Building the ultimate NOS DAC using TDA1541A - Page 360 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Line Level

Digital Line Level DACs, Digital Crossovers, Equalizers, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 9th February 2011, 08:35 PM   #3591
diyAudio Member
 
Join Date: Apr 2008
Quote:
Originally Posted by -ecdesigns- View Post
Hi maxlorenz,


Thanks for your reply,






The HDD "quality" doesn't matter at all here, since data is first buffered in RAM (Random Access Memory). When you play music trough a pc / mac, you are basically reading it from a semiconductor memory rather than a HDD. The HDD only fills the RAM buffer, it has plenty of time to do this. SATA connection stands for Serial ATA, it's a serial interface type that allows for faster data transfers.

To avoid quality loss, use either WAV or some type of lossless compression. WAV files take up more HDD space than the ones using lossless compression.

I use a mac music player with the Apple Lossless format. With ubuntu you could use flac lossless compression. There are many other different lossless formats like ALAC, FLAC, LA, LPAC, APE, MPEG-4 SLS, OFR, RAL, SHN, TTA, VW, WMAL and ADA.

Before you start storing CD images, it's a good idea to pick the lossless compression format that's best suited for the application you intend to use.




With better power supply, you could think of a higher quality power supply with a very silent cooling system. Most cheap pcs have crappy power supplies that are just capable of supplying enough current.

The pc motherboard caps are a different story, it seems at some time electrolytic capacitors with wrong electrolyte where used. These caps pop-open and loose their capacitance over time. It starts with a lot of high-pitched noise from the motherboard DC converters, and start-up problems. In the end both power supply and motherboard could fail. If you have a pc motherboard with these electrolytic capacitors, it's a good idea to replace them before they could cause problems.

You could also use high quality caps / decoupling close to the USB connector power supply (5V), so the connected USB / I2S / SPDIF devices receive a cleaner power supply.




Use System > Administration > network.

Ubuntu usually detects everything fully automatic, (I am using the latest Feisty Fawn version). If the internet is not "recognized" automatically, you may need to add a router.




What I also meant is that the DI 16 manages to provide similar sound improvements on completely different audio sets.

As for interlink materials / constructions, these both do have significant effect on sound quality. Interlinks need to transport ac signals without over-emphasizing different frequencies (flat frequency response). This is much more difficult than most people would think. Interlinks have a specific inductance, Eddy currents cause frequency-dependent losses in the interlink material, and conductive materials placed close to the interlink.

Eddy currents could also cause a lot of problems is speaker crossovers, and sound quality is accordingly.
Hello John

Would these newer media transports such as WD Mini , Popcorn Hour C-200 etc etc (they use realtek and Sigma based integrated CPU/GPU) Would they be any good to use a transport compared to a fully optimised PC ?

Regards Paul
  Reply With Quote
Old 10th February 2011, 09:13 AM   #3592
phi is offline phi  Sweden
diyAudio Member
 
Join Date: Feb 2002
Location: Stockholm, Sweden
Hello John,

Some questions about your latest schematics, the Mk6 version. It's fun to follow your progress, constant changes, never boring.

1. You did use extreme LC-filtering for the oscillator, up to 20:th order or something, and now just a 40 ohm, resistor, quite a change, why !?

2. This 40ohm, with 1uF will give approx. Fc of 4kHz. It is the same 5V feeding the 500ohm output resistor. Are you not modulating the oscillator with the analog output?

2. The flip-flops for WS, MCK, DATA are now feed by a 680ohm or 1.5k resistors from 5V, how does this work ?

4. 1uF ceramics for decoupling everywhere, analog as well as digital?
(except the 47nF's for the TDA1541)

//Per
  Reply With Quote
Old 11th February 2011, 12:59 AM   #3593
diyAudio Member
 
Join Date: May 2006
Location: Holland
Hi phi,

Quote:
1. You did use extreme LC-filtering for the oscillator, up to 20:th order or something, and now just a 40 ohm, resistor, quite a change, why !?
This filtering was an attempt to get rid of 78XX regulator large bandwidth noise. It didn't help much, once the interference is generated it is almost impossible to fully block it. So I tackled the problem by the source by using high resolution, large bandwidth, ultra low noise discrete voltage regulators.

Quote:
2. This 40ohm, with 1uF will give approx. Fc of 4kHz. It is the same 5V feeding the 500ohm output resistor. Are you not modulating the oscillator with the analog output?
All bit currents (current flowing through passive I/V resistor, and unused bit currents) flow back into +5V. The MK6 has no active output stage, the output voltage is generated across a 500 Ohm I/V resistor. The power amp "measures" the ac voltage across this I/V resistor. This means that ripple voltage due to the audio signal is extremely low and no problem for the masterclock.

