Notes from the Test Bench
By Bruce Hofer, Chairman & Co-Founder, Audio Precision
I’m in Copenhagen, at the excellent Radisson Blu (SAS) hotel, having just returned from a day of labs as part of Dr. Andersen’s industrial course on Class D amplifier design at the Technical University of Denmark.
Each group of students had an APx525 analyzer with the BW52 ultra-high bandwidth option on their workbench—there’s no better instrument for looking at Class D amplifiers.
The prototype amplifiers exhibited a range of switching frequencies, from as low as 300 kHz to some as high as 600 kHz.The BW52 option (extending the APx525’s bandwidth in excess of 1MHz, with 1 million point FFTs) was quite a hit, allowing the students to accurately measure and set the switching frequency with a precision that vastly exceeded their only other alternative: using an oscilloscope. Being able to see switching frequency shift along with sideband behavior really opened the eyes of the PhD student leaders, who had devised the various lab experiments in the first place.
The students also appreciated the graphic reports we were able to generate on the fly with the APx, which should supply some food for thought over the next few days.
Thank you to Dr. Andersen and his students for inviting AP to participate as they advance the state-of-the-art in Class D design. Very interesting and most worthwhile work.
And finally, tomorrow will be a big day for us here at AP, when our new APx515 audio analyzer ships to its first customers around the world. We’re quite proud of what we’ve created, and are confident you’ll agree that it takes production test to a whole new level of speed, performance, and economy.
Portable One Sale and End-of-Life
AP's Portable One Dual Domain and Portable One Access analyzers are being offered at 25% off list price until December 31st, after which time they will be discontinued.
In addition to new units, we also have a few refurbished Portable One Dual Domain analyzers available. These have a 20% refurbished instrument discount on top of the 25% End-of-Life discount, and include our standard three year warranty. Please note that the prices below are for the US and may vary in other countries.
The ATS-1 will remain in production. It is nearly identical to the Portable One, buts adds HP8903 emulation mode and a metal case for superior rejection of electromagnetic interference. Both the ATS‑1 and P1 are self-contained instruments that don't require a PC to operate.
Now is a great time to contact your AP representative or distributor, because remaining stock is limited.
Output: Testing PC Audio
by Eric Schultheis and Adam Liberman.
The excerpt below is the first part of a new AP Technote on Testing PC Audio, which includes an accompanying APx project and .wav test files. You can download the complete Technote package from AP.com, which includes further instructions on how to perform the tests.
This Technote discusses the techniques necessary to perform the “Basic Six” audio measurements on consumer-level PC audio devices. This category includes sound systems integrated into notebooks, netbooks, tablets, and PC motherboards, as well as expansion cards and some external devices. It doesn’t include USB-connected headsets, nor pro-audio devices with balanced input and output connections.
We’ll help you to work around some of these challenges to make the basic six audio measurements—level, frequency response, phase, THD+N, crosstalk, and dynamic range (see Technote 104: Introduction to the Six Basic Audio Measurements). These measurements are often enough to characterize the performance of an audio device.
This Technote does not delve into more comprehensive tests, like maximum and minimum input levels and anti-aliasing filter response, which can detect serious problems that might arise when using a device and when interfacing it to other audio devices. It also does not test for compliance with Microsoft’s WHQL (Windows Hardware Quality Labs) requirements to receive the "Certified for Windows" logo.
Make sure that all signal processing options in the software and device drivers, including EQ, echo, and simulated surround sound, are turned off before beginning testing.
Fig 1 PC Audio connections
Using good quality cables, connect the APx unbalanced analog outputs (left and right) to the DUT’s analog inputs. If the DUT has both microphone and line level inputs, then start with the line input and retest afterwards using the microphone input. The APx generator output should be set to 600 Ω impedance and AC coupling when feeding a consumer-type microphone input, and low impedance with either AC or DC coupling otherwise.
Connect the DUT’s analog outputs to the APx unbalanced analog inputs (left & right). If the DUT has both line and headphone outputs, test first with the line output, and then retest later with the headphone output driving a 32 Ω load. Using cable with two conductors plus shield will result in lower noise than using a single conductor plus shield. The shields should be connected only at one end.
A heavy ground strap (large gauge multi-strand copper wire) between the DUT’s chassis and one of the
ground lugs on the APx may reduce noise by minimizing ground potential between them. On the other hand,
sometimes it can make things worse by adding an additional ground path. Therefore, it’s necessary to try
both ways while measuring the noise, to determine the best method before performing any additional testing.
Plugging the PC and the computer into the same AC outlet may also reduce noise. Never remove or lift the AC
power plug's safety ground on the PC or the audio analyzer—it’s there to prevent you from getting a
shock should an electrical fault occur.
Setting the DUT Levels
If possible, we recommend setting the input and output level controls to their maximum levels. This takes their setting variables out of the equation, reducing the chance of input overload and resulting in more repeatable measurement results. The analyzer output level is then adjusted to obtain a recording level as close to full scale as possible. Alternative level-setting methods include setting the controls to a nominal position (like 70%), and using a fixed generator level (such as +10 dBV) while setting the controls for a proper recording level.
