Topic: R-09 self noise comparison at different settings

[SunWizard]

I ran a series of tests to see the R-09 self noise at all the settings.  This is useful info for when I record nature and acoustic shows.  I hooked up a mini-plug with a 3.3k ohm resistor across each channel.  Then recorded at steps of 5 gain on both inputs, and hi low settings, at 16/44.1.  Here are the results after analyzing them in Har-Bal, attachments are a spectrum plot comparing line-in at 8 (approx 0 gain) and mic-lo at 30 (max) and the files are attached incase you want to analyze yourself :

Here is the Average power figures from Har-bal to compare for all the tests I ran:
Gain   High     Low    Line
30     57.2    80.3   80.1
25     61.8    84.2   84.3
20     66.7    86.5
15     71.4    87.6
8                 88.5   89.0

What this shows that is helpful for me is that running the gain switch on high increases the noise too much.  So running high output mics like the AT-853 helps since they work great on line or mic low settings, both below -100db across the spectrum at 25 gain.  Low and Line are almost the same noise and gain, the difference is you can get plug power on the mic-in.  When using line or mic low the last steps from 25 to 30 is where you get the most added noise of 4db , so I will run at 25 or lower, line or mic low only, and add any makeup gain in post.

[SunWizard]

The other question this answered for me is whether its worth it to run an external pre.  Since my AT-853 will be run on the mic low setting, even if I set gain at 25 which would only be needed for very quiet stuff, adds only 3.7db of noise.  I don't know of any external pre plus cables I could use that would only add 3db of noise so not using one is best, and it sure is nice not to carry another box and cables.

[Blue Noise]

Thanks for the data, SunWizard. Very useful information.

[SunWizard]

I ran these same tests for line-in at 24/48 to see if it made any difference, and I get the same self noise results.

[udovdh]

Does anybody know how a 16-bit sample can give noise levels below -96 dB?

[SunWizard]

Does anybody know how a 16-bit sample can give noise levels below -96 dB?

Because thats not a SNR, its a level measurement which can go anywhere, doesn't depend on bit depth.

[guysonic]

Good question! 
Just got an R-09 a few days ago and have started a review. 

Input noise spectrums at different settings are finished, so take a look at the graphs, and the gain structure chart, and PLEASE allow for the very unfinished look, only 30% finished I think at this time.

Go to: http://www.sonicstudios.com/r-09revw.htm

[poorlyconditioned]

Good question! 
Just got an R-09 a few days ago and have started a review. 

Input noise spectrums at different settings are finished, so take a look at the graphs, and the gain structure chart, and PLEASE allow for the very unfinished look, only 30% finished I think at this time.

Go to: http://www.sonicstudios.com/r-09revw.htm

Excellent job!!

Lots of valueable info.  You've also got a great argument for an external preamp there.  Nice work...

  Richard

[SunWizard]

Excellent job!!
Lots of valueable info.  You've also got a great argument for an external preamp there.  Nice work...
  Richard

Yes very nice info page, thanks guysonic.  I would be interested to see a comparison of mic low at 25, which is where I will run based on my tests, to the figures you show.  Looks like it will be almost the same whether you do internal or external pre with that setting.  Because most of the noise is added when going from 25 to 30.  And it confirms my tests of not wanting to use mic high setting, thats where an external pre would be nice if your mics need lots of gain.

Also, did you test with lower settings on line-in than 18?  From my tests a setting of 8 was close to 0 gain, and it seems that when using an external pre with better specs that you may get even lower noise with a lower setting on the R-09, maybe even lower than 8?

[udovdh]

Good review. What about the mods for the light?
Either remove it, replace it or put a resistor in series so it generates less light?

[guysonic]

Good review. What about the mods for the light?
Either remove it, replace it or put a resistor in series so it generates less light?

I was thinking the same thing, but removing it is less attractive as knowing clearly when in pause mode saves occasional grief with lost recordings.  The service manual is being printed this week.  I have one on order for this and other potential improvements (like ability to directly power DSM mics).

Excellent job!!
Lots of valueable info.  You've also got a great argument for an external preamp there.  Nice work...
  Richard

Yes very nice info page, thanks guysonic.  I would be interested to see a comparison of mic low at 25, which is where I will run based on my tests, to the figures you show.  Looks like it will be almost the same whether you do internal or external pre with that setting.  Because most of the noise is added when going from 25 to 30.  And it confirms my tests of not wanting to use mic high setting, thats where an external pre would be nice if your mics need lots of gain.

Also, did you test with lower settings on line-in than 18?  From my tests a setting of 8 was close to 0 gain, and it seems that when using an external pre with better specs that you may get even lower noise with a lower setting on the R-09, maybe even lower than 8?

Gain change from MIC input set at LOW-to HIGH is about 24 dB, and it seems by the graphs posted in the review that the character of the noise plot does NOT change, only the level driven by the applied coarse gain set by the H/L switch (24 dB jump), and further modified with REC level adjustment in 1 dB steps. 

If this is as true as it seems, then any of the review's spectrum plots can be used to determine R-09 input noise at ANY REC level setting by subtracting exactly 1 dB x #30 minus #whatever-REC-#setting-you-want from the #30 (MAX setting) level depicted in the graphs. 

[SunWizard]

Gain change from MIC input set at LOW-to HIGH is about 24 dB, and it seems by the graphs posted in the review that the character of the noise plot does NOT change, only the level driven by the applied coarse gain set by the H/L switch (24 dB jump), and further modified with REC level adjustment in 1 dB steps. 

If this is as true as it seems, then any of the review's spectrum plots can be used to determine R-09 input noise at ANY REC level setting by subtracting exactly 1 dB x #30 minus #whatever-REC-#setting-you-want from the #30 (MAX setting) level depicted in the graphs. 

Thats not true as it seems, from my tests on mic-low the noise went up 4 db when changing settings from 8 to 25, then another 4db from 25 to 30.  So its not a linear function, looks more exponential.  Your formula is correct for mic-high setting.

[Gutbucket]

Excellent information SunWizard!

I just got back from a trip recording ambient nature stuff and acoustic music, testing matched 4060's>MMA6000>R-09.  I've no measured data, but also no question that the gain of the MMA6k was signifcantly cleaner than that of the R-09 line in (which wasn't bad).  I mostly kept the R-09 line in set at around 8-10 (per your unity gain measurment) and made up the gain I needed on the DPA pre. 

With playback via ER4s I could detect higher hiss levels when the R-09 was set to the mid-teen range, so I tended to keep it around 8-10 which allowed for enough gain even for very quiet sources with the MMA6k maxed out.

I'm quite impressed.

[Gutbucket]

Guysonic,

Nice review, thanks for your work!
Had a little trouble deciphering the table showing gain/noise levels, FWIW.