You also have to understand where slew rate is important and where it isn't. As an amplifier spec, it's pretty useless- if an amp has slewing, there's a distortion signature associated with it. Is the distortion there or not? In the case of a class D amp, the input is bandwidth limited, making the slew rate spec even more misleading (however you define and measure it).
You're certainly free to desire to run 10kHz square waves and use that as your acceptance criterion. It's your money that you're spending on an amp. But it has nothing to do with audio or sound.
You're certainly free to desire to run 10kHz square waves and use that as your acceptance criterion. It's your money that you're spending on an amp. But it has nothing to do with audio or sound.
A service manual is intended for low-level technicians (I think you could refer to it as "child's stuff").
Usually people do their testing based on actual formal testing standards (EBU, AES, etc.).
More importantly, you have to understand what you measure, and why. Measuring square wave response and slew rate without understanding the feedback architecture and parameters of an amp doesn't really tell you much.
Why do so many people misunderstand slew rate. It has very little to do with normal operation. It's a measure of the speed of recovery from the non linear overdriven condition to normal linear operation. An (almost) instant step voltage on the input causes that full voltage across the + - inputs of the input stage, instead of the normal mili/micro volts ( depending on the amount of negative feedback ). This overloads the input and basically turns the amp into a switch. It's no longer a linear device until it recovers. The speed of this recovery is directly related to slew rate. As long as the slew rate is 5 to 10 times faster (depends who you talk to) than the highest freq full power sine wave, there should be no difference in amp sound. This is obviously for non class d amps but it probably still applies ( don't know enough about there architecture, but they still use feedback).
They bandwidth-limit at the input, just as any other sampled system does, so the externally measured rise time is not a measure of slew rate. The OP seems to conflate these two characteristics.
There are several different possible feedback architectures for class D, but in general, your logic still applies.
You can't put a square wave on a LP either, the needle would effectively hit a brick wall during playback.
Exactly. Take these words and look from the other side to the spec.
We need to reproduce a certain frequency with a certain amplitude.
The less amplitude we need at high frequencies, the less slew rate we need.
Don't let you get tricked by definitions.
What you are demanding are slew rates which correspond to reproduce a 70kHz or more sine wave with full power.
There is nothing wrong in designing an amp which provides a 70kHz full power bandwidth - but why should anyone take it as the poorest acceptable value and rate it as the dominant criteria for sound quality, while music program has less than 10% of its power content above 10kHz?
Service Manuals are intended to check a specific amp model.
When everything is right, then you will get certain readings.
Requirements from service manuals are not a minimum spec which every amp model has to meet in order to be a good sounding amp - it is for checking if this specific amp model has a defect or not and for giving hints to find defects.
Depending on the amp model the requirements in the service manual can be better or worse than what you need for a good sounding amp.
Regarding rise times of 2us from 10%-90%:
Doubtless nice, especially if done without overshoot.
But you will still find people who insist that anything above 300ns is to slow and slew rates of 500V/us are must... And the same people will blindly buy any ultra fast amp and praise the sound and the next day wonder why their amp has burned after changing nothing than the speaker wire - not understanding that the changed capacity of the wires has shifted a pole in the loop gain and resulted in destructive HF oscillation....
(With Crown, I fully trust they stayed in the speed range which they really can handle at any reasonable complex load.)
" slew rate is dependant by output level, that's"
I don't agree. Slew rate for a given design is a constant measured in V/s, that may be reached during playback or not (better if not). Of course the likelyhood to run into slew rate limitation is bigger with higher signals. Input filter can be optimized to avoid slew rate under most conditions. But slew rate itself is not dependent on output level, as a given amplifier design will have a given slew rate (for example 20V/us), no matter if it outputs 10mW or 2000W.
If slew rate is reached or exceeded during playback, distortion is resulting. The distortion is less bad in nature compared to clipping. Since adding slew rate distortion for any given signal by nature equals differentiating said signal by time, then clipping the result (=hard limiting the highest / lowest parts that correspond to the steepest parts of the original signal), then integrating the result. Through the final integrating, the higher distortion frequecies get attenuated.
