I'm re-running the sim with a .5n maximum time step. It's taking a very long time to churn out the low speed answer, but the fast one came up with a few tenths nS more error.
Originally looking into it as a point of argument about the effects of carrier frequency, but the error at both speeds (with perfect signal impedance and power supplies to boot) looks pretty bad at ~-40dB. Of course this is with no feedback correction. What does the typical linear VAS have open loop? Don't know. Will try adding some capacitance from out to + in, but don't expect it to help much since the problem is when the output begins to budge not how fast the edges are.
Originally looking into it as a point of argument about the effects of carrier frequency, but the error at both speeds (with perfect signal impedance and power supplies to boot) looks pretty bad at ~-40dB. Of course this is with no feedback correction. What does the typical linear VAS have open loop? Don't know. Will try adding some capacitance from out to + in, but don't expect it to help much since the problem is when the output begins to budge not how fast the edges are.
Last edited:
There are two ways to teach people. One is based on making them repeat sentences like "crossover distortion is bad" or "A is better than B" or "24bit sounds better than 16 bit" like ****ing parrots, without having a clue about the physical phenomena they are naming. Then they will fight without even knowing why. We are suffering a world crisis in part as a consequence of decades of teaching people in that insane way. The creativity ( http://en.wikipedia.org/wiki/Neuroplasticity http://en.wikipedia.org/wiki/Synaptic_plasticity ) of the average individual is seriously limited.
The other way of teaching involves stimulating people to research, think and reach their own conclusions...
Last edited:
Thanks Eva for the summary. I am no expert at any one topology/technology, but I do know them all. 
other classes and topologies (A, B, D, G, H) will not cease to exist, but the playing field may look a bit different.
and now for some Dylan:
"Come gather ’round people
Wherever you roam
And admit that the waters
Around you have grown
And accept it that soon
You’ll be drenched to the bone
If your time to you is worth savin’
Then you better start swimmin’ or you’ll sink like a stone
For the times they are a-changin’"
this is key! (and is seen in all technology) It was indeed not so long ago that most all class-D was not done well. But things do change.
other classes and topologies (A, B, D, G, H) will not cease to exist, but the playing field may look a bit different.
and now for some Dylan:
"Come gather ’round people
Wherever you roam
And admit that the waters
Around you have grown
And accept it that soon
You’ll be drenched to the bone
If your time to you is worth savin’
Then you better start swimmin’ or you’ll sink like a stone
For the times they are a-changin’"
Dear Eclectic2k and Thedealer,
Though i haven´t heard all amps either class A/AB/B/D ect., I´m not sure class A (still) deserves its privileged placement among the other "lesser designs" concidered by many audiophiles.
Many so called class A amps i´ve heard were smooth and slow, almost to the point of boredome. The class D i´ve encountered weren´t perfect either, though to my ears more `correct´ though not as engaging as the dirtier sounding class B amp have running currently. One thing remains as ever important; the power supply! (nothing new here, of course) A good implementation of any class amp will probably sound really good, though some classes demands more from the PS in terms of noise sensitivity and current draw.
I do actually consider to implement class D modules in the future and am very interested in your perceptions the of different possibilities out there, and what you tested them against. My AR ICE amp do sound very clear and crisp, but seems to sit somewhat on the music. It all `looks´ pretty, but doesn´t feel that engaging. Would UCD be a better choice, in your minds? any thoughts?
Eva,
I´m maybe thickheaded, but are you affiliated with coldamp by any chance? Or are your designs in circulation in any way? -I´m quite interested in the different approaches to get the most out of things. Can you comment on how your designs sound compared to other class D (or any class actually)?
regards,
Though i haven´t heard all amps either class A/AB/B/D ect., I´m not sure class A (still) deserves its privileged placement among the other "lesser designs" concidered by many audiophiles.
Many so called class A amps i´ve heard were smooth and slow, almost to the point of boredome. The class D i´ve encountered weren´t perfect either, though to my ears more `correct´ though not as engaging as the dirtier sounding class B amp have running currently. One thing remains as ever important; the power supply! (nothing new here, of course) A good implementation of any class amp will probably sound really good, though some classes demands more from the PS in terms of noise sensitivity and current draw.
I do actually consider to implement class D modules in the future and am very interested in your perceptions the of different possibilities out there, and what you tested them against. My AR ICE amp do sound very clear and crisp, but seems to sit somewhat on the music. It all `looks´ pretty, but doesn´t feel that engaging. Would UCD be a better choice, in your minds? any thoughts?
Eva,
I´m maybe thickheaded, but are you affiliated with coldamp by any chance? Or are your designs in circulation in any way? -I´m quite interested in the different approaches to get the most out of things. Can you comment on how your designs sound compared to other class D (or any class actually)?
regards,
My only foray into class D was an IRS2092 amp.
This sounds very good, at least as good as my best class AB amp.
I wouldnt recommend building your own pcb/design for IRS2092 class d as they can be a pig to iron out all the problems, which I wont go into here.
But once working they sound good.
