Hi maxlorenz,
Sorry for the late reply,
Yes that sounds familiar, I am also waiting for my 2.2uF /200V Auricap capacitors.
Installing Ubuntu is very easy and straight-forward, you just need to download the ISO image and burn the installation CD.
Mac OSX is basically for mac computers only. Did you run XP on the same hardware that's now running OSX?
If you are running mac OSX (Tiger), you can use iTunes for both listening, and ripping CDs, format: Apple Lossless Encoder
When using Ubuntu, you can use Rhythmbox Music Player for listening, and Sound Juicer CD extractor for ripping CDs, format: FLAC Audio (lossless)
Use the native sampling speed (44.1 KHz), and set iTunes volume to maximum.
You could also use other programs for ripping like CD-paranoia, especially when the CDs are damaged.
Sorry for the late reply,
The built of my new 8*4TDA1543 DAC is at standby because of 100nf caps out of stock
Yes that sounds familiar, I am also waiting for my 2.2uF /200V Auricap capacitors.
Installing Ubuntu is very easy and straight-forward, you just need to download the ISO image and burn the installation CD.
Mac OSX is basically for mac computers only. Did you run XP on the same hardware that's now running OSX?
If you are running mac OSX (Tiger), you can use iTunes for both listening, and ripping CDs, format: Apple Lossless Encoder
When using Ubuntu, you can use Rhythmbox Music Player for listening, and Sound Juicer CD extractor for ripping CDs, format: FLAC Audio (lossless)
Use the native sampling speed (44.1 KHz), and set iTunes volume to maximum.
You could also use other programs for ripping like CD-paranoia, especially when the CDs are damaged.
maxlorenz said:
I am not surprised, win has a very "messy" build software-environment. ("you have to restart your computer to use this new program") Old fashioned if you ask me. MS delivers every time another newer version based on the old win95 (3.1?) it became an "stacked-on-eachother" system when it reached XP. And all service packs are bandages to fix it on eachother again, to get the system internet-safe.
Maybe another idea: buy a old apple G3 or so, and make it dual boot with Ubuntu for apple besides the OS-X.
Hi guys,
I did that but, then (I'm not a PC expert) the option to install it did not show at the "bios". I did not know if I had to reformat the disc...etc. So I went to my local expert for help.
Yes. An AMD sempron based PC. Strange, isn't it?
I do use iTunes, but the copied tracks seem to be MPEG???
Is Apple Lossless Encoder set by default or do I have to set it?
...and where do I have to set it?
I'm new to it...
Thanks for your help.
M
I did that but, then (I'm not a PC expert) the option to install it did not show at the "bios". I did not know if I had to reformat the disc...etc. So I went to my local expert for help.
Yes. An AMD sempron based PC. Strange, isn't it?
I do use iTunes, but the copied tracks seem to be MPEG???
Is Apple Lossless Encoder set by default or do I have to set it?
...and where do I have to set it?
I'm new to it...
Thanks for your help.
M
Hi maxlorenz,
When the ISO image is burned to the installation CD correctly it's bootable (contains boot record)
Next you must make sure that your first boot device is the CDROM (BIOS setting, boot sequence)
Then insert the Ubuntu install CD, and restart, the PC will now boot from CDROM instead of the HDD. Then just follow the installation steps.
No it's not that strange that mac-OSX can run on the correct PC hardware, I just meant it's not legal to do so.
The ripped tracks (Apple Lossless) have the extention .m4a
Ripped CD images are dumped in Home > Music > iTunes.
The desired format can be selected as follows:
Go to iTunes > Preferences > Advanced > IMPORTING
"On CD Insert:" Import CD and Eject"
"Import Using": "Apple Lossless Encoder"
You might also mark the following"
"Automatically retrieve CD track names from the Internet"
"Create file names with track number"
"Use error correction when reading Audio CDs"
Next you might select GENERAL (Advanced settings)
Set Streaming Buffer Size to "Large"
Finally under BURNING (Advanced Settings), you might want to choose:
"Disk Format" Audio CD
"Include CD Text"
If you want to add CD cover artwork, select all tracks of an album (left-click on the first track, press and hold shift, left click on the last track)
Now right-click on one of the selected tracks, and select "Get Info". now simply drag the (prepared) jpeg image to the small box, that's all.
I use image size of 500 x 500 pixels (filesize approx 40K), this will work very nice with full-screen coverflow (button on the bottom Right of iTunes graphics screen).