Fig 2 Windows mixer
Setting levels using the recording level meters in the recording software does not guarantee distortion-free
recordings. It is necessary to make a test recording to check for clipping. Note that some PC audio devices may
clip before reaching full-scale recording—this should be noted, as it indicates a defect in the
Launch your recording software, and open a new file for recording in uncompressed wave (.wav) format at a 48 kHz sample rate with 16-bit resolution. You may also wish to choose other settings, or make multiple test passes using different sample rates and bit depths for each one. The playback files included with the APx project that accompanies this Technote are recorded at 48 kHz/16-bit. You can use the APx Waveform Generator Utility to generate playback files at other sample rates and bit depths.
Open the APx project file “PC_Check.approjx” (provided in the Technote 108 download) and select Signal Path Setup in the #1_Record_to_File signal path. Set the DUT’s input and output levels to maximum. Turn on the APx Generator, and then start recording. While the signal is being recorded, you will be able to monitor the signal level through the DUT’s outputs.
Adjust the APx Generator level, first using the increment/decrement buttons, and then by typing a level into the dropdown box to adjust by tenth decibel increments. When the APx generator units are set to Vrms, the increment/decrement controls change the generator value by about 2 dB. If you switch the units to dBV, the increments are only 1 dB.
Getting the level as high as possible without any clipping will produce the best possible signal to noise ratio. Common practice is to set the level slightly below full scale, at -0.3 or -1.0 dBFS, because of the difficulty of precisely setting 0.0 dBFS, and because some converters show an increase in distortion and go into compression just before clipping. It may take a few iterations of this process to find the optimum level.
Fig 3 Adjusting levels to find the clipping point.
Figure 3 shows the graphic view of a file recorded using the procedure just explained. You can see where the level adjustments were made over time, and that they end just before 0 dB full scale. Zoom in on the recorded waveform (using zoom tool in your recording software) to look for clipping.
Fig 4 Clipped signal.
Figure 4 shows a clipped signal, and Figure 5 shows a clean (unclipped) signal.
Fig 5 Non-clipped signal.
When you are satisfied that your level is optimized, make note of the generator level, as well as any changes you made to the DUT level controls.
In all cases we will be using the analog inputs and outputs, which will enable us to test the A/D and D/A converters. If the PC audio device has S/PDIF digital I/O, you can test this as well, but we won’t be covering that here.
Download Technote 108 to read the continuation of this article and to get the APx project and test files.
Sound Advice: Graph Colors in AP2700 and ATS
By Bill Rich and Adam Liberman
The graph traces in the AP2700 and ATS control software are automatically assigned colors each time a sweep is performed. Additionally, you can manually set the color, style, and thickness of each trace after measurements are completed. In the following, we’ll explain how the colors are automatically chosen, and we’ll give you an AP Basic macro that lets you configure the traces exactly the way you want using one mouse click.
Overall settings for graph panel traces are found in Utilities | Configuration on the Graph tab under Trace Colors.
Utilities | Configuration, Graph tab
Reset Color Cycle
If Reset color cycle is checked, each new appended sweep will assign the same six colors to Data 1 through 6. If Reset color cycle is unchecked, each new appended sweep will use the color cycle, but the colors assigned to Data 1 through 6 will each be advanced by one color in the cycle.
The color assignment pattern is illustrated in the chart below. The last two columns show the colors that will be assigned if you do a typical two-channel sweep, either by setting up Data 1 and Data 2, or by setting up Data 1 and checking the Stereo Sweep checkbox.
Trace color assignments
Use test colors only
"Use test colors only" checked
AP Basic Macros
AP.Graph.Legend.LineColor(1, 1) = apbGreen
We’ve written a short stand-alone AP Basic macro that lets you set the graph trace properties quickly, even if you aren’t otherwise using macros or automation for your testing. By adding the macro to your Quick Launch toolbar, you can configure the graph and printout properties by simply clicking an icon after running your sweeps. If you wish to have an alternate set of properties, simply edit the macro as desired, resave it with a new name, and create another icon on the Quick Launch toolbar for it. Installation instructions are included with the macro download.
Other Recent Knowledge Base Articles:
Test Results: AP News & Events
APx500 v2.6 Software Release
APx500 v2.6 is being released tomorrow, August 27th, and will be available for download at /download/apx.
This latest release adds a derived measurements feature and an audible signal monitor, as well as support for the new APx515 production test audio analyzer. Download What's New in APx500 v2.6 for the full list of new features and improvements.
Welcome James Kelly, Audio Precision's UK Employee
AP Website Speed-up
We've been working hard on hardware and software upgrades to speed up our website. The improvements are now in place, so you should now experience a faster, more responsive site. If you haven't visited AP.com lately, go ahead and browse our products (including the new APx515), audio test solutions, knowledge base, downloads (utilities, Technotes, Application notes, etc.), support area, and online configuration and pricing.
©2010 Audio Precision Inc.