DIYHiFi.org

For the sake of audio
It is currently Wed Oct 22, 2014 7:55 am

All times are UTC




Post new topic Reply to topic  [ 114 posts ]  Go to page Previous  1 ... 4, 5, 6, 7, 8
Author Message
PostPosted: Thu Feb 17, 2011 7:48 pm 
Offline

Joined: Sat Mar 04, 2006 5:26 pm
Posts: 558
Location: Nova Scotia
One thing to keep in mind when you're analyzing this stuff: DACs act like frequency dividers.

Suppose you have a 11.2896MHz, 256Fs master clock driving a DAC. But someone badly messed up a word clock synchronization PLL, so your master clock has a nasty PLL phase noise spur at 1/256 this rate. Say, 40dB down at +-44.1KHz from the clock.

If you make the DAC output a 5KHz tone, you'll see the same spurs, 40dB down, at +-19.53Hz (5KHz/256). Heck, every tone that the DAC puts out will have the same spurious tones present on it.

I learned this lesson at work, when a VCXO supplier changed the part they were selling us from a overtone type to an analog multiplying type without telling us. The VCXO in question was driving a RF DDS chip, which is basically a high speed DAC that puts out a sine wave. The new VCXO had subharmonics at Fout/3, 2*Fout/3, 4*Fout/3, etc... and sure enough, the DDS put out the same spurious frequencies at N*Fout/3, at the same levels as the clock subharmonics.


Top
 Profile  
 
PostPosted: Fri Feb 18, 2011 3:38 pm 
Offline

Joined: Mon May 09, 2005 6:11 pm
Posts: 587
Location: Gran sasso
George,

What You say is totally true - for DDS chips..
But there is also an NCO included, doing the actual frequency synthesis.
Simple dacs are rather some kind of mixers. The sample clock modulation frequency shows up directly, as sidebands on the output signal.
Though the magnitude of it gets scaled, in function of the signal/sample frequency.

Ciao, George


Top
 Profile  
 
PostPosted: Fri Feb 18, 2011 9:29 pm 
Offline

Joined: Sat Mar 04, 2006 5:26 pm
Posts: 558
Location: Nova Scotia
Instead of using a NCO with your DDS, use two NCOs at different frequencies, say 2.5KHz and 5KHz, with their outputs added together. You'll see the same spurs, at +-19.5Hz from the 5KHz tone, and +-9.75Hz from the 2.5KHz.

Now replace the NCO output with a recording of a Hammond organ. Which is basically a big collection of NCOs (tonewheels) added together based on the key you hit and the drawbar position. Around each tone the organ generates, you'll see the same phase modulation spurs around them.

Adding more instruments won't remove any of this. Replacing a NCO tone with music doesn't make the effect disappear, It just means there's way more "tones" involved.

Hell, this technique is taken to the extreme for FM synthesis, used in old synthesizers/Adlib cards/etc.


Top
 Profile  
 
PostPosted: Sat Jun 25, 2011 12:54 am 
Offline

Joined: Mon May 09, 2005 6:11 pm
Posts: 587
Location: Gran sasso
Some time had passed since my last intervention - this is life eating up hobbies..:(

But recently I had managed to put my hands on an Art Legato - something I had desired from the very beginnings..
I would like to show here - no, not how does it perform in my measurement setup..
Rather I would like to use it for checking out my testing abilities.. :thumbsup:

I had used the same test setup described earlier: the same scope, with the same settings. It is easy, because the scope is keeping all configuration values together with the saved tracks, so I had just re-loaded everything.
The setup is: Legato 75ohm BNC - A 30cm high speed video cable with 75ohm BNC at one side, RCA at the other side - RCA-BNC converter-75ohm bias tee - a good 75ohm termination - osc channel on 1MOhm input setting.

For easier comparison, here are the earlier Hiface shots again:
The newer series original version:
http://postimage.org/image/3oc0qiys/
The external PS modified Hiface:
http://postimage.org/image/unog915w/

The Art Legato, with almost the same settings: caution, the jitter track resolution is doubled!
http://postimage.org/image/23dme145g/

The shot shows 5.7psec rms jitter, ~35psec peak to peak.
I think the spectrum plots / differences are talking for themselves..


