Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

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

Building the ultimate NOS DAC using TDA1541A
Building the ultimate NOS DAC using TDA1541A
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 25th July 2007, 05:26 PM   #1611
ecdesigns is offline ecdesigns  Netherlands
diyAudio Member
 
Join Date: May 2006
Location: Holland
Hi Terry Demol,


Quote:
How did you measure the jitter of the 74HC74 FF's? The jitter
induced by one of these, if correctly decoupled, should be very low.
I measured jitter with a 60 MHz oscilloscope, I reclocked a 2.8224 MHz BCK signal using a 48 MHz master clock. Timing errors (jitter) show up as (random) displacements of signal transients in the time domain. When timing errors (jitter) is very low, al subsequent clock signal transients are perfectly aligned, resulting in a single visible transient.

The 74xx74 chip was part of a shiftregister reclocker with a 48 MHz master clock. Decoupling: 100uF electrolytic capacitor, and 100nF very close to the chip power supply. Unused inputs connected to GND/VCC. The HF path was kept as short as possible by using a power supply series resistor.


Quote:
Why do figure the shift register of same logic family - say HC, has
lower jitter?
I already indicated this, it's probably caused by chip internal lay-out (location of traces / power supply routing to each separate MOSFET). MOSFETS draw peak currents during signal transients, causing short current surges. It's very likely that these peak currents affect circuit operation (phase shifts / crosstalk) depending on the chip lay-out.
  Reply With Quote
Old 25th July 2007, 05:37 PM   #1612
ecdesigns is offline ecdesigns  Netherlands
diyAudio Member
 
Join Date: May 2006
Location: Holland
Default Jitter measurement

Hi tubee,


Quote:
I managed to check an buffered output of diy 27Mhz clock for jitter with a humble but here available 20 Mhz scope, the signal is finer visible on the screen, on cristal itself the signal is more blurry. I even managed to look 36mhz with this scope.
Yes you will be able to measure jitter on a 20 MHz scope, I assume you expanded the timebase with the x10 setting. So each scale division represents 1/200,000,000 = 5nS.

I am using a 60 MHz scope with x10 timebase setting, resulting in 1/600,000,000 = 2nS or 2000ps. Each sub-division (approx. 1mm) represents 2000 / 5 = 400ps. With a low intensity / good focus setting, you can easily subdivide this by 10 (0.1mm), similar to reading a Vernier calliper. This will allow for measuring jitter peak-to peak amplitudes down to 40ps (14ps rms) or better. Then there is the trace brightness, if all clock transients align perfectly, trace brightness appears higher then when they are slightly spread. So a brighter (signal transient) trace also indicates lower jitter amplitude.

20 MHz scope, 100ps (35ps rms)
40 MHz scope, 60ps (21ps rms)
60 MHz scope, 40ps (14ps rms)
100 MHz scope, 20ps (7ps rms)
  Reply With Quote
Old 25th July 2007, 09:10 PM   #1613
Brent Welke is offline Brent Welke  Canada
diyAudio Member
 
Join Date: Jun 2004
Location: Victoria, BC
Default The reclocker is wonderful

The reclocker is wonderful. I can't listen without it now.

I have been using this reclocker for a couple weeks now on a couple single TDA1543 chip USB DACs. I added a jumper so that I can switch between reclocker on/off while the music is playing.

This is my first experience with how jitter sounds. It took a couple weeks for me to fully appreciate and become accustom to the difference between the two settings. I didn't want to fall into the trap where a difference in sound seems to be a step forward when in time it becomes apparent it is not.

Reclocked the sound is much cleaner sounding and brings more life to the music. The music even sounds louder. How jitter sounds is well described here http://www.jitter.de/english/engc_navfr.html

Included is a picture of how I implemented it. The clock beneath the stack of 4 SMD chips is the 48MHz clock with 1ps jitter. This used to be a 12MHz clock. I have at least a dozen DACs to modify. I will experiment with using two clocks (leaving the 12MHz clock alone) to see what affect that has. That would be a simpler mod then replacing the clock.

