• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Jitter stew

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
191
Location
Australia
Nice rhetoric but not convincing. I've never heard anything as irritating on CDs as a locked groove on an LP with music I really want to hear. A badly warped record that causes the stylus to lift off on every revolution is just as irritating.
You've just reminded of two CD "funnies" when I first started buying disks. One was a collection of "hits" by a female opera singer, and at the start of the first track, it was noisy - digitally noisy; I always noticed this peculiarity when this track played, and much more recently ripped the track and had a look at the waveform - yes, classic staircase waveform at the beginning; someone had finger problems when mastering, and got the gain levels wrong for the first 30 secs or so; never went back back and fixed it up. And the other was a jazz collection from the 50's - at the start there was an electronic "zzzzzzt", rang the label distributor to get feedback - I was an enthusiastic chappy back then! - and he sent out a replacement disk - same noise again! There was a mastering glitch, perhaps from the tape deck, embedded - not to worry!

I have yet to be irritated by non-optimised "digital sound". I guess that I haven't spent enough money on gear. :)
Main thing is that CD can be better than typically heard - add in all the good bits that people jump up and down about with regard to LPs, without losing any of the digital goodies ...
 

Frank Dernie

Master Contributor
Forum Donor
Joined
Mar 24, 2016
Messages
6,445
Likes
15,781
Location
Oxfordshire
You've just reminded of two CD "funnies" when I first started buying disks. One was a collection of "hits" by a female opera singer, and at the start of the first track, it was noisy - digitally noisy; I always noticed this peculiarity when this track played, and much more recently ripped the track and had a look at the waveform - yes, classic staircase waveform at the beginning; someone had finger problems when mastering, and got the gain levels wrong for the first 30 secs or so; never went back back and fixed it up. And the other was a jazz collection from the 50's - at the start there was an electronic "zzzzzzt", rang the label distributor to get feedback - I was an enthusiastic chappy back then! - and he sent out a replacement disk - same noise again! There was a mastering glitch, perhaps from the tape deck, embedded - not to worry!


Main thing is that CD can be better than typically heard - add in all the good bits that people jump up and down about with regard to LPs, without losing any of the digital goodies ...
How is the "classic stairform", which, I thought, is only revealed by not using a reconstruction filter caused by a period of "finger problems when mastering"?
Seeing a "stairform" in the output would be a fault in the replay DAC, I would have thought.
 

Old Listener

Senior Member
Joined
Apr 28, 2016
Messages
499
Likes
556
Location
SF Bay Area, California
How is the "classic stairform", which, I thought, is only revealed by not using a reconstruction filter caused by a period of "finger problems when mastering"?
Seeing a "stairform" in the output would be a fault in the replay DAC, I would have thought.

I agree, missing anti-aliasing filter came to my mind.

There is some irony here. One sect of audiophiles swears by NOS DACS without any anti-aliasing filter. More analog like, they think.
 

Frank Dernie

Master Contributor
Forum Donor
Joined
Mar 24, 2016
Messages
6,445
Likes
15,781
Location
Oxfordshire
I agree, missing anti-aliasing filter came to my mind.

There is some irony here. One sect of audiophiles swears by NOS DACS without any anti-aliasing filter. More analog like, they think.
Yes afaik the only way to get a stairform on the output analog waveform is to not use the reconstruction filter which removes it, thereby not conforming to red book (iirc).
True there are makers who deliberately omit the filter and people who like the sound it makes, perhaps because it it is not too disimilar from the high level of HF distortion they are used to from turntables??
 

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
191
Location
Australia
How is the "classic stairform", which, I thought, is only revealed by not using a reconstruction filter caused by a period of "finger problems when mastering"?
Seeing a "stairform" in the output would be a fault in the replay DAC, I would have thought.
No, this was literally flawed mastering - imagine reducing the first 30 secs of the track to 8 bit resolution, and then converting that back to 16 bits. All the information in the lower 8 bits has been lost, and at decent volumes this is very obvious, no dithering was applied so it doesn't sound 'analogue' noisy, but 'digital' in nature.

The staircase in the track waveform was because of the loss of those 8 bits of resolution - note, I'm pulling the number 8 out of the air, what it actually represented I don't know - suffice that it was well short of 16 bits.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Chances are there is filtering going on, just not explicitly... Can happen in the output buffer, cables, tweeters, ears... One of the "non-filtered" outputs I saw years back had a transformer-coupled output. Hmph. Look up how inductors pass high frequencies... It is also possible the images create distortion elsewhere in the system (including the tweeters) that is modulated down in frequency and "fills in" the audible band.

