• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

General debate thread about audio measurements

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
6,245
Likes
2,293
Location
Riverview, FL
To reach 67dB how many watts is that?
Guessing less than one dissipated at the speakers (each).

Using the MartinLogans (so I can measure speaker voltage), and seeing 0.7Vpk with a multimeter (with max hold), at 4 Ohms, that's 0.13 watts peak. (or some similar small amount).

From your own system's volume knob (pot), which position on a clock would that be?
On the preamp, it spins freely and is unmarked, but the display reads from 0 to 151.

At the moment, the preamp display is "30" (a unitless number, but something like .25 to .5dB per digit, never measured critically).

Is your volume control analog or digital?
Have both. Preamp has some kind of electronically selectable resistor set (analog), DAC and DSP recalculate the digits.
 
Last edited:

NorthSky

Major Contributor
Joined
Feb 28, 2016
Messages
4,999
Likes
413
Location
Canada West Coast/Vancouver Island/Victoria area
Cool, educational. That's why they say that the first watt is important. You want that watt, and below...0.1 watt, and just above, 3 watts or 5, clean.

I was just wondering if we can get a cleaner volume control in the analog or digital domain.

Ray, do you also listen with headphones sometimes?
 

trl

Senior Member
King of Mods
Joined
Feb 28, 2018
Messages
469
Likes
281
Location
Iasi, RO
Alright, how loud are we listening to, where on that dial are we turning up that volume control knob (potentiometer, volume attenuator)? How much break pedal (attenuation) are we releasing? What resistance is applied say when we are using normal power...less than 3 watts RMS? @ what average level are we listening to our music? What is the meter showing...0.3 watts? ...Little bit more...1.0 watts? It depends, passive or active listening...background music or front end serious music.
...From a normal stereo preamp, from a headphone amp.


@ low volume level listening we usually measure larger amount of THD.
And the more we crank the volume up the distortion level decreases, up to the power amp limit, where the distortion rises abruptly and clips.
With a quality amp the level of THD is low from 0.001 watts to say 5 watts.
Some amps have larger level of THD @ very low power...say less than 3 watts.

Using a quality amp, how much influence a volume pot can have on the sound output...measurable distortion level? Are there better potentiometers with less measured distortion, and if yes, what are the best ones and why are they measuring better?
The really potential issue might be with a DAC having built-in digital control is the noise level under really quiet environments and very sensitive speakers/headphones. The reason for this is the lack of gain attenuation between the input stage and output stage, basically the entire noise of the DAC's output + I/V stage + LPF stage + preamp stage + output stage is on your headphones/speakers. This noise only manifests with very sensitive IEM's for example or with very sensitive speakers (depending what you're going to amplify and drive). A regular potentiometer places between LPF stage and the preamp stage will lower both noise & volume and background noise will be virtually non-existent or outputs.

I've seen this behavior on Burson PLAY (ES9018K2M with built-in volume contro, 4V RMS/channel on DAC's outputs), but only with very sensitive 16 Ohms IEM's: placing a resistive divider or a regular 10 KOhms pot. "in the middle" reduced the background noise and the output volume, of course. Basically, "pot. in the middle" behavior is also why Objective2 has virtually nonexistent noise on outputs when driving sensitive IEM's.
 

esotechik

New Member
Joined
Mar 9, 2019
Messages
3
Likes
0
What about statistical (ACF) delta-sigma conversion errors?
Many R2R DACs has "no missing codes" in description.
Audio Precision with D-S ADC and long averaging cannot detect ACF errors.
AP is old instrument initial for analog circuits, with some digital features.
sdm_fail.png
https://en.wikipedia.org/wiki/Analog_signal_processing#Linear_time-invariant_(LTI)

"Linearity and time-invariance are important because they are the only types of systems that can be easily solved using conventional analog signal processing methods. Once a system becomes non-linear or non-time-invariant, it becomes a non-linear differential equations problem, and there are very few of those that can actually be solved. (Haykin & Van Veen 2003)"

Delta-sigma DACs is not LTI system, has all missing codes and significant noise shaping.

