Notes from the Test Bench
By Bruce Hofer, Chairman & Co-Founder, Audio Precision
First, I am very proud to announce that APx500 v2.8 will be released in early July. Version 2.8 supports our new APx Bluetooth module, as well as our updated HDMI module which adds audio return channel (ARC) capability.
Since we announced our Bluetooth module in May, we’ve previewed it to literally hundreds of engineers with a nearly universal positive response. APx is the only fully-fledged audio analyzer to integrate a Bluetooth radio, and the benefits in terms of set up speed and measurement results are clear.
Of course, connectivity is just the first step, so this issue of Audio.TST runs through a couple of real-world test scenarios—the first measuring the audio characteristics of a Bluetooth headset with an artificial ear, and the second testing the HDMI audio return channel of a newer flat screen TV. You can also download What’s New in APx500 v2.8 to see the other features we’ve added—many in response to customer requests.
Before I end this month’s Notes, I hope you will join me in a moment of reflection to mark the passing of two extraordinary analog engineers: Jim Williams and Bob Pease. Both men contributed to the field with a talent and dedication that will be sorely missed.
Output: Testing Bluetooth Enabled Devices with APx
Audio Precision’s new APx Bluetooth module, announced in the previous issue of Audio.TST, will be available early July in new instruments and as an upgrade to existing APx520, 521, 525, and 585 analyzers. In the following article, we go through the steps of testing a Bluetooth enabled headset.
The Bluetooth module in an APx525 analyzer.
Testing a Bluetooth enabled device is similar to testing other audio gear, once you’ve established a Bluetooth connection between the DUT (device under test) and the audio analyzer. The Bluetooth enabled headset that we will test contains two signal paths—a Bluetooth receiver feeding an earpiece, and a microphone feeding a Bluetooth transmitter. In the following tests, we are going to test the receiver-to-earpiece path through the device.
Signal path for testing the headset.
The APx Bluetooth module will be used to transmit test signals to the headset. The headset is placed on an ear and cheek simulator, which has a physical cavity that duplicates the response characteristics of a typical ear. A measurement microphone, placed where the eardrum would be located, is connected to an unbalanced analog input on the APx analyzer. For this article, we’ll skip the details of connecting and calibrating it.
G.R.A.S. 43AG Ear and Cheek Simulator.
Our intent here is to measure the performance of the entire headset. If we had wanted to analyze the device internally, we could have opened it up and connected directly to the speaker terminals or to the output of the Bluetooth receiver chip. With a fully equipped APx500 Series analyzer, it is possible to run tests using any combination of Bluetooth, analog (bal/unbal), digital (bal/unbal/optical), digital serial (I2S, etc.), and HDMI connections.
Testing a Bluetooth Headset
Before establishing a Bluetooth connection, it is necessary to pair both devices—in this case, the device under test and the analyzer. Pairing establishes a trust relationship so the devices know that it’s OK to connect to each other in the future.
Pairing the headset and the APx analyzer.
Once pairing is completed, a connection can be made. In this case, we will use the Bluetooth profile HSP with the CVSD codec, because this is what the headset supports. APx supports the HSP, HFP, A2DP, and AVRCP Bluetooth profiles, and the SBC, aptX, and CVSD codecs (data compression methods). If you want to learn more about Bluetooth profiles, there’s a good explanation and list at Bluetooth.org, (the website of the Bluetooth SIG).
Now that we’ve paired the two devices, we go ahead and make a connection.
Then there’s one more step—we need to open a SCO. Because this hands-free device acts like a phone receiver, we need to tell it that we are “on a call” if we want to send and/or receive audio.
Bluetooth Actions dialog.
Making the Measurements
The first measurement we’ll run is Continuous Sweep, which returns 14 different results. The maximum sweep frequency is set to 3.960 kHz because this device is limited to a sample rate of 8 kHz. Below is the frequency response, relative to 500 Hz after compensating for the transfer function of the ear and cheek simulator:
Headset frequency response using Continuous Sweep.
Equally useful is the APx Multitone Analyzer measurement. Because of the low 8 kHz sample rate of the Bluetooth connection to the headset, we created a special multitone with a frequency range of 100 Hz to 4 kHz. This took about a minute to do in APx500 using the Edit Multitone Signal Definition dialog. As you can see below, the results very closely match the Continuous Sweep measurement. Slight differences in response are due to the behavior of the CVSD codec under different signal conditions.
Headset frequency response using Multitone Analyzer.
Now, let’s look at THD+N. It’s extremely high at 20% (–14 dB). The distortion product ratio view shows that the third harmonic is the highest, followed by the second.
