Test Procedure for comparing 24 bit vs 16 bit audio

[Marc Hildebrant]

Dan,

The concept of "ENOB" revolves around the idea that an "analog" voltage is converted to a "perfect" digital result. Thus, the noise present will degrade the conversion to digital but once in the digital world, the signal isn't changed by most noise present. Each time that you move through the conversion, though, the effective number of bits can be impacted.

There are noise reduction methods that can help greatly with signal recovery in noise environment cases, but for audio and 16 bits, the actual number is much less than 16.

By the way, the D to A process uses precision resistors that convert the digital levels to analog levels. If you calculate the temperature change to these resistors value verses a couple of degrees, you will quickly find that your 24 bits isn't even 16 bits with normal temperature variation in a room.

In the past, I worked with precision resistors that were placed in a temperature controlled oil bath for accurate measurements. The resistors were made by Julie Research in New York City. The owner, Julie ???? , invented the original Operational Amplifier (OP).

Marc

[Craig Maier]

There are techniques to reduce temperature change effects on the gain accuracy of an A-D or D-A converter. It involves two things. One is to mount the divider components (having the same tempco in ppm/degree C) on the same thermal substrate and to regulate that substrate temp. The other is to measure the substrate temperature and to add a feedback loop to the converter to compensate for gain changes (or offset changes) due to delta T.

Craig

ps - as an aside, I own an early vacuum tube, chopper stabilized op amplifier that was part of the dosimeter system on a linear accelerator. It converted the dc signal to ac via a chopper, amplified the ac and then reconverted that ac back into dc again via synchronous amplification. I think it was a Philbrick USA-1, but not sure. It's in my collection of goodies somewhere in the basement.