ACF_random.png
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
17,405
Likes
14,859
Location
Seattle Area
AP is old instrument initial for analog circuits, with some digital features.
Everything we want to analyze is analog. Don't care what happens upstream. We want to know what the analog signal is doing. The AP digitizes this analog signal and then processes it in software.

On the rest of your post, I am not clear what the issue is. DACs that we deal with are multibit so limit cycles are much less of a problem with them. Your references seem old and deal with 1-bit quantizers????
 

esotechik

New Member
Joined
Mar 9, 2019
Messages
3
Likes
0
DACs that we deal with are multibit so limit cycles are much less of a problem with them
"Much less" is only words, not measurements.
IMHO we need true SAR ADC (no missing codes, without noise shaping) + Eye Diagram with meander signals, this is instrumental method vs. AK5394+FFT in AP.
https://www.diyaudio.com/forums/equ...rformance-audio-adc-project-ltc2380-24-a.html

My first (green) picture relating to fifth order sigma-delta modulator.

https://m.eet.com/media/1166738/295165-tmw_eye_freescale_fig4.jpg
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
17,405
Likes
14,859
Location
Seattle Area
"Much less" is only words, not measurements.
IMHO we need true SAR ADC (no missing codes, without noise shaping) + Eye Diagram with meander signals, this is instrumental method vs. AK5394+FFT in AP.
https://www.diyaudio.com/forums/equ...rformance-audio-adc-project-ltc2380-24-a.html

My first (green) picture relating to fifth order sigma-delta modulator.
Like you said, that is only words, not measurements. You are not showing what the Audio Precision analyzer can do. So let's set the conditions similar to yours and compare:
APX Loopback with 1 million point FFT.png


Your graph on the left is using relative dB but unfortunately 0 dB is not set to signal level (what good is relative dB then???). Mine is set to 0 dB so compensating for that, the APx is much, much better. There is only one harmonic component at whopping -157 dB down from signal. In your case, you are getting -130 dB (again, referenced to 0 dB, not -6dB). You also have fair bit of noise and distortion components beyond that harmonic.

How does the APx55 get such superlative performance? It does so by using two ADCs. One handles the signal and then the other notches it out and measure what is left. The second ADC therefore is operating with no distortion since the signal is all gone. Through signal processing, the output of the two ADCs are then summed together. This is how you build a high-performance audio analyzer. Not just by brute force but rather, engineering smarts.

Noise floor is also lower in APx555 analyzer:
APX Loopback noise floor with 1 million point FFT.png


The above with the signal generator on but set to zero which disadvantages the APx555. Still, it beats your solution by 15 dB.

All of this is beside the point: the analyzer only needs to be better than devices we are testing. And the APx555 does that well, able to differentiate between the highest performance audio components we have tested with it.
 

esotechik

New Member
Joined
Mar 9, 2019
Messages
3
Likes
0
Noise floor is also lower in APx555 analyzer:
1 000 000 point averaging is a bad idea for detect occasional "glitches".
Digital oscilloscopes has flash or SAR ADC, not delta-sigma.
Old phonograms (<1995 yr.), digitized via SAR ADC is more preferred vs. modern, generated by D-S ADC (with equal music styles and dynamic range).

This is other test system:
MSB_DAC-Measurement-Flow-Chart-Pico.jpg
 
Last edited:
Joined
Feb 24, 2019
Messages
87
Likes
122
Well, I've waded through this one.:confused:
I've really only got one point to make but it's really important; there is no music!
People keep talking about the effects of various parameters on the music. These boxes of electronics don't make music, they don't transport music, they don't decode music, they wouldn't know music if you took them to a concert. There is no music on the medium, no music comes out of the speakers. The music is all in your head.
It's signal processing. The measurements are about how well, or badly a particular unit does this. The comparisons are to other signal processing units. Some measure better than others. You may well not be able to tell the difference by ear. That's not the point.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
17,405
Likes
14,859
Location
Seattle Area
1 000 000 point averaging is a bad idea for detect occasional "glitches".
There is no million point averaging. The averaging and FFT lengths are set exactly to what was measured in the link you provided. For one test it is 20 averages and the other 10. And if that gets rid of occasional "glitches," that much the better. It makes the measurement system immune to issues you imagine to be there (but not demonstrated with any measurements).
 
Top Bottom