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

PC Based Computer music servers, crossovers, and equalization

IanCanada's Latest RPi GB Goodies Impressions... and your tweaks, mods and hints...
IanCanada's Latest RPi GB Goodies Impressions... and your tweaks, mods and hints...
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 14th June 2019, 06:17 AM   #501
Markw4 is offline Markw4  United States
diyAudio Member
 
Join Date: Jul 2016
Location: California
Quote:
Originally Posted by iancanada View Post

How did you see a 100MHz clock looks like a sin wave?
There are lots of ES9038Q2M dacs that come with 100Mhz clocks. in some cases we have experimented with replacing the original clocks with Crystek 575 100MHz clocks. For one example: ES9038 Q2M I2S DSD Optical Coaxial Input Decoder DAC Assembled Board DC9-15V 8751762607403 | eBay

Of course with a 100MHz scope, a 100MHz clock will look like a sine wave anyway, or pretty close to that. The first harmonic frequency of a 100MHz square wave is at 300MHz, well above the maximum frequency of a 100MHz scope. However, even using a 600MHz scope, Crystek 575 waveforms don't look very square. The blue trace below is the output of a 100MHz Crystek 575 100MHz clock, and the yellow waveform is the MCLK signal from a Sabre dac exported on a GPIO pin. The Crystek clock looks better than some cheaper competitors, but still not all that square. The dac GPIO MCLK output looks pretty much like a sine wav. Of course, its possible if using active differential probes it would be possible to see somewhat better looking waveforms, but not in all cases. Some ultra-low jitter clock manufacturers include output waveforms in data sheets, and they rarely look all that square at very high frequencies. It probably would make more sense to have LVDS clocks and dacs if we ever get to DSD2048. Even DSD1024 is really pushing it with data being clocking on both clock edges, IMHO.
Attached Images
File Type: jpg Crystek_vs_GPIO_MCLK.jpg (177.1 KB, 274 views)

Last edited by Markw4; 14th June 2019 at 06:22 AM.
  Reply With Quote
Old 14th June 2019, 12:22 PM   #502
iancanada is offline iancanada  Canada
diyAudio Member
 
iancanada's Avatar
 
Join Date: Dec 2009
Location: Toronto
IanCanada's Latest RPi GB Goodies Impressions... and your tweaks, mods and hints...
@Markw4

What's the bandwidth of your oscilloscope? And what's the bandwidth of your probe?

Regards,
Ian
  Reply With Quote
Old 14th June 2019, 03:19 PM   #503
Markw4 is offline Markw4  United States
diyAudio Member
 
Join Date: Jul 2016
Location: California
Both scope and probe are rated at 600MHz bandwidth.
Risetime of Crystek 575 is rated at 2ns max 20%-80% Vdd.

A 100HMz clock has a period of 10ns, which is to say a half-cycle is 5ns. With a 2ns risetime, there might actually be 3ns when it is at or above 80%Vdd (or at or below 20%), and of course we expect some loss of detail on that with only a 600MHz scope and 600MHz probe combination. Better to have a 5GHz scope, or faster.

In the above regard Crystek clocks are pretty good.
NDK NZ2520SD which are available up 80Mhz are rated for a rise time of 6ns max.

Hopefully we will always be fortunate, and have much less than max risetimes

What concerns me about reclocking as it is normally done, is that clocks are delayed maybe by NB3L553 with as much as 35ps output skew (typical), with the skew possibly the main timing complication/uncertainty if one output drives the dac MCLK, and another output drives the reclocker flip flop. Then there are further delays (and sometimes unknown capacitive loading) by cable and PCB traces, and signals are typically delayed by something like Potato PO74G74A which is maybe 2ns if not loaded with a lot of capacitance, but unbuffered they are not really 50-ohm transmission like drivers. In terms of maxmum frequency Potato parts are sensitive to capacitive loading.

Making sure the whole thing really works as intended could get rather complicated. Perhaps more so when there are multiple boards and interconnects involved. Just one loose or otherwise imperfect RF connection could cause unintended and or erratic operation.

