Dear all,
I need to order some TCXO for my 1541a RE-clock kit.
Can anyone told me that TCXO really good in re-clock circuit because its more expensive than standard 30ppm oscillator.
The best TCXO was less than <1ppm.
If I choose <2ppm or <5ppm TCXO will cheaper.
anyone can give me some opinions.
thx
thomas
I need to order some TCXO for my 1541a RE-clock kit.
Can anyone told me that TCXO really good in re-clock circuit because its more expensive than standard 30ppm oscillator.
The best TCXO was less than <1ppm.
If I choose <2ppm or <5ppm TCXO will cheaper.
anyone can give me some opinions.
thx
thomas
tube-lover said:
Hi,
The ppm level is not relevant as long as it is within the redbook specification (+/- 100pm).
You should look for the jitter specification.
best regards
Dear Guido Tent,
very thanks for your advise. I try to ask the factory already. But they only produce very high precision standard TCXO, XO in very low ppm. Very low ppm & high accurancy as 1/10000% resisters for measurement equipment. lots of parts produced from this factory was gold legs.
They( This Factory) did not produce any parts for audio use. As the NPC, not mainly in audio field. So they did not understand how jitter effect the sound quality.
BTW, this is sample of the coming precision ~2ppm TCXO from
-10~+85 stability TCXO. Hope diyers will like them.
I will try to put this into my coming re-clock kit. Hope the performance will improve more.
thx
thomas
very thanks for your advise. I try to ask the factory already. But they only produce very high precision standard TCXO, XO in very low ppm. Very low ppm & high accurancy as 1/10000% resisters for measurement equipment. lots of parts produced from this factory was gold legs.
They( This Factory) did not produce any parts for audio use. As the NPC, not mainly in audio field. So they did not understand how jitter effect the sound quality.
BTW, this is sample of the coming precision ~2ppm TCXO from
-10~+85 stability TCXO. Hope diyers will like them.
I will try to put this into my coming re-clock kit. Hope the performance will improve more.
thx
thomas
Attachments
Well, I don't think so.
PPM is very important to me I even use 45.1584MHz clock for my heavily modded SCD-1 by an external proffesional synthesizer.
http://www.programmedtest.com/pts160.html
Mine is oven controlled version, OCXO.
The fellow in here uses Cecium clock.
Rubidium is not good though Teac uses in their Esoteric G0-s.
My choice for the best combo will be a very good (but expensive) oven controlled x-tal osc. (will costs you more than $1000)
Kohjin
PPM is very important to me I even use 45.1584MHz clock for my heavily modded SCD-1 by an external proffesional synthesizer.
http://www.programmedtest.com/pts160.html
Mine is oven controlled version, OCXO.
The fellow in here uses Cecium clock.
Rubidium is not good though Teac uses in their Esoteric G0-s.
My choice for the best combo will be a very good (but expensive) oven controlled x-tal osc. (will costs you more than $1000)
Kohjin
Kohjin Yamada said:
You can purchase a used HP3586C (selective level receiver) with an ovenized 10MHz XO for around $100 -- only problem is that it needs 24 hours to warm up -- over a year they may wander a few PPM, NIST traceable etc. I use mine with a pair of HP3336B oscillators as the master clock -- you only need one for a whole mess of equipment!
You occasionally see Tracor Rubidium oscillators on the Bay -- not easy to fix, however.
MY CD clock -- garden variety !!!!! I just listen (sorry Elso).
Hi,
I always turned switch on the synthesizer of course, why not!
Forget about Rubidium OSC we tried and confirmed not good.
Kohjin
I always turned switch on the synthesizer of course, why not!
Forget about Rubidium OSC we tried and confirmed not good.
Kohjin
Jocko Homo
Dear Jocko,
Let me ask a silly question?
How can get standard XO in Low jitter. I saw the factory
selection for the crystal was use a equipment with XXXX.XXXXXXXXXXXXMhz to test & selection for them.
