You could try to find any patent applications for crystal oscillators that achieve low phase noise by using two crystals. Hopefully that will explain how they use the two crystals and why that helps, although many patent applications are utterly unreadable.
The oven part should be possible to DIY. Put the oscillator in a thermally insulated enclosure with a power resistor as heater and some sort of temperature sensor and make a feedback loop that keeps the temperature stable. Ideally you should try to get the temperature to settle at a value where the temperature coefficient of the crystal is zero.
The oven part should be possible to DIY. Put the oscillator in a thermally insulated enclosure with a power resistor as heater and some sort of temperature sensor and make a feedback loop that keeps the temperature stable. Ideally you should try to get the temperature to settle at a value where the temperature coefficient of the crystal is zero.
In fact the datasheet shows a block diagram:
http://www.ndk.com/en/ad/2013/001/pdf/c_NH47M47LA_e.pdf
One crystal is used in a crystal oscillator, the other in a crystal filter that filters off phase noise of the oscillator (for frequency offsets > 100 Hz, according to the datasheet).
http://www.ndk.com/en/ad/2013/001/pdf/c_NH47M47LA_e.pdf
One crystal is used in a crystal oscillator, the other in a crystal filter that filters off phase noise of the oscillator (for frequency offsets > 100 Hz, according to the datasheet).
If you have a low-impedance output on your oscillator that can drive a small capacitive voltage divider, you can probably make a filter like this:
capacitive voltage divider (C1, C2) - crystal - capacitor to ground (C3)
where (C1+C2)*C3/(C1+C2+C3) must be the crystal's load capacitance, and C3 must probably be a trimmer to fine-tune the filter's centre frequency.
As a netlist:
V1 1 0 // input voltage from the crystal oscillator
C1 1 2
C2 2 0
X1 2 3 // filter crystal
C3 3 0 // trimmer
The output is then taken from node 3, so across trimmer capacitor C3.
The trick is to dimension everything such that the crystal is driven hard, but not (far) above the allowable drive level.
The most difficult part is to convert the output signal of the filter into a square wave without messing up the phase noise too much. Fast bipolar comparator maybe?
capacitive voltage divider (C1, C2) - crystal - capacitor to ground (C3)
where (C1+C2)*C3/(C1+C2+C3) must be the crystal's load capacitance, and C3 must probably be a trimmer to fine-tune the filter's centre frequency.
As a netlist:
V1 1 0 // input voltage from the crystal oscillator
C1 1 2
C2 2 0
X1 2 3 // filter crystal
C3 3 0 // trimmer
The output is then taken from node 3, so across trimmer capacitor C3.
The trick is to dimension everything such that the crystal is driven hard, but not (far) above the allowable drive level.
The most difficult part is to convert the output signal of the filter into a square wave without messing up the phase noise too much. Fast bipolar comparator maybe?
The datasheet seems to imply that the second crystal (acting as a filter) improves things beyond 100Hz. The improvement below 100Hz comes from other things such as careful mechanical and thermal design. This is roughly what one might expect. I guess an expert on oscillator noise with good fabrication facilities might be able to DIY something similar. Your average audio DIYer? No chance!
The second crystal filters phase noise from 100Hz away, and further. A second crystal oscillator is no help, because it is an oscillator - not a filter.
The oven and mechanical design deal with closer phase noise by isolating the circuit and crystal from outside influence.
All this has already been said in this thread or in the datasheet - apart from the knowledge that a crystal filter and a crystal oscillator are not the same thing, which is basic electronics.
The oven and mechanical design deal with closer phase noise by isolating the circuit and crystal from outside influence.
All this has already been said in this thread or in the datasheet - apart from the knowledge that a crystal filter and a crystal oscillator are not the same thing, which is basic electronics.
I don't know what you mean. Crystals are available from crystal manufacturers. The best way of obtaining two identical crystals is to buy them together. When buying a crystal you obviously have to specify what you want, although a good manufacturer will be able to help you.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
- Home
- Source & Line
- Digital Line Level
- NDK DuCULoN: How it works? Possible to DIY?