There is something wrong with a crystal oscillator if you can measure its phase noise with a spectrum analyzer.
Phase detectors can generate high levels of phase noise. Long ago I used to use the MC4044 phase detector in RF synthesisers and it turned out that the D type based phase detector had a dead zone causing the thing to constantly hunt to and fro, modulating the VCO
Phase detectors can generate high levels of phase noise. Long ago I used to use the MC4044 phase detector in RF synthesisers and it turned out that the D type based phase detector had a dead zone causing the thing to constantly hunt to and fro, modulating the VCO
I don't own an FFT, can you suggest a suitable soundcard and the right software (cheap) to do the test? Or other inexpensive gear to do this?
I believe that a bandwidth from 1 Hz to 100 kHz is more than enough for audio.
We could use the Driscoll at 22/24 MHz with harmonic SC-Cut crystals, since they have a Q around 500k, a good starting point for the Driscoll oscillator that maximizes the loaded Q. Adding the VCO capability.
To keep the project not too complicated would be better starting with two frequencies (22/24) dividing by 2/4 and multiplying by 2/4. D-type flip-flop could easily do the division job. I ask Gerhard about the best way to do the multiplication job.
Phase noise measurement tool
I have just started a new thread about the opportunity of building a phase noise measurement system:
http://www.diyaudio.com/forums/equi...urement-system-digital-audio.html#post4812029
Please, use the above, leaving this thread related to oscillators only
I have just started a new thread about the opportunity of building a phase noise measurement system:
http://www.diyaudio.com/forums/equi...urement-system-digital-audio.html#post4812029
Please, use the above, leaving this thread related to oscillators only
Last edited:
Really high performance spectrum analyzers can show phase noise in inexpensive crystal oscillators but not at this level.
The phase detector I linked is a special double balanced mixer designed for high DC out per the data sheet so it should be well suited to this task.
You have to use an unbuffered '04 7V tolerant.
The parts value in the BOM are OK for 6.7V, power dissipated in the crystal is around 100uW.
Hi Andrea.
Kudos on your nice work.
The inverter-based Pierce is selling it short.
Logic devices are not designed for:
- linear amplification,
- low noise figures (as amps)
- or "buffering".
They should only be used for their designed purpose: high SR pulse generation.
The Pierce uses the crystal as a bandpass + feedback element: this is a powerful combination for lower frequency oscillators.
A correctly-designed Pierce will easily outperform even the Driscoll @ say <20 MHz, due to its circuit features.
Stay away from multiple cascaded or parallelled gates for "buffering". They are not buffers & will not work as such.
Keep up the good work gentlemen.
Last edited:
I just noticed a slight mistake in my text above.
What I meant to say was:
"A correctly-designed Pierce will even be able to outperform the Driscoll @ say <20 MHz, due to its circuit features."
This was actually tested at the lab back in 2010.
Measurements were also performed on an Inverter-based Pierce, (same xtal) and its performance was far worse.
-----------------------------------------------------------------
Keep up the good work gentlemen.
...
...
Alexiss .. I read what you are writing and I have no doubt that you have both experience and knowledge in this field ... But personally I have this feeling of being the rabbit tempted with a carrot that it cannot reach ... Any chance you can be specific about which alternatives you would suggest?
BTW I am concluding laying out a Pierce oscillator today (according to Andrea's design) and so if you have suggestions for changes I'd appreciate if you can share them in due time.
Cheers,
Jesper
...
...
...
Alexiss .. I read what you are writing and I have no doubt that you have both experience and knowledge in this field ... But personally I have this feeling of being the rabbit tempted with a carrot that it cannot reach ... Any chance you can be specific about which alternatives you would suggest?
BTW I am concluding laying out a Pierce oscillator today (according to Andrea's design) and so if you have suggestions for changes I'd appreciate if you can share them in due time.
Cheers,
Jesper
Thankfully there are many more knowledgable people here.
My recommendations on the Pierce is a discrete design, with a disrete buffer - yes, a buffer, not an inverter used as such.
I think I saw one in a book recently. Just google it & if u dont find anything, let me know.
The Inverter-based Pierce is ok for u-processors, but not for high performance design that people crave on this thread.
Pierce has, in my opinion, the best performance <20 MHz with normal transistors. It is an deceptively simple circuit, but dont be fooled.
Colpitts and Driscoll are very good but not as good as Pierce at low f.
I know Herbert does not agree, and i'm willing to listen to him, he is very knowledgable in osc design.
My recommendations on the Pierce is a discrete design, with a disrete buffer - yes, a buffer, not an inverter used as such.
I think I saw one in a book recently. Just google it & if u dont find anything, let me know.
The Inverter-based Pierce is ok for u-processors, but not for high performance design that people crave on this thread.
Pierce has, in my opinion, the best performance <20 MHz with normal transistors. It is an deceptively simple circuit, but dont be fooled.
Colpitts and Driscoll are very good but not as good as Pierce at low f.
I know Herbert does not agree, and i'm willing to listen to him, he is very knowledgable in osc design.
Hi Alexiss ... Thanks for your prompt response and suggestion.
Hmmm ... as it is I will admit to not having the discernment in this field that will (quickly) allow me to sort between the circuits I find when googling. Also, as it is the Pierce circuitry Andrea has suggested won't feed a very long path - it goes into a divider which I reckon also has quite a say in the final quality of the oscillator.
Here I use a FF - but am a bit unsure of which FF to choose - but guessing that one of the reasons why Herbert's oscillator achieves these fine phase noise figures may be due to the high voltages I most likely will use the same series FF (i.e. an HC74 version) and feed it a high voltage.
However, if you or one of the other "oscillator insighters" here has a better suggestion I'm all ears (although - as I wrote - I'm wrapping up this layout today).
Cheers,
Jesper
Hmmm ... as it is I will admit to not having the discernment in this field that will (quickly) allow me to sort between the circuits I find when googling. Also, as it is the Pierce circuitry Andrea has suggested won't feed a very long path - it goes into a divider which I reckon also has quite a say in the final quality of the oscillator.
Here I use a FF - but am a bit unsure of which FF to choose - but guessing that one of the reasons why Herbert's oscillator achieves these fine phase noise figures may be due to the high voltages I most likely will use the same series FF (i.e. an HC74 version) and feed it a high voltage.
However, if you or one of the other "oscillator insighters" here has a better suggestion I'm all ears (although - as I wrote - I'm wrapping up this layout today).
Cheers,
Jesper
Hi Alexiss ... Thanks for asking ...
I basically use the circuitry that Andrea has outlined in his TWTMC- D&D (link here:
http://www.diyaudio.com/forums/digi...itter-crystal-oscillator-123.html#post4759763 - top part of schematic on P. 2)
i.e one dual FF IC (74LVC74 or 74HC74) to make the division. The Pierce circuitry also is similar to Andrea's although with some layout changes. I don't use any resistors between the 74HC(U)04 and the FF input - the track length is about 34 mms. Following the FF there are resistors to the ADC inputs. Supply voltages are 6.7 - 6.8 VDC.
I hope this gives an idea about the setup I intend ... Might you have any ideas/suggestions?
God dag herfra
Jesper
The crystal is a 6.144 MHz AT cut from Laptech (courtesy Andrea) which I then divide by /2/2 (i.e. 4) to arrive at 1.536 MHz.
I basically use the circuitry that Andrea has outlined in his TWTMC- D&D (link here:
http://www.diyaudio.com/forums/digi...itter-crystal-oscillator-123.html#post4759763 - top part of schematic on P. 2)
i.e one dual FF IC (74LVC74 or 74HC74) to make the division. The Pierce circuitry also is similar to Andrea's although with some layout changes. I don't use any resistors between the 74HC(U)04 and the FF input - the track length is about 34 mms. Following the FF there are resistors to the ADC inputs. Supply voltages are 6.7 - 6.8 VDC.
I hope this gives an idea about the setup I intend ... Might you have any ideas/suggestions?
God dag herfra
Jesper
- Status
- Not open for further replies.