Make sure to select 44.1 KHz sample rate:
Go > Utilities > Audio MIDI Setup
Properties for: USB Audio DAC
Format: 44100.0 Hz
Make sure to put iTunes volume to maximum.
When the ISO image is burned to the installation CD correctly it's bootable (contains boot record)
Next you must make sure that your first boot device is the CDROM (BIOS setting, boot sequence)
Then insert the Ubuntu install CD, and restart, the PC will now boot from CDROM instead of the HDD. Then just follow the installation steps.
Yes. An AMD sempron based PC. Strange, isn't it?
No it's not that strange that mac-OSX can run on the correct PC hardware, I just meant it's not legal to do so.
The ripped tracks (Apple Lossless) have the extention .m4a
Ripped CD images are dumped in Home > Music > iTunes.
The desired format can be selected as follows:
Go to iTunes > Preferences > Advanced > IMPORTING
"On CD Insert:" Import CD and Eject"
"Import Using": "Apple Lossless Encoder"
You might also mark the following"
"Automatically retrieve CD track names from the Internet"
"Create file names with track number"
"Use error correction when reading Audio CDs"
Next you might select GENERAL (Advanced settings)
Set Streaming Buffer Size to "Large"
Finally under BURNING (Advanced Settings), you might want to choose:
"Disk Format" Audio CD
"Include CD Text"
If you want to add CD cover artwork, select all tracks of an album (left-click on the first track, press and hold shift, left click on the last track)
Now right-click on one of the selected tracks, and select "Get Info". now simply drag the (prepared) jpeg image to the small box, that's all.
I use image size of 500 x 500 pixels (filesize approx 40K), this will work very nice with full-screen coverflow (button on the bottom Right of iTunes graphics screen).
Make sure to select 44.1 KHz sample rate:
Go > Utilities > Audio MIDI Setup
Properties for: USB Audio DAC
Format: 44100.0 Hz
Make sure to put iTunes volume to maximum.
Hi again,
I swaped for "Apple Lossless Encoder" and it certainly improved the sound: more ambience ("air" and space sensation) and amazing increase in stage noises, like cough of things falling dawn ...not that these noises increase the musical experience much, but they are useful to state on what amount the "details", like bow sound, string plucking and breathing (amongst other), increase.
M
I swaped for "Apple Lossless Encoder" and it certainly improved the sound: more ambience ("air" and space sensation) and amazing increase in stage noises, like cough of things falling dawn
...not that these noises increase the musical experience much, but they are useful to state on what amount the "details", like bow sound, string plucking and breathing (amongst other), increase. M
Hi,
I used Windows media wich sounds flat an uninvolving...iTunes (MP3?) made the structure of the music more apparent, but Apple Lossless completed the picture, and the slow character that I found before is almost completelly gone...and I am not using your USB/DI2S module yet I hope another improvement there. (not to mention a good USB cable and a silent PC)
Strings sound like strings (I love strings)
Anyway, I espect the DI 8*4 to be even "faster" and to make images even more dense and detailed, wich is the experience with previous experiments...my DAC will be way hoter than my amps though...
I used Windows media wich sounds flat an uninvolving...iTunes (MP3?) made the structure of the music more apparent, but Apple Lossless completed the picture, and the slow character that I found before is almost completelly gone...and I am not using your USB/DI2S module yet
I hope another improvement there. (not to mention a good USB cable and a silent PC)Strings sound like strings (I love strings)
Anyway, I espect the DI 8*4 to be even "faster" and to make images even more dense and detailed, wich is the experience with previous experiments...my DAC will be way hoter than my amps
though...iTunes uses AAC, it's not MP3. Apple Lossless is Lossless AAC with DRM on it I think.
If you used Windows Media Player it might explain things... Foobar2000 or Winamp on Windows with a Kernel Streaming plugin should sound well.
The sound engine on Mac OS X is the same as one of those on Linux(Ubuntu) I think. You shouldn't see any difference between a Windows player with Kernel Streaming and MacOS, Linux or whatever. You can use other lossless formats than the DRM'ed Apple Lossless and WAV. You can use FLAC or WavPack too.
If you used Windows Media Player it might explain things... Foobar2000 or Winamp on Windows with a Kernel Streaming plugin should sound well.