The RC filter offers sufficient attenuation of the 11.2896 MHz clock signal. The 40 Ohm Mobius Honeycomb wire wound resistor adds very little noise, so inter-modulating noise spectra are minimized.

The +5V power supply is taken out of the signal path as I use +5V (or a derived voltage) as reference. The TDA1541A keeps the output current through the passive I/V resistor constant (high PSRR). So we basically have a clean ac voltage across a 500R passive I/V resistor, even if the +5V power supply would be a bit noisy.

Quote:
2. The flip-flops for WS, MCK, DATA are now feed by a 680ohm or 1.5k resistors from 5V, how does this work ?
I used these resistors in order to provide high series impedance over a very large bandwidth while keeping the circuit as simple as possible. Ferrite beads have problems with core saturation (DC current) and hysteresis.

The reclockers drive I2S attenuators, design is such that the reclocker only has to drive little over 12pF load. The signal amplitude equals 1.2Vpp instead of the usual 5Vpp. This not only reduces ground-bounce, it also reduces peak currents during signal transitions.

DATA and WS reclockers consume approx. 1.7mA each during normal operation (the reclockers run on approx. 2.4V power supply). Decoupling caps are not used here in order to minimize peak currents during signal transients. These can occur in combination with the TDA1541A I2S input circuit stray capacitance.

This greatly reduces crosstalk between BCK (that needs to remain as clean as possible) and WS / DATA signals.

The flip-flops remain stable under these conditions. Since the DATA and WS signals only need to be sampled during the positive going edge of BCK, fast transients are not required.

BCK is responsible for sample timing and needs to have fast transients. Because of the higher switching frequency and little over 12pF load, it consumes 3.8mA so 680 Ohm power supply series resistor is sufficient. Here a charge compensated 1uF SMD 1210 size film decoupling cap is used to supply required energy for fast transients. Charge compensation improves (film) cap small signal handling.

Quote:
4. 1uF ceramics for decoupling everywhere, analog as well as digital?
(except the 47nF's for the TDA1541)
Latest MK6 has (charge compensated) 1uF 1210 size SMD film decoupling caps. Active divider decoupling caps are now 330nF 1206 size SMD film caps. Circuit is built on HF board and TDA1541A is mounted on an "open frame" socket in order to increase isolation resistance between the active divider decoupling pins.

I attached a photograph of this MK6.

For the digital section (SD-transport) I still use 1uF 1206 size X7R decoupling caps but I also plan to test 1uF film caps here.
Attached Images
File Type: jpg mk6-hf.jpg (98.9 KB, 1028 views)
  Reply With Quote
Old 11th February 2011, 09:09 AM   #3594
fff0 is offline fff0  Singapore
diyAudio Member
 
Join Date: Jul 2008
Quote:
Originally Posted by -ecdesigns- View Post
...............
The HDD "quality" doesn't matter at all here, since data is first buffered in RAM (Random Access Memory). When you play music trough a pc / mac, you are basically reading it from a semiconductor memory rather than a HDD. The HDD only fills the RAM buffer, it has plenty of time to do this. .............
Hello John,

Does playing CDs thru a PC has such benefit?
  Reply With Quote
Old 11th February 2011, 12:11 PM   #3595
phi is offline phi  Sweden
diyAudio Member
 
Join Date: Feb 2002
Location: Stockholm, Sweden
Hi John,

Thorough answer as always, but it generates more questions

Quote:
DATA and WS reclockers consume approx. 1.7mA each during normal operation (the reclockers run on approx. 2.4V power supply). Decoupling caps are not used here in order to minimize peak currents during signal transients.
There are 1uF in the schematics (C29, C30) or are you refering to something else? I get a sense that you are talking about yet another unpublished schematic of the Mk6 version?

Quote:
BCK is responsible for sample timing and needs to have fast transients. Because of the higher switching frequency and little over 12pF load, it consumes 3.8mA so 680 Ohm power supply series resistor is sufficient.
My assumption was that you were 'balancing' the supply voltage for the flip-flops with the resistors, but it did not make sense with your published shematic. It has an error in the resistor values, the 680 ohm is on the WS but should be on BCK, ok fine, makes sense now.

Quote:
Charge compensation improves (film) cap small signal handling.
What is "Charge compensation" ?

Quote:
Circuit is built on HF board and TDA1541A is mounted on an "open frame" socket in order to increase isolation resistance between the active divider decoupling pins.
Why do you never solder the TDA1541 chip directly to the board? Are you making too many changes all the time?

