Using the Audio Precision APx585

The APx585 is a completely new and different measuring instrument from Audio Precision. To start with, for analog signals, it is an eight-channel unit capable of measuring all eight channels of a unit under test at once.
Publish date:
Social count:
The APx585 is a completely new and different measuring instrument from Audio Precision. To start with, for analog signals, it is an eight-channel unit capable of measuring all eight channels of a unit under test at once.

The APx585 is a completely new and different measuring instrument from Audio Precision. To start with, for analog signals, it is an eight-channel unit capable of measuring all eight channels of a unit under test at once. For digital testing, it is a two-channel unit. Notably absent from the front panel are XLR balanced connectors; the basic front panel analog I/O is via unbalanced BNC connectors. Balanced I/O is taken care of via the two DB-25 connectors on the front panel for input and output. Available from Audio Precision in the CAB-585 Cable Kit are accessory breakout cables terminating in eight XLR line connectors for input and output connection to the device under test. Eight channels is a logical number for such an instrument to have: it handles the current surround sound and DVD audio formats and is a nice logical number to measure multichannel studio gear with eight or more channels (which, if it has more than eight, is usually a multiple of eight). [Audio Precision says that an 8-out /16-in version is in the works - Ed.]


FAST FACTSApplications

Test and measurement

Key Features

Hardware interface; Windows software; 24-bit/192 kHz A/D-D/A; USB port; multiple acquisition modes


Starts at $21,000


Audio Precision | 503-627-0832 |

Aside from these fundamental physical aspects, the software and user interface is completely new and totally different from past Audio Precision machines. The design intent of the 585 was to make the user interface much more intuitive, easier to use, and, as a result, enable much faster testing. This is accomplished through a “Measurement Navigator window” approach that has all the common types of measurements listed. It is a measurement-oriented interface rather than one with an approach of controlling the machine's generator, analyzer, and graphing approach (as was used in prior Audio Precision instruments). In practice, one picks the tests desired by clicking on them and their associated subsets. Each one can be simply set up as to the particular parameters to be used. What is amazing is that one can click on the Run Sequence button at the top of the Measurement Navigator window and the machine does all selected tests in sequence and, if desired, generates a report of the results upon completion. This is far better suited to the production testing of audio gear and is much easier to set up as to limits of pass/fail for the particular tests. Each project is then a complete compilation of all the relevant tests and a setup of them in one project file. In effect, a complete procedure is in a single all-inclusive file — the particular project named and saved file. Note that the measured data itself does not get saved; the idea is to generate and save the report of the measurements. It goes without saying that loading a particular project will repeat all the tests within at some later time, which is a very relevant situation for production testing.

Like the compact Audio Precision ATS-2 instrument, this new machine is all-digital, meaning that the analog inputs are digitized by 24-bit/192 kHz A/D converters and all the output signals are converted to analog with 24-bit/192 kHz D/A converters. All the signal acquisition and generation is done in the digital domain. There are three basic methods of generating the tests: single value using one or two sine waves for stimulus, continuous sweep using a log “chirp” type signal, and stepped frequency or level utilizing a sine wave across a range of frequencies in discreet steps. In all cases, the analysis is done with FFT calculations on the acquired data. Also new in the APx585 is the method of connection to the host or control computer. It is now via a USB interface. Great — no more APIB PC interface card and cable to deal with!

