39 replies to this topic

[Pulse-R]

Be it in your home or in your car, your system has noise.
Whether or not you can hear it, and whether or not you like it, there will be some noise in your system.

There is a bit of discussion in the SQ forums at the moment about what is system noise, and how should it be measured or tested.
I'd like to present my thoughts on the source of system noise; how to identify it, and how to test it.

System noise is that feint "shhhh" sound that you can hear in the background while music is playing, and sometimes when there's no music.
System noise is not recorded as part of the music, it is generated in the playback system, and is presented to you along with the music.

What causes system noise? Almost every part of your system.
Active components such as power supplies, amplifiers, oscillators, processors, DACs.. Any 'active' component in the system will contribute to the level of noise in your system.

The history of "removing" system noise from car audio started many years ago, with some people going to great lengths to make the noise in their system inaudible.

In modern Car Audio products, most are fitted with a 'noise gate' which is like a switch - when the level of the sound on the CD drops below a certain threshold, the gate closes; stopping any sound from getting out of your CD player.

The most common form of noise gate in car audio is called "Zero Bit Mute" and is a hardware function of the DAC circuit. It tells the output stage "There is no audio data output". The Zero Bit Mute signal is fed from the DAC to the output section, to some switching circuit, and effectively shorts-circuits the outputs, preventing any signal from getting to the amplifier.

This Zero Bit Mute behaviour results in most, if not all, system noise being prevented from reaching the speakers. Good so far, except in the case of a recent discussion which resulted in the outcome of some tests being highly skewed by the use of the wrong test track to determine system noise.

There are two test tracks on the Sheffield Lab "My Disc/A2TB Test Disc" (also known as Autosound 2000 CD103) which can be used for measuring System Noise.

Track number 41 on "My Disc" is a Zero Bit track - it is designed to activate the Zero Bit Mute circuit in the CD Player (or DAC), and should result in no 'audible' sound being output from the system. There are many test discs which have a Zero Bit track, and this is the common method across the entire industry for determining system noise. The liner notes for track 41 state " This track can be used for evaluating the noise floor of components or a system".

Track number 42 on the same disc is a "High Bit" track. The High Bit track, because its sample values are not all zero, will open the mute gate and allow us to hear the 'true' system noise, although it will introduce a small amount of noise itself. This means that track 42, although theoretically silent, actually will cause a small amount of noise in a high resolution system which may otherwise have none. This really exposes the internals of the CD player to scrutiny for which it was not designed, and any noise source - as described above - will be bared for all to see (or hear, actually)! The liner notes for track 42 say this track is "Useful only with the proper test equipment".

What follows is a technical explanation of the differences between the two tracks.

What is the difference between track 41 and 42?
Music on a CD is recorded in LPCM format, with 2 channel 16bit resolution and 44.1kHz sample rate.
LPCM is a 2's complement signed integer format, which means if the first bit (MSB) is a 0, then the sample value is positive. It the MSB is a 1, then the sample is a negative number. We know audio signals consist of positive and negative values, to represent compression and rarefaction of air pressure - sound.

0000 0000 0000 0000 (16 bits all '0') represents digtial 0. The Zero Bit track consists of the entire track, with every sample value set to this 0 value (0000 hex)
0111 1111 1111 1111 is maximum output value, the leading zero tells the DAC that this is a positive number. (7FFF hex)
1000000000000000 is minumum output, the 1 at the beginning is for negative number, and 2's complement means we invert all the other bits, then add 1 to get the sample magnitude. (-8000 hex)

the High Bit track (track 42) contains alternating samples at 1111 1111 1111 1111 (FFFF hex) and 0000 0000 0000 0000 (0000 hex). This corresponds to sample values of -1 and 0. The amplitude step from 0 to -1 is very small (0.03dB), and the sample rate makes the audio frequency 22.05kHz. In this case there should still be no audio; just a very small (-96dB) signal which is well below a normal reproduced output range. And yet, not zero.

[Matt VIP]

in that case, why do some systems display audible noise when track 42 is played, and others do not? How does one go about reducing that noise floor so that it doesnt show up in competitions? is it just a matter of turning the gains down?

I'm not trying to wind you up, i'm interested in this as well as I'm fairly sure my system has noise using that track.

cheers!

[Pulse-R]

reducing "Track 42 noise" would consist of any/all of the following:

1. turn down the amp gains
2. modify the components and/or design of your amps, source(Cd Player), processor (if any), etc. to reduce their inherent noise.

[Marc ♫]

Track 42 is looking good for MEASQ '09.

[Luke352]

If I understand it track 41 is the most useful one in our situation due to it muting the HU's output, you will still hear if there is noise induced along your RCA's or noise induced in your amps, or just noisy amps, whereas 42 will show the internal noise of your HU which we shouldn't be judgeing, you could use it as a back up to find whines (alternator noise) but if all you get is a faint hiss with no change in pitch or volumne dependent on throttle, lights on/off etc... then all your hearing is internal noise of the HU which the competitor can't change unless he goes modifying DAC's etc... so we should not be marking them down because of something they can't do anything about mostly, unless they have there amp gains set way too high.

Luke

[Pulse-R]

It is possible that track 41 (and other Zero Bit tracks) will continue to be used for Intermediate (and novice) class.

Expert class may be using track 42 for the next season - depending on the outcome of current discussions.

Edit -
After looking over the scoring of some 'other' SQ formats, noise is measured using the equivalent of track 41 and 42.

Edited by Pulse-R, 25 November 2008 - 08:22 PM.

[jas]

simon what happens when you play track 42 (the non muted track) with all your gains set to 0 aka minimum???

[Pulse-R]

like anything - and source noise will be lower when you set the amp (or source) gain lower, same as turning the volume down.

[jas]

ok you didnt get what im on about.....

i was asking if a competitor had problems with noise on track 42 there could be a work around where they could set their amps to 0 gain and set levels via headunit or processor instead.

[Riley.]

jas, on Nov 26 2008, 04:31 PM, said:

ok you didnt get what im on about.....

i was asking if a competitor had problems with noise on track 42 there could be a work around where they could set their amps to 0 gain and set levels via headunit or processor instead.

jas thats how i get around noise on my disc - by turning my gains down

[Pulse-R]

jas,
If the noise source is the player, then turning the gains up in that source will have the same issue - just need a quieter source, or mod the one you have.

there is actually another way to eliminate noise from a 'non-zero' track, to use a digital noise gate which is set to ignore low-bit noise. you lose a little in dynamic range ( 1 dB or so) but then there's no noise again, same as a zero bit.

this won't work so well if the noise is entering after the noise gate, such as amp noise, or alternator whine.

guess we'll see soon enough.

[jas]

yes

but if as you say the source has a high noise floor then lowering the amps to 0 gain will have to help.

[Cyberpunky]

if the source has a high noise floor then the amp will amplify this regardless of its settings. Lowering the gains on amp can only reduce amps noise floor, and so would only prevent amps from adding to the sources original noisy signal

peace
Cyberpunky

[Pulse-R]

as mentioned in the original post, 'Track 42' noise can only be reduced by turning down the gain/volume or using components with lower inherent noise. This is where a lot of money can be spent.

As it turns out, a Zero-bit track is used for noise testing in MEASQ in all classes for 2009.

[ktvin09]

Guys i wonder why my system only get noise when the head lights are on, i mean, i can hear no noise when the car has started and the head lights off. I tried taking the headlights out but still the noise present if i hit the light switch.