Is that actually so? I would think just the opposite, that it's the effect of a complete truncation of upper harmonics.
If you simply chop out higher harmonics and leave everything else unchanged then you get symmetrical ripple - equal amounts of pre- and post-ringing. Jan's plots only show post-ringing. However, people do need to remember that ringing does not necessarily mean something has been added to the signal - subtraction can cause ringing too.
That's along with my line of thinking, looking at those scope pics - is it real ringing (an actual resonant condition), or is it Gibbs phenomenon? Maybe some of each?
I presume that square wave is 1kHz.
That's with digital filters - the "pre-ringing" building up before the transition is an artifact of sampled filters "looking ahead" at future samples, and can't happen with a traditional analog filter (unless it has a time machine).
Also, all the stated phase shift of this analog filter within its passband no doubt changes the resulting waveshape as well.
In fact, that can be what makes it audible, the resulting waveshape being different, causing a lower or (probably in this case) higher peak that might not be reproduced as accurately as the rest of the waveform. I read something about audible phase with speakers (maybe this thread?), with a 20Hz sawtooth being audibly different when inverted, due to the driver responding differently in the positive direction than the negative direction.
Yes.
Dave is right that in a true Gibbs, the ringing is symmetrical. Nonetheless, Jan's scope photo shows the ringing to be at the filter cutoff frequency, so probably not due to higher harmonics since the ringing seems to be a nearly pure damped sine wave.
No matter add you or remove something, wave shape is changed. When you add or remove something periodic, it will be changed periodically.
Not too surprising, if you have been listening to such a source for any longer period of time.
I'd call it "getting used to".
Listen to any type of sound you care to name long enough, and like it or not, you (we) get used to it, at least in part. In a while, it becomes one's reference, not meaning absolute quality, but what we are used to.
I am glad we agree, W., but this begs the logical question: why is this not recognized widely, instead of promoting only the first half of the whole?
We talk of symmetry, but in real life, we do not adhere to it.
As an example, Siemens Sikorel series of large power capacitors is THE ONE AND ONLY I have ever seen to quote in their specs the current slew rate, beside the voltage slew rate, which is also rarely seen.
Yes but those capacitor current and voltage slew rates are not a property of the cap, they are what the cap can handle without breaking.
Current slew into a cap and the resulting voltage are a function of whatever signal you impress on them (and some secondary effects like the resistance and inductance of cap connection wires and such).
jan
Yes the ringing is at the filter cutoff freq, I missed that. So, whatever the transient period (on the pic indeed a 1kHz square wave), the ringing is always at 15kHz then.
Makes it all the more surprising I heard a difference in 'sharpness'. I mean, I am pretty sure I'm rather deaf at 15kHz...
jan
No sure what you mean by this - in principle or just in audio power amps?
I can easily amplify voltage or current without amplifying power.
jan
Doh! Why don't I think before hitting 'post quick reply'?benb said:
PS that was a rhetorical question!
Completely agreed, but what's the sense in making a hell of fast amplifier, capable of say 200 V/uS as far as the electronics go, only to be slew limited to say 40 V/uS by the power supply capacitors?
This notwithstanding the accepted practice that you are done with slew rates if your slew rate is 0.5V/uS per every peak volt of output signal (SOURCE: National Semiconductor, Audio cookbook). This works out to 20V/uS at 20 kHz with nominally 100 Watts (40V) into 8 Ohms. And when do we ACTUALLY have a real world full power signal at 20 kHz?
And this is related to voltage; what about a power supply cap's capability to deliver current on demand, i.e. its current slew rate?
Don't you think all of the wonder capacitor manufacturers would just love to brag about these specs - if they could? Yet, they remain silent on technical aspects, and brag about silk and probably dairy milk they make them from.
These are unrelated issues. If you have an input LTP with a 2mA tail current and a 50pF compensation cap in the Vas, your max slew rate is 4V/uS. With a 10mA tail current and 10pF compensation your slew rate increases to 100V/uS.
But that has nothing to do with anything happening in the power supply caps. Those caps are re-charged just about once every 10 (or 8.25) mS for 1 or 2 mS duration with a very high current pulse, and discharged by the amp load current.
jan
But why very high slew rate is needed?
One example is Quad ESL.
Doe to very capacitive loading on amplifier full power bandwidth is just few kHz depending on amplifier current rating. That would translate to max. slew rate really low compared to today's standards. And, if you try to exceed it, amplifier's short circuit protection probably kicks in. Should be horrible?
Still some people think Quad esl are one of best sounding speakers.
So why bother with square waves ? IMO it is just useful to test stability of amp/preamps.
One example is Quad ESL.
Doe to very capacitive loading on amplifier full power bandwidth is just few kHz depending on amplifier current rating. That would translate to max. slew rate really low compared to today's standards. And, if you try to exceed it, amplifier's short circuit protection probably kicks in. Should be horrible?
Still some people think Quad esl are one of best sounding speakers.
So why bother with square waves ? IMO it is just useful to test stability of amp/preamps.
- Status
- Not open for further replies.