Topic: DPA 4060 (HEB) > D50

[Md2545]

I recently bought a pair of the Core-sound 4060 off of a member here. When I use them with my D50 running line in I have to set the record nob to about 7 or 8 to get it to peak at -12db, I was recording a loud jazz show and the recording turned out really low. I tried the mic in but everywhere I read people say to stay away from that, and when I did it seemed way to sensitive, I was going to try mic in with the 20db MIC ATT on. What are your thought on this. I'm thinking I have to much sensitivity for the mic in and not enough on the line in. Any help would be appreciated, there are alot of posts on 4060 here but not one that is specifically dealing with the D50

thanks,

MD

[Belexes]

Get a ST-9100 pre and power the mics from that, line-in.  That's what I used when I owned 4061's.

[Len Moskowitz (Core Sound)]

I recently bought a pair of the Core-sound 4060 off of a member here. When I use them with my D50 running line in I have to set the record nob to about 7 or 8 to get it to peak at -12db, I was recording a loud jazz show and the recording turned out really low. I tried the mic in but everywhere I read people say to stay away from that, and when I did it seemed way to sensitive, I was going to try mic in with the 20db MIC ATT on. What are your thought on this. I'm thinking I have to much sensitivity for the mic in and not enough on the line in. Any help would be appreciated, there are alot of posts on 4060 here but not one that is specifically dealing with the D50

Line level is roughly 26 dB below the Mic input level.  That's a lot quieter.

You should be using the low sensitivity (-20) setting for the Mic input.

Despite being marketed as a 24-bit recorder, the Sony -D50 has less than 16-bits of dynamic range, less than what DAT used to provide.  Its noise floors are also comparable to DAT, rather than true 24-bit recorders.

There are ways to turn your -D50 into a true 24-bit recorder with 110+ dB of dynamic range and a noise floor 20+ dB lower than the stock -D50.  The simplest is to use an external mic pre/ADC like our Mic2496 V2 or the Grace Lunatec V3 to feed the -D50's S/PDIF digital audio input. 

Using an analog mic pre feeding the Line inputs won't give you comparable performance.

[jlykos]

Despite being marketed as a 24-bit recorder, the Sony -D50 has less than 16-bits of dynamic range, less than what DAT used to provide.  Its noise floors are also comparable to DAT, rather than true 24-bit recorders.

Using an analog mic pre feeding the Line inputs won't give you comparable performance.

Are you sure about this?  Leonard from Sonic Studios found something very different in his tests and stated that the D50 does give true 24 bit performance with a very low noise floor from the line inputs, the best from any portable unit in its class (the FR2LE, PMD671, HD-P2, and the 722 were better).  Do you have any graphs or measurements to support your claim?

[Len Moskowitz (Core Sound)]

Are you sure about this?  Leonard from Sonic Studios found something very different in his tests and stated that the D50 does give true 24 bit performance with a very low noise floor from the line inputs, the best from any portable unit in its class (the FR2LE, PMD671, HD-P2, and the 722 were better).  Do you have any graphs or measurements to support your claim?

I'll try to explain.

Using the -D50's Mic inputs will give you less than 16-bits of dynamic range.  So no one (other than Sony marketers) would consider it a 24-bit recorder.

Using the -D50's Line inputs with microphones, you'll get a very low level signal unless you use an external analog mic pre to feed the Line inputs or you're recording a very loud sound source.  If the analog mic pre is not of very high quality, it will degrade the Line input's noise performance.  So simply adding a low cost external analog mic pre to the Line inputs will likely not significantly improve the -D50's performance over the Mic input, unless it is quite a special low cost analog mic pre. 

Do you know of any for which you can post noise and dynamic range data?  I'm definitely interested in hearing about one.

[Todd R]

Are you sure about this?  Leonard from Sonic Studios found something very different in his tests and stated that the D50 does give true 24 bit performance with a very low noise floor from the line inputs, the best from any portable unit in its class (the FR2LE, PMD671, HD-P2, and the 722 were better).  Do you have any graphs or measurements to support your claim?

