The Well Tempered Master Clock - Building a low phase noise/jitter crystal oscillator

Status
Not open for further replies.
Jump to Latest
A

Alexiss

Member
Joined 2003
1audio said:
It would be interesting to see what performance can be had using a Potato gate in a gate oscillator.
Click to expand...

Im wondering about that my self.
You hear so much about potato stuff these days.

It says 74gU04 on the package....dont know if that's a single stage or a cascaded triple stage inverter.

It is possible to build a really good gated oscillator, but due to their high gain-bandwidth product and low saturation voltage, the output wave contains a lot of harmonics that might affect the final PN....but i'm not so sure about this.

However, discrete oscillators dont have this problem.
 
A

Alexiss

Member
Joined 2003
Herbert Rutgers said:
Well, I meant as a squarer (74...04) not the JFET.
Herbert.
Click to expand...

Yes, then you are right.


Looking for example at the output of the RutgerS oscillator amp, it is a very clean sine wave, containing almost no harmonics.
This is good.
But looking at the Inverter-based Pierce, the output of the oscillator amplifier contains a lot of harmonics.

Since the oscillator amplifier feeds the sine-to-square converter later, it might not be that important.

I dont know how this affects the PN of the oscillator itself. It might be negative.
 
A

Alexiss

Member
Joined 2003
andrea_mori said:
...I performed the listening session and assigned a score from 0 to 10 points to each oscillator, reporting the listening impressions.

These are the oscillators that were compared, all at 22.5792 MHz:
- original Aune XO
- Crystek CCHD-957
- TWTMC-AIO Driscoll All in One with standard HC-49/U AT-Cut crystal from TFC
- TWTMC-AIO Driscoll All in One with cold welded HC-43/U AT-Cut crystal from Laptech
- TWTMC-P Pierce with cold welded HC-43/U AT-Cut crystal from Laptech

Following the score and the listening impressions for each oscillator:

1) Score: 9. The best performers are the TWTMC-P Pierce and the TWTMC-AIO Driscoll with the same Laptech HC-43/U AT-Cut crystal.
... I suspect that the Driscoll with harmonic SC-Cut will be the winner with a score of 10.
Click to expand...


I was just reading this very interesting test (thank you Andrea) and came upon the highlighted comment.
Do not be so sure about SC crystals scoring high Andrea.


I mentioned a while ago that I saw some measurements on the AT vs SC crystals.
They did not differ at close-to-carrier offset frequencies.

Very roughly:
the close-in PN is much more dependent on the ESR of the xtal, as this dictates the xtal self-noise.
Strangely not the Q of the xtal so much.

Here, the SC xtals are almost 10 times worse.

The close-in PN can be improved by increasing the current through the xtal (the Drive Level) and using AT cut xtals for their much lower ESR.
Both of which has been adressed in the RutgerS oscillator.

I seriously doubt that RutgerS oscillator's excellent performance at < 100 Hz will be improved upon by using SC-cut xtals.
Neither is it trivial to design a ”proper” mode-surpressor for SC based oscillators.

--------------------------------------
Keep up the good work Gentlemen.
 
Last edited:
Terry Demol

Terry Demol

Member
Joined 2002
Alexiss said:
I was just reading this very interesting test (thank you Andrea) and came upon the highlighted comment.
Do not be so sure about SC crystals scoring high Andrea.


I mentioned a while ago that I saw some measurements on the AT vs SC crystals.
They did not differ at close-to-carrier offset frequencies.

Very roughly:
the close-in PN is much more dependent on the ESR of the xtal, as this dictates the xtal self-noise.
Strangely not the Q of the xtal so much.

Here, the SC xtals are almost 10 times worse.

The close-in PN can be improved by increasing the current through the xtal (the Drive Level) and using AT cut xtals for their much lower ESR.
Both of which has been adressed in the RutgerS oscillator.

I seriously doubt that RutgerS oscillator's excellent performance at < 100 Hz will be improved upon by using SC-cut xtals.
Neither is it trivial to design a ”proper” mode-surpressor for SC based oscillators.

--------------------------------------
Keep up the good work Gentlemen.
Click to expand...

It will be interesting to see how SC cut perform in reality with listening.

Information from OCXO manufacturers is mixed but most point toward
SC-cut having better close in phase noise.

This also seems to be reinforced by close in PN performance of best
currently available OCXO's. Those that are better than -140 or so at 10Hz
(10MHz carrier) are all SC-cut.
 
