Topic: Edirol R09 - ADC not good enough for 24bit, right?

[poorlyconditioned]

I just got my Edirol R09 and did some tests.

Following Guysonic, I put a 1k resistor on the inputs, and recorded "silence" to see what the noise floor was.

I was suprised to learn the lowest floor is something like -90 to -92dB.  For example, both mic (lo sens) and line give this value for gains between #1 and #20.  Noise goes up to about -85dB at #30 gain.

The ideal noise floor for 16bits is -96dB, and for 24bits is -144dB (each bit is 6dB).  So, this is about 4-6dB *worse* than 16bits.  So, there is no point in recording anything in 24bit, because those extra 8 bits will be useless.

What I was hoping was for something like 17 or 18bits and the rest noise.  Ie., one or two bits better than 16bit.  But the current rig is comparabe to minidisc.  Approaching but not quite 16bits.

  Richard

[BayTaynt3d]

So, there is no point in recording anything in 24bit

Not sure, but I think there are more reasons than just that point. For one, no matter what the theoretical dynamic range could be, having 24 bits means more resolution with the sound you do have, and if you do any editing at all, having that resolution helps make quantization noise a non-issue, but not so at 16-bit. I'm sure there's more though, anyone?

[Nick's Picks]

I think it sounds great.  dont care how many bits its using.
my ears like it.

[poorlyconditioned]

OK, I tried the R09 out last week.  Franken-AT822 > batt box > R09 (mic/lo-sens level#25).

The sound on this thing is excellent!  The interface is great too.  So, it is a keeper.  I don't think recording at 24 bit has any gain, and certainly means less storage on a Flash card.  So, I'm going to stick to 16 bit.  But for that, it is a great unit.

  Richard

[spyder9]

The only thing I noticed when recording 24 bit:  Bass recorded seems thinner than it was at the show.  Noticed it in all 3 recorders when doing 24 bit: Tascam HD-P2, Edirol R09, and Zoom H4.  Anybody else?  Besides that, 24 bit is vundabar!   

[nihilistic0]

There may be more factors here


96dB for 16bit isn't 'ideal', it's the max.  I record 16bit into my JB3, but God knows I'm not seeing anything close to a 96dB noise floor, but that's due more to the lack of pre-amp than anything

[Sebastian]

Bass recorded seems thinner than it was at the show.  Noticed it in all 3 recorders [...]

Then it *must* be the ATs
(*ducks and hides*)

[Church-Audio]

So, there is no point in recording anything in 24bit

Not sure, but I think there are more reasons than just that point. For one, no matter what the theoretical dynamic range could be, having 24 bits means more resolution with the sound you do have, and if you do any editing at all, having that resolution helps make quantization noise a non-issue, but not so at 16-bit. I'm sure there's more though, anyone?

I think your 100% correct. Its about quantization and the bit path if this increases then the sampling information increase and you capture more of the wave form. Richard, the noise floor your talking about -148 or so is the "official" spec of 24 bit but very few 24 bit machines reach that because, you have to have a analog front end if you have a mic preamp. And your not going to get any analog front end that will do that unless you want to spend $1000's So I think your getting a huge advantage with 24 bit and I would stay where you are the edirol it is a great recorder imo. Its miles better then the microtracker for having less noise when we are talking about the front end of both units. 

[nihilistic0]

nothing does -144db noisefloor

I think the best we can hope for is 110-120

[Church-Audio]

nothing does -144db noisefloor

I think the best we can hope for is 110-120

Yep that's true.. I think -120 is pretty dam good. I mean if your recording a whisper its not great but for rock and roll you will never hear the difference between -120 and -140.. With a hot signal on your machine.

[poorlyconditioned]

nothing does -144db noisefloor

I think the best we can hope for is 110-120

Yep that's true.. I think -120 is pretty dam good. I mean if your recording a whisper its not great but for rock and roll you will never hear the difference between -120 and -140.. With a hot signal on your machine.

That is right, you can hope for something lower then -96, even -108 would be better, like two bits more than 16, or 12dB better.  But, alas, the *minimum* noise floor is -90 on the Edirol, so I don't see the point in grabbing 24 bits.

Anyway, as I said earlier, it sounds good, so I'll just take it as a 16 bit recorder.  It is also interesting that the Burr Brown spec sheet for the chip does not say 24 bit anywhere.

  Richard

[Carrera2]

<snip>

I was suprised to learn the lowest floor is something like -90 to -92dB.  For example, both mic (lo sens) and line give this value for gains between #1 and #20.  Noise goes up to about -85dB at #30 gain.

The ideal noise floor for 16bits is -96dB, and for 24bits is -144dB (each bit is 6dB).  So, this is about 4-6dB *worse* than 16bits.  So, there is no point in recording anything in 24bit, because those extra 8 bits will be useless.

<snip>

  Richard

Does anyone know how this would compare to a stock UA-5?  This is a function of the AD conversion, yes?

[Church-Audio]

<snip>

I was suprised to learn the lowest floor is something like -90 to -92dB.  For example, both mic (lo sens) and line give this value for gains between #1 and #20.  Noise goes up to about -85dB at #30 gain.

The ideal noise floor for 16bits is -96dB, and for 24bits is -144dB (each bit is 6dB).  So, this is about 4-6dB *worse* than 16bits.  So, there is no point in recording anything in 24bit, because those extra 8 bits will be useless.

<snip>

  Richard

Does anyone know how this would compare to a stock UA-5?  This is a function of the AD conversion, yes?

Its a basic function of the A/D but it starts at the front end if the noise floor on the front end of the preamp is -120 then no mater what converter is after it the noise floor is still -120. So it begins and ends with the analog mic preamp in the product your using. This is what really determines noise floor because no amount of bit rate or sample rate after that is going to change shit. It will just do a better job of chopping up the wave form and recreating it when it gets put back out into analog.


Chris Church

[Javier Cinakowski]

For me, 24bit makes no sense.  It takes up more HD space, it takes longer to process, convert and transfer. Also, the quietest thing I record is still fairly loud.  Also my UA5 (or ST9100) does not have specs that warrant 24bit.  I also do no editing in post, except for an occasional normalize.  Also, I distribute and listen to my recordings in 16bit.

[Church-Audio]

For me, 24bit makes no sense.  It takes up more HD space, it takes longer to process, convert and transfer. Also, the quietest thing I record is still fairly loud.  Also my UA5 (or ST9100) does not have specs that warrant 24bit.  I also do no editing in post, except for an occasional normalize.  Also, I distribute and listen to my recordings in 16bit.

16 bit has been around for a long time I use 24 with my sound analyzer program only because I am looking at waveforms but for playback so much of the things around the house are still 16 bit I don't really find the need for it in every day life of listening to music maybe if I was some kind of audiophile I would.