John Curl's Blowtorch preamplifier part II

[Wavebourn]

We loose quality in a/d conversion, then in storage-transmission, then in d/a conversion. It is obvious that for better quality delta-sigma modulation simplifies requirements to d/a, then between a/d and d/a we can use whatever conversion we need for better storage and transmission, with redundant coding and appropriate buffering. In reproduction latency does not matter... Any world length can be used, no need to set dependence between word length and analog quality. But for better quality of conversion I don't see anything better than plain old sigma-delta modulation that can be recreated digitally back before d/a conversion.

[jcx]

um - no one is proposing Recording/Mastering in 16 bit - 24 bit ADC, using the ~20 bits of dynamic range practically available in better converters today is good Recording practice where noise floor, dynamic headroom, clipping issues once encoded in digitial are errors forever

DAW work in high intermediate bit depth to accomodate many layers of processing without rounding/truncation error creeping up to audible levels, likewise to provide adequate headroom to avoid clipping by filters, EQ boosting the source

but once the uncertainties of capture, all of the processing is done the level can be shifted up to the 0 dB line as a part of word length/bit rate reduction

RedBook CD 16/44 is to our best knowledge a adequate post production recorded music Delivery format - proper noise shaped dither can encode the final Musical Recording Mastered Product - with full linearity, and greater than 96 dB S/N in our most sensitive hearing frequency range

[bcarso]

Originally Posted by SY

OK, my mike (nothing exotic) has a sensitivity of 2.6mV/Pa. Got a mike preamp that will get me better than -90dB noise wrt 1 Pa?

What's the mic's self-noise, assuming a still atmosphere at about 300 Kelvin?

[SY]

About 1.6uV, 20-20k, unweighted. I think you have the point.

[bcarso]

Originally Posted by jcx

um - no one is proposing Recording/Mastering in 16 bit - 24 bit ADC, using the ~20 bits of dynamic range practically available in better converters today is good Recording practice where noise floor, dynamic headroom, clipping issues once encoded in digitial are errors forever

DAW work in high intermediate bit depth to accomodate many layers of processing without rounding/truncation error creeping up to audible levels, likewise to provide adequate headroom to avoid clipping by filters, EQ boosting the source

but once the uncertainties of capture, all of the processing is done the level can be shifted up to the 0 dB line as a part of word length/bit rate reduction

RedBook CD 16/44 is to our best knowledge a adequate post production recorded music Delivery format - proper noise shaped dither can encode the final Musical Recording Mastered Product - with full linearity, and greater than 96 dB S/N in our most sensitive hearing frequency range

A good overview.

I once (many years ago) made a point-of-sale motion-detector-annunciator which used uncompressed 8 bit audio. The initial reaction from most upon hearing it was very favorable. When I played it for Steve Dove he was delighted, and said that the hiss was doubtless from an analog cassette source. In fact it was a very carefully tuned random noise source on the record end, without which the lower level material got very strange-sounding indeed. Overall, the reproduction, which was also bandlimited to about 8kHz, sounded like a pretty decent AM radio.

The sound quality ran circles around a competing approach of the day, an "analog EEPROM" from Information Storage Devices.

[RNMarsh]

slowly slinking off without the rest seeing me ----

Certain personalities remind me of: You might be an engineer IF:

1. You can translate English into Binary.
2. You have no life --- and you can Prove it mathematically.
3. It's sunny and 75 F outdoors and you are working on a computer.
4. You know vector calculus but you cant do long hand division.
5. Know the second law of thermal dynamics but not your shirt size.

--- are they still fighting?

[bcarso]

Originally Posted by SY

About 1.6uV, 20-20k, unweighted. I think you have the point.

Yes, as I remarked to Scott elsewhere, I was rather disappointed when, telling John Eargle of my ideas for lower-noise condenser mic preamps, he pointed out that the limitations of thermomechanical noise and air molecule bombardment were already setting lower limits to mic noise, even for the quietest of recording spaces.

Reminds me of when I house-sat for a composer who had four keyboard instruments: An Ahlborn-Galanti synthesized "pipe" organ (with a complete console with two keyboards and pedals), a 9 foot Bechstein, a single-manual harpsichord (Italian style, very nice instrument), and a clavichord. The neighborhood was in the San Fernando Valley and relatively quiet after, say, 3AM, for a while. But it was still often too noisy to hear the clavichord very well, while sitting at it and playing. One really has to get out of the city altogether, and hope when you do your ears are not ringing too much

[simon7000]

Originally Posted by SY

OK, my mike (nothing exotic) has a sensitivity of 2.6mV/Pa. Got a mike preamp that will get me better than -90dB noise wrt 1 Pa?

SY,

You know it is an impedance question. A 5nV/hz preamp would get you a noise level around 23db. Drop the impedance and you can go quieter. Of course you can also move the mic closer to the source. But that would bring up the tangent of recording techniques.

A symphony at 105 would need 125 for good headroom. So with a 23 db noise floor you just miss 16 bits. At only 10db headroom you would make it, having heard what 10 sounds like, it ain't bad, but you can hear it.

[bcarso]

Originally Posted by RNMarsh

slowly slinking off without the rest seeing me ----

Certain personalities remind me of: You might be an engineer IF:

1. You can translate English into Binary.
2. You have no life --- and you can Prove it mathematically.
3. It's sunny and 75 F outdoors and you are working on a computer.
4. You know vector calculus but you cant do long hand division.
5. Know the second law of thermal dynamics but not your shirt size.

--- are they still fighting?

My favorite along those lines: How can you tell if an engineer is an extrovert?

Answer: He is staring at your shoes.

[jcx]

some of us “egg heads” occasionally put our hands on hardware - or our lives in the hands of the hardware we have placed

was climbing outdoors on real rock yesterday, may again tomorrow

I spend 3-4 nights a week in the climbing gym – my impression is that at least locally near half of the climbers are engineers, grad students, post docs, researchers, professors... technical/scientific professionals

dining after climbing in the gym this last thursday the table had 2 associate professors, 2 software engineers, one who also runs a martial arts studio, a biology post doc, a RN and myself


wrung about as much info as I wanted from my IXYS kV depletion mode Mosfet casoded opto isolator electrostatic amp's point-to-point proto last week - sims and 'scope agreed remarkably well on compensation overshoot, ringing damping with the 20 yr old DIP op amp from my parts drawer in the front end

next step for the amp will be PCB to use faster, better op amps available in smt only - sim looks better with 90 MHz GBW op amp - need to remember how to use Eagle again

30 years of EE employment has given me the impression that good theory is good for practice

and stereotypes, caricatures aren't often that useful dealing with real people, technical arguments, or hardware