FWIWFM, I use the term aliasing to refer to signals that are aliased to baseband around the sampling frequency and its multiples. An anti-alias filter at the input of an ADC is used to suppress signals that would otherwise be aliased. The output of a DAC produces images at multiples of the sampling frequency so the term I use for the filter after the DAC's output is anti-imaging filter, or just image filter.

My 0.000001 cent (microcent) - Don
 

Frank Dernie

Master Contributor
Forum Donor
Joined
Mar 24, 2016
Messages
6,445
Likes
15,781
Location
Oxfordshire
No, this was literally flawed mastering - imagine reducing the first 30 secs of the track to 8 bit resolution, and then converting that back to 16 bits. All the information in the lower 8 bits has been lost, and at decent volumes this is very obvious, no dithering was applied so it doesn't sound 'analogue' noisy, but 'digital' in nature.

The staircase in the track waveform was because of the loss of those 8 bits of resolution - note, I'm pulling the number 8 out of the air, what it actually represented I don't know - suffice that it was well short of 16 bits.
I think you are wrong in your speculation.
If you remove the lower 8 bits of data, and I have done this, what you get is more noise. Most music has less than 8 bits of dynamic range so the most marked, ie noticeable, effect is a raised noise floor. In fact, a 24/96 file converted to 8/96 with noise shaping are almost indistinguishable from each other by ear.

I can speculate that considerably reducing the sampling rate may produce a visibly stepwise analogue output waveform from a standard CD with normal reconstruction filter. Reducing the bit rate would not.
 

Frank Dernie

Master Contributor
Forum Donor
Joined
Mar 24, 2016
Messages
6,445
Likes
15,781
Location
Oxfordshire
Chances are there is filtering going on, just not explicitly... Can happen in the output buffer, cables, tweeters, ears... One of the "non-filtered" outputs I saw years back had a transformer-coupled output. Hmph. Look up how inductors pass high frequencies... It is also possible the images create distortion elsewhere in the system (including the tweeters) that is modulated down in frequency and "fills in" the audible band.

FWIWFM, I use the term aliasing to refer to signals that are aliased to baseband around the sampling frequency and its multiples. An anti-alias filter at the input of an ADC is used to suppress signals that would otherwise be aliased. The output of a DAC produces images at multiples of the sampling frequency so the term I use for the filter after the DAC's output is anti-imaging filter, or just image filter.

My 0.000001 cent (microcent) - Don
Certainly agree that anything with transformers in will be filtering the high frequencies, still don't see that £200,000 is justified for a NOS DAC with no output filter, even if the coupling transformers are silver:) and isn't the first significan peak due to no filter is at 66kHz. the 3rd harmonic of the sampling frequency?
I can see the logic in calling the DAC output filter an image filter though here in the EU reconstruction filter is the commonly used term for audio IME (I was taken to task about the term by a Dutch engineer).

I was the first person to use a digital recorder to measure data on a Formula 1 car, in 1982 iirc. Tape recorders were not much good because of the environment and I was keen to know what was going on. I was not trying to record things at high frequencies and had never read anything about digital recording of AC, my ignorance about aliasing was embarassingly total in retrospect, I was just wanting to measure suspension travel, throttle opening and output shaft speed. All this was very low frequency but I never converted back to an ac signal, just looked at sample values.
In fact the first commercially available data recorder for racing cars, which made the man who did it very rich eventually, had no provision to deal with aliasing for several generations.
My custom first in-house designed recorder, started in 1986, was actually developed as a controller for the Williams Active Suspension which I had designed, the recording aspect was just a secondary feature. By then removing aliasing effects had become essential, of course.
All this had to be do-it-yourself since there was zero suitable commercially available kit. And the fastest data transfer we had was wired RS232 when the car was in the pits and a 256kB memory chip was about £1000!
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Haven't really followed the thread, sorry, and did not know nor made any comment upon the price. Clearly I am in the wrong business.

The first image is actually "wrapped around" one-half the sampling frequency (the Nyquist frequency). That is, when sampled at 44.1 kS/s, a 20 kHz output signal produces a DAC image at 24.1 kHz. Images will occur at multiples of the Nyquist frequency for as long as the bandwidth and edge rates allow, subject to sinc(x) filtering and the output pulse shape and so forth.

Reconstruction filter is another term for image filter. Image filters are also used to refer to RF filters used to suppress image spurs from mixers, and there is a branch called image-parameter filter design, so reconstruction filter might be a better choice. Depends on where you were weaned, I supposed. I do not recall hearing the ADC's input filter being called anything other than an anti-alias filter but there are probably other terms.

Must have been very exciting working on those race cars (worthy of another thread IMO). A senior scientist at the place I began my career many years ago developed the electronics for the Nissan racing cars. Impressive stuff.

FWIWFM - Don
 
Top Bottom