Keep on the good work John. It gets simpler and simpler and I like that.

//Per
  Reply With Quote
Old 12th February 2011, 01:08 PM   #3596
diyAudio Member
 
Join Date: May 2006
Location: Holland
Hi phi,

Quote:
There are 1uF in the schematics (C29, C30) or are you refering to something else? I get a sense that you are talking about yet another unpublished schematic of the Mk6 version?
Yes I was talking about a MK6 that's assembled differently as a result of fine tuning.

Quote:
It has an error in the resistor values, the 680 ohm is on the WS but should be on BCK, ok fine, makes sense now.
Seems I didn't fix this error in the schematic, thanks for pointing this out.

Quote:
What is "Charge compensation" ?
Capacitors appear to have a charge "threshold". This means that applied charge on one capacitor plate has to exceed a certain threshold level before the other capacitor plate "follows". Compare it with a kind of mild cross-over distortion or electronic backlash. The audible effect is distortion and loss of resolution. Transformers offer better handling of low level signals, but have issues with higher signal amplitude.

The compensation consists of a carefully chosen bypass resistor that passes very small ac signal changes directly to the second plate.

Quote:
Why do you never solder the TDA1541 chip directly to the board? Are you making too many changes all the time?
Easy DAC chip swapping for testing different chips. Verifying correct circuit operation prior to installing the DAC chip. Easy trouble shooting (damaged DAC chip). Easy access to the parts that are mounted directly underneath the DAC chip such as decoupling caps. Cooling, the socket allows airflow underneath the chip and helps to keep chip temperature more constant. Varying chip temperature affects bit accuracy (passive dividers). Avoiding thermal stress that would occur during soldering. This thermal stress could affect chip (long term) accuracy and reliability.

Quote:
It gets simpler and simpler and I like that.
The DAC evolved from a rather complicated design with 8 DAC chips and tube output stage to this simple straight-forward design. During development the main objective was best approximation of live sound quality.

Every part adds distortion because it isn't perfect. The fewer parts, the less distortion is added, and fewer flaws need fixing.
  Reply With Quote
Old 12th February 2011, 09:17 PM   #3597
phi is offline phi  Sweden
diyAudio Member
 
Join Date: Feb 2002
Location: Stockholm, Sweden
Thank you John

If you by 'Charge compensation' mean 'damper', a resistor i parallell with the capacitor that dampens high frequency ringing, then I follow you...

//Per
  Reply With Quote
Old 13th February 2011, 06:52 PM   #3598
diyAudio Member
 
Join Date: Jan 2006
Location: Italy
A little progress - I was able to reclock BCK and the two DATA lines from the SAA7000. I had to connect BCK at pin 5, not 3, and feed reclockers with this signal and then DATA output to DACs again form pin 5. But, unfortunately, no way to get WS working!

One question about DEM reclock. With the TDA1540s even with 820pF caps disconnected I don't get distorted sound (?) like the 1541, so it's difficult to undestand if reclock is working or not. Bad thing. I connected two 12K resistors after each BAT17 diode in order to feed both TDA1540s, then a 2K2 resistor goes to -17V on each DEM pin. Is this theoretically correct?
Is it possible to get instrumental confirmation that DEM reclock is working?

Thank you.

Last edited by fralippo; 13th February 2011 at 06:55 PM.
  Reply With Quote
Old 14th February 2011, 03:48 PM   #3599
diyAudio Member
 
Join Date: Jan 2006
Location: Italy
I got the oscilloscope today. Problem is that 74AUC74 does NOT output nothing similar to WS signal at input (probably nothing). And this 74 is confirmed to be working with the DATA signal!
What should I do next?
  Reply With Quote
Old 15th February 2011, 09:27 AM   #3600
oshifis is offline oshifis  Hungary
diyAudio Member
 
Join Date: Mar 2004
Location: Budapest, Hungary
Quote:
Originally Posted by -ecdesigns- View Post
Cooling, the socket allows airflow underneath the chip and helps to keep chip temperature more constant. Varying chip temperature affects bit accuracy (passive dividers). Avoiding thermal stress that would occur during soldering. This thermal stress could affect chip (long term) accuracy and reliability.
John, have you considered a more effective cooling of the DAC chip by using a heatsink on the top? I observed the TDA1541A gets rather hot during operation; cooling it surely has benefits... I use two copper heatsinks like on the picture, connected to the ground.
Attached Images
File Type: jpg Thermaltake.jpg (6.2 KB, 772 views)
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



New To Site? Need Help?

All times are GMT. The time now is 09:12 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2