Yes, you read it right: I now own a second and the latest state-of-the-art analyzer from Audio Precision, the APx555. She is sitting here next to her elder unit to the left I have been using until now:
Avid readers know that I have been getting a lot of mileage out of my much older AP2522 unit. Despite being nearly 20 years old, it has had excellent performance. Great analog design and power of signal processing had kept it relevant for such use even now.
Alas, a few reasons prompted me to buy the new unit. First and foremost is the fact that the software for the new model allows me to directly talk to computer audio products through ASIO interface. You may have seen my difficulty in using my older analyzer there when a DAC doesn't have USB input. With APx555, I am able to now drive those products over USB just as well as it can using S/PDIF, Toslink, and analog output out of the analyzer. This makes measurements now the same whether I use USB or any of these other interfaces.
Second reason is interchange with the industry. Often people want to repeat my measurements and they cannot if they have the newer unit than mine. Now I am able to do that since I am using the latest and current products from Audio Precision.
Audio Precision was kind enough to loan me both the APx525 (USD $18,000) and APx555 (USD $28,000). This was important to me as I wanted to make sure my workflow would be preserved or improved and that performance of the new units could beat what I already had.
With respect to APx525, it only managed to match the performance of my 2522. It had slightly less distortion but somewhat higher noise in ultrasonics due to noise shaping of its ADCs. As a single figure THD number, it was 1 dB better than mine using the audio bandwidth (and a db or so worse in wider bandwidth). So as much as I wanted to save money, I begrudgingly passed on the APx525.
The APx555 did manage to beat my analyzer because of a neat and unique capability. Typical audio analyzer is an analog to digital converter (ADC) with a bunch of software. In the case of Audio Precision analyzers, they also have excellent auto-scaling "gain" stages that can amplify or reduce the levels of signals to allow very wide dynamic range. When fed strong signals though, the ADC can produce its own noise and harmonic distortion which can be confused with the device under test.
The late Dr. Tom Kite explains and demos this well in this short video:
Technically, this is the simplified architecture of APx555:
We see two parallel paths for the audio signal. One goes through an ADC followed by a bandpass filter which narrowly extracts the excitation sine wave (i.e. 1 kHz).
The other path does the reverse, attempting to nearly completely remove our source signal (again, 1 kHz in the above example). What is left now are the distortion and noise products. The very strong source signal is removed completely. An amplification stage now boosts the levels of very low noise and distortion products by 22 dB and feeds that to a second analog to digital converter. Since this ADC is not seeing the original strong source signal, it no longer produces much distortion of its own.
Signal processing is the used to stitch the two ADC digital outputs, giving the presentation of a single ADC, even though two are working together this way.
In addition to above analysis hardware, the generator is also improved to have lower distortion. Ironically, this is an analog solution that beats the digital one (a DAC)!!! Sometimes old technology beats new.
Combined these too and looping the APx555 to itself produces this on my bench:
We see that with a 1 volt output, THD+noise is at nearly -119 dB. Spectrum analysis on top right only shows our main tone with no visible distortion/harmonic distortion.
Same test on my older AP2522 shows this:
So about 6 dB lower. That may not sound like much but gains like this are hard, hard to come by.
This is class leading performance as As AP shows in this graph:
The APx555 is the black graph at the bottom.
Note that for testing USB DACs, we don't use the AP's generator. But for upcoming amplifier/pre-amp tests, we do and there, I love the fact that the high-performance generator and analyzer have essentially no contributions to beyond the audible limit of 120 db.
There are other goodies such as bandwidth of 1 Mhz, allowing one to measure such things as switching distortion of class-D amplifier as Dr. Kite shows in this video:
FFT resolution for frequency domain analysis has been substantially enhanced to support up to 1.2 million points. Mine stops at 32 thousand points. While mine is plenty good for anything audible, the much finer resolution FFT in the APx555 will allow me to zoom way in, and test any and all audiophile tweaks for the slightest improvements in such things as jitter and distortion.
Chirp-style log sweeps are provided which allow fast measurements of many parameters, ala what REW does for acoustics. This will allow me to measure and fully characterize loaned equipment and save that in archive for future referral (rather than the few measurements presented).
The software interface of the system is of course radically different. It took some getting used to but overall is intuitive. I have not needed any help in making measurements even though AP folks are standing by ready and very willing to help. Support on that front has been excellent and something that I am very appreciative of. I love the display above where you always see the time domain/oscilloscope type response. It lets you easily verify that everything is working (or not).
It will take me some time to fully port all of my tests from the older machine to new. While I will strive to make them comparable, in some cases the new machine architecture is different so some measurements may differ a bit. That is unavoidable but sets us up for the long term given the newness of the APx555.
I like to take the opportunity sincerely thank Tony Spica of Audio Precision for working with me for some two years (!) on purchase of this gear. The level of support is what I would expect if I still worked at Sony and such, not an independent person who is buying just one unit.