The sound engine on Mac OS X is the same as one of those on Linux(Ubuntu) I think. You shouldn't see any difference between a Windows player with Kernel Streaming and MacOS, Linux or whatever. You can use other lossless formats than the DRM'ed Apple Lossless and WAV. You can use FLAC or WavPack too.
Thermal Memory
Hi maxlorenz,
If Windows media sounds flat and uninvolving, the digital sound data was altered in some way. I noticed a similar effect with upsampling.
I have no explanation why MP3 on mac OSX sounded better than Windows media. For me MP3 sounds flat, uninvolving, and distorted.
All I can say is that I get the same sou-d quality from both Ubuntu and mac OSX, I also noted this earlier on this thread. Both were compared with a reclocked I2S output from a CD transport as reference.
All you need is correct digital audio data (no errors), native sample rate (no up-sampling), and a very low jitter timing signal (reclocking). It's as simple as that.
This week I noticed something important. The real effects of Thermal Memory on audio signals. Compared to Thermal Memory, jitter isn't the main problem.
Reducing Thermal Memory can result in significant sound improvements. The main problem is the (differential) input stage of OP-amps and power amplifiers. Thermal memory causes the typical harsh semiconductor sound. Even worse is that Thermal Memory errors are amplified by the amount of feedback. So theoretically an OP-amp buffer (gain = 1) will produce highest Thermal Memory Distortion.
I noticed that these errors become even more apparent with digital sound sources, NOS DACs introduce an extra problem due to the rapidly changing step signals.
Thermal memory is caused by (small) temperature fluctuations in semiconductors, and all contacts / traces on it. These temperature fluctuations cause thermal-voltages across all contacts that consist of conductive materials with different properties. These thermal-voltages distort the passing audio signal, and add a lot of unwanted harmonics. The negative feedback loop amplifies these thermal-voltages, this is one of the factors why negative feedback loops degrade sound quality.
The reason why the DI 8M sounds so wonderful is that Thermal Memory is halved by using a tube / semiconductor hybrid stage, while THD is also very low. Tubes have no thermal memory and therefore are ideal for High-End audio, the "warm" sound is basically caused by reduction of unwanted harmonic distortion, this in turn is a result of the absence of Thermal Memory. The problem with tubes (power amplifiers) is high THD.
But there is a cure, operating the differential input pair of a semiconductor amplifier with constant power. This takes some extra components, but it sure is worth the effort.
I modified my symmetrical MOSFET cascode power amplifiers so their input stage has constant power.
I added a picture of the mod (the two small piggyback PCBs).
They now have 12 transistors / JFETs in the differential input stage. This resulted in the most dramatic increase in sound quality so far. The MOSFET power amps no longer sound as semiconductor amplifiers, in fact, they are now extremely transparent, without even the slightest harshness or edge. I could best describe it as "silent sound", the (dynamic) distortion is now so low, that the sound level appears to be much lower than before, using identical signal amplitude.
My next plan is to construct a discrete OP-amp with very low thermal memory for use in the D-I DACs.
Hi maxlorenz,
If Windows media sounds flat and uninvolving, the digital sound data was altered in some way. I noticed a similar effect with upsampling.
I have no explanation why MP3 on mac OSX sounded better than Windows media. For me MP3 sounds flat, uninvolving, and distorted.
All I can say is that I get the same sou-d quality from both Ubuntu and mac OSX, I also noted this earlier on this thread. Both were compared with a reclocked I2S output from a CD transport as reference.
All you need is correct digital audio data (no errors), native sample rate (no up-sampling), and a very low jitter timing signal (reclocking). It's as simple as that.
This week I noticed something important. The real effects of Thermal Memory on audio signals. Compared to Thermal Memory, jitter isn't the main problem.
Reducing Thermal Memory can result in significant sound improvements. The main problem is the (differential) input stage of OP-amps and power amplifiers. Thermal memory causes the typical harsh semiconductor sound. Even worse is that Thermal Memory errors are amplified by the amount of feedback. So theoretically an OP-amp buffer (gain = 1) will produce highest Thermal Memory Distortion.
I noticed that these errors become even more apparent with digital sound sources, NOS DACs introduce an extra problem due to the rapidly changing step signals.
Thermal memory is caused by (small) temperature fluctuations in semiconductors, and all contacts / traces on it. These temperature fluctuations cause thermal-voltages across all contacts that consist of conductive materials with different properties. These thermal-voltages distort the passing audio signal, and add a lot of unwanted harmonics. The negative feedback loop amplifies these thermal-voltages, this is one of the factors why negative feedback loops degrade sound quality.
