Oscilloscopes and audio Analog to Digital Converters

[Fantasma25]

I don't understand this. Most audio ADCs have a resolution of 16 to 24 bits (and some of them even 32bits), but when it comes to the scopes that are supposed to be super precise and reliable measurment tools, they only have like 8 or 12 bit ADC (with a much much higher sample rate, of course). If we only need 12 bits of resolution to represent a signal with fidelity, then why audio products have such a high resolution? 

[madshaman]

Most high-speed ADCs are lower resolution, this is sometimes a direct result from how the ADC is implemented (iterative process).

There are ways to get more than 8-bits of effective resolution from an 8-bit ADC, however these also involve lowering your effective time resolution.

Audio DACs/ADCs are really in a category of their own because solve small signal problems in the real world at audio frequencies is quite different than say RF speeds.

Take a look at the specs for instrumention, audio and high speed ADCs on digikey, they all focus their strengths somewhere different.

This is all just my humble opinion.

[awallin]

There are a lot of charts online with speed vs. resolution for ADCs, for example:
http://www.analog.com/library/analogdialogue/archives/39-06/architecture.html

If your scope screen doesn't have more than 256 pixels vertically it makes little sense to digitize at more than 8 bits
If you need 16-bit resolution there are e.g. DAQ cards (PCI or PCIe) that allow you to collect at maybe 500 kS/s. Displaying all that data in real-time might be difficult - and it also takes a lot of disk space.

[madires]

You need a high resolution for audio signals because they got a high dynamic range and the human ear is quite sensitive to that. For speach 8 bits are sufficient (telephone) but not for music or other complex signals.

[SeanB]

Generally the audio ADC's are a lot looser on things like monotonicity, absolute error and drift. Many are very plainly marked in the data sheets that they have a low frequency cut off of around 5Hz or so, and this is because they drift so much around DC. On a scope used to measure DC you will not be happy with a trace that varies slowly by around 2 divisions of zero, is non linear and has portions where the digital output goes down for an increasing input voltage, where 1V of change is shown as a variable digital difference depending on the DC offset and where it slowly drifts as well.

The only areas they are good at is that they generally have noise below 16 bits ( even in the 24 bit devices) and have a set of digital and analogue filters built in to reduce artefacts from frequencies fed to them outside the audio range. That and they are mostly 2 channel devices that make a specific audio based data stream with only needing a single fixed frequency clock.

[Christe4nM]

The main difference comes from the frequencies that the ADC/DAC must be able to convert. There are different internal architectures involved in data converters that each have their own strengths and shortcomings. Generally the higher the conversion speed, the harder it becomes to get many bits of resolution.
At relatively low sampling frequencies for audio signals the type of ADC today is probably always of the Sigma-Delta kind. Those are able to be designed with quite high resolutions at good accuracy and speed.
An oscilloscope on the other hand will use a (pipelined) flash architecture. The flash architecture is the one for very high speeds. It is however difficult (and very expensive) to make them at higher resolution than say 8 bits.

BTW, a good read on ADC architectures is this one: Which ADC Architecture Is Right for Your Application?

[Simon]

from memory my scope has less pixels on the vertical axis than 255 so why have a better resolution sampling system ? it can be handy for zooming into a captured wave form but meh, in my rigol to get 1GS/s they had to use a bank on 10 ADC's firing off 1/10 of a sample take after each other to get the full time res so no hope of getting better vertical res.

[BravoV]

Same question when I was a noob back then, and remembered really well what my mentor taught me with this examples and I found it really makes sense.

The analogies used :
- Resolution -> Power
- Frequency -> Speed/Agility

A big truck -> Big power (High Resolution) but low speed (Low Frequency) , and its capable of carrying heavy load.
A sport car -> Low power relatively to big truck (Low resolution) but high speed and acceleration (High Frequency), also can make tight turn at high speed because its weight less compared to a fully loaded truck.

Now, normally you just can not ask the best of both worlds, like you want a car as fast as the sport car "AND" capable to carry tons of stuffs like a truck, and also it must able to make a really sharp turn while fully loaded at high speed. 

The audio ADC is like the heavy duty truck, and its enough since our ears have a limit, while scope's ADC is like the exotic sport car. 

Ok, what if you keep insisting a truck capable of carrying tons of loads while have the acceleration and maneuverability of a sport car ?   

Well, maybe you need a special designed truck with huge rockets installed, and capable of firing those crazy powerful rockets like the one that used to lift Apollo 11 to the moon, and also have them installed at all sides of the truck so it can stop or accelerate like a sport car, and also the truck can float to reduce drag while those installed gigantic rockets firing at the right angles and sequences to make those cool tight turn. 

Now, do you have your money ready to buy it ? 

