What are the usual suspects with AD converter noise? How to improve converters?

[canidoit]

Hi All,

When it comes to Analogue to Digital Converters, what are the usual causes when the noise floor(pink or white noise) increases/exists on the unit?

Does the type of noise indicate a particular problem?

When DIY mod their converters for improvement, what do they usually do - ie. change to metal film resistors with low tolerance, higher capacitors, in what area of the circuit?

Thank you.

 

[user 130681]

Noise is analogue, digital is either on or off.
Distortion however, is a different matter.

 

[Newmarket]

Hi All,

When it comes to Analogue to Digital Converters, what are the usual causes when the noise floor(pink or white noise) increases/exists on the unit?

Does the type of noise indicate a particular problem?

When DIY mod their converters for improvement, what do they usually do - ie. change to metal film resistors with low tolerance, higher capacitors, in what area of the circuit?

Thank you.

Do you have a case where your converter noise has increased ?

 

[abbey road d enfer]

When it comes to Analogue to Digital Converters, what are the usual causes when the noise floor(pink or white noise) increases/exists on the unit?

Noise almost invariably comes from the analogue circuitry. It covers not only the I/V converter, but the power rails and the clock. Of course layout is paramount.

When DIY mod their converters for improvement, what do they usually do - ie. change to metal film resistors with low tolerance, higher capacitors, in what area of the circuit?

They may do that but the improvements often are the results of wishful thinking. The only significant point of improvement, for a given D/A chip is in the opamp used in the I/V converter.
The current high-performance opamps are quite good, but there is always room for improvement, particularly the slew-rate/transient response.

 

[Newmarket]

Noise almost invariably comes from the analogue circuitry. It covers not only the I/V converter, but the power rails and the clock. Of course layout is paramount.

They may do that but the improvements often are the results of wishful thinking. The only significant point of improvement, for a given D/A chip is in the opamp used in the I/V converter.
The current high-performance opamps are quite good, but there is always room for improvement, particularly the slew-rate/transient response.

Yes. All that. But the OP is asking wrt ADC rather than DAC.

 

[living sounds]

Older converters can get out of spec and show increased noise. This is a maintenance or even repair issue, but not a general "DIY improvement" or "mod" kind of thing. It depends on the converter itself. I've got older Sony units that use a certain converter chip that produces spurious tones that were not present originally, so I guess these have to do with deterioration of the chip itself (I've done recaps and adjustments with no luck). I've also got fully discrete DCS ADs that show increased noise, but without schematics there isn't much chance of fixing things.

Generally, I would look at the power supply lines and measure for ripple or noise there. A recap can usually fix that, regulators get hot and performance can worsen, too.

 

[ruffrecords]

I suspect ADC limitations are due to the supply voltage used. A 24 bit ADC has a theoretical dynamic range of 144dB. A typical 5V rail limits the audio to +2dBu at most so the convertor resolution is equivalent to an inpt level of -142dBu and there is no op amp in the world that comes anywhere near that. Now if you could arrange the convertor to accept a +20dBu input then its resolution is equivalent to -124dBu and we can make input circuits quieter than this when fed from a 150 ohm source.

Cheers

Ian

 

[living sounds]

I suspect ADC limitations are due to the supply voltage used. A 24 bit ADC has a theoretical dynamic range of 144dB. A typical 5V rail limits the audio to +2dBu at most so the convertor resolution is equivalent to an inpt level of -142dBu and there is no op amp in the world that comes anywhere near that. Now if you could arrange the convertor to accept a +20dBu input then its resolution is equivalent to -124dBu and we can make input circuits quieter than this when fed from a 150 ohm source.

Cheers

Ian

The ADC itself cannot get anywhere near theoretical 24 bit performance. Modern converters perform all kinds of tricks to increase SNR (not always to the benefit of SQ, IMO, but that's another issue). There are also multi-stage designs, which have their own problems.

Elevated noise over spec is indicarive of a problem, but noise at the level of currently availible converters is far exceeding "good enough" IMO. We usually have a lot more noise in the analog domain, and great records have been made despite the limitations of tape machines.

I'd rather use a noiser converter that sounds good than an anemic ultra-low-noise one. There are plenty of the latter IMO.

 

[abbey road d enfer]

Yes. All that. But the OP is asking wrt ADC rather than DAC.

Yes, sorry. But most of my comments are valid there, except the opamps are in the input stage instead of the I/V converter.