The reason why the DI 8M sounds so wonderful is that Thermal Memory is halved by using a tube / semiconductor hybrid stage, while THD is also very low. Tubes have no thermal memory and therefore are ideal for High-End audio, the "warm" sound is basically caused by reduction of unwanted harmonic distortion, this in turn is a result of the absence of Thermal Memory. The problem with tubes (power amplifiers) is high THD.
But there is a cure, operating the differential input pair of a semiconductor amplifier with constant power. This takes some extra components, but it sure is worth the effort.
I modified my symmetrical MOSFET cascode power amplifiers so their input stage has constant power.
I added a picture of the mod (the two small piggyback PCBs).
They now have 12 transistors / JFETs in the differential input stage. This resulted in the most dramatic increase in sound quality so far. The MOSFET power amps no longer sound as semiconductor amplifiers, in fact, they are now extremely transparent, without even the slightest harshness or edge. I could best describe it as "silent sound", the (dynamic) distortion is now so low, that the sound level appears to be much lower than before, using identical signal amplitude.
My next plan is to construct a discrete OP-amp with very low thermal memory for use in the D-I DACs.
Attachments
discrete OP-amp
Hi EC,
I would be happy to test your discrete OP-amp design. Funny, just yesterday I was thinking about tesing discrete OP-amps but didn't know where to start or even why.
By the way... tomorrow I have parts arriving to test reclocking the PCM2706/7 using two clocks, 12MHz and 48MHz.
Cheers,
Brent
Hi EC,
I would be happy to test your discrete OP-amp design. Funny, just yesterday I was thinking about tesing discrete OP-amps but didn't know where to start or even why.
By the way... tomorrow I have parts arriving to test reclocking the PCM2706/7 using two clocks, 12MHz and 48MHz.
Cheers,
Brent
Hi -ecdesigns-
About sample rate, I failed to find where to set 44100Hz on OSX
That "Thermal Memory correction circuit" that you propose is very interesting. I hope you will share more info/schematics...(you can use mail if it is secret )
Do you think that soldering is better than using adaptors for opamps, in the thermal memory aspect?
Talking about piggyback, I have not finished my DI8*4 DAC yet and I am already thinking a way of implementing an DI8*8 DAC!
Could one piggyback the mentioned buffers to increase the output current and, perhaps, be able to improve jitter figures???
Sorry to bother with intuitive thinking...
Keep on with the good work!
M
PS: I can really enjoy now my contemporary chamber music CDs
You were right, each CD sounds different from each other: you can see the hand of the recording engineer.
About sample rate, I failed to find where to set 44100Hz on OSX
That "Thermal Memory correction circuit" that you propose is very interesting. I hope you will share more info/schematics...(you can use mail if it is secret
)Do you think that soldering is better than using adaptors for opamps, in the thermal memory aspect?
Talking about piggyback, I have not finished my DI8*4 DAC yet and I am already thinking a way of implementing an DI8*8 DAC!
Could one piggyback the mentioned buffers to increase the output current and, perhaps, be able to improve jitter figures???
Sorry to bother with intuitive thinking...
Keep on with the good work!
M
PS: I can really enjoy now my contemporary chamber music CDs
You were right, each CD sounds different from each other: you can see the hand of the recording engineer.
PCM2706 reclocked with two clocks
I am happy to report that reclocking the BCK line from the PCM2706/7 chip with a separate 48MHz clock works/sounds very nice.
This means it is very easy to add this reclocker to many existing DACs. For example no need to replace the 12 MHz clock and no need for the divide by 4 chip.
Here is my latest Digikey parts list (better parts then my last list):
CTX315LVCT-ND OSC CLOCK 48.0000 MHZ 3.3V SMD
568-2620-5-ND IC 8BIT SHIFT REGISTER 14-SSOP
728-1022-1-ND IC REG LDO 3.3V SOT-89-3
728-1024-1-ND IC REG LDO 5.0V SOT-89-3
493-2352-1-ND CAP TANTALUM 10UF 10V 20% SMD
NC7SZ04M5XCT-ND IC INVERTER SGL TINYLOGIC SOT-23
If anyone wants to try it and is not up to soldering SMD parts point-to-point then message me.
I think once more people hear this reclocker they will have a new appreciation for EC's designs and kits.
I am happy to report that reclocking the BCK line from the PCM2706/7 chip with a separate 48MHz clock works/sounds very nice.