A few pages back I posted the Digikey parts list I used if anyone is interested in trying it on their DAC.

The reclocker makes a very welcome improvement to the sound. Now I cannot help but think is there even more improvement waiting to be had in reclocking.

I wish I had the test equipment to see exactly what is going on. However I can visualize the reclocker is not ideal. This is not news but I would like more discussion about it and so I go on about it...

The raw BCK and the master clock (48MHz) are not in sync and not exact multiples. The master clock is almost 17 times BCK's frequency. Reclocked BCK now falls on one of the 17 master clock transitions. The majority of the time 17 master clock cycles will pass between BCK pulses but as the clocks beat there will be times when the reclocker jumps to 16 or 18 clock cycles and then continues to be 17 until the next time they beat. This would add 1/48MHz jitter to that sample.
Since the raw BCK is jittery this event would likely affect a few BCK samples before proper reclocking is resumed.

Listening to the reclocker is proof that it spends the majority of its time providing low jitter clocking. At no time can I hear the moment(s) the reclocker is adding jitter. It either happens too often, too quickly, once an hour or something like that.

Perhaps someone with a quick mathematical mind can provide some numbers to the amount of jitter that occurs when the reclocker beats, how long it beats and the percentage of time reclocker is working correctly.
Attached Images
File Type: jpg reclocker.jpg (90.2 KB, 1504 views)
  Reply With Quote
Old 26th July 2007, 08:25 AM   #1614
Terry Demol is offline Terry Demol  Australia
diyAudio Member
 
Terry Demol's Avatar
 
Join Date: Apr 2002
Location: *
Quote:
Originally posted by -ecdesigns-
Hi Terry Demol,

I measured jitter with a 60 MHz oscilloscope, I reclocked a 2.8224 MHz BCK signal using a 48 MHz master clock. Timing errors (jitter) show up as (random) displacements of signal transients in the time domain. When timing errors (jitter) is very low, al subsequent clock signal transients are perfectly aligned, resulting in a single visible transient.

The 74xx74 chip was part of a shiftregister reclocker with a 48 MHz master clock. Decoupling: 100uF electrolytic capacitor, and 100nF very close to the chip power supply. Unused inputs connected to GND/VCC. The HF path was kept as short as possible by using a power supply series resistor.

Hi EC,

2.8224MHz and 48MHz arent syncronous or integer multiples
so I don't quite understand how this can work properly.

I would imagine you will be generating sidebands and
running into DFF setup / timing issues.

Maybe this is the 'jitter' that you are seeing?

cheers

Terry
  Reply With Quote
Old 31st July 2007, 02:33 AM   #1615
dsiroky is offline dsiroky  United States
diyAudio Member
 
Join Date: Mar 2007
Default My (almost completed) DI16

Attached are some pictures of my almost completed DI-16.

I used a left over piece of oak I had to make the case, and plan to fold some perforated stainless steel over the top as a cover that will provide ventilation.

I haven't hooked up the balanced outputs yet.

I also need to hook up a small USB cable to the pass through USB connector before I can close it up.
  Reply With Quote
Old 31st July 2007, 02:41 AM   #1616
dsiroky is offline dsiroky  United States
diyAudio Member
 
Join Date: Mar 2007
Default My (almost completed) DI16

Attached are some pictures of my almost completed DI-16.

I used a left over piece of oak I had to make the case, and plan to fold some perforated stainless steel over the top as a cover that will provide ventilation.

I haven't hooked up the balanced outputs yet.