Last edited by Markw4; 14th June 2019 at 03:41 PM.
  Reply With Quote
Old 14th June 2019, 05:19 PM   #504
atom6422 is offline atom6422  South Korea
diyAudio Member
 
Join Date: Nov 2009
Default Nyquist Sampling Theorem and Stupid Question

Quote:
Originally Posted by Markw4 View Post

Of course with a 100MHz scope, a 100MHz clock will look like a sine wave anyway, or pretty close to that.
IMHO, I think 100MHZ scope can not detect 100MHZ square wave and sine wave based on nyquist sampling theorem and my experience on my 20MHZ analog scope and 500MHZ digital scope.

Could you please let me know how to detect 100MHZ signal using 100MHZ scope ?

Please overlook my stupid question, I have no theoretical background about electronics.
  Reply With Quote
Old 14th June 2019, 05:34 PM   #505
randytsuch is offline randytsuch
diyAudio Member
 
Join Date: Jan 2003
Location: LA
Answer is to take out a 2nd mortgage, and get a 5GHz scope and active probe

They are seriously expensive, so with the test equipment more people here would have I expect there to be doubt when you start looking at 100 MHz waveforms.
__________________
My system is here
http://randytsuch-audio.blogspot.com/2005/10/my-system.html
  Reply With Quote
Old 15th June 2019, 01:54 AM   #506
Markw4 is offline Markw4  United States
diyAudio Member
 
Join Date: Jul 2016
Location: California
Quote:
Originally Posted by Markw4 View Post
What concerns me about reclocking as it is normally done, is that clocks are delayed maybe by NB3L553 with as much as 35ps output skew (typical)
Sorry. Not sure what I was thinking when I wrote the above. 35ps is too little for most of what I was talking about. There is another dac project I have where I see sound quality sensitivity related to I2S timing on the scale of 100ps or less, but not an issue for what I was attempting to address earlier today.

Again, sorry if any confusion.
  Reply With Quote
Old Yesterday, 04:48 AM   #507
iancanada is offline iancanada  Canada
diyAudio Member
 
iancanada's Avatar
 
Join Date: Dec 2009
Location: Toronto
IanCanada's Latest RPi GB Goodies Impressions... and your tweaks, mods and hints...
Quote:
Originally Posted by Markw4 View Post
It depends. If computer software does any resampling for any reason, and sometimes operating systems such as Windows will do it without telling you, and or if resampling occurs in a player such as HQplayer, that processing can and will change the bits and potentially make quite a large difference in sound quality. At other times it can sound slightly bad somehow, but it may be less obvious that the bits have been changed.

Then there are the mechanisms that Greg mentioned. There is ground common mode noise that can turn into a ground currernt. There is HF and RF noise from various sources in a computer that can couple into signal and ground connections, etc. And, of course there is what ESS refers to as 'transport jitter.'

Some people have believed that galvanic isolation and 'reclocking' could eliminate all such issues, but as you have already heard, it is really an issue of attenuation of noise that occurs. How much attenuation is necessary? Basically, until you can't hear any of effects of it, and can't measure much or any of it remaining.

Optical isolation is good because it eliminates current flowing between devices, especially if the optical cable is long enough to keep sending and receiving devices far enough apart such that electromagnetic coupling between devices is also strongly attenuated.

However, optical interfaces increase jitter. If there is too much jitter, ASRC may be needed to recover the signal before it can even be run through a separate 'reclocker.'

I would also point out that most reclocker circuits are not designed to work well at very high data rates. That's because propagation delay effects between clocks, dacs, and I2S signals are not usually accounted for to close tolerances. Delays through chips such as clock buffers, D-flip flops, etc., and delays of cable lengths, etc., typically work out well enough time aligned at low to moderate data rates, but can be fairly skewed relative to the timing constraints of high sample rate data and clock frequencies.