Hope this helps.

[fcb]

If we only need 12 bits of resolution to represent a signal with fidelity, then why audio products have such a high resolution? 

Others have addressed the reasons 'scopes tend to be 8 bits.  I'll address your question of fidelity - 12 bit is quite tiring to listen to.

[Kompost]

I disagree, 12 bit audio is actually hard to discern from 16 bit on some genres (especially heavily compressed music).

Discerning 16 from 24 bit recording is harder still and generally requires very good listening conditions.

[Fantasma25]

Well I found this.
This is probably why audio applications need higher resolution.

You need a high resolution for audio signals because they got a high dynamic range and the human ear is quite sensitive to that. For speach 8 bits are sufficient (telephone) but not for music or other complex signals.

I think you're right. Actually, if you try to look an audio signal in an oscilloscope, it will look like noise. It is neither a periodic signal, nor a signal with discrete or predefined values, and that is the opposite of the type signals that are normally fed to a scope. That's probably why you can have a scope with 8 bit ADC and have no issues.

[tszaboo]

Audio ADCs are usually limited to 24 bit, 192KHz. That is 4,6 MBAUD data. While scopes easily can measure 1GHz, 8 bit, which is 8000 MBAUD. So you get much more information about your signal with a scope.
However, they are not interchangeable. And if you measure an audio signal with a scope for THD, SNR or stuff, that is usually a big mistake, as it is not meant to be used like that.

[madshaman]

I disagree, 12 bit audio is actually hard to discern from 16 bit on some genres (especially heavily compressed music).

Discerning 16 from 24 bit recording is harder still and generally requires very good listening conditions.

The extra bits give more dynamic range, so, for example, a real quiet bit will still have many bits, as will a quiet bit mixed with a loud bit.

This isn't as useful for music listening because more music is highly highly compressed (amplitude compression, not data) which normalises the dynamic range, but for movies played over a home theatre in a box-like setup, there is much more dynamic range and extra resolution is needed.

[madires]

The extra bits give more dynamic range, so, for example, a real quiet bit will still have many bits, as will a quiet bit mixed with a loud bit.

This isn't as useful for music listening because more music is highly highly compressed (amplitude compression, not data) which normalises the dynamic range, but for movies played over a home theatre in a box-like setup, there is much more dynamic range and extra resolution is needed.

The killer application for high resolution audio is classic music. And a lot of studios aren't even capable to use their audio compressors properly and we get audio CDs with compressed rock/pop music and clipping. Yep, Sony and all the other big players are selling bad CDs. It's called the Loudness War. 

[madshaman]

The extra bits give more dynamic range, so, for example, a real quiet bit will still have many bits, as will a quiet bit mixed with a loud bit.

This isn't as useful for music listening because more music is highly highly compressed (amplitude compression, not data) which normalises the dynamic range, but for movies played over a home theatre in a box-like setup, there is much more dynamic range and extra resolution is needed.

The killer application for high resolution audio is classic music. And a lot of studios aren't even capable to use their audio compressors properly and we get audio CDs with compressed rock/pop music and clipping. Yep, Sony and all the other big players are selling bad CDs. It's called the Loudness War. 

I wouldn't call them "bad CDs", compression usually improves the listening experience of most music, some music, like most symphonies have a large dynamic range which can be a problem, but I would say most audio engineers can handle that fairly well.

Having most sections of a CD at close to the same "loudness" also saves you from constantly adjusting the volume on your stereo system.

[madires]

I wouldn't call them "bad CDs", compression usually improves the listening experience of most music, some music, like most symphonies have a large dynamic range which can be a problem, but I would say most audio engineers can handle that fairly well.

I don't got a problem with compression of music but I find it a shame that even with compression I regularly see clipping. Audio compression done correctly doesn't cause clipping at 2^16 - 1. If the music exceeds the CD's 16 bits the CD is bad and someone has done a lousy job.

Having most sections of a CD at close to the same "loudness" also saves you from constantly adjusting the volume on your stereo system.

That's called normalization.

[madshaman]

I wouldn't call them "bad CDs", compression usually improves the listening experience of most music, some music, like most symphonies have a large dynamic range which can be a problem, but I would say most audio engineers can handle that fairly well.

I don't got a problem with compression of music but I find it a shame that even with compression I regularly see clipping. Audio compression done correctly doesn't cause clipping at 2^16 - 1. If the music exceeds the CD's 16 bits the CD is bad and someone has done a lousy job.

Having most sections of a CD at close to the same "loudness" also saves you from constantly adjusting the volume on your stereo system.

That's called normalization.

I'm actually not talking about normalisation, I'm talking about very dynamic sections of music where you have alternating loud and soft parts.  That's compression.