Last edited by Joseph K on Sat Jun 25, 2011 1:05 am, edited 1 time in total.

Top
 Profile  
 
PostPosted: Sat Jun 25, 2011 1:00 am 
Offline

Joined: Mon May 09, 2005 6:11 pm
Posts: 587
Location: Gran sasso
As you might have noticed, the cable applied was not the best..
Here it is my best effort, with a ~similar length, 40cm, Belden RG59 cable assembled with good 75ohm BNC connectors.

Caution, the settings are not the same, they are a bit better optimized.
http://postimage.org/image/23drcn5ms/

The slight difference in sigma is real, caused by the cable difference.
It's 4.9psec rms, ~28psec peak to peak.


Top
 Profile  
 
PostPosted: Sat Jun 25, 2011 1:32 am 
Offline

Joined: Mon May 09, 2005 6:11 pm
Posts: 587
Location: Gran sasso
The results are so clean, that .. maybe.. it would even be possible to look at cable / connector differences?
Here I would like to show some shots - maybe- interesting?

I had cheated a bit. Up above you had seen what it is like when everything is "perfect".
Now I had changed the scope termination to 50ohm. This creates -20% reflection at the "far", oscilloscope end.
On the ART output we still have it's very good output termination, hopefully creating less than a percent reflections, that is, "eating up" the crud coming back to it.
Also, the cable is still that short cable - the consecutive reflections are decreasing down faster.
So, it's still quite a "not bad" situation!

Here it is that fast video cable, with RCA connector at the scope end, and the obligatory rca-bnc converter.
Everything is strongly zoomed in, so as to see the differences. Also, I am looking at the lower, audio range, (0-50kHz),
so lowering the sampling rate, with which also the resolution is decreased, as explained/ had shown earlier.
So, the higher picosec distribution values are a consequence of all these factors together.

http://postimage.org/image/9qzw6tus/

Here it is the same setup, with the Belden cable, all 75ohm BNC:

http://postimage.org/image/9r6ic7tw/

And here it is the same Belden cable, with a 10dB 75ohm attenuator applied:

http://postimage.org/image/9r85vkbo/

I would leave to You the conclusions..


Top
 Profile  
 
PostPosted: Sat Jun 25, 2011 1:45 am 
Offline

Joined: Mon May 09, 2005 6:11 pm
Posts: 587
Location: Gran sasso
Or, rather not - would not want to leave out some explanations..
As it was shown earlier, data dependent jitter, DDj, always appears in the FFT spectra as spurs.
The smoother the floor is, the less DDj is present.
One effect of the attenuator is that the DDj spurs are getting visibly lower.
There are a lot present, in the low audio range, in the shot with the RCA connector.
It gets a bit better with full BNC connectors, though the 50ohm mismatch still remains and causes problems.
Finally the attenuator helps to restore a cleaner situation.
And this hash visible here is in the audio band!! That spur at the upper end of the spectra, not diminishing, is correlated to the 44.1kHz sampling frequency.
An another effect is also visible: The rms noise gets a bit higher, by applying the attenuator. The double peaks in the first distribution (no attenuator) are a bit "sharper", then the resulting one peak with the attenuator.

Ok, so much for now - have fun!

Ciao, George


Top
 Profile  
 
PostPosted: Sat Jun 25, 2011 11:20 pm 
Offline
User avatar

Joined: Tue Dec 14, 2004 12:26 pm
Posts: 4927
Location: Lisbon, Portugal
Back to your stuff, George? :D
Excellent work, as always.
Thanks for posting.

_________________
Carlos Filipe

"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius - and a lot of courage - to move in the opposite direction." Albert Einstein


Top
 Profile  
 
PostPosted: Sun Jun 26, 2011 6:37 pm 
Offline

Joined: Mon May 09, 2005 6:11 pm
Posts: 587
Location: Gran sasso
Ola' Carlos!

Just wanted to say: thanks for the drink / and good suggestions! It was a real pleasure to talk to You!
Tróia is really beautiful. As is Portugal.. lucky guy.

Ciao, George


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 114 posts ]  Go to page Previous  1 ... 4, 5, 6, 7, 8

All times are UTC


Who is online

Users browsing this forum: No registered users and 4 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group