This means it is very easy to add this reclocker to many existing DACs. For example no need to replace the 12 MHz clock and no need for the divide by 4 chip.
Here is my latest Digikey parts list (better parts then my last list):
CTX315LVCT-ND OSC CLOCK 48.0000 MHZ 3.3V SMD
568-2620-5-ND IC 8BIT SHIFT REGISTER 14-SSOP
728-1022-1-ND IC REG LDO 3.3V SOT-89-3
728-1024-1-ND IC REG LDO 5.0V SOT-89-3
493-2352-1-ND CAP TANTALUM 10UF 10V 20% SMD
NC7SZ04M5XCT-ND IC INVERTER SGL TINYLOGIC SOT-23
If anyone wants to try it and is not up to soldering SMD parts point-to-point then message me.
I think once more people hear this reclocker they will have a new appreciation for EC's designs and kits.
Hi Brent Welke,
Glad to hear that. I have EC's USB/DI2S but I am reserving it for the DI8*4 DAC. SMD should be better but will increase kit's price
There's the main problem, I think. The skeptical despise believers like yours truly...they know better. That way they forget that this is an empirical hobby.
I think I'll try to DIY this circuit also for my other DAC's.
Thanks for sharing.
M
Glad to hear that. I have EC's USB/DI2S but I am reserving it for the DI8*4 DAC. SMD should be better but will increase kit's price
There's the main problem, I think. The skeptical despise believers like yours truly...they know better. That way they forget that this is an empirical hobby.
I think I'll try to DIY this circuit also
for my other DAC's.Thanks for sharing.
M
Hi Brent Welke,
You can find a lot of discrete OP-amp schematics on the internet. I started experimenting with a simple all JFET design with 0nly 7 JFETs, similar to the Forssell Class A JFET OP-amps. It sounds open and natural, but I could still hear distortion. So the first mod was adding a constant power input stage. Next I added a current mirror, and a cascode current sources.
DC offset remains a bit problematic with these designs, so I decided to add a DC servo. This discrete OP-amp can be also converted to a quite impressive MOSFET power amplifier by adding only a few components.
I will post some schematics of both the experimental discrete OP-amp and power amplifier soon.
The 48 MHz shiftregister reclocker masterclock doesn't need to be synchronized to the PCM2706/7 12 MHz clock in order for the reclocker to function correctly. The 12 MHz crystal oscillator is only used for USB communications and has little or no effect on jitter. PCM2706/7 BCK / SCK clock signal is generated by a built-in SpAct (Sampling period Adaptive controlled tracking) PLL, similar to the circuit used in some SPDIF receivers.
However, I rather have one crystal oscillator than two that could intermodulate. With the DI 8M I get slightly better perceived sound quality when using just a single 48 MHz oscillator with a divide-by 4 circuit for the PCM2706.
The shiftregister reclocker can basically do the same with ALL I2S sources, so a jittery CS8412 I2S output can be reclocked in the same way.
Yes this is correct, another advantage of this reclocker is that standard 48 MHz low jitter (<1ps) / low cost crystal oscillator modules can be used, without the need for superclocks.
When the reclocker is placed close to the DAC chip(s), you have similar performance as with a slaved transport.
By the way, I (sort of) completed my new audio set, well you know how it is with diy projects, there is always room for improvement. I added a photograph of it. It consists of two 80W rms monoblock amplifiers (two outer modules), a passive volume and input selection unit (with remote control), and of course the DI 8M. The remote control housing consists of a piece of CNC machined teak wood and some anodized aluminum sheets. The keys are made of clear plastic with engraved symbols.
You can find a lot of discrete OP-amp schematics on the internet. I started experimenting with a simple all JFET design with 0nly 7 JFETs, similar to the Forssell Class A JFET OP-amps. It sounds open and natural, but I could still hear distortion. So the first mod was adding a constant power input stage. Next I added a current mirror, and a cascode current sources.
DC offset remains a bit problematic with these designs, so I decided to add a DC servo. This discrete OP-amp can be also converted to a quite impressive MOSFET power amplifier by adding only a few components.
I will post some schematics of both the experimental discrete OP-amp and power amplifier soon.
The 48 MHz shiftregister reclocker masterclock doesn't need to be synchronized to the PCM2706/7 12 MHz clock in order for the reclocker to function correctly. The 12 MHz crystal oscillator is only used for USB communications and has little or no effect on jitter. PCM2706/7 BCK / SCK clock signal is generated by a built-in SpAct (Sampling period Adaptive controlled tracking) PLL, similar to the circuit used in some SPDIF receivers.
