DIYHiFi.org

I know that. First place that I found it, was your post @ Diya called "low noise oscillator" (if my memory serves me right) . Since I add CFP - can you call me a "audio designer" ??

Jocko
What kind of parameters, levels do you use when looking at 1/f;
F cutoff of?, what are you considering to be a low level? 2,3,4 pole roll off, etc.

Mike

Not sure that I fully understand your question. I just measure the noise, period. Trying to be better than a 3-terminal regulator. Maybe you just have to measure one to know how bad one really is, and go down from that point.

Jocko

I've long wondered what the hell people mean when stating a clock's regulator is "different" than say one for an op amp for an input stage, aside from the obvious voltage difference and potential need for an opposite polarity with similar performance.

From this discussion it seems "noise". Is that not also a desirable factor to optimize for an input stage? Please clarify this for me.

Clock-spikes perhaps?

Jocko means the self noise generated by the regulator. It is certainly drastically inferior to a 3 terminal regulator, from the numbers he has posted.
This is not the same thing as PSRR. That was not his main goal (it was made to be used with a 9V battery).
Otherwise he recommends the use of a pre-regulator.
I feel like I'm repeating his words. Guys, go back and read the first page of the thread again.

_________________
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

"It is certainly drastically inferior to a 3 terminal regulator, from the numbers he has posted."


Yar?

Sadly the question was not specific to this new regulator, my bad. I also hadn't remarked the statement about the PSRR (smells like cap multiplier).

It's just often a repeated them when it comes to regulator discussion, that they just can't be the same, somehow they are too different, yet I can't help but think if I had a regulator that excelled in all critical areas it would work equally well for either a clock or an input stage, and why not.

The 1/f part is what has me interested. Measuring broad band noise is doable. Measuring 1/f, say 1 to 100 Hz is another challenge entirely. The reason for my asking.

What bandwidth(s) for noise is most critical for a clock? Or is it low noise across all bandwidths, which is what I've been figuring.

Story
Many years ago I worked for a biomed company that made SIDs monitors (sudden infant death monitors) The front end used TLO72s. We had to test each chip due to high 1/f and popcorn noise. The front end gain was so high that noise spikes would create false triggers. One thing we found was that if the chips were run through the wave solder they had more noise than if the chips were soldered in by hand. I don't know this but we figured that the higher temps of wave soldering affected the front end of the opamp some how.

I use a really expensive HP FFT machine. Good down to......some ridiculous mHz range. Yeah, takes time to get LF readings.

Short on time this week (got up at 5:30 AM, and am just now getting into my daily grind...........after spending all day in some hospital waiting room. Perfect waste of a day. No, it wasn't serious, and no, wasn't a family member.)

Anyway.............

The Q of the crystal is responsible for what is called "FM flicker noise". Once you get far enough away from the carrier frequency, modualation on the rails won't have as much effect.

(You will also notice that most every high-$ oscillator is spec'ed for jitter at frequencies > 1 kHz or so. Because you can't do much to 'eff up that spec.)

Down at 100 Hz............certainly at 10 Hz.............1 Hz!...........you can blow the spec out of the water by having a noisy supply. They know this, so that is a big reason why it is not spec'ed.

Unfortunately, for digital audio, the close-in jitter is the most detrimental. So, you gotta have something really quiet.

Why do you think that some clock mongers use a Wenzel-style noise reduction circuit? Go to their site, and see what one is. (BTW, it needs to be adjusted. Which means that you need lots of gear to do it. Not DIY friendly.)

Jocko

_________________
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

Thanks Jocko, Carlos
Seen the Wenzel, played with it in spice but never built it.

Have to think about this some.
Mike

> (You will also notice that most every high-$ oscillator is spec'ed for jitter at
> frequencies > 1 kHz or so. Because you can't do much to 'eff up that spec.)
>
> Down at 100 Hz............certainly at 10 Hz.............1 Hz!...........you can blow
> the spec out of the water by having a noisy supply. They know this,
> so that is a big reason why it is not spec'ed.

There are exceptions:
http://www.mti-milliren.com/ocxo_260_ocxo.html#

I was lucky to get a few of them at ebay.

regards, Gerhard

_________________
Everything has been said already - but not yet by everyone. (Karl Valentin)

Thanks, that brings quite a few things together.

Yeah, that is one helluva an exception. Lucky, indeed. Jitter works out to be <1 pSec for any of them.

Now, can anyone other than Gerhard (obvious), Elso (probable), and Carlos (possible) explain the differences between the data for the AT and SC cut versions?

Jocko

Let me try:



That means - SC has higher Q of resonance - less influence from PSU.

Correct ?