G

gabor80

Member
Joined 2012
andrea_mori said:
I have just compared a few oscillators in a listening session. The comparison was done in the same audio chain, simply replacing the oscillator board at each session. I'm tweaking a chines SD card player for a friend, so I used it to compare the oscillators. For this listening session I have used the S/PDIF output of the player (Aune X5 that provides also I2S output), since the DAC (NOS with AD1865) accepts S/PDIF input only.

I performed the listening session with a friend. During the comparison I knew which oscillator was playing at every moment, while for my friend this was a blind session. At the end of the session we have assigned a score from 0 to 10 points to each oscillator, reporting the listening impressions.
I don't like to publish listening impressions, since they are subjective and not measurable. So, I invite everyone to do his own comparison to evaluate the difference. What I can claim is that there are clearly audible differences, in most cases the difference is huge.
It's very impressive that replacing the clock in the same audio chain, the sound changes totally.

These are the oscillators that were compared, all at 22.5792 MHz:
- original Aune XO
- Crystek CCHD-957
- TWTMC-AIO Driscoll All in One with standard HC-49/U AT-Cut crystal from TFC
- TWTMC-AIO Driscoll All in One with cold welded HC-43/U AT-Cut crystal from Laptech
- TWTMC-P Pierce with cold welded HC-43/U AT-Cut crystal from Laptech

Following the score and the listening impressions for each oscillator:

1) Score: 9. The best performers (ex-aequo) are the TWTMC-P Pierce and the TWTMC-AIO Driscoll with the same Laptech HC-43/U AT-Cut crystal. Performance of these oscillators are very similar and far better than the others. There are only a little differences between the two oscillators, not more than a little shades, that's the reason they win the comparison ex-aequo. Bass are deep and controlled, the Pierce is a little less deeper but more dumped, the Driscoll is a little more "rounded". Vocals are spectacular in both cases, with the Driscoll a little smoother, although a little backward. The upper octaves are very good for both oscillators, the Pierce is a little more detailed and a little less smoother. The soundstage is very wide and high, the Pierce reach the maximum extension. Unplugged music is so realistic with both the oscillators, with the Pierce more detailed and the Driscoll a little smoother. Hard to choose between the two oscillators, after a long debating I chose the Driscoll while my friend opted for the Pierce. But as I said before they are simple shades, both oscillators perform very good. I suspect that the Driscoll with harmonic SC-Cut will be the winner with a score of 10.

2) Score: 7.5/8-. The second place was assigned to the TWTMC-AIO Driscoll with the TFC HC-49/U AT-Cut crystal. Its performance is similar to the Driscoll with Laptech HC-43/U crystal, although the differences are audible. Bass are a little lighter but very controlled. Vocals are a little more thin compared to its big brother, but very nice anyway, never aggressive. High are very detailed and smooth. The soundstage is a little compressed compared to best performers, the image is stable and realistic. Where the difference is clearly audible is about the complex music such as the full Orchestra, in this case the result is a little confused, while the Pierce and the big Driscoll don't suffer of this issue. All in all the sonic performance of this oscillator is very respectable , considering that uses an inexpensive crystal. As I stated many times the crystal makes the difference.

3) Score: 5. With the Crystek oscillator the differences are huge and clearly audible after the first notes. If the Driscoll with the budget crystal can be defined similar to the best performers of the group, although the differences exist and they are audible, the Crystek sounds harsh and confused. Bass lacks of thickness and is boomy. Vocals are confused and aggressive, they get you tired after a few minutes of listening. The soundstage is narrow and shallow, the image is less stable than the previous oscillators. Orchestra is totally confused and kneaded. All in all the resulting sound is very poor, far worse than the previous 3 oscillators.

4) Score: 4. The original Aune oscillator is the worst of the group, although not much worse than the Crystek. It has the same shortcomings of the Crystek in all respects although more prominent.
Click to expand...


Do you test anytime the Clapp oscill. (TWTMC-C) in this audio chain, if not, will you?
I saw the PN of Driscoll AT-cut measurement wasnt so good like Clapp (see Rutgers measurement). Driscoll with SC cut xtal will it improve enough to overtake the Clapp type boards?
 
A

Alexiss

Member
Joined 2003
9
zenelectro said:
Information from OCXO manufacturers is mixed but most point toward
SC-cut having better close in phase noise.
Click to expand...

Keen observation.

This is because their circuit design is very different from the simple oscillators we use here.
They often utilize much lower frequency xtals with very low close-in PN charcteristics and reach target frquency via balanced active up-converters.
This gives desired frequency while retaining exellent close-in performance of lower f xtals.

At lower f even SC-cut xtals operate at fundamental-mode.
For each additional overtone mode, the ESR increases.

---------------------------------------------------------
Keep up the good work gentlemen.
 
Last edited:
Status
Not open for further replies.
Top Bottom