I don't agree. Slew rate for a given design is a constant measured in V/s, that may be reached during playback or not (better if not). Of course the likelyhood to run into slew rate limitation is bigger with higher signals. Input filter can be optimized to avoid slew rate under most conditions. But slew rate itself is not dependent on output level, as a given amplifier design will have a given slew rate (for example 20V/us), no matter if it outputs 10mW or 2000W.
If slew rate is reached or exceeded during playback, distortion is resulting. The distortion is less bad in nature compared to clipping. Since adding slew rate distortion for any given signal by nature equals differentiating said signal by time, then clipping the result (=hard limiting the highest / lowest parts that correspond to the steepest parts of the original signal), then integrating the result. Through the final integrating, the higher distortion frequecies get attenuated.
Definitely no.
As long as you operate it in its linear region (no internal overdrive of any gain stage), you will get a constant rise time, but slew rate will grow with the amplitude of the signal.
At least that is what all amps I ever measured did show and from linear filter theory can be expected.
Slew rate in amp specs is defined to be measured with full amplitude rectangles in order to see the limits of slewing, no matter if internal gain stages are already heavy in clipping or not during this measurement.
Does somebody have schematic from K-Series ?
have check PCB
Atmega64 Risc processor
Fixed clock pwm TL074
74HC04 driving HCPL0611
HCPL0611 out driving TC4427 Mosfet Driver
TC4427 out driving Totem Pole 2 x IRFR9014
Totem Pole driving 3 pair IXFR44N50Q output devices each channel
from Hardware cost K series nothing special, overpriced amp with (fake) peak power ratings
,
have check PCB
Atmega64 Risc processor
Fixed clock pwm TL074
74HC04 driving HCPL0611
HCPL0611 out driving TC4427 Mosfet Driver
TC4427 out driving Totem Pole 2 x IRFR9014
Totem Pole driving 3 pair IXFR44N50Q output devices each channel
from Hardware cost K series nothing special, overpriced amp with (fake) peak power ratings
,
Last edited:
The power ratings are called out as EIAJ on Powersoft's spec sheet. They appear to meet and exceed them. K series distortion, noise, and frequency response are also quite good. You may not like EIA's J method, so perhaps they are the ones you should be complaining about.
http://www.diy.poweraudio.ro/albums/userpics/10001/BENCH_COMPARISON_TEST.pdf
http://www.diy.poweraudio.ro/albums/userpics/10001/BENCH_COMPARISON_TEST.pdf
What did you think/expect?
This and size&weight are the points which make classD attractive, same situation like comparing SMPS with old fashion 50Hz transformers.
High prices as per today are just a consequence of high R&D efforts + low volumes + offering advantage to the customer.
The prices are already coming down and will come down much further in future. Of course it will also depend on the reputation if an amp or brand is known to perform well in the field and being reliable.
For getting a first impression without buying it - DJs movies are a valuable source of information regarding power & speed.
Also helpful is this page (not only classD):
Amplifier Testing :: ABELtronics
Here are a part of amplifiers repaired or tested by me.
No account needed, to watch.
www.forum.poweraudio.ro • Vizualizare forum - Laboratorul lui LECO
No account needed, to watch.
www.forum.poweraudio.ro • Vizualizare forum - Laboratorul lui LECO
I do those tests to share with other people, real acting or performing of various power amplifiers on dummy load, load as the most constant impedance (resistance in that case), not for me, I don't need them I know it deja, I'm dealing with power amplifiers since I had 7 YO that means 36 years ago when I made mine first amplifier exceeding 50 watts/8 ohms.
So, if you'll watch all movies posted there, you will understand basically, old school amplifiers performs good in squarewave slew rate tests, offers clean output power and constant power, and respect as much possible, specifications regarding power and freq response from manual.
So, if you'll watch all movies posted there, you will understand basically, old school amplifiers performs good in squarewave slew rate tests, offers clean output power and constant power, and respect as much possible, specifications regarding power and freq response from manual.
Last edited:
"
Originally Posted by ViennaTom
But slew rate itself is not dependent on output level, as a given amplifier design will have a given slew rate (for example 20V/us), no matter if it outputs 10mW or 2000W.
Definitely no.
As long as you operate it in its linear region (no internal overdrive of any gain stage), you will get a constant rise time, but slew rate will grow with the amplitude of the signal.
