I see Elso Kwak's much rated 'Kwak Clock 7' design: A Colpitts oscillator using a J309 FET decoupled to an AD8561 fast comparitor.
Q1) Why is this comparitor not arranged to use hysteresis - aka why is it not arranged as a schmitt trigger?
Q2) What happened to the Kwak Clock 8 design.
Q3) Out of interest: Has Elso left these forums?
Many thanks,
Andy
Q1) Why is this comparitor not arranged to use hysteresis - aka why is it not arranged as a schmitt trigger?
Q2) What happened to the Kwak Clock 8 design.
Q3) Out of interest: Has Elso left these forums?
Many thanks,
Andy
underwurlde said:
Hi Andy, you can reach Elso at elsokwak225athotmail.com
Replace at by monkeytail (@)
I am sure he will patiently answer all your questions.
Q3: he did not leave; he was banned.
Thanks chaps.
My thinbking was surely not implementing a schmitt trigger would cause jitter?
I think I am missing something here (hence the question)...
Andy
My thinbking was surely not implementing a schmitt trigger would cause jitter?
I think I am missing something here (hence the question)...
Andy
Re: Re: Kwak Clock & Schmitt
You have hysteresis when you have slow rising signals (with noise) and a much faster comparator. Must first thought is to use a schmitt trigger when I'll convert a sinus to square wave.QSerraTico_Tico said:
underwurlde said:
A schmitt trigger has hysteresis, and hysteresis couples energy back to the input. This will screw up the timing if it occurs at the decision point of the comparator.
Not so obvious perhaps?peranders said:My second thought is also that many parts in an oscillator aren't not so obvoius.😕
And why then?
underwurlde said:
As you observe, the Quack clock is nothing special in itself.
It is the sum of the parts which makes it better - regulation of the power rails, layout, use of a high-speed comparator instead of a schmitt trigger.
Ray (on the CD63/67 thread - http://www.raylectronics.nl/ ) made a single rail version which works well. I have seen other versions with a high-speed schmitt as well.
Some people swear by the Tent clock and similar - but what's in the can ????
In any case, the external clock is almost certainly fed into the DAC chip via a 'slow' cmos gate !!!
My advice - try it and see !!
Andy
Andy
Re: Re: Kwak Clock & Schmitt
Does it matter? Tent special order them with his requirements for frequency and jitter, and that's what matters. I am not sure he knows himself what is actually inside the cans.
poynton said:
Does it matter? Tent special order them with his requirements for frequency and jitter, and that's what matters. I am not sure he knows himself what is actually inside the cans.
Re: Re: Re: Kwak Clock & Schmitt
I was not refering to the Tent clock in particular - only as an example of a 'canned' type clock.
Of course, I know that these can be ordered in various grades and specifications.
The original question was - why use a comparator not a schmitt.
I was only trying to point out that many implementations use a schmitt already.
Andy
Christer said:
I was not refering to the Tent clock in particular - only as an example of a 'canned' type clock.
Of course, I know that these can be ordered in various grades and specifications.
The original question was - why use a comparator not a schmitt.
I was only trying to point out that many implementations use a schmitt already.
Andy
Re: Re: Re: Re: Kwak Clock & Schmitt
OK, so you were perhaps rather curious about whether there might be schmitt triggers inside the cans?
poynton said:
OK, so you were perhaps rather curious about whether there might be schmitt triggers inside the cans?
Re: Re: Re: Re: Re: Kwak Clock & Schmitt
I am almost certain that some cans do contain a high-speed cmos solution.
Andy
Christer said:
I am almost certain that some cans do contain a high-speed cmos solution.
Andy
Re: Re: Re: Re: Kwak Clock & Schmitt
Hysteresis introduces a phase shift dependent on signal amplitude when you're using it to square up a sine. Which can cause power supply sensitivity, and possibly/probably other bad stuff.
Hysteresis doesn't really offer much benefit in a crystal oscillator circuit anyway.
It's useful for avoiding spurious comparator output when you're comparing signals with noise, but if the slew of the oscillator's sine is faster than the instantaneous noise slew (which is not hard to achieve) you'll be fine. Hysteresis is also handy for avoiding spurious in cases such as a disconnected output, again, irrelevant for a crystal oscillator circuit.
Hysteresis introduces a phase shift dependent on signal amplitude when you're using it to square up a sine. Which can cause power supply sensitivity, and possibly/probably other bad stuff.
Hysteresis doesn't really offer much benefit in a crystal oscillator circuit anyway.
It's useful for avoiding spurious comparator output when you're comparing signals with noise, but if the slew of the oscillator's sine is faster than the instantaneous noise slew (which is not hard to achieve) you'll be fine. Hysteresis is also handy for avoiding spurious in cases such as a disconnected output, again, irrelevant for a crystal oscillator circuit.
Hysteresis also creates an uncertainty in the point at which the actual transition occurs from cycle to cycle and that is a contributor to overall clock jitter, and that is one of the reasons why the Kwak clock does not use it. Incidentally this is only an issue with signals that have relatively modest slopes, once squared up a Schmitt trigger is not going to do much if any additional harm as the rise time and fall times are short. (Not the case with a sinewave, iirc the Kwak clock has a sine output. Colpitts?)
- Status
- Not open for further replies.