However, I rather have one crystal oscillator than two that could intermodulate. With the DI 8M I get slightly better perceived sound quality when using just a single 48 MHz oscillator with a divide-by 4 circuit for the PCM2706.
The shiftregister reclocker can basically do the same with ALL I2S sources, so a jittery CS8412 I2S output can be reclocked in the same way.
Yes this is correct, another advantage of this reclocker is that standard 48 MHz low jitter (<1ps) / low cost crystal oscillator modules can be used, without the need for superclocks.
When the reclocker is placed close to the DAC chip(s), you have similar performance as with a slaved transport.
By the way, I (sort of) completed my new audio set, well you know how it is with diy projects, there is always room for improvement. I added a photograph of it. It consists of two 80W rms monoblock amplifiers (two outer modules), a passive volume and input selection unit (with remote control), and of course the DI 8M. The remote control housing consists of a piece of CNC machined teak wood and some anodized aluminum sheets. The keys are made of clear plastic with engraved symbols.
Attachments
DI DAC presented on HI-FI 2007 / DI DAC mods
Hi maxlorenz,
Sorry for the late reply, I plan to demonstrate my new audio set on HI-FI 2007 in Belgium next week:
http://hifishows.be/content-12.html
So I spend a lot of time boosting performance of my audio set. The modifications are now down to component level, most significant improvements were made this month. I will share these mods, as I think they are very important to achieve ultimate (DI DAC) performance.
It all started with the Vishay S102J resistors in the I/V stage. It seems that low temperature coefficient, non-inductive and non-magnetic resistors perform best. The top performer is the Vishay VSMP bulk metal foil precision resistor with only 0.2ppm !
I made a list of popular resistors, based on temperature coefficient for comparison:
1) Vishay VSMP bulk metal foil precision, (non-inductive), 0.2ppm !
2) Vishay SMR1D series bulk metal foil (non-inductive), 0.6ppm
3) Vishay S102J bulk metal foil precision (non-inductive), 15ppm
4) Rhopoint wirewound (inductive), 15ppm
5) Arcol MRA0207 0.1%, tyco YR1, RC55Y, 15ppm
6) Mills MRA5, MRA12 (non-inductive), 20ppm
7) Vishay SFM Tantalium (nitride), Shinkoh tantalium? , 25ppm
8) Caddock MK132, MS223 (non-inductive), 50ppm
9) Metal film (different brands), 50ppm
10) Caddock MK132 0.1%, 50...80ppm
11) Wirewound (different brands), 75ppm
12) Holco H2, H4P, H4, H8, 100ppm
13) PRP precision (non-magnetic), 100ppm
14) Thick film resistors, 100...150ppm
15) Metal oxide resistors, power metal film, 300...350ppm
16) Carbon composition CBT50, -200...+600ppm
17) Carbon ceramic (Riken ohm?) CBT25, -400...+700ppm
I personally prefer Vishay bulk metal foil (Z-foil) resistors for critical applications like feedback loops and input circuits, and the Arcol MRA0207 / tyco YR1 with 15ppm for less critical applications.
I already mentioned forcing LM4562 into class A. The best performance by far is achieved with a 3.3mA cascode current source (JFET). It pushes THD way below 0.00003%. These low THD values are almost impossible to achieve with discrete OP-amps. The best discrete OP-amps still produce around 0.001% THD
Transformer (I/V) circuits are champions in producing distortion (approx. 1% !!). I ran some tests with these recently, they cause extreme sound colouration, attenuate trebles and have significant difficulties with (subsonic) bass. Subsonic bass response is necessary to get realistic impact (slam), like with a symphony orchestra or drum solo for example.
Next mod was the coupling capacitors, I decided to try some "transparent" audiophile capacitors, so I bought Auricaps. This has resulted in the biggest downgrade of my audio set since years, the advised 200 hours burn-in didn't help either. In my personal opinion Auricaps aren't transparent at all, they boost trebles, attenuate midrange and boost bass. In the process they also add plenty of (phase) distortion. They can't handle (high) power, so they are not very suitable for passive speaker crossovers either.
