Hi soundcheck,
Just a test, set volume control to a level where sound is barely audible, then walk over to one of the speakers and listen. Is the sound crystal clear, revealing every single detail, and is hum / noise completely, and I mean completely inaudible now?
If yes, you can do without the described volume control.
When you do some practical measurements (at the speaker terminals), resolution is more like 12 ... 16 bits at maximum volume setting. The S/N ratio required for true 24-bit resolution simply cannot be maintained in a conventional audio set, resolution is already lost when the signal leaves the DAC PCB (provided the DAC PCB properties even enabled achieving true 24-bit resolution).
Example, your power amp produces approx. 7 watts rms in 8 Ohms (average volume setting). This translates to approx. 21Vpp at the speaker terminals. At 24 bit resolution, LSB change equals 21 / 2^24 = 1.25uVpp. You won't even be able to measure this with a plain oscilloscope set at maximum sensitivity using a 1:1 probe.
This is AFTER amplification in the power amp (usually 50 ... 100x), meaning that the same 1 LSB change at the power amp input represents 12.5 ... 25nV. This is AFTER the signal passed interlinks (pre-amp) and volume control.
Perhaps this illustrates the problem a bit better.
Same example with 16 bit resolution, 21 / 2^16 = 0.32mV, this is just barely measurable with a plain oscilloscope and a 1:1 probe. LSB change at the power amp input equals 3.2uV ... 6.4uV, even this is almost impossible to achieve considering the signal path.
Now the same with max. volume setting, 70 watts rms in 8 Ohms. This translates to approx. 67Vpp and a LSB change (16 bits) of 67 / 2^16 = 1mV. At the amplifier input (50x gain like with my amp), it results in 204uVpp LSB change and 1.34Vpp at the DAC output. Even the 240uVpp is already problematic, and requires very careful design and a clean mains voltage.
Yep, recently had my mains voltage tested in all rooms (expected too high interference levels @ 1 KHz ... 10 MHz). Well, average reading should have been around 25 GS units (acceptable), at my place the meter went off-the-scale (>1999 Gs units) and was finally damaged (overloaded). Time for some action.
After installing 15 filters at strategic points (and repairing the meter that appeared to have a blown 10nF multilayer cap), readout dropped to around 30 ... 40 GS units in every room.
The filters appear to be 15uF caps with a 47K bleeder resistor, and perhaps some other parts.
Here is the website of the manufacturer:
http://www.stetzerelectric.com/filters/
Also have a look under "Graham Stetzer research"
Bottom line is that mains voltage is pretty polluted nowadays and this will only get worse. This can be easily verified and quantified by measurements. So perhaps it's time for battery power supplies.
Batteries have relative low noise, and most important, offer a true floating power supply / reference voltage, and don't contain pollution like present on the mains.
My favorites are plain alkaline batteries and lead acid batteries.
The SD-player is designed to run on a single 5V source (approx. 250mA), so it can run on a battery power supply. Meaning both digital audio source and DAC can run on a clean floating battery power supply, and when using this setup, ground-loops with connected equipment can be completely avoided.
Just a test, set volume control to a level where sound is barely audible, then walk over to one of the speakers and listen. Is the sound crystal clear, revealing every single detail, and is hum / noise completely, and I mean completely inaudible now?
If yes, you can do without the described volume control.
When you do some practical measurements (at the speaker terminals), resolution is more like 12 ... 16 bits at maximum volume setting. The S/N ratio required for true 24-bit resolution simply cannot be maintained in a conventional audio set, resolution is already lost when the signal leaves the DAC PCB (provided the DAC PCB properties even enabled achieving true 24-bit resolution).
Example, your power amp produces approx. 7 watts rms in 8 Ohms (average volume setting). This translates to approx. 21Vpp at the speaker terminals. At 24 bit resolution, LSB change equals 21 / 2^24 = 1.25uVpp. You won't even be able to measure this with a plain oscilloscope set at maximum sensitivity using a 1:1 probe.
