advertisement


Best CD clock upgrade?

Carl,

The bag of parts came today to my postbox.
I have to start dealing them to the group-buy guys.
Looks good! Thanks very much indeed!

Special thanks to Martin for the manual comes from
Finnish "Torvihörhö" of the group of buyers.
Torvihörhö means Hornfreak in English and I have
told the guy Martin is in same hornfreak league with him.
Thanks Martin!

Thanks Andy!

Oz
 
Yo,

Nice, nice, nice ! I'm looking for power supply designs for my next DAC...

http://audio.peufeu.com

...I'm not interested in the PCB (coz I'll design a PCB for the whole thing) but I like your design. It's elegant.

Two little questions...

- Has anyone done a listening test compared to a much cheaper TL431 shunt reg fed from a current source ?

- What is the output inductor ? A real inductor or a ferrite ? Have you measured the output impedance ?

Have a nice day !
 
Hey there, welcome to Pink Fish :)

In answer to your questions,

1. not that i know of - certainly nobody has posted results
2. it's a ferrite bead. In fact, the radial version of these: http://www.rapidonline.com/producti...omponents&tier4=Bead+inductors&moduleno=29157

This circuit works very nicely on the Vddd pin of a TDA1305T DAC (pin 28), I use one in addition to the one I use for the clock:
twin-fleas.jpg


A cheap and cheerful way to give very clean 5V power to your DAC :)
 
Welcome indeed.

Has anyone done a listening test compared to a much cheaper TL431 shunt reg fed from a current source ?
No, but I've just collected a tubeful of TL431s to play with so I might get to try this soon.

There's more on the application of this circuit in a more recent thread here. I haven't measured output impedance directly, but there's some relevant notes on limitations of driving more than a handful of mA in the last couple of pages of the thread.
 
what voltage should my clock output peak to peak? 5v right? I got my scope and running tonight and had a look. I suspect the thing isnt very well calibrated. ANyway, I decided to look at the clock signal.

should i be using the x1 setting or the x10 setting on the probe.

Because they display completly different wave forms. At x10 there is loads of extra harmonics on the edge of each pulse and x1 the signal is fairly rounded. It also looks as if the wave form is not evenly square eg the up or down part is longer than the other.

I also notice that on the power rails there is about 0.15v pp noise coming throught. The main part seems to be from the clock. But other harmonics are present but the whole thing seems to bae farily regular (11mhz, surprise!) Is this an acceptable figure?
 
You should be using the x10 setting, and getting something like this, running rail to rail (so ~5v Pk-Pk)

tentout.jpg


The reason the 'x1' setting rounded-off the appearance is that the clock is being loaded with the scope's input capacitance (usu. 15pF) directly, and that of the probe cable. In fact at 'x1' setting the probe will only have a usable bandwidth of about 5-6Mhz, and so the scope effectively displays a low-pass filtered result. Even at 'x10' the display isn't quite right, because the clock isn't driving a matched impedance (50ohm input) on the scope, and because the probe only has a bandwidth of 50-60Mhz or so (as in the chaep + crappy one I used above). With this it can only show the first 5 harmonics, not enough to show a clean squaewave and leaving significant ripple on the display waveform

...But don't worry about it. Actually seeing what the clock is really doing is not so simple at all. Even a scope with adequate bandwidth, and suitable probes to allow you to exploit it (expensive!) will tell you nothing about the resulting sound. In fact the key parameter that a 'good' clock addresses, jitter, is the variation in timing between pulses, and you'll never see this on a scope. For the TentLabs module the specified jitter would represent an average variation in the position of the vertical lines above of just about ... 1/30,000 the width shown above.

Given the oscillator is chopiing between the rails at high speed and driving current into the load (and scope probe..) it should be no surprise that the supply rail carries noise at clock-related frequencies. This high frequency noise which get everywhere and can only be controlled by passive components (principally, the tiny SMD ceramic cap underneath the Flea board across the clock module pins).

0.15v sounds reasonable if a touch high (10-20mV can be achieved); but - again- don't worry, this could be due to something as simple as picking the wrong place to put the ground clip, the length of cable on the ground clip, using too wide a bandwidth setting (if your scope has one) and so on. Scopes are great fun, but remember they lie all the time...

A really good primer on getting the most out of your scope is Tektronix'
XYZs of Oscilloscopes. HTH!
 
One more thing I should have made clearer above. Getting a clean display on a scope of a 12Mhz sine is relatively simple - it has no harmonics. Although the clock is 11.28Mhz, and so might seem well within the capabilities of (say) a 60 or 100Mhz scope, the signal content is far higher in frequency because a square wave is composed of an infinite series of odd harmonics... but the scope can only display the first few.

- Hence the display will look slightly rippled even if your technique and 'matching' is perfect.
 
The Audiocom clock requires more current and at a higher voltage input than flea can provide, so your plan for a good quality separate PSU is the way forward. The link looks OK, using a 3-pin regulator would be also work well.
 
would you suggest using the regulator (i have a few lm317's spare out my A5) set it to 15v and do that zener thing with it then send it to the gyrator circuit?

Are more stages of filtering usually better or can you end up with funny things happening.

Also, the transformers i have are those torroidals from RS (nuvotem). I think the effieceincy rating isnt great (i expected something in the high 90's but these seems to be rather lower). Do you think they'll be up to the job, or is the gyrator etc good enough to be able to ignore these effects.
 
Andy, Martin,
VERY nice job!
I'll experiment this supply as soon as possible with some clock modules!
I never thought about powering them with the o/p of a '797.

Now, about the bypass cap, I quote some infos from G.Tent about his XO:
"Do NOT place the parallel cap, a special RF cap is integrated in the XO"
"the canned oscs have 1uF internally"

He often recommend to use a Oscon or similar close to the XO.
I didn't personally check what there's inside a Tent XO, so take this info as it is.


Bye
Roberto
 
Thanks for the info Roberto, it measures like there's something about 0.1uF inside the can. I have tried measuring noise with and without and external bypass cap, and get slightly lower noise with one.... *

What you use doesnt seem to matter too much though; oscons are good (I've tried up to 10uF), tantalum types (tried up to 47uF) or the small SMT ceramic type Andy specced, which is the most econmic - and best at RF.


*note also that 'dead-bug' prototypes with the AD797 tended to oscillate without this extra cap - all solved in the PCB/schmatic layout though.
 


advertisement


Back
Top