There are two simulataneous conversations going on in this thread!
When talking about active XO, there is no way to get a square wave through both halves because we are intentionally frequency limiting. Now, you may be interested in the summed response looking like a square wave. In addition to preserving the phase relationship, you want a high "slew rate" for the high-pass capacitors. This is NOT a problem at audio frequencies.
The low pass is a little trickier but generally it is possible to have a low enough high pass on the coupling cap that there is not much tilt. It is questionable how low you want to go because you will just waste power and potentially get IMD from subsonics.
In conclusion, for AUDIO FREQUENCIES, there is not much problem with coupling caps if chosen properly.
Guido,
I think we are in agreement. I was just trying to steer the thread back to what I thought was the original subject.
When talking about active XO, there is no way to get a square wave through both halves because we are intentionally frequency limiting. Now, you may be interested in the summed response looking like a square wave. In addition to preserving the phase relationship, you want a high "slew rate" for the high-pass capacitors. This is NOT a problem at audio frequencies.
The low pass is a little trickier but generally it is possible to have a low enough high pass on the coupling cap that there is not much tilt. It is questionable how low you want to go because you will just waste power and potentially get IMD from subsonics.
In conclusion, for AUDIO FREQUENCIES, there is not much problem with coupling caps if chosen properly.
Guido,
I think we are in agreement. I was just trying to steer the thread back to what I thought was the original subject.
A 8 said:
Hi all,
Yes, confusing
C coupling and its' habbits principally do not depend on frequency, however the appplication, reaction and sensitivity does.
Yes, I measured a lot, but do not directly give away the results, if you do not mind, as lots of time and effort where put in.
all the best
Who among you actually carried out easurements on.......
I have, but I no longer have access to all the equipment that you still do, Guido.
So I have to come up with different approaches. In any case, you are right.....it is a lot of work, and most people here don't have our capabilities.
Jocko
I have, but I no longer have access to all the equipment that you still do, Guido.
So I have to come up with different approaches. In any case, you are right.....it is a lot of work, and most people here don't have our capabilities.
Jocko
Sometimes passing perfect square waves is not the best thing to do. That means higher EMI, higher transient currents, etc. that could create their own problems. I have experienced cases where a bit of slew rate limiting has resulted in better system performance, though not specifically in audio, though the output gates on some logic families (on the output of the XO) are prone to ground bounce and a little bit of slew rate limiting could help the final effect.
However, what the issue here may be is more a matter of implementation. If the output of the XO is simply a sine wave (which internally they all are close to at some point), then AC coupling and limiting the bandwidth is not really an issue.
A 8, did you have a specif application or instance in mind?
Alvaius
However, what the issue here may be is more a matter of implementation. If the output of the XO is simply a sine wave (which internally they all are close to at some point), then AC coupling and limiting the bandwidth is not really an issue.
A 8, did you have a specif application or instance in mind?
Alvaius
I don't have the equipment to carry out jitter measurements either, but I do have a spectrum analyzer and I've been experimenting with the oscillator Jocko posted (the one with the common-base amp to pick off the signal) and the output Q is higher without the coupling caps. I was using a 20pF mica, but if I just jumper it the Q improves. Presumably a higher Q on a sine oscillator is always desirable.
The coupling cap is not needed as long as the common mode signal is within the input range of the next thing downstream. In this case it is a 5V comparator and the common mode is right in the middle at about 2.5V.
-jwb
The coupling cap is not needed as long as the common mode signal is within the input range of the next thing downstream. In this case it is a 5V comparator and the common mode is right in the middle at about 2.5V.
-jwb
alvaius,
I just finished my new dac CS8420/DF1704/PCM1704K and made a 24.576 Mhz clock and shapers based on singlegate logic with individual decoupling.
I want to change the clock to a better one and need to decide what to consider.
The clock is directly connected to bitclock in on the dac chips so I expect the clock quality to have a very direct impact on the final result.
I just finished my new dac CS8420/DF1704/PCM1704K and made a 24.576 Mhz clock and shapers based on singlegate logic with individual decoupling.
I want to change the clock to a better one and need to decide what to consider.
The clock is directly connected to bitclock in on the dac chips so I expect the clock quality to have a very direct impact on the final result.
I think the only argument for very high bandwidth in jitter-sensitive application is that the time spent in the transition region is very short, leading to better rejection of various types of signal modulation. But of course this causes the problems stated. I think it is all a tradeoff.
Murata makes some special ferrite beads that are designed to pass up to a given bandwidth and then sharply reject higher frequencies--they are intended specifically for EMI limiting without screwing up the transitions. I've never tried them in a jitter-sensistive clock path though.
Murata makes some special ferrite beads that are designed to pass up to a given bandwidth and then sharply reject higher frequencies--they are intended specifically for EMI limiting without screwing up the transitions. I've never tried them in a jitter-sensistive clock path though.