Distortion product ratio.
Next, we check dynamic range, using the APx “Dynamic Range—AES17” measurement:
Dynamic Range AES17 measurement.
And last, using Measurement Recorder, we play a tone through the device and monitor THD+N over a specified time period to check for dropouts. Any dropouts will show up as spikes in the THD+N result.
Signal dropouts will show up as spikes in the THD+N reading of Measurement Recorder.
There are some special considerations that need to be taken into account when testing over a Bluetooth connection. One is the delay caused by the transmission and encoding process. In the case of the particular device we just tested, it’s pretty low at around 10 ms. Sometimes though, it can be larger. The acquired waveform and impulse response results in Continuous Sweep allow us to verify that the delay has been accounted for and that the entire acquisition has been measured. If necessary, we can tell the analyzer to wait briefly for the signal to arrive by setting a delay in signal path setup. The image below shows the acquired waveform before and after the delay has been applied.
Acquired waveform, with no delay and 110 ms delay added in Signal Path Setup.
Another consideration is the effect of the codec. Codecs reduce the data rate by altering the signal according to defined algorithms. Therefore, you may get different measurement results when you change settings, such as generator level and sweep time, because they affect how the codec processes the audio signal. The graph below shows seven appended Continuous Sweep measurements on a device, with sweep times ranging from 200 ms to 2.5 s. Note how the low and high frequency response varies.
Frequency response variations with seven different sweep times.
And finally, due to the noise-gating built into some digital converters and input stages, it may be necessary to measure noise in the presence of signal to get accurate results. This can be done in APx500 using either the Dynamic Range, THD+N, OR Multitone measurements. With either measurement, be sure to set the fundamental level low enough so that distortion products are buried in the noise floor of the device, yet are high enough to open any gating or muting.
Sound Advice: Testing HDMI ARC with APx
AP’s optional HDMI module for the APx585 has been upgraded to include the Audio Return Channel (ARC) feature that’s part of the HDMI 1.4 specification. In the following article, we run some tests using the new module with APx500 v2.8 software.
In a typical multimedia setup, a disc player is connected to a multimedia receiver, which plays the surround sound over its attached speakers while sending the video signal on to a TV over an HDMI cable. However, when the TV is the source of the audio (for example with over-the-air broadcast or direct input from cable), the TV instead must send audio to the receiver. Rather than using a separate SPDIF digital audio cable, HDMI ARC uses a previously unused pair of wires on the HDMI cable to accomplish this function, sending audio down the cable in the opposite direction of the regular HDMI audio and video flow.
HDMI-ARC eliminates a separate SPDIF digital audio cable.
First, we will test the HDMI ARC receiving capability of a multi-media receiver, using the APx analyzer to emulate a TV. We connect a cable from the HDMI ARC Tx (transmitter) jack on the APx analyzer to the HDMI Output jack on the receiver. This can be a little confusing, but since the Audio Return Channel flows backwards, it is correct to go into outputs and out of inputs! From the speaker outputs on the receiver, we go into the Analog Unbalanced inputs on the APx analyzer.
Connections between APx and the receiver.
Even after hooking up the cables, the audio can’t flow until the two devices negotiate a connection using CEC (Consumer Electronics Control) commands. The receiving device is the one that initiates and terminates the connection. The HDMI ARC transmitter and receiver settings dialogs in APx500 let you send CEC commands, as well as force a connection if necessary.
CEC command dialog.
As HDMI ARC is a SPDIF connection, the APx generator can send two channels of linear audio from its generator or from a .wav file. The APx generator can also play a Dolby or DTS encoded file that contains up to six channels. In this case, we will just send two channels of audio using the Multitone Analyzer measurement. As can be seen in the result below, the response is flat.
Frequency response through receiver.
Now, let’s have the APx analyzer reverse roles and emulate a receiver. We connect HDMI source on the APx to HDMI IN 1 on the TV. Then, we connect HDMI IN 2 from the TV to HDMI Rx on the APx. We use HDMI IN 2 because the label on the TV tells us that this particular HDMI input has ARC.
Connections between APx and the TV.
Our test results show flat response and low distortion, although running the Digital Error Rate measurement shows us that the audio signal is not bit-accurate and is altered somehow on its way through the TV.
Digital Error Rate bit accuracy test through TV.
As you can see, testing HDMI ARC is pretty straight-forward, once you get over the confusion of the backward signal flow!
Test Results: AP News & Events
©2011 Audio Precision, Inc.