NOTE: THIS PAGE IS JUST A DRAFT AND NEEDS A LOT OF WORK! DON'T HESISTATE TO ADD YOUR STUFF THAT MIGHT BE USEFUL TO OTHERS!

The RTX6001 is a high-performance audio analyzer that was developed by JensH with feedback from the diyAudio community. The development is described in a long thread. The analyzer connects to a computer using a USB 2.0 connection. It works as a standard USB Audio Device and can therefore be used with all audio testing software that will work with a USB soundcard. Technical details of the analyser are available in the RTX6001 data sheet and in the manual, which is available as PDF download at the RTX website.

The RTX6001 manual contains a chapter with the title "Performing Measurements". However this chapter is extemely short. This is it (yes, that's the complete chapter):
The exact operational procedure to use when performing measurements depends on the analyzer SW used. Refer to the supplier’s user manual anual for the selected analyzer program for further information. If very low distortion is to be measured, it is generally best to operate the RTX6001 Audio Analyzer below -10dBFS, since the distortion will increase slightly when approaching full scale.

This is where this WIKI page comes into play. The idea is to describe information and real-world examples on the usage of the RTX6001 audio analyser. The aim of this page is to help novices get good measurement results out of their RTX6001. While targeted at the RTX6001, the information on this page may also be useful with other audio analysers or soundcards. The set up and configuration of the software is left to the user and is beyond the scope of this WIKI page.

• The Tektronix Cookbook of Standard Audio Tests, Tektronix Company. This classic manual has great descriptions of some of more common audio test procedures.
  
• Sound System Interconnection, Rane Company. This must read describes how (not) to connect audio inputs and outputs, avoid ground loops, and interface balanced and unbalanced, etc.
  
• Understand SINAD, ENOB, SNR, THD, THD+N, and SFDR so You Don't Get Lost in the Noise Floor, Analog Devices. A nice description of the different noise levels involved in digital audio analysis.
  
• FFT Scaling for Noise, Audio Precision. FFT noise floor vs SNR in spectral analysis, including the effects of windowing functions.
  

This drawing shows the general setup used to test the audio performance of a device under test (DUT):



The computer software sends the test signals to the RTX6001 output(s), and receives the DUT response from the RTX6001 input(s). Some tests work by directly connecting the DUT to the RTX6001. Others may require additional buffers (e.g., a power amplifier driving a loudspeaker) or sensors (e.g., a microphone recording the acoustic output of a loudspeaker).

The RTX6001 uses balanced audio inputs and outputs with XLR connectors, with the GND pins connected to the chassis. There are also unbalanced outputs using BNC connectors, with shields connected to the chassis. Unbalanced output is also possible by using XLR-to-RCA converters on the XLR outputs.

The following drawing shows a (simplified) schematic of a typical set up used for testing amplifiers or similar DUTs with single ended inputs:



Note how the safety earth connections of the RTX6001 and the DUT may form a ground loop. Such ground loops may introduce unwanted noise and hum to the measurement. While it is best to avoid such ground loops, it's sometimes not easy (if not impossible).

Here are a few notes that may be useful:

• Do not mess with the mains connections if a DUT does have an earth connection.
  
• A ground loop breaker circuit in the DUT may be beneficial to reduce the noise and hum.
  
• A signal transformer inserted between the RTX6001 output and the DUT input will break the ground loop. While such transformers may be of very high quality, they may affect the test signal and hence the analysis results.
  
• If the RTX6001 output is connected to the DUT using a balanced connection to a DUT with a balanced input, the ground loop is separated from the audio signal. The loop still exists, but runs through the shield of the balanced connection, which is not connected to the audio signal.
  
• Note how the DUT output is connected to the balanced input of the RTX6001. As shown in the drawing, there is no connection of the DUT GND/chassis to the GND/chassis of the RTX6001 to avoid another ground loop running through the connection between the DUT output and the RTX6001 input.
  

The RTX6001 has switches to select discrete voltage levels at the audio outputs and inputs. This is a considerable advantage over the potentiometer controls that are commonly used with conventional USB soundcards, which make it difficult to achieve accurate and repeatable voltage levels. There are three switchable output voltage levels (0.1 V, 1 V, 10 V; 20 dB steps) and seven input levels (100 V, 31.6 V, 10 V, 3.16 V, 1 V, 0.316 V, 0.1 V; 10 dB steps). Note that these voltages (as labelled on the frontplate) are RMS amplitudes corresponding to a sine signal. For any signals that are not shaped like a sine curve (there will be many of these in everyday work!), it is best to convert these levels to absolute peak amplitudes by multiplying them by 1.414 for zero-to-peak values or 2.828 for peak-to-peak values. The following table shows the conversion between the front-panel sine-RMS labels to the peak amplitudes corresponding to 100% full scale (FS) of the DACs (outputs) and ADCs (inputs):

sine-RMS (front plate labels)	zero-to-peak (0% to 100% FS)	peak-to-peak (+/-100% FS)	notes
100 V	141 V	282 V	input only
31.6 V	44.7 V	85.4 V	input only
10 V	14.1 V	28.2 V	input and output
3.16 V	4.47 V	8.95 V	input only
1 V	1.41 V	2.82 V	input and output
316 mV	447 mV	854 mV	input only
100 mV	141 mV	282 mV	input and output

Note: These values refer to the voltages present between the pins 2 and 3 of the XLR connectors. The single ended BNC output gives only half the amplitude of the balanced XLR output.

