Looking for low noise crystal ref cct's
Very few (1 or 2 companies) can achieve this type of result and it will cost S10k+ per osc
This a request or an offer?
I am not offering. I am requesting information or tips for designing a circuit that will have very low phase noise & noise.
This company says they make the lowest phase noise and noise floor osc in the industry
http://www.wenzel.com/oscillators.htm
I built while ago a very low phase noise 10MHz crystal oscillator with these performances.
The main schematic was based on this reference oscillator recommended by Analog Devices:
http://www.analog.com/UploadedFiles/...13367an419.pdf
I have used a 100MHz overtone crystal followed by a divider with 10.
The divider not only divide the frequency, but also divide the base frequency band FM and PM capability. The phase noise reduces 6 dB each time the frequency is divided.
The Q of the LC components, the NF of the LO transistor, and the quality of the crystal is very important getting a very good phase noise.
What about the noise floor? The challenge seems to be low phase noise 1Hz away and overall noise floor 1k-100k away
Indeed, getting the phase noise below -160dBc is challenging. I just wanted to add this link, where I'd put a schematic+photo. Phase noise measurement plot is available upon request.
https://www.edaboard.com/viewtopic.p...206&highlight=
rfmw, please post the plot if you don't mind.
With these things there are always tradeoffs. If you improve the close-in phase noise you raise the noise floor and the same happens when you drop the noise floor. Another problem to deal with is the crystal current. Driving the crystal harder improves both PN and NFloor but it will up the crystal's aging rate and long term drift. A typical 10MHz crystal has a Rs of between 35-75 Ohm. For low aging such a crystal should not dissipate more than 1mW ever and preferably around 100-500uW. This itself is a challenge. For crystals above 10MHz the drive level gets even smaller as the Rs goes down with increased frequency unless you use overtones.
If this was easy then there will be a lot more companies selling them and the price will be much lower. If you order one of these ultra low noise references, you pay up to S20k and delivery is about 2 months minimum. So that tells me it is not easy to achieve and the % yield is very small. My friends company ordered 3 units and waited more than 2 months for delivery. Also the osc company stated that you need to run the osc for at least 30 days before it settles and the specs will be achieved.
Here is the plot using Agilent's SSA.
Here is test data for the Wenzel osc 5MHz and 100MHz
http://www.wenzel.com/documents/ulntests.html
To get the kS performance a standard crystal is no where near good enough. These very high performance units use a BVA resonator. Don't ask me what BVA means, I don't know. They are a block of quartz with no attached metalisation. That is deposited on to two other pieces of quartz that are close to but not touching the resonator. Do a Google search for BVA resonator and it will lead you to a few ( very few ) papers and some patents.
I have spoken to someone in the past who needed A BVA oscillator for a time standard, ( part of a radio telescope I think ) and was told that the resonator takes ages to make & they don't always work as expected when they are made.
I doesn't help you with a circuit for your super oscillator, but you may find that the circuit is not much different from that in any text book, just the high performance resonator.
There are some good Ebooks on oscillator design here on EDAboard that will give you a start for the circuit. Getting the BVA, that's a different matter, I doubt if the few companies ( I know of only one ) that make them will sell you one.
Peter
Yes, I am aware that the Q of the crystal itself is a major factor and just one of the many facing the designer. It makes sense that a bigger "lump" of crystal will also allow higher driving currents. Close in the phase noise will be determined by resonator Q and far out by the noise floor of the circuit. Thanks for the info
Added after 9 minutes:
I wonder if they can achieve a low noise floor. A look at this page shows that they have very low aging and low allan variance but the noise floor at -145 is not great.
http://www.oscilloquartz.com/file/pdf/8607.pdf and the Wenzel units hit noise floor before 1kHz offset.
I don't know. I'm not sure how hard you can drive a BVA or what loss it has. Perhaps a combination of a low noise 'standard' crystal oscillator locked to a BVA might work to get the floor down, but would need to be carefully designed to keep the close in noise.
If your not looking for many oscillators with this performance then the kS may be well spent, and your time put to better use, but only you can answer that one.
Thanks
Peter
My friend's company sells a lot of time frequency standard systems which they manufacture. He owns the company. His customers are the like of Jet prop lab, USA navy, Boeing, observatory's etc. Part of the box is the reference osc. He has asked me to investigate the technology with the aim of producing them in his factory. The price according to him is not the major obstacle, it is the long delivery times which customers don't like.
Iam also working on low phase noise crystal oscillator.Anybody have a ckt for this pls help me.