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

Status
Not open for further replies.
Jump to Latest
#3,224
kinsei said:
hello
1) what supply voltage should be used for the DBM boards for best PN performance?
2) will the adoption of exotic parts (like Vishay z-foil resistor) further improve the PN performance?
Click to expand...

We have measured all devices (oscillators and frequency doublers) with 3 different power supply:
- TWRPS-LBS LiFePo4 battery supply at 16.5V (series of 5 batteries)
- TWRPS-pp push-pull regulator at 16.5 V
- TWRPS-UGL linear regulator at 15V

The phase noise performance of all the devices are almost the same for all the above power supply, the LiFePo4 power supply is slightly better so we recommend this for the ultimate performance.

Exotic parts would be a wast of money, these are RF devices and we have used the best parts available for such that application (NP0-C0G capacitors, low flicker noise active devices, ceramic core Coilcraft and Epcos single layer inductors for the highest self resonant frequency as possible)
And moreover we had taken the maximum care to the PCB layout.
 
#3,230
andrea_mori said:
I does not work as a multiplexer between the two clocks because the output of the LTC6957 can not be put in a high impedance (3-state) condition.

You can only use the LTC6957 as fanout buffer to get 2 square wave outputs from a single sine wave oscillator.

If there will be enough interest I could design a dedicated board to switch between 2 clocks with enable inputs.

Please, let me know.
Click to expand...

A switch board would be awesome to get the best out of the 5 and 6 mHz clocks for 44.1-48khz files and switch to the doublers for 192khz.
 
#3,233
andrea_mori said:
We had taken care of components tolerance, the components listed in the BOMs were chosen with the right tolerance, you can find 1% capacitors and 2% inductors.

We recommend to get exactly the components listed in the BOM, no replacement parts at all.
Click to expand...
Thanks for confirming this Andrea. I had decided to implement your innovation exactly as specified. I respect your performance measurements. Why fix it if it is not broken?
 
#3,234
Supersurfer said:
A switch board would be awesome to get the best out of the 5 and 6 mHz clocks for 44.1-48khz files and switch to the doublers for 192khz.
Click to expand...

This is another thing.
If you would get the best form the 5/6 MHz oscillators I suggest to wait for the new discrete DAC Lite.

The board I was talking about would be designed with the following features:
- Input: 2 sine wave oscillators, one for each sample rate families)
- sine to square converters (LTC6957)
- double output (fanout, 2 output for each sample rate families to distribute the clock)
- or swithed output, single sample rate family at a time with parallel output (fanout to distribute the clock)

Double output and switched output would be alternatives to each other.
Switched output would require external control with the following options:
1) fixed High/Low control (Hiigh x48, Low x44.1)
2) 2 x fixed enable controls (high enabled, low disabled)
) 2 x pulsed enable controls, 30 ms min. (one for x48 + one for x44.1)

Please, let me know if there is enough interest on the above board (at least 10 pcs to start the design and the production).
 
#3,235
wlowes said:
Thanks for confirming this Andrea. I had decided to implement your innovation exactly as specified. I respect your performance measurements. Why fix it if it is not broken?
Click to expand...

Sorry, with "replacement parts" I meant "cross reference" in the BOM if some components were not in stock from Mouser.
In other words I meant that no parts in the BOM should be replaced with similar, you should get exactly the listed parts.
 
#3,236
andrea_mori said:
This is another thing.
If you would get the best form the 5/6 MHz oscillators I suggest to wait for the new discrete DAC Lite.

The board I was talking about would be designed with the following features:
- Input: 2 sine wave oscillators, one for each sample rate families)
- sine to square converters (LTC6957)
- double output (fanout, 2 output for each sample rate families to distribute the clock)
- or swithed output, single sample rate family at a time with parallel output (fanout to distribute the clock)

Double output and switched output would be alternatives to each other.
Switched output would require external control with the following options:
1) fixed High/Low control (Hiigh x48, Low x44.1)
2) 2 x fixed enable controls (high enabled, low disabled)
) 2 x pulsed enable controls, 30 ms min. (one for x48 + one for x44.1)

Please, let me know if there is enough interest on the above board (at least 10 pcs to start the design and the production).
Click to expand...

I am interested. (would use 2 x fixed enable controls, high enabled, low disabled)
 
Last edited:
#3,237
wlowes said:
Thanks for confirming this Andrea. I had decided to implement your innovation exactly as specified. I respect your performance measurements. Why fix it if it is not broken?
Click to expand...

i did not mean any fix, what i actually mean is, say you need two (not just one) 1K 1% R, i can buy 20 of them and it should not be difficult to hand pick just two which is well within 0.1%.
but as Andrea has confirmed, just stick with specified. thanks Andrea.
 
Status
Not open for further replies.
Top Bottom