• Mac OS X: From the software programming point of view, there are different ways of accessing audio devices. Some methods do not allow setting the sample rate and bit depth reliably by the audio analyser software. To get full control of the sample rate and bit depth used, always check those settings in the Audio Midi application.
  
• Windows: [add Windows notes here]
  
• Linux: [add Linux notes here]
  

• ARTA: [put ARTA specific details here.]
  
• Audio Tester: [put Audio Tester specific details here.]
  
• HpW: [put HpW specific details here.]
  
• MATAA: Configure the RTX6001 in the same way as any other USB audio device. Note that recent versions of MATAA include the files for absolute calibration of the test signal levels corresponding to the different level-switch settings, and for both the balanced and single-ended outputs or the RTX6001 (as given in the above table). On Mac OS X, remember to check the settings related to the sample rate and bit depth in the Audio Midi program, which take precedence over the settings used during sound I/O with MATAA.
  
• REW: [put REW specific details here.]
  
• jaaa signal generator and spectrum analyser (Linux): you will need to figure out the ALSA device identifiers for the RTX6001 output and input, and specify these when running jaaa from the command line. The commands "aplay -l" and "arecord -l" are your friends (https://superuser.com/questions/5395...t-which-to-use).
  

The loopback test is the most basic test. It's just a connection "looping" one of the RTX6001 audio outputs to one of the inputs. This is a useful test to understand how the RTX6001 analyser performs. It may be useful to run this kind of test when getting acquainted with the instrument, or to test if the instrument performs correctly.

The photo below shows a loopback setup using an XLR cable to loop the left balanced output the left balanced input. Both the output and the input are set to the 1V ranges.



The plot below shows the FFT spectrum obtained from a loopback test. The software (MATAA) was set up to play a sine signal with a frequency of approximately 1 kHz through the RTX6001 outputs, and do a Fourier (FFT) spectrum analysis of the signal at the input (from the loopback cable). The signal voltage levels were calibrated using the voltage levels corresponding to the voltage ranges set on the RTX6001 front plate (output: 1 V range, input: 3.16 V range). The harmonic distortion products are in the micro-Volt range; the 3rd harmonic (peak at 3 kHz) is about -108 dB relative to the fundamental. The noise floor appears at about 30 nV/rt-Hz (this depends on the length of the FFT, which was 131072 points in this example).



Note that the full dynamic range of the RTX6001 can only be achieved if the computer software is set to use a bit depth of at least 24 bits for both the RTX6001 outputs and inputs. With the typical 16 bit setting ("CD quality"), the signal/noise ratio is limited to about 96 dB. The plot below shows the same loopback measurement as before, but with bit depth set to 16 bit instead of 24 bit. The noise floor is clearly higher. Always make sure the computer software is set up to use 24 bits or more (there may be more than one such setting; check the audio analysis software and the settings of the operating system).

This example shows the analysis of harmonic distortion and noise of an integrated amplifier (an old Pioneer A-445).

The amplifier speaker output was connected to a dummy load resistor as shown in the photo below. The dummy can be configured as a 8Ω, 4Ω, or 2Ω load.



The balanced input of the RTX6001 was connected across the dummy load by using the POS and NEG pins of the XLR connector; the GND (chassis) pin of the XLR input was left unconnected. The RTX6001 output was connected to the line-level RCA input of the amplifier using an XLR-to-RCA converter and an RCA cable (the unbalanced BNC output would work the same).

Note that the XLR-to-RCA converter connects the RCA shield to the GND (chassis) of the RTX6001 and therefore provides a GND connection between the analyser and the amplifier (the shield/ring of the BNC connector would also provide GND). The amplifier has a two-prong mains cable (no safety earth pin), and the speaker output is not connected to GND at the speaker output / dummy load (remember that the GND pin at the RTX6001 input XLR was left unconnected). The RCA cable between the RTX6001 output and the the amplifier input is therefore the only GND connection between the two units, so there should be no ground loop in the test setup.

The software was set up for a 1 kHz sine test signal at the RTX6001 output. The amplitude of the test signal was adjusted to give 2.84 V-RMS at the speaker output (corresponding to 1W into an 8Ω load).

The plot below shows the FFT spectrum of the amplifier output with dummy loads of 8Ω, 4Ω, or 2Ω. With the 8Ω load the harmonic distortion products (peaks at 2 kHz, 3 kHz, 4 kHz, etc.) are about 100 dB or more below the fundamental. With the 4Ω and 2Ω loads, the harmonic distortion is higher. There are also a number of peaks at 50 Hz, 100 Hz, 150 Hz, etc. This might point to pickup of mains power noise in the amplifier or by the measurement setup (although there should be no ground loop in the test setup). Some of these peaks are higher with lower dummy loads. This might point to increased ripple in the linear power supply of the power amplifier when the output current is higher with lower dummy loads.

TO DO: show an example of how to analyse frequency-dependent impedance of a DUT (example: test inductance of a coil, speaker impedance). Describe things that can go wrong, and how to avoid them.

TO DO: different sections showing tests with involving microphones or similar (example: speaker testing). Is it possible to connect a measurement microphone directly to the RTX6001 input without using a dedicated mic-pre. How? Is this a good idea? Describe things that can go wrong, and how to avoid them.