[f] AMB on Facebook

ο1 precision master clock development

omicron1: A low-jitter, low phase noise clock generator for upgrading digital audio devices

Re: ο1 precision master clock development

Postby amb » November 12th, 2016, 1:20 pm

Let me build and test one first. :)
User avatar
amb
Site Admin
 
Posts: 10379
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: ο1 precision master clock development

Postby Balcmeg » November 12th, 2016, 6:56 pm

amb wrote:Let me build and test one first. :)


Ever noticed that there are some eager people here when a new project start? ;)
/Stefan

poseidonsvoice likes this post.
Balcmeg
 
Posts: 560
Likes: 24 posts
Liked in: 8 posts
Joined: September 6th, 2014, 6:54 pm
Location: Hong Kong
Country: Hong Kong (hk)

ο1 prototype PCBs are here

Postby amb » November 20th, 2016, 12:26 am

The PCBs are here! Here are some photos. Sorry about the blurry photos, I should have set the camera up a bit differently.

The top side:
Image

Bottom side:
Image

I fired up the soldering iron and built a single-clock version for use with my Behringer DCX2496.
Top view:
Image

Bottom view:
Image

I used a bench power supply (dialed to 5VDC output) to power the ο1, and measured the output with my Tektronics TDS2014 100MHz DSO, with a special coiled probe ground. Here is a screen capture of the waveform right at the U.FL connector (J2).
Image

I then connected a 100mm long U.FL shielded cable and captured the waveform at the end of that cable. Note a bit of overshoot at the leading and trailing edges, probably because the scope probe end of the U.FL cable is not terminated with the proper impedance.
Image

These are beautiful waveforms. Unfortunately I do not have the equipment to check the phase noise or jitter, but since the output is directly from the oscillator, we will have to rely on the specifications on its datasheet. For the Crystek CCHD-957, the performance is impressive. There is no question that it should be vastly superior to the original master clock circuit in the DCX2496.

I have not yet integrated it in the DCX2496, that is to be done next.

2 people like this post.
User avatar
amb
Site Admin
 
Posts: 10379
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

ο1 installation into Behringer DCX2496

Postby amb » November 22nd, 2016, 5:04 pm

Original master clock waveform

I have successfully installed the ο1 in my Behringer DCX2496. But before doing so, I took a scope shot of the original master clock circuit's output, and this is what it looks like. A bit rounded, but not too bad. I have no way of knowing its jitter or phase noise performance though:

Image

Build configuration

For the DCX2496, the ο1 should be built in the following configuration:
- Single frequency configuration
- Use CCHD-957-25-24.576 for X1
- Use ADM7150ARDZ-3.3 for U1
- Do not populate X2, U2, R2, R3, R4, C7, C8 (leave empty)

Installation procedure

Here is a view of the bottom side of the DCX2496 DSP board (click to enlarge), where most of the work will be done:

Image

What I describe below pertains to the new style DCX2496 without PCMCIA slot. On older DCX2496s and on DCX2496LE, the part IDs may be different, but the photos should illustrate what needs to be done. On DCX2496s with PCMCIA slot, the standoff for mounting the ο1 needs to be taller in order for the ο1 to clear the PCMCIA slot mechanism and the Q1 crystal, and use the right angle Molex KK 3P header listed as (alt) in the parts list so that it would fit under the top cover.

- Remove the following parts from the bottom of the board (see photo below): C77, C78, R38, R34
- Remove Q3 (need to desolder the body from the top side of the board, and desolder the pins on the bottom side)

Image

The following photo shows the top side of the board with Q3 removed:

Image

- Short circuit the pads for C78 (previously removed) with a small piece of resistor lead and solder.
- Determine the wire lengths needed for the power/ground wiring and U.FL shielded cable.
- Strip the power/ground wires
- Strip the unterminated end of the U.FL cable and separate its shield from the center wire. Twist the shield together and tin it with solder.
- Install the power/ground wires and U.FL cables as shown in the photo below:

Image

Image

- Apply hot glue to the wires to help insulate them and to prevent them from working loose:

Image

- Install Molex KK crimp pins on the other end of the power and ground wires, and push them into the Molex KK plug housing. Do not reverse the VCC and GND wires! See the PCB silkscreen to verify.
- Check your wiring for shorts
- Plug the Molex connector and U.FL cable to the ο1 PCB. Note that the U.FL connector is only rated for a few connect/disconnect cycles, so do not remove it unless absolutely necessary.
- Reinstall the DSP board, but leave off the mounting screw adjacent to the removed Q3. Instead, use a male-female M3 metric standoff. I also used a lock washer (see photo below). As mentioned above, use a taller standoff than shown if you have a DCX2496 with PCMCIA slot.