I'll try to explain.

Using the -D50's Mic inputs will give you less than 16-bits of dynamic range.  So no one (other than Sony marketers) would consider it a 24-bit recorder.

I don't think what is being requested is an explanation, its the data to back up your statements.  You keep saying the D50 has only 16bits of dynamic range, but that's what I'd like to see data on. 

From guysonics testing below, it looks like the D50 with the mic power off, measured from 20Hz to 20kHz has a noise level at or below -114db.  This is about 19 bits of dynamic range, not 16 bits.  Do you have other measurements that demonstrate that the D50 only has 16 bits of dynamic range?  I don't recall the Sony literature addressing the noise level on the mic inputs, but am interested to know about other testing that's been done.

[Len Moskowitz (Core Sound)]

I don't think what is being requested is an explanation, its the data to back up your statements.  You keep saying the D50 has only 16bits of dynamic range, but that's what I'd like to see data on. 

Sony's specs are that it has a 93 dB S/N ratio at 1 KHz, when set to 24-bit.  That's A-weighted, so it's probably slightly worse unweighted.  (In the graph you posted, you can clearly see the degradation at low frequencies -- A-weighting removes that extra noise from the spec calculation.)

http://pro.sony.com/bbsc/ssr/product-PCMD50/

I could be wrong, but that seems to say that it has less than 16-bits of dynamic range.  What does that spec say to you?

From guysonics testing below, it looks like the D50 with the mic power off, measured from 20Hz to 20kHz has a noise level at or below -114db.  This is about 19 bits of dynamic range, not 16 bits.

That data is taken with an FFT size of 64k, and so the noise floor that you're seeing is in a very narrow band.  That doesn't translate into a wide band noise spec of -114.

Have a look at the Mic2496 V2's noise graph, measured with an FFT size of 32K, and you'll see a noise floor of around -138 -- it would be even lower with a 64K FFT.  The Grace Lunatec V3 has similar performance.

[Todd R]

The sony info you link to is from their web overview of the D50.  They have the same 93db figure in the D50 manual, but they are more clear that the 93db performance is for their line input, not their mic input.  I would imagine the mic input would be no better and perhaps worse, but still -- not any information on the mic input itself.

I'd rather see more actual data and testing provided, not just specs.  Sony might just be stating a conservative spec'd number so they could not be held liable if the D50 didn't perform up to spec.  I.e., they might expect much better performance, but they don't want to reflect that in their specs.

I'd still like to see more testing and data like Guy's graphs, but I appreciate your input on what similar testing shows for the Mic2496 and Lunatec V3.  Overall though, with little but Guy's testing to go on, I'm not ready to say for sure that the D50 has worse than 16 bit performance.

[Len Moskowitz (Core Sound)]

...Overall though, with little but Guy's testing to go on, I'm not ready to say for sure that the D50 has worse than 16 bit performance.

Off the top of my head and without going through the calculation, the -114 dB noise number taken with a 64K FFT, comes out to a broadband noise spec of something in the high 80's.

You can draw your own conclusions.

[guysonic]

Have a look at the Mic2496 V2's noise graph, measured with an FFT size of 32K, and you'll see a noise floor of around -138 -- it would be even lower with a 64K FFT.  The Grace Lunatec V3 has similar performance.

Unfortunately Core's graph is the one printed in the application/data sheet for the convertor device depicting optimized maximum performance breadboard and is the most ideal implementation (rarely if ever achieved in real world products), and is NOT the performance graph of Core's actual product. 

So we know nothing about his products implemented performance, only how good it might get if someone using the device covered all bases and then some to get near this documented ideal performance.   

It does look good for marketing purposes, but in reality is NOT his product's performance as is often stated here and elsewhere.