I really think there will be Class D designs in the near future that will sound as good as or better than the older technologies. But the question was has "Class D reached Pass level yet". I for one look forward to the day when it does. I thought we might be there now, thats why I purchased the Bel Canto, NuForce and Audio Zone gear in 2010. But after auditioning this group of Class D amps against well designed Class A and A/B amplifiers, everyone I invited to listen (audiophiles, recording engineers, and untrained listeners) to the different topologies, they all picked the Class A or A/B designs (to me this is night and day). I am sure with the group passionate Electronic Engineers and Designers such as Eva and Roberto, we shall get there sooner than I think.
It's not as if the "crossover distortion" issue is new, or not understood . . . Doug Self discusses it in detail, and concludes (for the reasons you discuss) that properly implemented Class B is better than AB. And the 3886 demonstrated years ago that it is possible to (mostly) overcome the problem with device matching and careful bias control. That anyone would be surprised when another topology that actually eliminates the problem sounds good amazes me.
Once the crossover problem is gone even a relatively simple device like the IRS2092 can produce an amplifier with essentially no audible distortion . . . not "perfect", to be sure, but better than almost everything up to now . . . and it's really down to mastering the intracies of the details (at which you seem to be exceptional) to get the job done right.
Oh, and you seem to me to be a *great* teacher . . . I've learned more from your comments here than all others together.
Completely agree!
I think this is make or break. And seems to be the downfall of so many...
I'm a little biased toward UCD. I'm certainly not convinced it is the *only* way. (I've not recently listened to ICEpower or the latest Nuforce) I'm also not really averse to any topology. I settled for now on Class-D in part for the efficiency. My amp weighs as much as many A/B, (the linear supply) but it is in a smaller enclosure with no heatsinks other than the chassis, and does not need any special considerations for staying cool in a vented cabinet. I would not have settled for this if I did not think it sounded good.
I have parts laying around to build some chipamps, some tripath, and some tubes, (but only tube pre's I'm not interested in tube power stages
)
+1
(in general on the class-D forum here)
c'mon, you'd ask a parent how great their kids are?
Would be nice if Eva weren't bound by NDA's on a project one of these days. Then we could all know and go listen
Ok this is strange. When paying better attention to the switching point (I mean the start of a transition rather than 90% settling or something), the high angle crossings are getting about 5nS output delay while the low ones are getting ESP, meaning the output is switching Before the compare signals actually cross.
20k sine, 40k triangle, both 4VP-P, .5nS max timestep, LT1016, +/-5V supplies, no output load.
I admit I don't know what the heck this result means. I'd like to try this stuff with a real comparator and a scope but I guess that's my problem.
20k sine, 40k triangle, both 4VP-P, .5nS max timestep, LT1016, +/-5V supplies, no output load.
I admit I don't know what the heck this result means. I'd like to try this stuff with a real comparator and a scope but I guess that's my problem.
Last edited:
+1
(in general on the class-D forum here)
c'mon, you'd ask a parent how great their kids are?
Would be nice if Eva weren't bound by NDA's on a project one of these days. Then we could all know and go listen
Well good parants tend to listen to their kids hence they should know them for both good and bad
I wouldn´t expect Eva to disclose anything violating her agreement, though a little insight from listening experiences shouldn´t do that, I hope...
Yes, this is a very informative forum indeed. Great reading, and thanks for the responses. It seems UCD is still the safe bet regarding class d, no?
BTW i wondered about the no-filter approach inspired by Eva´s insightful lecures and found this:
An externally hosted image should be here but it was not working when we last tested it.
And the related reading:Class D Fundamentals of Operation | Class D Amplifiers Shop OEM Products DIY Audio
Is this something in line with your thinking Eva?
Very interesting indeed
cheers,
It means that the "pulse skew" (the parameter you are looking for) is below simulation and model accuracy.
To eclectic2k:
Icepower and UcD are almost the same thing. They are both phase-shift oscillators. The difference is that in ICEpower they didn't dare to take full feedback from the output (probably due to high EMI issues and the fact that almost nobody had done it before), feedback is split instead, being LF taken after the filter and HF before the filter (a step further from current IRAUDAMP which is full pre-filter f.b.), while in UcD they were able to get the layout and the switching good enough to get a clean carrier residual that could be fed directly to the comparator.
Last edited:
LOL...........
So show me some proof that anybody can. I can provide evidence that people can't.
w
Isn't pulse skew simply an asymmetry in the prop delay of the part? I wouldn't think that would cause this issue on the sim? (haven't touched spice in long time) Is the LT1016 LATCH pin defined?
Thanks! I'm surprised the split feedback doesn't introduce more problems than is solves. (well maybe it does
) I'll have to give that some thought...I don't know much about the ICEpower designs. I've studied UCD a bit, but in general, I'm still learning about self oscillating designs. I might if pressed be able to design a rudimentary amp based on an external oscillator though ;-)
interesting full bridge design. and single rail too. and signal detect. curious... must sill be embedded design.
my understanding of filterless designs is that they're limited to embedded low power apps like cell phones and small devices as the low power and very short distance to the speaker limits the EMI liabilities. Also, the speaker is a well defined part of the design and can be controlled. In a stand alone amp, the speaker (most often a system with crossover) can vary quite wildly in it's parameters and phase response. Not to mention completely uncontrollable amounts of speaker cable "antennae"...
-CK
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.