At least that is what all amps I ever measured did show and from linear filter theory can be expected."
Signal dv/dt will grow with the amplitude of the signal. Not slew rate. Slew rate ist the highest possible dv/dt that the amp can produce, before leaving linear operation.
Originally Posted by ViennaTom
But slew rate itself is not dependent on output level, as a given amplifier design will have a given slew rate (for example 20V/us), no matter if it outputs 10mW or 2000W.
Definitely no.
As long as you operate it in its linear region (no internal overdrive of any gain stage), you will get a constant rise time, but slew rate will grow with the amplitude of the signal.
At least that is what all amps I ever measured did show and from linear filter theory can be expected."
Signal dv/dt will grow with the amplitude of the signal. Not slew rate. Slew rate ist the highest possible dv/dt that the amp can produce, before leaving linear operation.
Yes, and the power measurements are useful and interesting, so thanks for sharing them!
I guess I am a later starter then, I only made my first amplifier when I was 15, 38 years ago, but that got me hooked and lead me to study electrical engineering at university...
Indeed - the point we are making is that square wave response doesn't really tell you anything about a class D amplifier, unless you are analyzing specific properties of the feedback architecture.
Man, feedback architecture is not important to me, even if is pre filter, post filter or both of them.
What is important for me is final resultat wich goes from OUTPUT on the speaker to match maximum as waveform to input waveform used, wich in that case and the next one is completely mess.
In most cases of classD amps looks the output filter is underdamped, affecting alot feedback for high freqs, leaving on output lots of switching residues and distortion as you will see down on pictures.
As you see also on Mr. Abeltronics page, himself has tested some classD amplifiers.
Funktion-One, FFA F60Q/FFA6004 Amplifier Test Results :: ABELtronics
So what he says, I quote him:
Good performance before clipping and at low frequencies, very high distortion at mid-high frequencies and potential reliability issues. Bad clipping behaviour, and internal heating of components when idle. Good looking, nicely engineered front panel, very deep casework (approx 20"). The amplifier performance is satisfactory if it is not used full range or at near full-power . Mains voltage did not fall below 239.6V. 05-06-2010.
Why that?
29.10kHz square wave, 8R, threshold of clipping. Very distorted and lots of overshoot.
And here on that amplifier looks even better at 10Khz square than powersoft
http://www.abeltronics.co.uk/amptesting/Funktion-One_F60Q_-_FFA6004/29.jpg
30.10kHz square wave, 8R, low level. Switching residual rides over waveform. 50% overshoot despite being very low level (approx 10V p-p).
exactly as powersoft, alot of switching residues
http://www.abeltronics.co.uk/amptesting/Funktion-One_F60Q_-_FFA6004/30.jpg
What do you have to say now?
What is important for me is final resultat wich goes from OUTPUT on the speaker to match maximum as waveform to input waveform used, wich in that case and the next one is completely mess.
In most cases of classD amps looks the output filter is underdamped, affecting alot feedback for high freqs, leaving on output lots of switching residues and distortion as you will see down on pictures.
As you see also on Mr. Abeltronics page, himself has tested some classD amplifiers.
Funktion-One, FFA F60Q/FFA6004 Amplifier Test Results :: ABELtronics
So what he says, I quote him:
Good performance before clipping and at low frequencies, very high distortion at mid-high frequencies and potential reliability issues. Bad clipping behaviour, and internal heating of components when idle. Good looking, nicely engineered front panel, very deep casework (approx 20"). The amplifier performance is satisfactory if it is not used full range or at near full-power . Mains voltage did not fall below 239.6V. 05-06-2010.
Why that?
29.10kHz square wave, 8R, threshold of clipping. Very distorted and lots of overshoot.
And here on that amplifier looks even better at 10Khz square than powersoft
http://www.abeltronics.co.uk/amptesting/Funktion-One_F60Q_-_FFA6004/29.jpg
30.10kHz square wave, 8R, low level. Switching residual rides over waveform. 50% overshoot despite being very low level (approx 10V p-p).
exactly as powersoft, alot of switching residues
http://www.abeltronics.co.uk/amptesting/Funktion-One_F60Q_-_FFA6004/30.jpg
What do you have to say now?
Last edited:
- Status
- Not open for further replies.