In addition, audio clocked around 50MHz or 100Mhz using LVCMOS devices starts to look a lot more like sin waves than square digital pulses. That makes data setup and hold timing constraints very hard to assure, say, as data is clocked into a dac chip. It can also make it complicated to know if reclocking is really working as intended.
Hi Markw4,

Your 100MHz clock waveform doesnít look that correct. I cannot see a clear rising edge or falling edge, as well as ringing, overshoot and reflections. Looks like the bandwidth is not sufficient though you said you oscilloscope has 600MHz bandwidth. Just hope you can figure out whatís the problem of the measurement.

The attached picture is the waveform of a typical 98.3040MHz audio clock I measured while back. There are roughly 1ns rising and falling time, as well as ringing and everything. It doesnít look like a sin wave. My oscilloscope has 1GHz bandwidth with sampling rate up to 8GS/s. It doesnít superior too much than yours.

If you cannot see the true waveform, what you get from your measurement will mislead your analyzing result. For example, when you talking about the 100MHz clock, you are worry about the setup time and hold time because of that waveform you saw. Actually Tsu and Th have nothing to do with the accuracy and sound quality. Itís all about the tolerance of your digital system. For a synchronized logic system, they describe the relationship between the original clock and date (or generated clock). If the Tsu or Th condition (depend on the logic chips) doesnít meet, the output will be wrong, and you will hear noise or get malfunction (So you will clearly know something wrong). If that is the case, it will be totally unacceptable for audio application. Otherwise, the digital music will be 100% correct. Itís black and white and nothing in between.

In the digital domain, the only thing impacts the sound quality is the jitter of the MCLK. Propagation delay and skew are all totally different concept to the jitter. They will have no any business with sound quality as long as all timing requirements are met.

Clock frequency doesnít have direct relationship with the clock quality. Both 100MHz and 50MHz clock can have same raising and falling time, same overshoot, same reflection. The difference could be only the plateau time. Actually, if they have same phase noise numbers, the higher frequency ones will have less jitter.

So, higher frequency clock closing to 100MHz doesnít look like a sin wave and it doesnít degrade the sound quality (may upgrade) under the same condition (as long as your DAC can work with).

Click the image to open in full size.
CLOCK98.3040MHz
by Ian, on Flickr

Good weekend,
Ian
  Reply With Quote
Old Yesterday, 10:36 AM   #508
Supersurfer is offline Supersurfer  Netherlands
diyAudio Member
 
Join Date: Feb 2013
Hi,

I want to connect two battery boards to the dac. Using the 13v rails of the 2 boards in parallel to lower ps impedance to the iv stage. In the manual there is only mention of putting the batteries in series, not in parallel.
Is there a downside of using them in parallel?

I also have a 7,7A 19v switching ps, will this suffice for feeding two boards in parallel? The manual states a minimum of 3,5A but the charge current on the display is around 6A max (is this the current at 3,3V?)
  Reply With Quote
Old Yesterday, 04:02 PM   #509
Markw4 is offline Markw4  United States
diyAudio Member
 
Join Date: Jul 2016
Location: California
Hi Ian,
Thank you for posting that scope picture. Its enough to make me want to go back and double check the some things.
  Reply With Quote

Reply


IanCanada's Latest RPi GB Goodies Impressions... and your tweaks, mods and hints...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

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Detailed specifics, hints and tweaks in designing a sealed sub Squeak Subwoofers 1 17th July 2015 03:27 PM
The latest tweaks :) lowtherdream Everything Else 6 13th January 2009 05:43 PM
S5 K-16LS mods, tweaks? hartze11 Tubes / Valves 3 1st December 2008 05:13 AM
NAD C541i: possible tweaks and mods? Karlosak Digital Source 4 23rd November 2006 06:20 PM
Any mods/tweaks for a Panasonic DVD S-49 mikee55 Digital Source 4 19th March 2006 07:04 AM


New To Site? Need Help?

All times are GMT. The time now is 10:54 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio
Wiki