frequencies, but the measurements were unweighted. And there's the fact that noise can be steady and smooth, or full of momentary peaks ("shot noise"), or anything in between; so if a noise level is given as, say, 10 mV what do you know about its actual audibility with your microphones in regard to a given maximum SPL? I'd say it's somewhere between "not enough information" and "next to no information."

--best regards