This is AFTER amplification in the power amp (usually 50 ... 100x), meaning that the same 1 LSB change at the power amp input represents 12.5 ... 25nV. This is AFTER the signal passed interlinks (pre-amp) and volume control.
Perhaps this illustrates the problem a bit better.
Same example with 16 bit resolution, 21 / 2^16 = 0.32mV, this is just barely measurable with a plain oscilloscope and a 1:1 probe. LSB change at the power amp input equals 3.2uV ... 6.4uV, even this is almost impossible to achieve considering the signal path.
Now the same with max. volume setting, 70 watts rms in 8 Ohms. This translates to approx. 67Vpp and a LSB change (16 bits) of 67 / 2^16 = 1mV. At the amplifier input (50x gain like with my amp), it results in 204uVpp LSB change and 1.34Vpp at the DAC output. Even the 240uVpp is already problematic, and requires very careful design and a clean mains voltage.
Yep, recently had my mains voltage tested in all rooms (expected too high interference levels @ 1 KHz ... 10 MHz). Well, average reading should have been around 25 GS units (acceptable), at my place the meter went off-the-scale (>1999 Gs units) and was finally damaged (overloaded). Time for some action.
After installing 15 filters at strategic points (and repairing the meter that appeared to have a blown 10nF multilayer cap), readout dropped to around 30 ... 40 GS units in every room.
The filters appear to be 15uF caps with a 47K bleeder resistor, and perhaps some other parts.
Here is the website of the manufacturer:
http://www.stetzerelectric.com/filters/
Also have a look under "Graham Stetzer research"
Bottom line is that mains voltage is pretty polluted nowadays and this will only get worse. This can be easily verified and quantified by measurements. So perhaps it's time for battery power supplies.
Batteries have relative low noise, and most important, offer a true floating power supply / reference voltage, and don't contain pollution like present on the mains.
My favorites are plain alkaline batteries and lead acid batteries.
The SD-player is designed to run on a single 5V source (approx. 250mA), so it can run on a battery power supply. Meaning both digital audio source and DAC can run on a clean floating battery power supply, and when using this setup, ground-loops with connected equipment can be completely avoided.
Hi yygomez,
The oscillator runs on 5V (approx. 10mA), the 1.5V is used as reference voltage to bias the JFETs in the master clock.
The oscillator is basically a buffer with a phase shifter and a crystal filter in the feedback loop. This results in a pure, very low distortion 5Vpp sine wave at the amplifier input.
I already constructed a prototype with 8 crystals in the feedback loop (very difficult to tune), I now use simpler versions with 2 or 3 crystals that are less critical and more easy to build.
I picked the signal from the JFET input, using a second Buffer circuit.
I attached an oscillogram of a 2-crystal version running on a dirty 5V power supply and using a 1.5V alkaline reference voltage. I included the scope in the picture to verify settings. I used a 1:10 probe, and time base X10 setting. Frequency equals 11.2896 MHz.
The sine wave can be used to trigger a suitable high-speed comparator with high input impedance and low input capacity. but amplitude is also high enough to directly trigger connected loads (after unity gain buffering).
The oscillator runs on 5V (approx. 10mA), the 1.5V is used as reference voltage to bias the JFETs in the master clock.
The oscillator is basically a buffer with a phase shifter and a crystal filter in the feedback loop. This results in a pure, very low distortion 5Vpp sine wave at the amplifier input.
I already constructed a prototype with 8 crystals in the feedback loop (very difficult to tune), I now use simpler versions with 2 or 3 crystals that are less critical and more easy to build.
I picked the signal from the JFET input, using a second Buffer circuit.
I attached an oscillogram of a 2-crystal version running on a dirty 5V power supply and using a 1.5V alkaline reference voltage. I included the scope in the picture to verify settings. I used a 1:10 probe, and time base X10 setting. Frequency equals 11.2896 MHz.
The sine wave can be used to trigger a suitable high-speed comparator with high input impedance and low input capacity. but amplitude is also high enough to directly trigger connected loads (after unity gain buffering).
Attachments
Perhaps you remember: I am using batteries fon almost everything for quite a while - my amp also runs on batteries ( This way I can even run my amp DC coupled - which makes a huge difference).
Strictly seperated the analog and digital supply. I get along with 12V throughout the chain.
(Which requires a high SPL speaker of course if you want to avoid DC/DC converters).
The great thing I am mainly talking about "The First Watt" only to stay in Nelson Pass terms. This gives me much more flexibility then anything else. I don't enter "distoration critical" phases in any situation. I run low gain amplification etc. That's IMO more worth than anything else.
I selected one of the Northstar batteries ( I don't know any better choice) since there was no other giving me low ESR of 2mR.
I guess there are not very many PS supplies which can compete with this. These batteries are quite expensive nowadays. People will think twice to spend 250$ for one battery.
The 24bit thing refers to volume control on my PC without impacting the signal a lot. I received some feedback from a dutch friend last week, that he finally took out his passive control, to connect the DAC rigth to the amp. He won't switch back. Me neither.
Cheers
Strictly seperated the analog and digital supply. I get along with 12V throughout the chain.
(Which requires a high SPL speaker of course if you want to avoid DC/DC converters).
The great thing I am mainly talking about "The First Watt" only to stay in Nelson Pass terms. This gives me much more flexibility then anything else. I don't enter "distoration critical" phases in any situation. I run low gain amplification etc. That's IMO more worth than anything else.
I selected one of the Northstar batteries ( I don't know any better choice) since there was no other giving me low ESR of 2mR.
I guess there are not very many PS supplies which can compete with this. These batteries are quite expensive nowadays. People will think twice to spend 250$ for one battery.
The 24bit thing refers to volume control on my PC without impacting the signal a lot. I received some feedback from a dutch friend last week, that he finally took out his passive control, to connect the DAC rigth to the amp. He won't switch back. Me neither.
Cheers
EC
Thanks for the answer.
If you are interested I can help you to measure the Phase Noise of the master clock/Buffer, I have access to a Symmetricon 5021A Phase noise test set, In such way you can fully characterize the jitter of your master clock.
Attached some screen pictures:
Attachments
Hi EC,
Thanks for your reply (post #2818).
Glad you are making progress
I have learnt to fear your long silences
Just when I was prepearing myself to build a "one box solution" D1M (or D4M or D16M) with trans-impedance circuit and variable I/V resistor as volume selector, with class A power buffer (whose schematic you would kindly care to provide ) you came up with this other possibility of power resistor to attenuate musical signal pre-speaker
Frankly the power resistor in series is not very practical to me, safe on my full-range speakers, probably. Isn't it better to change I/V resistor as was your first idea?
About your SD-card player, my canines are already this long hearing about it
Won't you at least show some picks to calm our hunger?
Cheers,
M
Off topic post scriptum: for those interested in classical music from the past, please check out http://www.europarchive.org/
Thanks for your reply (post #2818).
Glad you are making progress
I have learnt to fear your long silences
-ecdesigns- said:
Just when I was prepearing myself to build a "one box solution" D1M (or D4M or D16M) with trans-impedance circuit and variable I/V resistor as volume selector, with class A power buffer (whose schematic you would kindly care to provide
) you came up with this other possibility of power resistor to attenuate musical signal pre-speaker Frankly the power resistor in series is not very practical to me, safe on my full-range speakers, probably. Isn't it better to change I/V resistor as was your first idea?
About your SD-card player, my canines are already this long hearing about it
Won't you at least show some picks to calm our hunger?Cheers,
M
Off topic post scriptum: for those interested in classical music from the past, please check out http://www.europarchive.org/
-ecdesigns- said:
I am slightly astonished, that nobody questions these statements, if they were correct, it would be a waste of time to make ad/da converters or music files with more than 16bits.
Here dynamic range and resolution have been mixed up.
Of course we can not hear the lower digital binary numbers, because they disappear into hiss or hum, but we certainly can hear the resolution, that means the difference between a number and the next higher one. If the resolution power of the system is good, this can certainly be heard, because it is above the noise floor. Let us do an analogy with decimal numbers:
Lets say our system is capable of 1000 different volumes, 1 to 1000, the resolution power is 1/1000.