Will this selection can easy choose out the low jitter one??
thx
thomas
Dear Jocko,
Let me ask a silly question?
How can get standard XO in Low jitter. I saw the factory
selection for the crystal was use a equipment with XXXX.XXXXXXXXXXXXMhz to test & selection for them.
Will this selection can easy choose out the low jitter one??
thx
thomas
Well, for starters...........you need a very quiet power supply.
The circuit is important, although something as simple as a Pierce can work well.
Most "standard" XOs have poor jitter, because the circuit is not optimised for that, or really anything else. Except price.
High accuracy ones are optimised for accuracy. Some might spec jitter, but I doubt it.
The long term drift of a crystal is mostly a function of how clean the quartz is.
Jocko
The circuit is important, although something as simple as a Pierce can work well.
Most "standard" XOs have poor jitter, because the circuit is not optimised for that, or really anything else. Except price.
High accuracy ones are optimised for accuracy. Some might spec jitter, but I doubt it.
The long term drift of a crystal is mostly a function of how clean the quartz is.
Jocko
Kohjin Yamada said:
Hi
If you care or the DC component of the spectral part of the jitter, ppm spec is important.
I do however not mind that DC part, as long as the ppm value is within the redbook spec
Questions to all contributors
- Do we know what jitter is
- Do we know how to measure it and do we have a preference for the unit of measurement
- Do we know how jitter affects DA conversion and what jitter properties of a clock count
It seems many of us are ivolved in the dicussion on a toppic they do not (fully) grasp
Answers to these questions are required to improve the discussion
Guido
Guido Tent said:
All you need to know about jitter -- courtesy of Tektronix:
"Understanding and Characterizing Timing Jitter"
http://www.tek.com/Measurement/scopes/jitter/55W_16146_1.pdf
jackinnj said:
For those who think they can accurately measure time using a scope, dream on.
My range of interest is ps, and scope timebase jitter is 20 ps. (e.g. Tektronix 7104, which I use at my Philips job).
For those really interested, look at the Wavecrest 2075, for example.
cheers
Guido Tent said:
Hi Guido,
this comment does little to clarify the measurement problem for many users. The TDS7104 is specified at 8ps rms trigger jitter which will probably give 20ps pk-pk trigger jitter for many types of signal. BTW, the newer members of the family are four times better than this.
However, you're missing the point. Trigger jitter is NOT important for *some* of the more modern jitter measurement packages. They trigger only ONCE and then fill their (large) memory with data points. This data is then analysed off-line to provide the jitter numbers. The jitter of the first and only trigger point is completely irrelevant.
Before all of you grab your mice or catalogs to research this, a few points to note:
(1) You need a very good scope to start with (read very expensive)
(2) The scope must be able to sample the signal in real-time (no repetitive interleaved sampling allowed).
(2) You need to load it up with megabytes of memory ( at least 16 to 32Mbytes ). Scope memory from either HP or Tektronix is not cheap - you'll be paying the price of a small car
(3) The jitter analysis packages themselves aren't cheap.
Just for the record, I've used the Wavecrest box and IMHO it was very overpriced. The hardware and software was flakey and the box needed to be recalibrated frequently.
James
You want to ask them about the dBc (this value describes the differance in level between the level for the nominal freq and the level for the noise (unwanted) at a specific frequency. Noise is always higher near the nominal frequency (the clock freq-fo) and falls in level with rising freq. The freq fo...+20khz is of cource the range You are interested in as this will have direct corellation to the jitter in the audible range. Try to get something like -140dBc or better at 20 hz and You have a good clock ;-). PPM as already stated is uninportant but sadly often mentioned in discussion.
Further, to really get a jitter free design there are other important aspcts but it will never get better than the clock that feeds the D/A. Some decoders has a tendency to generate jitter related to sub-codes (which actually is one reson why a duplicated CD migth sound better as the burning software dosent write them to the duplicates. Also, power (DC) could have mains "dirt" coming through so You get 60-120-240 hz jitter.
