That's how I'd describe the BelCanto but I still prefered it to the Anthem.
What?
Have you poked around inside a Class-D amp with a probe? I'm sure you have. You've only seen it on the output stage? And feedback fixed it?Again, color me puzzled. I see noise all over the place, like on the ground. It goes all over the amp. Can you site some examples of designs or amps that don't do that? I'm genuinely curious.
(Sorry for the late reply, I've been away)
I agree on all 6 points and would even add two more.
...in fact I just could not resist to search the old measurements, when I saw you promoting a trivial sagging of 2.7V at a 500W burst
It was to simple.
I know well how easy it is. maybe you're referring to experiment or diy smps or a linear regulator. i mean, please show dynamic measurements of commercial products, or products that are fitted with a smps designed for audio. if it is as simple as I can not see?
I do not want a competition, right to defend myself from these attacks.
Looks good what say others diy ... smps? no problem, huge nfb on class D, cleans everything.
Measure on this? no one. SIN measure on amplifier ... a lot hahaha.
hi Kal
I'm using a DIY UCD180ST implementation with a SMPS made for audio.
just received an Audio Refinement Complete (YBA design) amp for evaluation. it was praised in many reviews, including by Stereophile. judging by the measurements, it seems to be a low NFB design. I'm curious how it compares to the switching amp. I wish it wasn't -3dB @20kHz, but... who knows.
Audio Refinement The Complete integrated amplifier Measurements | Stereophile.com
Last edited:
It is definitely not wrong to pay attention to a good PSU. One has to keep in mind that with a class-d amp one switches directly between rails. There is no intrinsic rejection of hash riding on the PSU rails compared to an AB amp with common emitter output stage for instance, at least not at the output stage to be exact.
The open-loop gain of a class-d amp is directly proportional to the supply voltage and therefore some measures have to be taken. One such measure is the modulation of the triangular wave in proportion to the supply rails. With a self-oscillating topology this is already intrinsic to the working principle.
With a carrier-based amp one has to rely on the feedback factor (which is usually not that high) and the use of additional circuitry (to achieve carrier modulation for instance) in order to get good PSSR.
What does that tell us ? PSU IS ipmortant for a class-d amp. But if an amp has to rely on 150'000 uF per rail then that wouldn't be considered good engineering IMHO. It is a complete misuse of resources. It is almost as bad engineering as a Bosshoss motorcycle.
And the other extreme case of a 500 Watt amp for instance with 2200uF PSU caps that is fully relying on a compensation scheme wouldn't be considered wise either.
Good engineering is trying to make good use of resorces by searching for the best trade-off. I.e. using materials in a reasonable way and making the best use of what the physics offer you
Regards
Charles
The open-loop gain of a class-d amp is directly proportional to the supply voltage and therefore some measures have to be taken. One such measure is the modulation of the triangular wave in proportion to the supply rails. With a self-oscillating topology this is already intrinsic to the working principle.
With a carrier-based amp one has to rely on the feedback factor (which is usually not that high) and the use of additional circuitry (to achieve carrier modulation for instance) in order to get good PSSR.
What does that tell us ? PSU IS ipmortant for a class-d amp. But if an amp has to rely on 150'000 uF per rail then that wouldn't be considered good engineering IMHO. It is a complete misuse of resources. It is almost as bad engineering as a Bosshoss motorcycle.
And the other extreme case of a 500 Watt amp for instance with 2200uF PSU caps that is fully relying on a compensation scheme wouldn't be considered wise either.
Good engineering is trying to make good use of resorces by searching for the best trade-off. I.e. using materials in a reasonable way and making the best use of what the physics offer you
Regards
Charles
Last edited:
Hi,
Another confusion with PSRR.
the drop in voltage transients (perfectly audible to an audiophile, which certainly does not want to hear), has nothing to do with the PSRR.
some audio measurements are ... intelligent. does not show in the case of FFT, to segment under 1Khz, or 10Hz-50Khz at 65% power referred to the class D obvious.
if it is a good amp, it is normal that I can use in loudspeaker crossover, perhaps high cost. then it is very easy to listen to 75% modulation with high output current, depending on the type of music I'm listening.
Or, why the manufacturer of the amp must say that it is perfect, but only if coupled directly to a speaker (multi-way).
With the words, we're all good, this is the problem.
In many smps, I see how easily, the engineer has the 2200uF filter cell high voltage, it is necessary for me to explain what catastrophe happens to the AC line? it is "power factory (PF), or" cosine of f ".
@ Choco: I think that can be constructive for others who read, if you show the dynamic measurements on smps fast that you have. We discuss the problems that can be solved (if any).
regards
Another confusion with PSRR.
the drop in voltage transients (perfectly audible to an audiophile, which certainly does not want to hear), has nothing to do with the PSRR.
some audio measurements are ... intelligent. does not show in the case of FFT, to segment under 1Khz, or 10Hz-50Khz at 65% power referred to the class D obvious.
if it is a good amp, it is normal that I can use in loudspeaker crossover, perhaps high cost. then it is very easy to listen to 75% modulation with high output current, depending on the type of music I'm listening.
Or, why the manufacturer of the amp must say that it is perfect, but only if coupled directly to a speaker (multi-way).
With the words, we're all good, this is the problem.
In many smps, I see how easily, the engineer has the 2200uF filter cell high voltage, it is necessary for me to explain what catastrophe happens to the AC line? it is "power factory (PF), or" cosine of f ".