My last hope was using it as tubediff coupling cap, but even there it managed to completely wreck audio performance. Now I am temporarily using an Intertechnik Audyn polypropylene cap with 4.7nF silvered mica bypass, this combination outperforms the Auricaps with the greatest ease. Auricap's can't be improved by using bypassing caps, tried that in a last desperate attempt, it even made matters much worse.
The best solution so far (that actually does sound very similar to DC coupling, without burn-in) are the Intertechnik Audyn polypropylene plus, and the Intertechnik Audyn TIN foil capacitors (tweeter) with a 10nF polystyrene or 4.7nF silvered mica bypass cap. The reason why Audyn polypropylene plus performs so well is because it can handle extreme power / impulses, and has an induction-free winding. The tin foil (tweeter filter) has high mass, and reduces movement (resonance) of the foil layers. This reduces distortion, it's especially suitable for higher frequencies.
The next spectacular mod was using a CTX315LVCT-ND (Digikey) 48 MHz SMD oscillator module with less than 1ps ! jitter. Needless to say what impact such extreme low jitter values have on sound quality. I used a rather unusual construction to reduce external resonances affecting oscillator performance.
I added a photograph of this mod. The DI 8M now performs like never before. My CD transports / changers have now become obsolete.
Now I am concentrating on interlinks, these have a major impact on sound quality (only when the audio set is reaching top performance). The usual screened cables sound muddy, grainy, and take away a lot of detail. Even worse is that large "audiophile" RCA connectors seem to introduce higher Eddy currents, adding extra distortion. I severely underestimated the importance of interlink quality, in fact they are just as critical as coupling caps.
If you want to construct a transparent, high-resolution interlink, special attention has to be paid to the effect of Eddy currents. The induced alternating magnetic fields, cause losses in the conductive material (copper), as usual these losses vary with different frequencies. This results in added noise and distortion.
At the moment I am using a rather extreme construction of a semi-balanced litz wire interlink. It consists of two identical litz wires (72 insulated wires for each conductor), all hand-made. The individual wires must be magnetically coupled. The idea is to significantly reduce the AC current in each wire, also reducing Eddy currents. The alternating magnetic field is now spread evenly over the litz wire cross section, instead of concentrating it in a single conductor core. Positive side effect is the extreme low inductance.
Each conductor is placed in a teflon tube (RS components), the teflon tubes are then placed in a braided copper wire screen (I used coax cable screening). This screen is only connected to GND on one side. The impact is simply staggering, detail is increased, there is absolutely no grain, and the music flows like liquid. The dynamic range is also extended. The impact is very similar to different coupling capacitor types, so even a plain copper wire isn't neutral / transparent at all.
My conclusion is that in order to achieve ultimate performance, utmost transparency / neutrality is required. Sound colouring should be avoided at all times.
I don't use it any more, it seemed to limit dynamics. You can find more information about thermal memory, and some practical circuits here:
http://peufeu.free.fr/audio/memory/memory-4-circuits.html
Hi maxlorenz,
Sorry for the late reply, I plan to demonstrate my new audio set on HI-FI 2007 in Belgium next week:
http://hifishows.be/content-12.html
So I spend a lot of time boosting performance of my audio set. The modifications are now down to component level, most significant improvements were made this month. I will share these mods, as I think they are very important to achieve ultimate (DI DAC) performance.
It all started with the Vishay S102J resistors in the I/V stage. It seems that low temperature coefficient, non-inductive and non-magnetic resistors perform best. The top performer is the Vishay VSMP bulk metal foil precision resistor with only 0.2ppm !
I made a list of popular resistors, based on temperature coefficient for comparison:
1) Vishay VSMP bulk metal foil precision, (non-inductive), 0.2ppm !
2) Vishay SMR1D series bulk metal foil (non-inductive), 0.6ppm
3) Vishay S102J bulk metal foil precision (non-inductive), 15ppm
4) Rhopoint wirewound (inductive), 15ppm
5) Arcol MRA0207 0.1%, tyco YR1, RC55Y, 15ppm
6) Mills MRA5, MRA12 (non-inductive), 20ppm
7) Vishay SFM Tantalium (nitride), Shinkoh tantalium? , 25ppm
8) Caddock MK132, MS223 (non-inductive), 50ppm
9) Metal film (different brands), 50ppm
10) Caddock MK132 0.1%, 50...80ppm
11) Wirewound (different brands), 75ppm
12) Holco H2, H4P, H4, H8, 100ppm
13) PRP precision (non-magnetic), 100ppm
14) Thick film resistors, 100...150ppm
15) Metal oxide resistors, power metal film, 300...350ppm
16) Carbon composition CBT50, -200...+600ppm
17) Carbon ceramic (Riken ohm?) CBT25, -400...+700ppm
I personally prefer Vishay bulk metal foil (Z-foil) resistors for critical applications like feedback loops and input circuits, and the Arcol MRA0207 / tyco YR1 with 15ppm for less critical applications.