And oh yes my AP 2522 that I have been using is up for sale. No, it is not cheap. These hold their values a lot. Contact me if you are interested.
Avid readers know that I have been getting a lot of mileage out of my much older AP2522 unit. Despite being nearly 20 years old, it has had excellent performance. Great analog design and power of signal processing had kept it relevant for such use even now.
Alas, a few reasons prompted me to buy the new unit. First and foremost is the fact that the software for the new model allows me to directly talk to computer audio products through ASIO interface. You may have seen my difficulty in using my older analyzer there when a DAC doesn't have USB input. With APx555, I am able to now drive those products over USB just as well as it can using S/PDIF, Toslink, and analog output out of the analyzer. This makes measurements now the same whether I use USB or any of these other interfaces.
Second reason is interchange with the industry. Often people want to repeat my measurements and they cannot if they have the newer unit than mine. Now I am able to do that since I am using the latest and current products from Audio Precision.
Audio Precision was kind enough to loan me both the APx525 (USD $18,000) and APx555 (USD $28,000). This was important to me as I wanted to make sure my workflow would be preserved or improved and that performance of the new units could beat what I already had.
With respect to APx525, it only managed to match the performance of my 2522. It had slightly less distortion but somewhat higher noise in ultrasonics due to noise shaping of its ADCs. As a single figure THD number, it was 1 dB better than mine using the audio bandwidth (and a db or so worse in wider bandwidth). So as much as I wanted to save money, I begrudgingly passed on the APx525.
The APx555 did manage to beat my analyzer because of a neat and unique capability. Typical audio analyzer is an analog to digital converter (ADC) with a bunch of software. In the case of Audio Precision analyzers, they also have excellent auto-scaling "gain" stages that can amplify or reduce the levels of signals to allow very wide dynamic range. When fed strong signals though, the ADC can produce its own noise and harmonic distortion which can be confused with the device under test.
The late Dr. Tom Kite explains and demos this well in this short video:
Technically, this is the simplified architecture of APx555:
We see two parallel paths for the audio signal. One goes through an ADC followed by a bandpass filter which narrowly extracts the excitation sine wave (i.e. 1 kHz).
The other path does the reverse, attempting to nearly completely remove our source signal (again, 1 kHz in the above example). What is left now are the distortion and noise products. The very strong source signal is removed completely. An amplification stage now boosts the levels of very low noise and distortion products by 22 dB and feeds that to a second analog to digital converter. Since this ADC is not seeing the original strong source signal, it no longer produces much distortion of its own.
Signal processing is the used to stitch the two ADC digital outputs, giving the presentation of a single ADC, even though two are working together this way.
In addition to above analysis hardware, the generator is also improved to have lower distortion. Ironically, this is an analog solution that beats the digital one (a DAC)!!! Sometimes old technology beats new.
Combined these too and looping the APx555 to itself produces this on my bench:We see that with a 1 volt output, THD+noise is at nearly -119 dB. Spectrum analysis on top right only shows our main tone with no visible distortion/harmonic distortion.
Same test on my older AP2522 shows this:
So about 6 dB lower. That may not sound like much but gains like this are hard, hard to come by.
This is class leading performance as As AP shows in this graph:
The APx555 is the black graph at the bottom.
Note that for testing USB DACs, we don't use the AP's generator. But for upcoming amplifier/pre-amp tests, we do and there, I love the fact that the high-performance generator and analyzer have essentially no contributions to beyond the audible limit of 120 db.
There are other goodies such as bandwidth of 1 Mhz, allowing one to measure such things as switching distortion of class-D amplifier as Dr. Kite shows in this video:
FFT resolution for frequency domain analysis has been substantially enhanced to support up to 1.2 million points. Mine stops at 32 thousand points. While mine is plenty good for anything audible, the much finer resolution FFT in the APx555 will allow me to zoom way in, and test any and all audiophile tweaks for the slightest improvements in such things as jitter and distortion.
Chirp-style log sweeps are provided which allow fast measurements of many parameters, ala what REW does for acoustics. This will allow me to measure and fully characterize loaned equipment and save that in archive for future referral (rather than the few measurements presented).
The software interface of the system is of course radically different. It took some getting used to but overall is intuitive. I have not needed any help in making measurements even though AP folks are standing by ready and very willing to help. Support on that front has been excellent and something that I am very appreciative of. I love the display above where you always see the time domain/oscilloscope type response. It lets you easily verify that everything is working (or not).
It will take me some time to fully port all of my tests from the older machine to new. While I will strive to make them comparable, in some cases the new machine architecture is different so some measurements may differ a bit. That is unavoidable but sets us up for the long term given the newness of the APx555.
I like to take the opportunity sincerely thank Tony Spica of Audio Precision for working with me for some two years (!) on purchase of this gear.
The level of support is what I would expect if I still worked at Sony and such, not an independent person who is buying just one unit.And oh yes my AP 2522 that I have been using is up for sale. No, it is not cheap. These hold their values a lot. Contact me if you are interested.
Last edited:
