Audio Amplifier Performance Assessment Program (AAPAP)
(This introduction must be read in conjunction with the accompanying Excel spreadsheet)
General use and objective of AAPAP
The program looks at several important amplifier performance areas and provides an objective method of assessing the design and engineering effort brought to bear in reducing or eliminating shortcomings that will materially affect electrical performance and therefore sound quality negatively. The philosophy behind this approach is that, for example, hum and noise or TIM/SID cannot bring positive benefits to any amplifier either electrically, or when assessed separately, for sound quality. These are problems that require design effort to eliminate, and this is what AAPAP addresses.
The assessment process can be used during the design phase to guide the designer towards improved electrical performance and/or to assess finished designs, be they DIY or commercial.
For assessment and scoring, the amplifier under test (AUT) is treated as a black box except for the protection features scoring where the AUT specification is consulted.
The following areas of electrical performance are covered by AAPAP:-
The tool distinguishes between different amplifier technologies and assesses performance within the capabilities of the designer's chosen technology. The technologies currently covered are: -
Tool outputs
A report card and audio analyser plots against which the parameters listed above are scored with a top-level score that places the amplifier in one of 6 categories as detailed in the Excel spreadsheet attached below.
(This introduction must be read in conjunction with the accompanying Excel spreadsheet)
General use and objective of AAPAP
The program looks at several important amplifier performance areas and provides an objective method of assessing the design and engineering effort brought to bear in reducing or eliminating shortcomings that will materially affect electrical performance and therefore sound quality negatively. The philosophy behind this approach is that, for example, hum and noise or TIM/SID cannot bring positive benefits to any amplifier either electrically, or when assessed separately, for sound quality. These are problems that require design effort to eliminate, and this is what AAPAP addresses.
The assessment process can be used during the design phase to guide the designer towards improved electrical performance and/or to assess finished designs, be they DIY or commercial.
For assessment and scoring, the amplifier under test (AUT) is treated as a black box except for the protection features scoring where the AUT specification is consulted.
The following areas of electrical performance are covered by AAPAP:-
- Output Power
- Capacitive load drive capability
- Overdrive recovery
- Frequency response into 8 Ohms resistive load and complex load
- Complex load drive capability vs distortion at high power levels
- THD + N at 1 watt into 8 Ohms (0 dBV) and at rated power (dBr measurement)
- IMD 19+20 kHz at 80% of rated power into 8 Ohms
- Noise floor of mains-related noise and hum, including cross-channel ground loop noise
- Speaker and amplifier protection assessment
- $ Cost per stereo watt (not scored, but must be stated in the assessment)
The tool distinguishes between different amplifier technologies and assesses performance within the capabilities of the designer's chosen technology. The technologies currently covered are: -
- Linear solid-state amplifiers with global feedback
- Linear solid-state amplifiers without global feedback
- Vacuum tube amplifiers
- Class D amplifiers will be added to the program once agreement is reached on how certain parameters will be scored
Tool outputs
A report card and audio analyser plots against which the parameters listed above are scored with a top-level score that places the amplifier in one of 6 categories as detailed in the Excel spreadsheet attached below.
Last edited:
THD is more like the color of your socks. Hard to make it useful as part of a larger metric.
How about RF susceptibility or emission?
Or weight/size per watt?
How about RF susceptibility or emission?
Or weight/size per watt?
Maybe we should add that the PSU measurement must always be wideband? Say 20 Hz to 80 kHz?
I like the concept, but rolling this up into one number is less useful than having each parameter tested to a standardized method, and listed as a summary. Picking an amplifier is more like buying tires, what parameters are more important for a given customer. Mileage, noise, wet traction, cost, etc.
I like this idea. We should give it a go. My only complaint is that point 11 dismisses build-quality of the amplifier. I understand the thought behind it and it is not a make-or-brake part of the rating. It takes away from gear that is built really well, like Accuphase. Does it give me the greatest bang for a buck in terms of power? No. But it will outlive me and my children will enjoy it while I'm six feet under.
What we need now is a manual for testing, so even less experienced users can do it.