 

[Newmarket]

Yes, sorry. But most of my comments are valid there, except the opamps are in the input stage instead of the I/V converter.

Understood. That's what I meant by "All that". There are indeed some points to be considered wrt driving the inputs of ADC devices. And I'm unclear as to whether the OP is experiencing a specific problem. And whether by noise they mean "hiss/pink/white" noise or clock related artefacts or other. Question references noise floor as I see it but there are other issues that might be considered such as IMD.
Agree that discrete component substitution unlikely to give real benefit. Unless something is "broken" or there is some fundamental design weakness.

 

[daveb]

Another possible culprit for adc noise is nyquist zone noise folding. I’m not familiar with audio specific adc design, but noise at higher frequencies (hundreds of kHz) can be downconverted in the analog front end of an ADC and show up in the audio band.

 

[gswan]

For ADC a noisy or bad voltage reference will cause distortion of the conversion, as will a poor clock.

 

[abbey road d enfer]

Another possible culprit for adc noise is nyquist zone noise folding. I’m not familiar with audio specific adc design, but noise at higher frequencies (hundreds of kHz) can be downconverted in the analog front end of an ADC and show up in the audio band.

That is correct; AFAIK it's not intrinsic to the converter chip, but due to the analog signal it is being fed with. Probable causes would be improper front-end design ( amongst others noisy opamps, wrong Low Pass filter, out-of-band noise in input signal, wrong source impedance, slow opamp that does not provide low enough impedance to the chip's inputs).
It's all basic analogue stuff, but must be taken in the context of converter's specific constraints; it's generally well detailed in manufacturer's datasheets.

 

[canidoit]

Hmm, well I was recently comparing AD converters and realised how noisy this well regarded ADC was. When I first purchased the ADC over 10 years ago, I thought the noise was normal like analogue gear, and did not question it because I did not know better. The pink noise is fairly loud when comparing to other ADC that are silent. I can't hear the noise though once a signal passes through.

It is my main stereo ADC which I use to track everything with. I don't know whether this noise is a good thing, I mean I do hear that my mixes tend to sound harsher but I thought that is just digital, or my poor mixing skills?

 

[ach]

The best ADCs are differential and have a high analog voltage. PCB layout is critical to separate the analog and digital return paths. Dynamic range is one thing, but digital hiss is completely unacceptable!

 

[ccaudle]

realised how noisy this well regarded ADC was.

What analog level input does it use? Does it have adjustable input gain?
As with any equipment proper gain staging should be used to optimize noise, but it is also possible that your "well regarded" ADC is just not designed very well.
Do you have a way to measure the noise level with no input signal, and the inputs terminated? The usual way to terminate inputs would be to solder a 100 Ohm resistor across pins 2 and 3 of an XLR connector, and insert that connector into the input connector of the ADC. Then record the output digital signal and check the noise level with a program that can measure the RMS value of the noise floor. I can recommend a couple of linux programs, but I don't know what would be best for Windows or MAC. Maybe Room EQ Wizard (REW)?
Also verify that the spectrum of the noise floor is flat, or only has a slight rise at very low frequencies. Rising high frequency noise or a peak in mid or high frequencies is more annoying that flat noise spectrum (but don't worry about a rise below a couple hundred Hz, all electronic devices tend to have a rising noise spectrum at very low frequencies).

 

[Newmarket]

The best ADCs are differential and have a high analog voltage. PCB layout is critical to separate the analog and digital return paths. Dynamic range is one thing, but digital hiss is completely unacceptable!

"Digital hiss" ???

 

[ach]

"Digital hiss" ???

When noise from the digital side makes it into the analog signal.

 

[Newmarket]

When noise from the digital side makes it into the analog signal.

Okay. Asked because ime "Hiss" is usually used to refer to broadband white or pink type noise. Usually from analogue / linear electronics. Whereas 'Digital' noise is generally related to clock and data signals and results in clicks and glitches etc

 

[michinger]

I´m also curious about A/D noise. I like to capture in 192khz and at that samplefreq it makes that increase of noise from 50khz above. It does not distrub me and it is not in lower samplereq, but I would like to know where it comes from. Maybe there is the lowcut filter not active anymore because it is not relevant at that high samplefreq?

It´s the same if the converter is Master or WC slave. It´s directly connected from the converter to RME AES-soundcard.