@ Choco: I think that can be constructive for others who read, if you show the dynamic measurements on smps fast that you have. We discuss the problems that can be solved (if any).
regards
75% modulation? with real speakers and real music?
I have done my measurements and I don't think that ever happens with real music. I think 75% modulation would be very visible even on quick visual inspection of the waveform.
I'm not saying it can't happen but I'd have to come up with the worst case conditions to achieve that.
I do not understand why you want to push me, to explain old things in some way resolved by the manufacturers of professional audio, already many years ago, the class AB.
Problems increase when the speed is added to the current.
at 30w, what issues you want to see?
When I listen to a good jazz in my living room (at volume that do not disturb), the B & W 802 II, passing 8-10Amp on load, repetitive peak.
Bass and vocals are warm and dry attacks really.
This effect of realism gives you immense pleasure. How do I get this? ... smps normal or psu with simlpe 2x10.000uF?, or with nfb?
In what measure audio traditional you can understand this?
Explain to me please
sorry, i mean 75% of modulation or 65% (volume amplitude) just create current at output.
Last edited:
@phase
I would very much expect duty cycle to change with power supply sag I'd be worried if it didn't, because likely I'd be able to hear it. but that is PSRR too, isn't it?
@AP2
out of curiosity, I have measured instantaneous current with my speakers using a shunt. I have to push them to the limit (way above decent levels) to reach the theoretical predicted value of 4 amps. and that is using full scale orchestral music, like Reference Recordings' version of Bolero, taiko drums etc. speakers have 83dB W/m sensitivity and max tolerable levels in my living room are reached @ 100-150W in 4 ohms.
looks like phase accurate which is obviously more schooled in EE and signal theory compared to me failed to understand exactly what you meant.
the 802's are known to be a difficult load. nevertheless, a good amp should be able to drive them too.
I would very much expect duty cycle to change with power supply sag
I'd be worried if it didn't, because likely I'd be able to hear it. but that is PSRR too, isn't it?@AP2
out of curiosity, I have measured instantaneous current with my speakers using a shunt. I have to push them to the limit (way above decent levels) to reach the theoretical predicted value of 4 amps. and that is using full scale orchestral music, like Reference Recordings' version of Bolero, taiko drums etc. speakers have 83dB W/m sensitivity and max tolerable levels in my living room are reached @ 100-150W in 4 ohms.
looks like phase accurate which is obviously more schooled in EE and signal theory compared to me failed to understand exactly what you meant.
the 802's are known to be a difficult load. nevertheless, a good amp should be able to drive them too.
Last edited:
Hi Roberto,
you did not look seriously to my posted screen shot.
It shows multiple dynamic load steps.
The 900W are applied for 1.8ms, then 3ms no load, then again 900W for 1.8ms...
The red line is showing the AC-voltage of the rail with 200mV/grid, in order to magnify the load step response.
The loop gain structure is PD, no integrating portion. Consequently the rail saggs a little bit (less than 500mV) and remains there.
As you can see the regulation reaches the stable situation within few hundret us almost free of regulation over shoots.
If you are asking for more measurements - I cannot serve. I just designed and built a single proto for a friend and handed it over.
In any case I cannot involve deeply into public discussions about SMPS any more. My work has drifted more and more into this direction, so I must limit my contributions to a very basic level.
Attachments
..to avoid misunderstandings.
At the beginning of the screen shot there is no load.
The dips of the red trace are the times when the smps is loaded with 900W.
The dip stabilizes fast with moderate undershoot and remains stable until the load is switched off. When the load is turned off, the rail steps up back to the no load level without any overshoot.
At the beginning of the screen shot there is no load.
The dips of the red trace are the times when the smps is loaded with 900W.
The dip stabilizes fast with moderate undershoot and remains stable until the load is switched off. When the load is turned off, the rail steps up back to the no load level without any overshoot.
EDN beginning a series of articles on Class-D:
New approaches to switched-mode audio power amplifiers (Part 1) | EDN
New approaches to switched-mode audio power amplifiers (Part 1) | EDN
I promised to give some comparison results and I guess it'll be obvious that bias is not a factor.
in short, the ARC wipes the floor with the UCD180 modules. night and day difference. I'll refrain from blurting the usual audiophile adjectives, because the difference is way too big.
only thing in favor of the UCDs I can find is that with some very hot mixes it sounds better. but with the majority of the music, not. at first listen the ARC seems on the euphonic side, but the more you listen you realize it's the exact opposite, recordings start to sound more different from each other than with the UCD.
and boy, does it measure horribly...
too bad I'm an engineer and that makes me even more curious as to what explains the major difference
although this is an old post, i can tell you that if there is a difference to be heard between any amplifier, i can hear it. they all sound different to me. i have several power amps that i swap into the system periodically using the pre-out/main-in connectors and each one sounds different. they all sound good, but they all sound different. none are driven to ear-bleed levels, just 'normal' listening levels, whatever this is.
and yes, audio is mostly BS. but some people actually can hear. never listened to a class-d full range, though. love to.
It is in the nature of Class D to go louder with less power, due to the nature of their design. The side effect is that the signal coming out of one is never going to be the same, even the hardest you try. As you may know, Class D generally modulates a sound into pwm, then smoothens it with a cap, low passes it, decouples with a cap, and then high passes it. At least, that's generally a reconstruction. Class D amps, for the reason of the manipulation, will sound damn close to the original to most ears; but never for everyone.
I personally use only a class D for my subwoofer, and the rest is a class AB.
I personally use only a class D for my subwoofer, and the rest is a class AB.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.