Using a Soundcard For Test & Measurements

Using a Soundcard For Test & Measurements

This document is intended as a suppliment to the Help files for various DazyWeb Laboratories software products such as the DS-2002 oscilloscope, SA-3002 spectrum analyzer, TMS-1 test and measurement system and the SG-2102a signal generator applications. All rely on the pc soundcard as the hardware platform to obtain audio measurement data or produce test signals. The soundcard for a PC has been designed as a consumer-grade media product, not unlike your Hi-Fi receiver or VCR. The inputs and outputs are minimally protected and are intended to hook to PC speakers, a microphone and possibly a TV capture card. To use them for general test purposes requires care and some knowledge of measurement techniques.

The soundcard has, at a minimum, a stereo line input and a monophonic microphone input for recording and a stereo line output sufficient to drive amplified loudspeakers or headphones directly. Many newer soundcard also have multichannel outputs that rely on proprietary decoding techniques such as Dolby Digital 5.1 from a DVD player to fully utilize those extra outputs. The microphone input typically has poorer signal to noise and frequency response than the line inputs so will not be discussed in this text, nor will extra multichannel outputs. If your purpose is to do microphone-based frequency response measurements, then buy the best mic you can and consider one with a built in preamp such that you can still use the line inputs. If your soundcard has digital input or output jacks, they may be utilized directly if connected to a preamp or processor that also has digital input and output jacks. The soundcard will convert to and from the SPDIF digital standard interface and provides all the buffering needed in that case.

In figure 1 below is a typical hookup for measuring a small signal product such as a mixer panel or a Hi-Fi preamplifier. If the device uses vacuum tubes then treat it as a power amplifier as output voltages may be in the tens of volts. Most mixers and preamps are capable of up to 7V rms out but it is limited by a 1000-2000 ohm resistor in those devices and will not hurt your soundcard even though the level technically exceeds the 2-4V maximum input of the soundcard.

Just monitor the incoming waveform from the preamplifier using one of the apps with a scope built in (TMS-1, DS-2002, SA-3002) and find the point where flat-topping occurs, no matter what the recording input level is set to on your mixer panel on the pc. That point then is the maximum you can drive the soundcard directly without an attenuator. Note the settings of the preamp (or mixer) gain controls and the pc mixer panel settings, and any settings in the app being used that provide the drive signals.

FIGURE 1.

For measuring power amplifier at the speaker terminals or vacuum tube preamplifiers or equipment you are unsure of the level of it's output, see Figure 2 below. The Soundcard Isolation Circuit is detailed in Figure 3. The Souncard Isolation circuit only shows 1 of 2 channels and you will need to build a version for each input channel. There is a company in Australia called JAYCAR Electronics that makes a kit with all parts that performs the same function, although I have not gotten one and verified it's operation. Here is the link:

http://www1.jaycar.com.au/index.asp

The model number is the KA1811 and it cost about $26.00 US at the time of this writing.

Figure 2.

Figure 3.

For general purpose probing done cheaply, or as a quick divider off of a power amplifier binding post, you can use the resistor divider shown in figure 4. Just build that into a cable between the soundcard line in and the speaker output on the amplifier and this will alleviate overload on amplifiers of 100W or less. For 100-400W amplifiers, use a 20:1 divider by increasing the 10K to a 20K resistor in the top of the divider leg. Any soundcard including any add-on isolation or divider circuits will need to be calibrated if absolute voltage measurements are desired. You will need to own or borrow a digital voltmeter of good AC accuracy. My reference standard is a Fluke 8060A. If only relative frequency response or relative displays of waveforms are needed then calibration is not required.

Figure 4.

The line output is generally good for 2V rms output into fairly low impedance loads. Walkman style headphones are 39 ohms per side and it drives them very well. An output buffer is generally not required. If you need to drive insensitive circuits that need 3-12V peak drive or desire maximum buffering of your soundcard then you may add an output buffer as shown in Figure 5 below.

Figure 5.