Maybe a measurement database would be a nice thing? Users could upload their measurement, which would be easily available to others. I might be able to create and maintain it.
What we need now is a manual for testing, so even less experienced users can do it.
Maybe a measurement database would be a nice thing? Users could upload their measurement, which would be easily available to others. I might be able to create and maintain it.
You are on a noble quest! 🙂
I would suggest some weighting on the harmonics profile, low order better than high order etc.
After liking to some amps that measured terribly, I don't hold THD as very important, I usually 'visualize' distortion with multitone measurements, and amps with a flat 'distortion floor' have generally been more to my liking. Many high loop gain amps tend to have a rising distortion towards the highs, Loudspeakers tend to generate least distortion in the treble, so this could be where it's most noticeable.
Maybe using multitone 20-20k and highest point taken from the 'distortion floor', with something like 5W output would be more interesting? Harmonic profile could be with single 1k at maybe 1-5W.
Trying to find some link between my own measurements and listening impressions, I have the following thoughts:
-Low output impedance at LF seems to give tighter bass/better control. CFP outputs seem good at this as an example.
-Amps with less loop gain (20-30dB) and higher output impedance seems to sound 'sweeter' and more relaxed especially midrange treble, even if measured distortion is similar and -80dB or lower.
When it comes to output impedance I think it should be measured, or simulated with PS. Seems small caps in PS can give rising output impedance in LF, if loop gain is on the lower side. Can also cause modulation with PS ripple at LF etc.
I would suggest some weighting on the harmonics profile, low order better than high order etc.
After liking to some amps that measured terribly, I don't hold THD as very important, I usually 'visualize' distortion with multitone measurements, and amps with a flat 'distortion floor' have generally been more to my liking. Many high loop gain amps tend to have a rising distortion towards the highs, Loudspeakers tend to generate least distortion in the treble, so this could be where it's most noticeable.
Maybe using multitone 20-20k and highest point taken from the 'distortion floor', with something like 5W output would be more interesting? Harmonic profile could be with single 1k at maybe 1-5W.
Trying to find some link between my own measurements and listening impressions, I have the following thoughts:
-Low output impedance at LF seems to give tighter bass/better control. CFP outputs seem good at this as an example.
-Amps with less loop gain (20-30dB) and higher output impedance seems to sound 'sweeter' and more relaxed especially midrange treble, even if measured distortion is similar and -80dB or lower.
When it comes to output impedance I think it should be measured, or simulated with PS. Seems small caps in PS can give rising output impedance in LF, if loop gain is on the lower side. Can also cause modulation with PS ripple at LF etc.
This is a good list.
I would add idle power consumption above 50W to penalize inefficient amplifiers.
Mechanical hum would be a showstopper.
Stray magnetic fields are also a showstopper.
I would not try to combine the metrics. The problem is inherently multi-dimensional.
Ed
I would add idle power consumption above 50W to penalize inefficient amplifiers.
Mechanical hum would be a showstopper.
Stray magnetic fields are also a showstopper.
I would not try to combine the metrics. The problem is inherently multi-dimensional.
Ed
How about stress testing. Load fault response. Then there is elevated temp, line voltage at full power.
Good point. Maybe it’s a graph showing the parameter assessments plus then a top level score.
I already disagree with point 1; 20 W per channel is more than enough for me.
I have not played with class D, but if an unregulated SMPSU is being used, are there no mains harmonics present on a full load output test?
Thinking of the amp in black box without knowledge of its design implementation should be a goal. The implementation will likely reveal itself in the testing results, but implementation should not guide the testing.
That’s why I’ve avoided operating class, feedback (none or with), topology, solid state or tube etc.
Then your amp just earned 2 points. I’m sure if it’s a good amp Marcel it would more than make up for it in other areas of AFOM.
Seems a tad harsh. Perhaps there might be arguments for not including #1 as score but banding amps into categories based on wpc?
It is harsh, but low power amps, if engineered well, can score big in other areas. For example, my kx2 would only get one point for power, but would get 4 for hum and noise, and 3 for IMD. The problem here is you have 5-10 watt amps masquerading as hi-fi that possible score very highly in every category, but really are limited to the speakers they can drive.