It is not surprising that all this great new capability requires a Herculean powered computer to run it efficiently since processing eight channels takes somewhat more computing power than for two channels. Minimum recommended attributes of the control computer are as follows: a 2 GHz processor or faster, 2 GB of RAM, 2 MB of Level 2 Cache, and Windows XP with Microsoft .NET 2.0 Framework (included with the APx500 software). Since I didn't have any machine with quite those capabilities, Audio Precision kindly loaned me a great new IBM Thinkpad laptop with the necessary computing power. As a related difference between the APx585 and prior Audio Precision machines, all of the data processing is in the control computer for the new APx585 with the hardware designated to acquire data only. In contrast, prior machines had a lot of processing done with the unit's machine hardware with much-relaxed requirements for the control computer. For instance, my SYS-2722 is run by a Tiger Direct Sempron 2400 computer with 480 MB of RAM; it runs this machine very nicely and responsively. While not necessarily mentioned by Audio Precision, the demo mode of the APx500 software can be run on a much less capable machine. I installed it on my lab computer and it ran just fine. Those interested in exploring the program interface and features should be encouraged to download the software for a try.

In Use

Turning our attention to the software interface, see Figure 1 for the basic window layout of the software. This is the default way the program comes up (at least with the high screen resolution of this control laptop). As can be seen, the program window area is divided up into several sections with the program navigation portion being at the upper left. The vertical width of the Measurement Navigator window and the area below it is adjustable; I personally liked the display set with the width maximized. See Figure 2 for a larger view of the navigation window. Here, some of the tests have had their associated + symbols clicked to open their subsets. The little boxes just to the right of the individual subset test check boxes will show whether or not a test was within set limits or not when run. The signal path setup has been highlighted, resulting in the pictorial diagram in the upper right of the program window showing the particular physical arrangement of input/output selected — in this case, unbalanced analog I/O. Between the Measurement Navigator window and the physical setup path picture are the actual signal path setup variables. The default setup is eight-channel unbalanced I/O. Other choices for the output configuration are analog balanced, digital electrical (BNC connector), digital optical (TOSlink), and none (external). The latter implies input from the DUT (device under test) only. Choices for input configuration are unbalanced, balanced, digital electrical, and digital optical.

At the bottom of this section are generator controls for generator amplitude, frequency, and the single channel at a time to be driven. The available variables in the section differ according to the particular test highlighted in the navigation window. At the lower left is a window that can be display one of three kinds of view: an oscilloscope for waveforms versus time, a spectral display of amplitude versus frequency, or a set of level meters for all channels. The default spectrum analyzer is shown. Taking up the rest of the area of the program window space is the measurement view window. The results of a particular measurement will show here with the axes and form of presentation being appropriate for the particular test run. The default display here assumes an incoming level for the DUT of rms output volts for each of the eight channels. The measurement view can be one of four types: the usual x and y axis plots like amplitude versus frequency, a tabular form of the measured data, horizontal or vertical bar graphs of various things like level versus channel number in horizontal form (or perhaps signal harmonic amplitudes versus harmonic number as vertical bars), and a 3D view that can show a x-y plot of a variable versus channel number like frequency response versus channel number on the z axis.

The loaned control computer was running a version of the newer v1.1 software in beta form, as it has not yet been released [Editor's note: The v1.1 software is now shipping]. This version adds support for playback only devices, Dolby/DTS Confidence Testing, and API Programming. After spending several days learning how to run the machine, I made numerous different kinds of measurements with the APx585 on a number of devices including a small digital switching power amp, an old solid state stereo power amplifier, some played back WAV files using Win Amp on the built-in sound card on the control laptop computer, and an eight-channel equalizer.

Figure 3 is a plot of the THD+N ratio versus power output at a 48 kHz sampling frequency for the small switching amp for a 1 kHz test signal with the 20 kHz measurement filter in (bottom trace) with the full measurement bandwidth of 80+ kHz (middle trace) and for a 5 kHz test frequency with the 80+ kHz measurement bandwidth (top trace). Note that the APx585 doesn't have a data smoothing function (like the earlier Audio Precision instrument software), which would have been nice for the lower curve. The data can be exported to a spreadsheet or other program for such smoothing if necessary.

