Why AC coupled XO?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Disabled Account
Joined 2002
We are talking digital electronics here and in digital electronics a square wave is important. The question is if a XO ( digital clock module ) should be coupled capacitively or not.

XO's are precision low jitter clock modules for use in cd-players etc.
 
A 8 said:
I see most of you guys use ac coupled output on your XO, is this only to remove dc offsets between XO and receiver or are there other reasons?

I assume it would be difficult getting a nice square wave through a cap!

dear all,

We do not need to transfer the DC part to get low jitter, as it is the slope of the square wave which is important.

However, we need to set, maintain or define the DC level of the next gate.

This can be done by the gate itself (you often see a feedback resistor in the range of 100 k-ohm to 1 M-ohm), or by the sending side, the clock, that is, when it is DC coupled.

I did not yet do extensive measurements on both options, though most people DC couple the XO's that I sell which very good results.

all the best
 
As far as my knowlegde goes a cap will always have a problem passing a steep slope.
I think skipping the cap would be the default choice to get the steepest slope possible (and reduce jitter effects in the next stage) but it seems to be the other way around:scratch:

Even commercial XO's with integrated shapers have a ac couple output so I am thinking there have to be a good reason....just dont know what it is
 
A 8 said:
As far as my knowlegde goes a cap will always have a problem passing a steep slope.
I think skipping the cap would be the default choice to get the steepest slope possible (and reduce jitter effects in the next stage) but it seems to be the other way around:scratch:

Even commercial XO's with integrated shapers have a ac couple output so I am thinking there have to be a good reason....just dont know what it is


Hi

Capacitors exhibit a very low impedance for high frequencies: They are excellent devices on passing on high slopes, cap coupling is also called highpass filtering

Increasing the slope, as I wrote earlier, also increases the RF energy (current), which in turn may, or very surely will lead to increased groundbounce, yielding higher jitter as well.

For this reason I always advice the use of series resistors

My XO's are DC coupledbest regards
 
Guido,
I agree as long as we are discussing sine wave but when it comes to square wave I do not agree that a cap can pass it unchanged, a caps "slew rate" is limited.

Not sure I completely understand "ground bounce" but assume it somehow relates to inferior grounding techniques?
Anyway my point of view is based on an more or less ideal situation.
 
A 8 said:
Guido,
I agree as long as we are discussing sine wave but when it comes to square wave I do not agree that a cap can pass it unchanged, a caps "slew rate" is limited.

Not sure I completely understand "ground bounce" but assume it somehow relates to inferior grounding techniques?
Anyway my point of view is based on an more or less ideal situation.

Hi

A square wave is just a bunch of sine waves, no difference here. Why are caps slew rate limitted ? We are talking coupling capacitors, not decoupling capacitors......

Groundbounce appears when currents run through ground: You do not need inferior grounding technique, any RF current through a non zero impedance (which is ground) causes a voltage gradient across that ground, called bounce.

best regards
 
tiroth said:
Depending on value Wima FKP have their series-resonant frequency up into the tens of MHz. To pass a decent 20kHz square wave you need only 140kHz of bandwidth (7th harmonic).

A good capacitor will have no problem with this.

dear Tiroth

Your assumption is right, but we are talking clock frequencies with fundamental in the range of 10 or 20 MHz

My measurements however show that most ceramic capacitors do this job quite well, and small polypropylene caps (like 10 nF) have their resonance very high as well.

best regards
 
I am really not an expert, but what I have understand is that a cap independent of wether it is a coupling cap or de-coupling cap will at the moment of "charge" behave as a factor dependant on the source resistance and cap value, thus creating a time factor limiting the speed at which it will change its electrical potential. (Yes I understand it is more of a moving target when the cap is used a coupling cap)
In order to accurately pass a square wave it would need to pass low frequencies as well as very high, which will lead to a higher cap value and ultimatly work against our goal of minimizing rise time.

If this is not the case I would really like to understand why, which kind of was my initial question.

BTW I know all commercial product typically uses a small ceramics cap in many application and as I said I just don't understand why it is used if it does more harm then good:confused:
 
A 8 said:
I am really not an expert, but what I have understand is that a cap independent of wether it is a coupling cap or de-coupling cap will at the moment of "charge" behave as a factor dependant on the source resistance and cap value, thus creating a time factor limiting the speed at which it will change its electrical potential. (Yes I understand it is more of a moving target when the cap is used a coupling cap)
In order to accurately pass a square wave it would need to pass low frequencies as well as very high, which will lead to a higher cap value and ultimatly work against our goal of minimizing rise time.

If this is not the case I would really like to understand why, which kind of was my initial question.

BTW I know all commercial product typically uses a small ceramics cap in many application and as I said I just don't understand why it is used if it does more harm then good:confused:


Hi,

When it comes to jitter the slopes are important, and a stable DC operating point of following gates. The latter can be made several ways.

Who among you actually carried out easurements on

- The transfer characteristics of RC coupled networks (with various qualities of components)?
- Jitter as function of the various ways of coupling and biasing ?

best regards
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.