Because of noise numbers 1 to 200 will disappear, we can just hear the numbers from 200 to 1000, but the resolution power is still there, is still 1/1000.
We can explain it in binary numbers as well:
010000000000000000000000
This is a 24 bit/digit binary number, its value is 6db below maximum, because it is shifted one digit to the right, no noise floor here of course! If we add 1, we have the resolution:
010000000000000000000001
If the system can differentiate this, it can sound probably more lifelike/ detailed.
regards
Andre
Thanks EC for your reply (post 2***)
class A power buffer (whose schematic you would kindly care to provide).
I could post a schematic. The problem with this configuration is that the buffer (cross-over) distortion, and noise / hum is dumped on the speakers without attenuation. With the new volume control, this distortion (and noise / hum) is attenuated, and signal amplitude is always maximum. This lifts the signal well above the noise floor, and provides much better clarity at the volume settings that are most often used.
I planned to use a singled ended class A power buffer (to avoid crossover distortion) and a regulated supply (to avoid humm). I found Pavel Macura's "Mosfet Power Follower" as per the description on Elliot's site:
http://sound.westhost.com/project83.htm
Yeah, it is weird...we should challenge common practice more often.
I could attenuate at the digital level, but I am not as smart as Soundcheck to do this without fear of lost of resolution...
What do you say? I love messy prototypes
Oops! I totally missed that one...
Time to catch up...
Cheers,
M
Hi John,
The idea that you can simply add an attenuator between the amp and the speaker is very attractive BUT unfortunately, it runs into many problems - in short, the amplifier and speaker are a SYSTEM and work together because of the difficult and varying speaker parameters that change with driver impedance, power, freq, phase, transient response, compression, Xover, etc, plus the amps ability to control the back emf and avoid loading the neg feedback networks, output stage damping, etc, etc. Adding a FIXED value resistive attenuator network into the middle of this is a relatively simple thing (driver eq) but even this produces unexpected complications and so a full freq range, varyible load attenuator will introduce enormous difficulties (even with a reduced operating volume range), particularly with very high quality sound reproduction. Perhaps it would be better to use this higher voltage output from the DAC and adjust the gain of the amplifier for the required volume or perhaps a higher voltage attenuator with an output power buffer - a much simpler thing to do.
The idea that you can simply add an attenuator between the amp and the speaker is very attractive BUT unfortunately, it runs into many problems - in short, the amplifier and speaker are a SYSTEM and work together because of the difficult and varying speaker parameters that change with driver impedance, power, freq, phase, transient response, compression, Xover, etc, plus the amps ability to control the back emf and avoid loading the neg feedback networks, output stage damping, etc, etc. Adding a FIXED value resistive attenuator network into the middle of this is a relatively simple thing (driver eq) but even this produces unexpected complications and so a full freq range, varyible load attenuator will introduce enormous difficulties (even with a reduced operating volume range), particularly with very high quality sound reproduction. Perhaps it would be better to use this higher voltage output from the DAC and adjust the gain of the amplifier for the required volume or perhaps a higher voltage attenuator with an output power buffer - a much simpler thing to do.
Hi John,
I also wonder why you are doing this.
At low volumes, so high series resistor versus load, you will create a current source in stead of a voltage source.
We know that loudspeakers have very bumby impedance curves, hence we need voltage source. By driving current, your frequency response will follow the impedance curve, which happens to be highest at bass resonance frequency and the high frequencies.
basically you create a loudness effect at low volumes. Is that why is sounds so nice ?
On top, you listen to different frequency respons at every position of the volume controll
Unless you complete flatten the impedance curve (what can be done) this is no option imo
If this is done in your system, that is fine than, so in that case let this be a warning to others, not to implement it without any thought. It is not a general solution
If your fear is, that you amplify noise etcetera, why not put volume controll directly for the Power amplifier last stage (1x gain) ? In that case you fullfill your requirements as well. Of course you need a good power supply, but I have never had problems with that on the last stage introducing humm or noise (audible)
I also wonder why you are doing this.
