hi all,
after hacking up (then shorting out the power supply of) my Sony DVP-NS500V DVD/SACD player, i've moved on to modding my C222ES SACD changer. it has a much nicer layout that's more open to mods than the NS500V. i've got ideas for the analog stage but i'm wondering about the clock circuit... one nice thing is that Sony put the crystal right on the DAC/analog board, very close to the DACs, instead of on the digital board. however i'm not so sure about the oscillator circuit employed... take a look below. the hex inverter is used for the oscillator, as well as to feed three DSD1702 DAC's and the main digital board. i'm not sure how the jitter performance of this circuit is but i'm thinking it could be improved. suggestions?
after hacking up (then shorting out the power supply of) my Sony DVP-NS500V DVD/SACD player, i've moved on to modding my C222ES SACD changer. it has a much nicer layout that's more open to mods than the NS500V. i've got ideas for the analog stage but i'm wondering about the clock circuit... one nice thing is that Sony put the crystal right on the DAC/analog board, very close to the DACs, instead of on the digital board. however i'm not so sure about the oscillator circuit employed... take a look below. the hex inverter is used for the oscillator, as well as to feed three DSD1702 DAC's and the main digital board. i'm not sure how the jitter performance of this circuit is but i'm thinking it could be improved. suggestions?
Attachments
dorkus said:
Yuck !
TOO many inverters in the circuit, and they are in parallel .....
Yes, the crystal may be where it should be, the design however does not have low jitter in mind.
Get rid of them, as much as possible, and use the lowest nr of gates possible to distribute the clock
I myself never encountered an 04 based clock circuit that measured / sounded well, so throw it out
succes
I would leave IC304 in place for buffering clocks to distant / noisy circuitry, but wire the replacement oscillator's output directly to the DAC if possible. IC304 is inverting, so if you remove C303, R323 and X301, then inject your clock into IC304 pin 1, this will pretty closely preserve the clock timings and hopefully prevent problems. You can simply lift the IC304 pins which are outputs to the DACs, and instead wire those pads directly to the oscillator. A canned oscillator will probably do a fine job to test if it will work, and then I'd move to a more sophisticated oscillator later.
hifiZen said:
My oscillators drive 15 pF, that is 3 or 4 gates, assumed little wiring capacitance is present.
Good suggestion, directly drive the DAC where conversion takes place, distribute other clocked gates using buffers. I always advise the use of series resitors, by the way.
My oscillator is not an off the shelf type, though it may look like this.
It has been selected among over 20 oscillators. After I measured all of them, and listened to them too, I selected this one, went to the manufactuer, sat down and talked about improvements. Once these where caried out I was really satisfied, as all of my customers are.
best regards
-
Guido
> I always advise the use of series resitors, by the way.
yes, good to damp the lines. here is one area where Sony appears to have really done their homework - you can't see them in the schematic but there are series resistors before all the clock lines to the DACs, and they are different values depending on the length of the line - 22, 33, and 47 ohms for the shortest to longest PCB traces. a nice touch.
as much as we scoff at the mass-market nature of Sony products, i have to admit there is a lot of solid engineering behind them (Nelson Pass has mentioned this himself). in particular, the diagnostic modes on the player are unbelievable - it's like The Terminator or something.
yes, good to damp the lines. here is one area where Sony appears to have really done their homework - you can't see them in the schematic but there are series resistors before all the clock lines to the DACs, and they are different values depending on the length of the line - 22, 33, and 47 ohms for the shortest to longest PCB traces. a nice touch.
as much as we scoff at the mass-market nature of Sony products, i have to admit there is a lot of solid engineering behind them (Nelson Pass has mentioned this himself). in particular, the diagnostic modes on the player are unbelievable - it's like The Terminator or something.
dorkus said:
Hi
In my humble opinion, they should be at the transmitting side, that is, if you only use one series resistor. Situation changes if you fully work with a characteristic system, but there is no need for that as long wiring is kept short.
Sony engineers are talented, but designing audio equipment requires a lot of skills, as most among us (including yself) find out in daily practice.......
enjoy
yes, the series resistors are near the HC04 chip, before the PCB traces to the DAC chips. i assume that's what you mean by transmitting side. i dunno if the values were chosen arbirtraily, or if the actually modelled the impedance of the board. would be impressed if it were the latter.
guido do you have a web page for your clock? how much is the module?
guido do you have a web page for your clock? how much is the module?
dorkus said:
Hi
Then the Sony engineers did the right trick. I myself see 47 to 100 ohm a s a right value to start with. If you choose a too high value, the slope of the 0/1 transition will be to low, which might lead to jitter.
The final impedance is a tradeoff between
- slope
- RF current reduction alongside the board
The slope obviously is important wrt jitter, the current amplitude and frequency are too (lees obvious)
Currents across the PCB yield voltage build-up in th egroundplane, which in turn leads to induced jitter at the receiving end as well.
Looking at average PCB designs, with thin traces (heavy traffic....) board trace impedance might be something between 100 and 200 ohms
As for the XO info: You may look at the site of my HK or USA distributors:
http://www.southernelectricaudio.com
http://www.southernelectricaudio.com
and look for XO or accesoires
all the best
jcarr said:
Hi
There are 2 issues
- the fact that inverters are paralleled
- the family of logic
I do not like parallelling inverters for several reasons, so always try to use only 1. If you want the non-plus-ultra, just use picogates !
Yes, you are right about VHC or HCU logic, it is better than HC, but has lower fanout: no free lunch !
enjoy
While my knowledge of digital audio is limited it seems like one thing that hasn't been mentioned is that most of the inverters that are run in parallel are for the six channel audio. If the CD player is going to be used for stereo the clock only needs to be sent to the DAC and Main board. I don't know if you would want an inveter for the line from the audio board to the main board, but that should be the only one you need...
Doug
Doug
dcole said:
Hi all
We need to discuss clock distribution here....
I'd suggest to
- Place the clock as close as possible to the DAC chips
- Directly feed the stereo channels
- Use 2 single inverter buffers (picogates) to drive the other 4 channels
Th measter clock here needs to feed 4 gates, which is feasable to my practice
best regards
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.