Image

- Install the ο1 on the standoff and tighten the M3 mounting screw. The finished product should look like this:

Image

Image

New master clock waveform

The new master clock waveform at the ο1 U.FL connector:

Image

And at the R34 location (including about 8cm of U.FL cable):

Image

Conclusion

The DCX2496 with the ο1 upgrade works great and sounds excellent. I have not done a A/B listening comparison of before/after modification (because it is not possible with only one DCX2496), but I did run a series of sanity tests on the Audio Precision audio analyzer. No problems were found and all performance aspects are as good as my previous measured results.

This is my "final frontier" for DCX2496 upgrades. The AES/EBU input and CS8420 receiver/ASRC chip are still suboptimal but I don't currently use that portion of the circuit in my system. So I'll just leave well enough alone there.

Use in other devices

A similar upgrade can be done with the ο1 in many other devices, including CD players, media streamers, ADCs and DACs. You need to determine whether you need a single-frequency master clock or a dual-frequency, and if the latter, the logic level for the SEL function. You also need to know what clock frequency (or frequencies) are needed, and choose X1/X2 accordingly. Remove/disable the existing master clock circuit, build the ο1, choose a convenient location to mount it and wire it up. If you need 5V logic level master clock output, then use a ADN7150ARDZ-5.0 for U1 and a 5V oscillator. The VCC input voltage should be at least 1V higher than your desired supply voltage, and no more than 15V.

2 people like this post.
User avatar
amb
Site Admin
 
Posts: 10379
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: ο1 precision master clock development

Postby poseidonsvoice » November 22nd, 2016, 6:31 pm

Yippee! I have the older PMCIA slot version of the DCX2496, so I'll need to use a right angled Molex header (J1'), and taller male/female standoff.

I'm ready to buy the board, like yesterday! Sign me up!

Best,
Anand.
User avatar
poseidonsvoice
 
Posts: 109
Likes: 68 posts
Liked in: 9 posts
Joined: April 21st, 2012, 11:09 pm
Country: United States (us)

Re: ο1 precision master clock development

Postby amb » November 22nd, 2016, 10:47 pm

ο1 PCBs are now available at AMB audio shop. Note that I only have a small number available (currently at 9 boards). If these sell out quickly then I will order more.

poseidonsvoice likes this post.
User avatar
amb
Site Admin
 
Posts: 10379
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: ο1 precision master clock development

Postby amb » November 25th, 2016, 1:59 am

The official ο1 website is now up:

http://www.amb.org/audio/omicron1/

Take a look!

poseidonsvoice likes this post.
User avatar
amb
Site Admin
 
Posts: 10379
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: ο1 precision master clock development

Postby poseidonsvoice » November 25th, 2016, 3:37 am

Ti,

It might be the blurry photos but in the DCX2496 application, am I correct in assuming that R3 should be shorted or just left unpopulated?

Thanks,
Anand.
User avatar
poseidonsvoice
 
Posts: 109
Likes: 68 posts
Liked in: 9 posts
Joined: April 21st, 2012, 11:09 pm
Country: United States (us)

Re: ο1 precision master clock development

Postby amb » November 25th, 2016, 3:38 am

poseidonsvoice wrote:It might be the blurry photos but in the DCX2496 application, am I correct in assuming that R3 should be shorted or just left unpopulated?

Both R3 and R4 should be unpopulated in single-frequency configuration.
User avatar
amb
Site Admin
 
Posts: 10379
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: ο1 precision master clock development

Postby LV26 » December 7th, 2016, 11:47 am

AMB,
CCHD-957 specified for 15 pF (max) load. Cable should be short as possible.
I think o1 trace with connector have ~ 5 pF, and DCX input ~ 5 pF. Total 10 pF.
Only 5 pF for Cable.
Also I think that 10 Ohm serial termination very low. I think 22 or 33 Ohm will be better.
And you need faster oscilloscope for testing termination.
Everything should be made as simple as possible, but not simpler. // Albert Einstein
User avatar
LV26
 
Posts: 431
Likes: 55 posts
Liked in: 16 posts
Joined: May 7th, 2014, 1:46 am
Location: UKRAINE
Country: Ukraine (ua)

PreviousNext

Return to ο1 precision master clock

Who is online

Users browsing this forum: No registered users and 0 guests

Quick Links

AMB audio DIY resources
AMB audio shop

cron