[Church-Audio]

Are you sure about this?  Leonard from Sonic Studios found something very different in his tests and stated that the D50 does give true 24 bit performance with a very low noise floor from the line inputs, the best from any portable unit in its class (the FR2LE, PMD671, HD-P2, and the 722 were better).  Do you have any graphs or measurements to support your claim?

I'll try to explain.

Using the -D50's Mic inputs will give you less than 16-bits of dynamic range.  So no one (other than Sony marketers) would consider it a 24-bit recorder.

Using the -D50's Line inputs with microphones, you'll get a very low level signal unless you use an external analog mic pre to feed the Line inputs or you're recording a very loud sound source.  If the analog mic pre is not of very high quality, it will degrade the Line input's noise performance.  So simply adding a low cost external analog mic pre to the Line inputs will likely not significantly improve the -D50's performance over the Mic input, unless it is quite a special low cost analog mic pre. 

Do you know of any for which you can post noise and dynamic range data?  I'm definitely interested in hearing about one.

My 9100 preamp has a wide band noise level of -109 db so I am pretty sure it will not degrade the specs of the D-50.. and my specs are based on measuring the "actual" product output not a factory data sheet  But please continue with your education on noise specs......

[scottwu]

There aren’t really any true 24bit IC A/D’s. I played with a rack based unit from Panasonic R&D once, a professor friend had an endowed chair at MIT from them and they gave him one. They considered it un-manufacturable. It actually measured -130dB on the CCIR 19k/20k IMD test, even this is not 24 bits.

FFT’s are nice but they are useless for noise specs without full disclosure of the details, which are way over the heads of most users. So they end up being pretty pictures with big numbers. Showing noise on a log (frequency) scale can be misleading too. The gentle rise below 100Hz is pretty benign. Take for instance a 30-40pF mic capsule and compute sometime how big the bias resistor would have to be to make the noise below 100Hz be around a few nano-Volts like a decent FET circuit or pre-amp. You would be surprised. And the bias noise rises at 20dB/decade while most preamps rise at 10dB/decade. This is in general, one can find pathological cases of any badness.

[Len Moskowitz (Core Sound)]

[Unfortunately Core's graph is the one printed in the application/data sheet for the convertor device depicting optimized maximum performance breadboard and is the most ideal implementation (rarely if ever achieved in real world products), and is NOT the performance graph of Core's actual product. 

The graph on our site is for a real Mic2496 V2. What you see there is the performance graph of the actual product, end to end, from the mic inputs to the digital output.

It hasn't been "optimized" (whatever you mean by that).

If you have an Audio Precision System II, you could duplicate that measurement yourself.

[Len Moskowitz (Core Sound)]

There aren’t really any true 24bit IC A/D’s. ..

Scott knows his stuff better than just about anyone.  He's got (as Richard Lee would put it) a brain the size of a planet.

I'd enjoy discussing noise specs and the need for better ways to spec electronics to reflect perceptual realities. 

Under certain conditions, we can in fact hear a signal recorded at -140 below another signal, both of them encoded with a currently available 24-bit A/D IC.  That evidence taken by itself would support that we really do have true 24-bit A/Ds, but that the specs we use don't reflect the real characteristics of auditory perception.

Perhaps we could start a thread on the Pro Audio mailing list or the Sursound list?

[scottwu]

From guysonics testing below, it looks like the D50 with the mic power off, measured from 20Hz to 20kHz has a noise level at or below -114db.  This is about 19 bits of dynamic range, not 16 bits.  Do you have other measurements that demonstrate that the D50 only has 16 bits of dynamic range?  I don't recall the Sony literature addressing the noise level on the mic inputs, but am interested to know about other testing that's been done.


[/quote]

Audition unfortunately lacks the option for normalizing the equivalent noise BW and giving you a dB re nV/rootHz Like a real spectrum analyser. I just started playing with "Visual Analyser" a freeware PC scope and confirmed that it will talk to an external A/D maybe it can be configured to do this.