At low volumes, so high series resistor versus load, you will create a current source in stead of a voltage source.
We know that loudspeakers have very bumby impedance curves, hence we need voltage source. By driving current, your frequency response will follow the impedance curve, which happens to be highest at bass resonance frequency and the high frequencies.
basically you create a loudness effect at low volumes. Is that why is sounds so nice ?
On top, you listen to different frequency respons at every position of the volume controll
Unless you complete flatten the impedance curve (what can be done) this is no option imo
If this is done in your system, that is fine than, so in that case let this be a warning to others, not to implement it without any thought. It is not a general solution
If your fear is, that you amplify noise etcetera, why not put volume controll directly for the Power amplifier last stage (1x gain) ? In that case you fullfill your requirements as well. Of course you need a good power supply, but I have never had problems with that on the last stage introducing humm or noise (audible)
Hi rolls,
High resolution recordings are required (digital mixing) so these aren't a waste of time.
I am referring to digital audio playback only, and the practical limitations of audio sets at moderate volume settings.
This article might be interesting:
http://www.proavmagazine.com/industry-news.asp?sectionID=1765&articleID=596561
Seems 16 bits (93dB dynamic range) is plenty for a practical High-End digital audio playback system.
The dynamic range of an audio set at a given volume setting, and listening distance / position from the speakers can be tested with the Sheffield lab test CD "My Disc A2TB.
http://www.amazon.com/Sheffield-A2TB-Test-Disc-My/dp/B000V93NKY
High resolution recordings are required (digital mixing) so these aren't a waste of time.
I am referring to digital audio playback only, and the practical limitations of audio sets at moderate volume settings.
This article might be interesting:
http://www.proavmagazine.com/industry-news.asp?sectionID=1765&articleID=596561
Seems 16 bits (93dB dynamic range) is plenty for a practical High-End digital audio playback system.
The dynamic range of an audio set at a given volume setting, and listening distance / position from the speakers can be tested with the Sheffield lab test CD "My Disc A2TB.
http://www.amazon.com/Sheffield-A2TB-Test-Disc-My/dp/B000V93NKY
-ecdesigns- said:
Hello John
You and the guy in the article still mix up dynamic range and resolution.
You can have a dynamic range of 93db only and still a resolution power of 24bit, it means that the steps are finer, one step is 1 divide by 2 to the power of 24.
In other words, the binary word is not zero when it "leaves" noise.
High resolution recordings are not made for mixing purpose only, you can download them from different sources. I agree that probably we will never have a system with a dynamic range of a complete range of 24bit numbers, 144db.
Oh yes, I did !Hi Bernhard,
Project is on target, the main objective is almost achieved. The design process is well documented on this thread. The preceding years were necessary to systematically locate bottlenecks in the design and eliminate them.
There is a practical result, and a conclusion.
The conclusion is that one of the major problems is the digital audio source. There will be no ultimate DAC without an ultimate digital audio source. I wasn't able to find a suitable digital audio source, so both my brother and I designed one from scratch.
The practical result is the SD-player.
I can't help it if you are puzzled, and yes the TDA1543 is capable of producing crystal clear sound, it's just a matter of design.
I attempted to explain why I did this, I can't help it if you can't understand.
Project is on target, the main objective is almost achieved. The design process is well documented on this thread. The preceding years were necessary to systematically locate bottlenecks in the design and eliminate them.
There is a practical result, and a conclusion.
The conclusion is that one of the major problems is the digital audio source. There will be no ultimate DAC without an ultimate digital audio source. I wasn't able to find a suitable digital audio source, so both my brother and I designed one from scratch.
The practical result is the SD-player.
I can't help it if you are puzzled, and yes the TDA1543 is capable of producing crystal clear sound, it's just a matter of design.
I will be soon implementing the passive volume control on the output of my power amplifier
I attempted to explain why I did this, I can't help it if you can't understand.