Oscilloscope probes designed for hardware oscilloscopes are generally 10:1 with a 1:1 divider option on a lot of the models. The 10:1 relies on the fact that an oscilloscope has a 1M input impedance. Any soundcard out there will be in the 10K-50K range for input impedance and will not be usable directly with a store bought scope probe. To provide a decent signal to noise and frequency bandwidth, the isolator shown above has only a 10K input impedance so it also won't work directly with a scope probe. I suggest using DVM meter probes instead and being careful what you probe. Although most all soundcards are ac coupled as is the Soundcard Isolation Circuit shown above, voltages greater than +/-15V in circuits may exceed the ratings of the coupling capacitors in the soundcard or Isolation Circuit. When using unshielded DVM probes, take care to not route the probe near devices containing high magnetic fields (power transformers, soldering irons) as hum will be induced into your measurements. Connect the ground connection to the nearest ground point in the circuit under test. The PC always has a 3-wire cord with an earth ground. If the circuit or device you are measuring also has a 3-wire cord with a ground prong then you may get excess hum in your measurements. Use ground isolators from the hardware store at your own discretion, preferably on the device under test NOT the PC. Alternatively an isolation transformer may be used on the device under test. These may be obtained from Mousser Electronics. Get one without the safety ground routed through to the output side.

FREQUENCY MEASUREMENTS: The maximum frequency that a soundcard can measure or reproduce is one half the sample rate. If you are using 44,100 as the sample rate, then you can measure up to 22,050 Hz. Many new soundcards claim 96 kHz sample rates, but read the specs carefully. This may only apply to the digital in/outs and not the analog record ins. Often they are limited to 48 kHz sample rate or 24,000 Hz maximum measureable frequency. Since fewer samples are available as you approach one half the sample rate, the oscilloscope waveform will degrade rapidly. Likewise, the accuracy of level measurements suffers. Soundcards vary in low frequency performance. The Soundblaster 16 is rolled off over 5 dB at 20 Hz whereas a Lynx Studio One is flat to within 0.1 dB at 10 Hz. You will be able to get something out of any soundcard down to 3-4 Hz with sine waves but the max level out will not be as great.

NOTE: SOUNDCARDS WILL NOT WORK TO MEASURE DC VOLTAGES DIRECTLY*. EVEN IF THE AC COUPLING COMPONENTS ARE BYPASSED, THE D/A SECTION OF THE CODEC IS GENERALLY NOT DESIGNED FOR DC OPERATION.

*A chopper circuit can be used in combination with measuring the new ac peak level, as a means to measure dc. The current software, as provided, does not integrate the chopped signal back to a fixed dc level for viewing, however. Also, due to the 20-20kHz range of the soundcard, both droop below 200 Hz and overshoot above 2 kHz will complicate a chopper method. The software will have to window the center of the chopped square wave (frequency ideally at 1 kHz) for best accuracy. A voltagr to frequency convertor is another design method that could be used. It will not be possible to accurately see an ac waveform riding on a dc waveform since either the chopper method or the ac coupled method would be used but not both. Sometimes there is no substitute for a real hardware oscilloscope.

Notice: No liability is assumed by the author or DazyWeb Laboratories for the integrity of your soundcard or PC, based on recommendations in this document. All circuits and discussions are based on good engineering practices and hands-on practical experience in the state of the audio art. However, even in the best of hands, accidents can happen during test and measurement. Power should be disconnected from devices under test when they are hooked up to the soundcard and the PC should be powered down when cables are hooked directly to the soundcard. Care should be taken during any measurement sessions and all hookups rechecked before power is reapplied to devices under test. If you have an assistant (or are the assistant) have someone else double check your work before re-applying power.

Considering that you are working with nearly free test techniques (the soundcard was already there on a PC you owned anyway and the DazyWeb Laboratories software is free), you now have means to do some measurements that previously was prohibitive to those of us on limited (hobby) budgets. Good oscilloscopes cost $700-1000 minimum and new Audio Precision test systems start at about $14,000. Any links or recomendations for additions or corrections to this document would be most welcome and you may E-mail the author at:

vic@infinitespectra.com

document revision 07-May-2010

Author: Vic Richardson