Figure 4 is a plot of THD+N versus frequency at a power output level of 1W for sampling frequencies of 48 kHz and 96 kHz for the same amplifier. Figures 3 & 4 illustrate a feature of adding comment text to a measurement graph. It should be noted that this particular little amp is a digital-input-only design with no output digital reconstruction filter as such and has the usual two-pole analog output low-pass filter with a cutoff frequency out of the audio range. These two figures were done using the stepped level and stepped frequency tests, respectively. Going on to the eight-channel equalizer and an eight-channel balanced I/O signal path setup, a different set of EQ curves was set up for each channel and a frequency response was done with the frequency response test mode. A sequence for the checked components was run; the gain versus frequency measurement is shown in Figure 5. The ripples in the response curves are artifacts of combining the adjacent filter bands in the equalizer. The measurement result reports are exportable in various forms including PDF. A report of this sequence was generated and exported as a PDF file. Page 3 of the report is shown for this measurement in Figure 6. The report is first viewable before exporting in the measurement view portion of the program window and is scrollable and steppable through the various pages. The saved page three of five of the report in the figure is typical of the appearance of the other pages of reports.

In Use: WAV Files

Measurements from a playback-only device have been greatly enhanced and made easier and more reliable in the APx500 software. Audio Precision has created a series of WAV files of different fixed frequencies, amplitudes and various frequency sweeps at different sample rates and bit densities. Further — and most important — the tests in the playback-only input mode of the APx585 are intelligent in knowing the characteristics of these files such as what the frequencies are and the various timings within, all of which will make these kinds of tests more reliable to run. These test files will be available as test CDs and DVDs for such testing of various devices with the APx585. As mentioned above, the control laptop has a small set of these files loaded into the Win Amp player application. Using the stepped frequency test, a 31-point frequency file was played back and level, relative level, deviation from flat in a 20 Hz to 20 kHz band, THD+N ratio, and THD+N level were all measured as a result of the single pass of the played file. The relative level frequency response of the two channels is plotted in Figure 7. With the Measurement Recorder mode, I then measured the amplitude versus time of a test signal that produced decreasing amplitudes from 0 dBFS down to some very low level in 5 dB steps. The noise level of the sound card in this test environment appears to limit the response to about - 80 dB. This measurement appears in Figure 8.

I have a few nits to pick with the machine. First, it does not have any monitor outputs for viewing the measured waveforms and their distortion products on an oscilloscope. Having been taught to always monitor what I was measuring with an oscilloscope (from the first days of my audio learning way back in antiquity by my mentor, Gordon Mercer) and always having done so subsequently as a matter of course, I do miss that. Another thing that is quite a departure from past Audio Precision practice is that the data values are not saved when a project is run and then saved. However, this was a deliberate system design choice with the saved report being the prime means to convey test results. This does make sense, as virtually anyone can read a PDF file. In going through the playback only measurements, I found that the means to measure something versus signal amplitude was not nearly as sophisticated as measuring something versus frequency. These kinds of measurements are done with what is called the Measurement recorder and simply plots the level, THD+N ratio, or THD+N level versus time.


I could go on and on with more measurements of various IM distortions, crosstalks, phases, and such, but I think I have given some idea of the APx585's capabilities and its program interface. In learning the machine, I quickly grew quite fond of it. If I had lots of extra bucks, I would buy one. While it won't do some things I like and need to do with my SYS-2722, it does a lot, if not most of the things I routinely do in my various testings. Further, it does them in a most expeditious and beautifully graphed manner.

In conclusion, this instrument appears to be very well conceived for its intended purpose. I am sure a lot of them are going to be purchased and put to good use in a great variety of audio testing duties in the very near future.

Bascom King is a consultant to the hi-end audio industry. Recently co-designed with Arnie Nudell most of the new speakers in the Genesis Advanced Technologies line, In addition to measuring digital devices and reviewing test gear for Pro Audio Review, he has been measuring preamps and power amplifiers for the on-line magazine SoundStage!, writing equipment reviews for The Audiophile Voice, and evaluationg and testing switching power amplifiers for various manufacturers.