All in all, what the clock looks like at the clock pin on Your D/A chip is what matters.
Good luck !
/
Further, to really get a jitter free design there are other important aspcts but it will never get better than the clock that feeds the D/A. Some decoders has a tendency to generate jitter related to sub-codes (which actually is one reson why a duplicated CD migth sound better as the burning software dosent write them to the duplicates. Also, power (DC) could have mains "dirt" coming through so You get 60-120-240 hz jitter.
All in all, what the clock looks like at the clock pin on Your D/A chip is what matters.
Good luck !
/
Hi everybody,
I though I share some of my thoughts:
Ovenizing the oscillator helps with the long term frequency stability (drift) not the jitter (short term stability).
On the other hand ovenized oscillators are made to more stringent specifications
and this includes usually also their phase noise so going with ovenized XO is a good choice.
What factory is using for selection does not prove much.
Manufacturing equipment must have an order of magnitude better accuracy than the product to be meaningful.
Moreover what they should be measuring (and what we are intersted in) is the phase noise.
In the lab you are going to use spectrum analyzer for that.
Another thing: anytime you hear "frequency synthesizer" think PLL (or at least some type of VCO, VXO etc)
This tunability means aditional components introducing more noise compared to oscillator optimized for single frequency.
At a time I was interested in building a low phase noise XO (CD upgrade 😉 ) myself and I dug some info on a subject:
Generally to lower the phase noise of the oscillator you need to reduce the interference
signal injected into oscillator resonator element (LC, crystal etc)
The sources of interference are multiple:
amplifier active component noise
power noise
amplifier nonlinearity
passive component thermal noise
From my understanding in circuit terms it means:
clean supplies (mentioned already in previous posts),
low noise active components,
circuit topology to achieve as high as possible Q of resonator element
(high load impedance for parallel resonator, low for serial), high Q helps filter out noise and harmonics
amplifier working in a class with a low distortion (note that most of the oscillators operate in class C)
or with soft clipping (likely FET is better than BJT), low parasitic capacitances (they are generally nonlinear)
so high frequency transistors are preferred
See below for a reference to a good summary with some more suggestions.
Cheers,
Przemek
The best summary on the subject I have found
http://www.odyseus.nildram.co.uk/Systems_And_Devices_Files/PhaseNoise.pdf
In summary, in order to minimise the phase noise of an oscillator we therefore need to ensure
the following:-
(1) Maximise the Q.
(2) Maximise the power. This will require a high RF voltage across the resonator and will be
limited by the breakdown voltages of the active devices in the circuit.
(3) Limit compression. If the active device is driven well into compression, then almost
certainly the noise Figure of the device will be degraded. It is normal to employ some form of
AGC circuitry on the active device front end to clip and hence limit the RF power input.
(4) Use an active device with a low noise figure.
(5) Phase perturbation can be minimised by using high impedance devices such as GaAs
Fet’s and HEMT’s, where the signal-to-noise ratio or the signal voltage relative to the
equivalent noise voltage can be very high.
(6) Reduce flicker noise. The intrinsic noise sources in a GaAs FET are the thermally
generated channel noise and the induced noise at the gate. There is no shot noise in a GaAs
FET, however the flicker noise (1/f noise) is significant below 10 to 50MHz. Therefore it is
preferable to use bipolar devices for low-noise oscillators due to their much lower flicker
noise, for example a 2N5829 Si Bipolar transistor, has a flicker corner frequency of
approximately 5KHz with a typical value of 6MHz for a GaAs FET device. The effect of flicker
noise can be reduced by RF feedback, eg an un-bypassed emitter resistor of 10 to 30 ohms
in a bipolar circuit can improve flicker noise by as much as 40dB.
(7) The energy should be coupled from the resonator rather than another point of the active
device. This will limit the bandwidth as the resonator will also act as a band pass filter.
I though I share some of my thoughts:
Ovenizing the oscillator helps with the long term frequency stability (drift) not the jitter (short term stability).
On the other hand ovenized oscillators are made to more stringent specifications
and this includes usually also their phase noise so going with ovenized XO is a good choice.
What factory is using for selection does not prove much.
Manufacturing equipment must have an order of magnitude better accuracy than the product to be meaningful.
Moreover what they should be measuring (and what we are intersted in) is the phase noise.
In the lab you are going to use spectrum analyzer for that.
Another thing: anytime you hear "frequency synthesizer" think PLL (or at least some type of VCO, VXO etc)
This tunability means aditional components introducing more noise compared to oscillator optimized for single frequency.
At a time I was interested in building a low phase noise XO (CD upgrade 😉 ) myself and I dug some info on a subject:
Generally to lower the phase noise of the oscillator you need to reduce the interference
signal injected into oscillator resonator element (LC, crystal etc)
The sources of interference are multiple:
amplifier active component noise
power noise
amplifier nonlinearity
passive component thermal noise
From my understanding in circuit terms it means:
clean supplies (mentioned already in previous posts),
low noise active components,
circuit topology to achieve as high as possible Q of resonator element
(high load impedance for parallel resonator, low for serial), high Q helps filter out noise and harmonics
amplifier working in a class with a low distortion (note that most of the oscillators operate in class C)
or with soft clipping (likely FET is better than BJT), low parasitic capacitances (they are generally nonlinear)
so high frequency transistors are preferred
See below for a reference to a good summary with some more suggestions.
Cheers,
Przemek
The best summary on the subject I have found
http://www.odyseus.nildram.co.uk/Systems_And_Devices_Files/PhaseNoise.pdf
In summary, in order to minimise the phase noise of an oscillator we therefore need to ensure
the following:-
(1) Maximise the Q.
(2) Maximise the power. This will require a high RF voltage across the resonator and will be
limited by the breakdown voltages of the active devices in the circuit.
(3) Limit compression. If the active device is driven well into compression, then almost
certainly the noise Figure of the device will be degraded. It is normal to employ some form of
AGC circuitry on the active device front end to clip and hence limit the RF power input.
(4) Use an active device with a low noise figure.
(5) Phase perturbation can be minimised by using high impedance devices such as GaAs
Fet’s and HEMT’s, where the signal-to-noise ratio or the signal voltage relative to the
equivalent noise voltage can be very high.
(6) Reduce flicker noise. The intrinsic noise sources in a GaAs FET are the thermally
generated channel noise and the induced noise at the gate. There is no shot noise in a GaAs
FET, however the flicker noise (1/f noise) is significant below 10 to 50MHz. Therefore it is
preferable to use bipolar devices for low-noise oscillators due to their much lower flicker
noise, for example a 2N5829 Si Bipolar transistor, has a flicker corner frequency of
approximately 5KHz with a typical value of 6MHz for a GaAs FET device. The effect of flicker
noise can be reduced by RF feedback, eg an un-bypassed emitter resistor of 10 to 30 ohms
in a bipolar circuit can improve flicker noise by as much as 40dB.
(7) The energy should be coupled from the resonator rather than another point of the active
device. This will limit the bandwidth as the resonator will also act as a band pass filter.
and for those with an allergy to the forum search button:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=25405
http://www.diyaudio.com/forums/showthread.php?s=&threadid=25405
Re: Wow PB!
That post helps in building oscillators, and indeed is a concise wrapup of what counts, allthough the slicer part is missing to make a square out of a sine wave.
It was about jitter measurements, which follows after you built one......
cheers
Robert Morin said:
That post helps in building oscillators, and indeed is a concise wrapup of what counts, allthough the slicer part is missing to make a square out of a sine wave.
It was about jitter measurements, which follows after you built one......
cheers
- Status
- Not open for further replies.