I already mentioned forcing LM4562 into class A. The best performance by far is achieved with a 3.3mA cascode current source (JFET). It pushes THD way below 0.00003%. These low THD values are almost impossible to achieve with discrete OP-amps. The best discrete OP-amps still produce around 0.001% THD
Transformer (I/V) circuits are champions in producing distortion (approx. 1% !!). I ran some tests with these recently, they cause extreme sound colouration, attenuate trebles and have significant difficulties with (subsonic) bass. Subsonic bass response is necessary to get realistic impact (slam), like with a symphony orchestra or drum solo for example.
Next mod was the coupling capacitors, I decided to try some "transparent" audiophile capacitors, so I bought Auricaps. This has resulted in the biggest downgrade of my audio set since years, the advised 200 hours burn-in didn't help either. In my personal opinion Auricaps aren't transparent at all, they boost trebles, attenuate midrange and boost bass. In the process they also add plenty of (phase) distortion. They can't handle (high) power, so they are not very suitable for passive speaker crossovers either.
My last hope was using it as tubediff coupling cap, but even there it managed to completely wreck audio performance. Now I am temporarily using an Intertechnik Audyn polypropylene cap with 4.7nF silvered mica bypass, this combination outperforms the Auricaps with the greatest ease. Auricap's can't be improved by using bypassing caps, tried that in a last desperate attempt, it even made matters much worse.
The best solution so far (that actually does sound very similar to DC coupling, without burn-in) are the Intertechnik Audyn polypropylene plus, and the Intertechnik Audyn TIN foil capacitors (tweeter) with a 10nF polystyrene or 4.7nF silvered mica bypass cap. The reason why Audyn polypropylene plus performs so well is because it can handle extreme power / impulses, and has an induction-free winding. The tin foil (tweeter filter) has high mass, and reduces movement (resonance) of the foil layers. This reduces distortion, it's especially suitable for higher frequencies.
The next spectacular mod was using a CTX315LVCT-ND (Digikey) 48 MHz SMD oscillator module with less than 1ps ! jitter. Needless to say what impact such extreme low jitter values have on sound quality. I used a rather unusual construction to reduce external resonances affecting oscillator performance.
I added a photograph of this mod. The DI 8M now performs like never before. My CD transports / changers have now become obsolete.
Now I am concentrating on interlinks, these have a major impact on sound quality (only when the audio set is reaching top performance). The usual screened cables sound muddy, grainy, and take away a lot of detail. Even worse is that large "audiophile" RCA connectors seem to introduce higher Eddy currents, adding extra distortion. I severely underestimated the importance of interlink quality, in fact they are just as critical as coupling caps.
If you want to construct a transparent, high-resolution interlink, special attention has to be paid to the effect of Eddy currents. The induced alternating magnetic fields, cause losses in the conductive material (copper), as usual these losses vary with different frequencies. This results in added noise and distortion.
At the moment I am using a rather extreme construction of a semi-balanced litz wire interlink. It consists of two identical litz wires (72 insulated wires for each conductor), all hand-made. The individual wires must be magnetically coupled. The idea is to significantly reduce the AC current in each wire, also reducing Eddy currents. The alternating magnetic field is now spread evenly over the litz wire cross section, instead of concentrating it in a single conductor core. Positive side effect is the extreme low inductance.
Each conductor is placed in a teflon tube (RS components), the teflon tubes are then placed in a braided copper wire screen (I used coax cable screening). This screen is only connected to GND on one side. The impact is simply staggering, detail is increased, there is absolutely no grain, and the music flows like liquid. The dynamic range is also extended. The impact is very similar to different coupling capacitor types, so even a plain copper wire isn't neutral / transparent at all.
My conclusion is that in order to achieve ultimate performance, utmost transparency / neutrality is required. Sound colouring should be avoided at all times.
I don't use it any more, it seemed to limit dynamics. You can find more information about thermal memory, and some practical circuits here:
http://peufeu.free.fr/audio/memory/memory-4-circuits.html
