Never mind, a little search gives me this: B82789C0513N002 EPCOS / TDK | Mouser Italia
It means there is no production consistency from Laptech ?
Or a detail changed in your setup as a passive component on the oscillator like a type or brand of caps for instance or a diiferent smd resistor ??
Is it a hearing feeling or do you check an important difference via your scope?
Agging or break in of the previous sc cut you have ?
Anyway thank you for your inputt
I did not change any component in the oscillator circuit except the crystal.
I did not check with a scope. It's just me, hearing an improvement.
Phase noise is a feature of the oscillator, so it's normal you cannot find anything about phase noise related to the crystal on Laptech's website.
The specs for a crystals are those I published for each batch (Q, motional capacitance, ESR and so on).
A little difference between crystals in a batch or in different batches is normal. Although we are talking about very very precise mechanical work, it's practically impossible to get exactly the same specs for each crystal.
The manufacturer guarantees minimal values for the most important specs, for example about the 5.6448MHz SC-Cut 3rd overtone Laptech guarantees a Q greater or equal to 2M, and as you can see in the measured data the Q of the batch varies from 2.024M to 2.409M.
Hi Andrea
As a curiosity, can we figure out which of the crystals listed in the Laptech test batch we have? I looked at the list of 45 MHz crystals and find no id on the crystal I have in hand that relates to one of the crystals listed in the test. Am I missing something? Not important, as I can assume it is somewhere within the bounds of the test bunch. Just neat to know exactly the spec if it is readily available.
Also for my understanding... what is the most important for sound quality? PPM or Q or some combination of these? If this is already described I am sorry.
I think your crystal is from a previous batch, please search in the GB thread, you should find data from all batches.
In a Driscoll oscillator the most important spec should be the Q, this is the starting point to get low phase noise oscillator. All Laptech crystals have high Q for each frequency. But Laptech aside the greater Q for a crystal can be reached around 5MHz, that's why I have suggested to use 5/6 MHz, maybe followed by a frequency multiplier if one needs absolutely higher frequencies.
Are there phase noise measurements of your designs available? Maybe not the exact clock frequency but something near so we can see in what ball park we are with our clock units.
Not yet measured, the gear to measure the phase noise are very expensive.
I'm building my own phase noise measurement system, as soon as it will be ready I will measure each oscillator.
PPM is not useful in digital audio, you have to look for short therm stability that means phase noise (frequency domain).
Also the jitter (time domain) is not useful because it depends on the integration of the signal measured. If the signal integrated starts at 100Hz from the carrier you probably get picosecond of femtosecond jitter, but if the signal starts at 1Hz from the carrier you get microsecond.
The only important measurement for digital audio is the phase noise close to the carrier.
Just for comparison, please take a look at the pictures below. They are the phase noise plot of two oscillators, the Crystek CCHD-957 at 22.5792MHz and the MTI OCXO 260 at 5MHz.
As you can see the phase noise measured at 10Hz from the carrier is a bit different: -97dBc for the Crystek and -140dBc for the MTI. Also if you add -6dBc to the plot of the Crystek (dividing down to the same frequency of the MTI - 5MHz), the performance of the two oscillators are very different.
This means that the Crystek seems to be a poor oscillator while the MTI is a SOTA oscillator.
This also means that the 5MHz is the ideal frequency to get the best phase noise performance.
As you can see the phase noise measured at 10Hz from the carrier is a bit different: -97dBc for the Crystek and -140dBc for the MTI. Also if you add -6dBc to the plot of the Crystek (dividing down to the same frequency of the MTI - 5MHz), the performance of the two oscillators are very different.
This means that the Crystek seems to be a poor oscillator while the MTI is a SOTA oscillator.
This also means that the 5MHz is the ideal frequency to get the best phase noise performance.
Attachments
Just got the PCB for the 16 bit design, now I start building the prototype.
For the 24 bit design I think it will take around six months to build a prototype, this second design has a very complex front-end.
I find the plethora of specs presented for clocks a bit confusing. I do understand that the MTI graph is better than the Crystek graph. What I am attempting to understand is where in this spectrum is the Laptech 45mHz crystal?
Firstly you cannot measure the phase noise of a crystal, you can measure the phase noise of the oscillator using the crystal.
Then I'm not able to measure the phase noise of an oscillator, as I said such that gear are very very expensive, only for professionals and not for hobbyists. Take a look at the price of an Agilent E5052A to understand what kind of price I mean.
- Status
- Not open for further replies.