I imagine I am typical of many DIY audio enthusiasts in that I build something and hope that subjectively it sounds good. However, I am also curious about the objective measurements of the equipment. I thought this level of analysis would be beyond my ability and budget, but an interesting thread appeared on DIYAudio.com recently, concerning a cheap method of measuring distortion. Essentially you can discover the harmonic distortion signature and the noise floor of your amplifier.
As forum member XRK971 details in the first post in the thread, all that is needed at the most basic level is a decent internal PC sound card or external USB Audio interface, some free software, three resistors and some wire. Rather than repeat everything that appears in that thread, I thought I would just outline the equipment and components I used and how I set it up.
The software, RoomEQWizard (REW), is primarily used for measuring and analysing room and loudspeaker responses, but it can also be used for measuring distortion in amplifiers. Although the software is free, if you find it useful, please donate to the developer. Install the software and on the opening screen click on ‘preferences’ to configure your external interface once the device is connected to the laptop. The input and output devices should be the sound card. The choice of interface for me was simple as I already own an external soundcard, a Behringer UCA202, which can be bought for under £20. A review including measurements of this interface by the respected engineer NwAvGuy shows that for the price, this is a reasonable piece of equipment.

At this point, with only the interface connected to the PC (and possibly just running off the battery if you are using a laptop), you can check the internal noise of your device. From the main REW screen, click on the RTA icon to bring up the Real Time Analysis window. This window has it’s own options on the right hand side and the image below shows how I set mine up, along with the display of the UCA202 internal noise.


As you can see from the above images, the UCA202 reports self-noise, and thus the noise floor is at about -130dBFS with a 50Hz spike and harmonics when connected to the mains supply. You can perform a ‘loopback’ test to measure the harmonic distortion of your own audio interface by connecting an RCA cable between the output and input and using REW to generate a 1kHz tone (use the generator icon on the main window). It is worth calibrating your device in the soundcard preferences panel before commencing testing. As you can see from the image below, the device isn’t bad, with THD at 0.0045% and THD+N at 0.020%. These figures can be improved upon if required by using a better interface (Focusrite Solo has been mentioned as good and relatively inexpensive) and also an external low noise tone generator.

The amp being measured needs to drive a load, so a dummy load simulating your speaker load needs to be made. All that is required is a resistor that has the same impedance as your amplifier and is rated to handle the amp’s power output. I chose a 8Ω 50W power resistor from RS Components. You may wish to heatsink this but I bolted it directly to a metal case. The dummy load can connect directly to the sound card but it is prudent to attenuate the signal before it reaches the audio interface. This can be achieved using a voltage divider and it was recommended that a 10:1 voltage drop would be suitable. I used 22K and 2.2K resistors as these were at hand. Make sure you get the resistors the right way round (see image below) or the ratio will be out. In true (mis) Adventures in HiFi tradition, I got them the wrong way round to begin with.
You can just attach everything together temporarily but I made a little project box with permanent connectors for the amp inputs and the signal output as I had these already. The enclosure came from the Maplins closing down sale and cost a couple of quid.

The various components now need to be connected together, in effect forming a loop. Perhaps the easiest way to demonstrate this is by using a block diagram of my setup. In addition, a multimeter set to AC Volts should be placed across the dummy load terminals so that the output of the amplifier can be measured



Once everything is connected together, turn on the amplifier and generate a 1kHz tone through REW. Adjust the level of the sine wave using a combination of the amplifier volume pot (if available) and the controls in REW until the voltmeter reads 2.83V which equates to 1W into 8Ω. An example of the preferences panel for REW and the UCA202 is shown below. This generated the plots for an old Cambridge Audio A1 amplifier that I have, which are also shown below.

The image below shows the amplifier connected and powered on but with no signal. As you can see, there is considerable noise being generated at 50Hz and at intervals thereof. At this stage I do not know if this is the environment or the amplifier. I will be redoing these tests in the future and will attempt to find the cause.


The next image shows the result of the generated 1kHz sine wave. This shows a dominant 2nd order harmonic but also odd and higher order harmonics. The inset box in the top left shows the THD, which at 0.012% isn’t particularly low by modern standards.
I will be running some tests on my DIY amps in the near future, at this stage I am just getting to grips with measuring, and more importantly, understanding what the measurements mean.
EDIT: There is a more recent post on this subject, with extra information, here.
Links to some interesting articles on harmonic distortion:
Understanding FFT diagrams for beginners
The sound of distortion (powerpoint file).
Audio distortion and feedback by Nelson Pass
Discussion on Gearslutz forum
Cool post, I have been trying something more or less similar!
A couple of observations:
– I have grown skeptical about using the same interface for generation and acquisition. This alters the impedance to ground of the amplifer outputs and I guess a BTL amplifier may not like it.
– could compensating the attenuator be beneficial? I haven’t do the math actually; maybe with a low impedance source this is not an issue.
Hi, thanks for the comment.
I have just got a separate generator as a kit and built another attenuator but haven’t had the time to play around with them yet. I’ll be posting a page on that very soon.
It will be interesting to compare the results of using internal and external signal generation.
Hi,
Could the same approach be used for measuring frequency response of an amplifier?
Does the voltage divider add any influence to the impedance seen by the amplifier – I guess not?
Finally could the same thing be repeated with an operation speaker that has varying impedance and see how the amplifier drives the real load?
Hi Richard,
I’m afraid I will have to direct you to the DIYAudio thread. There you’ll find people who can provide much more detailed advice than I could hope to give.
Great post about speaker dummy load, Thanks! I was encouraged to make something similar.
Regarding, the noise being generated at 50Hz and multiples of it, I’m sure that those can be eliminated , if you power both your PC and the audio interface with DC battery independently. I use VAGABOND MINI for my audio interface (Focusrite Scarlet) and my laptop PC is driven by it’s own battery. Only -120dB level 60Hz fundamental is remained.
Glad you found it useful.
You are, of course, correct regarding the noise being mains frequency, but from what I remember, it was present when the laptop was powered by battery too.
I did move onto a Focusrite Scarlet too and went a little more in depth in a follow-up post and there are a couple of measurements of headphone amplifiers.
I like your post!
RE: 50Hz noise
Try disconnecting the 8 ohm return from the 2.2K ground inside the box. Let the Vout/24K ohm current flow through the Behringer (116uA at 2.83V). As drawn, there could be a ground loop from the power amp and wiring to the power amp through the Behringer. 116uA through 20 milliohm (guess) in the Behringer would be 2uV or 120dB down from 2.83V.
After the change, the RCA ground should measure 24K to the speaker dummy load black with the Power Amp disconnected.
Thank you.
I’m pretty sure I made an adjustment to the dummy load following a suggestion on DIYAudio, I obviously forgot to document this alteration in my follow-up post. I went on to the Scarlett 2i2 and don’t have the problem anymore.
Thanks for the advice though, much appreciated.
Cheers