24.576 MHz low jitter Clock
Hi A8,
You can built my clock i.e the KWAK-CLOCK. If you use the email button below this post I will sent you the schematic.
Only a fundamental mode crystal will work like this one f.a.:
http://www.digikey.com/scripts/us/dksus.dll?Detail?Ref=22873&Row=336843
A 8 said:
Hi A8,
You can built my clock i.e the KWAK-CLOCK. If you use the email button below this post I will sent you the schematic.
Only a fundamental mode crystal will work like this one f.a.:
http://www.digikey.com/scripts/us/dksus.dll?Detail?Ref=22873&Row=336843
Jean-Paul:
I was refering to the tons of test equipment it takes.
JWB:
I have versions with and without the coupling cap. The cap was used to prevent long cables terminated in 75 ohms from mucking things up. Along with build out resistors.
When I get everything here back up and working, I will post some .wav files of the various clock thingies that I have around here. Should prove enlightening.
Ferrite beads act as resistors, that are frequency selective.
Bandwidth can be detrimental, as the peak currents give rise (inadvertant pun!) to ground bounce and crap like that.
I use a sine wave clock, as I pump it all over the lab, and square it up internally.
Jocko
I was refering to the tons of test equipment it takes.
JWB:
I have versions with and without the coupling cap. The cap was used to prevent long cables terminated in 75 ohms from mucking things up. Along with build out resistors.
When I get everything here back up and working, I will post some .wav files of the various clock thingies that I have around here. Should prove enlightening.
Ferrite beads act as resistors, that are frequency selective.
Bandwidth can be detrimental, as the peak currents give rise (inadvertant pun!) to ground bounce and crap like that.
I use a sine wave clock, as I pump it all over the lab, and square it up internally.
Jocko
alvaius said:
True about the EMI and bounce, and yes there are perfect trade offs. I have stated that many times, see also:
http://members.chello.nl/~m.heijligers/DAChtml/Supply_decoupling.pdf
I would not reccomend distributing a sine wave as that is far more susceptible due to the lower dV/dt.
My XO's put out a square wave
all the best
The problem with square waves.......
I use a cheap TV in my lab to detect EMI problems. A lot of stuff that I see generates tons. With all the clock signals I have going every which way, unless it was a ridiculously low level, it would mask the EMI the I am trying to detect. So sine wave it is. I have yet to see any phase noise detriment this way, and since I don't have a $40K 'scope to measure jitter with..........
The all time winner for generating EMI was the Audio Alchemy "Jitterbug". Took every VHF channel off of the air.
The original Arcam "Black Box" was a close second. Never measured the subsequent versions.
Jocko
I use a cheap TV in my lab to detect EMI problems. A lot of stuff that I see generates tons. With all the clock signals I have going every which way, unless it was a ridiculously low level, it would mask the EMI the I am trying to detect. So sine wave it is. I have yet to see any phase noise detriment this way, and since I don't have a $40K 'scope to measure jitter with..........
The all time winner for generating EMI was the Audio Alchemy "Jitterbug". Took every VHF channel off of the air.
The original Arcam "Black Box" was a close second. Never measured the subsequent versions.
Jocko
Re: The problem with square waves.......
Hello Jocko,
Does your TV "spectrum analyser" have an aerial antenna or cable ? In the latter case I am pretty sure the Audio Alchemy and the Arcam are not meeting international EMC requirements.....
It is my experience that low EMI also contributes to signal integrity and inmproves sound quality.
all the best,
Guido
Jocko Homo said:
Hello Jocko,
Does your TV "spectrum analyser" have an aerial antenna or cable ? In the latter case I am pretty sure the Audio Alchemy and the Arcam are not meeting international EMC requirements.....
It is my experience that low EMI also contributes to signal integrity and inmproves sound quality.
all the best,
Guido
It uses a dipole......
Which can be adjusted.......or moved close to the offending item.
No, I think that it is safe to say that they do not comply. Which is why they sometimes find their way into my shop when a disgruntled owner complains about them.
Yes, it always seems to make things sound better if you fix those problems.
I was unable to do much with the "Jitterbug". It used a DSP chip that caused most of the problems.
Another trick used by us ham radio types is to hit the TX button on a 2 meter handheld rig, and see what happens. Some pieces go nuts.
Jocko
Which can be adjusted.......or moved close to the offending item.
No, I think that it is safe to say that they do not comply. Which is why they sometimes find their way into my shop when a disgruntled owner complains about them.
Yes, it always seems to make things sound better if you fix those problems.
I was unable to do much with the "Jitterbug". It used a DSP chip that caused most of the problems.
Another trick used by us ham radio types is to hit the TX button on a 2 meter handheld rig, and see what happens. Some pieces go nuts.
Jocko
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.