I also need to hook up a small USB cable to the pass through USB connector before I can close it up.
Attached Images
File Type: jpg img_0892.jpg (80.1 KB, 1373 views)
  Reply With Quote
Old 31st July 2007, 02:42 AM   #1617
dsiroky is offline dsiroky  United States
diyAudio Member
 
Join Date: Mar 2007
Default pic #2

second picture
Attached Images
File Type: jpg img_0893.jpg (46.1 KB, 1236 views)
  Reply With Quote
Old 31st July 2007, 02:46 AM   #1618
dsiroky is offline dsiroky  United States
diyAudio Member
 
Join Date: Mar 2007
Default pic #3

picture 3
Attached Images
File Type: jpg img_0896.jpg (90.8 KB, 1290 views)
  Reply With Quote
Old 31st July 2007, 03:39 PM   #1619
ecdesigns is offline ecdesigns  Netherlands
diyAudio Member
 
Join Date: May 2006
Location: Holland
Default Jitter / ripple

Hi Terry Demol,


Quote:
2.8224MHz and 48MHz arent syncronous or integer multiples
so I don't quite understand how this can work properly.

I would imagine you will be generating sidebands and
running into DFF setup / timing issues.

I ran tests with both 48 and 47.9808 MHz, the oscilloscope shows very little difference between the BCK signal reclocked by either 48 or 47.9808 MHz.

The sound quality (sound depth, transparency, details, midrange and bass) I am getting with the DI 8M after the latest mods (Vishay bulk metal foil I/V resistors and 3.3mA bias current for the LM4562), show that jitter isn't causing any audible problems whatsoever.

More disturbing is the fact that sample amplitude ripple during the time the sample is output, causes similar problems as jitter. If you managed to get jitter values very low, you're only halfway.

The TDA1541A is one of the very few chips that have decoupling caps on the active current divider outputs. This enables the possibility to achieve a very low ripple signal on the current output. By increasing both DEM clock rate and decoupling capacitor values, the ripple voltage on the output can be reduced to very low levels.


I think that the jitter frequency components (the way the clock signal phase is modulated) have the biggest impact. If we take a perfect clock signal (square-wave) with zero jitter, all subsequent signal transients are perfectly time-aligned. When viewed on the oscilloscope, we should see a clear square-wave signal with very clean signal transients.

So I would say that it's most important to have as many clock signal transients time-aligned as possible. When a clock signal is phase modulated with noise, almost all signal transients take place with slightly different time intervals, and are by no means time-aligned. The clock signal shows hazy transients on the oscilloscope, the exact clock signal transients can no longer be determined.

With asynchronous reclocking, using two cascaded D-flip flops, we get two time displaced clock signal transients, both have the same brightness when viewed on the oscilloscope. So approx. 50% of the transients are time aligned in one "cluster" the other 50% in the second "cluster". Since the transients within a cluster are still perfectly aligned, both clock signal transients can still be determined accurately. Despite of the much higher time displacement, this clock signal will actually produce a sound quality that doesn't suggest high jitter is present.

The shiftregister reclocker is basically a digital one-shot circuit that generates a well defined pulse after each positive going edge of BCK. The pulsewidth is always exactly 7 masterclock pulses wide (BCK input = 2.8224 MHz / masterclock = 48 MHz). In order to maintain sync with the BCK input signal, the time between two subsequent fixed pulses may vary, but since both BCK and the masterclock are almost perfect multiples, there are few corrections needed. The shiftregister reclocker is also able to phase-lock to the BCK input signal, within the available time resolution.

The BCK output of the shiftregister reclocker looks like a single square-wave on the oscilloscope (unlike an asynchronous reclocked BCK signal). When the brightness is turned-up, a very dim second clock signal transient becomes visible, this is caused by the sporadic corrections needed to maintain perfect sync.
  Reply With Quote
Old 1st August 2007, 04:19 PM   #1620
maxlorenz is online now maxlorenz  Chile
diyAudio Member
 
maxlorenz's Avatar
 
Join Date: Oct 2003
Location: osorno , Chile
Hi -ecdesigns-,

The built of my new 8*4TDA1543 DAC is at standby because of 100nf caps out of stock

My computer expert advisor installed me OSX (MAC) on my PC instead of Ubuntu because he lacked experience on the later. Sound improvement is significant compared to XP.
What software do you recommend to listen and rip music? (lossless format)

Thanks,
M
  Reply With Quote

Reply


Building the ultimate NOS DAC using TDA1541AHide 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

Forum Jump


New To Site? Need Help?

All times are GMT. The time now is 06:59 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio
Wiki