2. amps-preamps
  3. 189095-189286

A listening test of two power amps

Hello, 

It's my first post here. I've been using two power amp setups for my main stereo and I've been curious to see if I can really discern any acoustic difference between the two. One setup involves a bi-wired high-powered stereo power amp and the other uses a pair of identical lower-powered amps with which the speakers, a pair of Tannoy System 12 DMT II monitors, are vertically bi-amped.

I decided to devise a listening test involving a mono acoustic recording made with a valve-condenser mic positioned at my usual listening position. I've used a relatively simple method to ensure that the recordings are level-matched. I've chosen a mono recording method since my goal is, principally, to evaluate the "tone" of the two recordings. I've been inspired to do this test after reading W. A. James' eBook "High end audio on a budget". The aim of the listening test is to try and discern which power amp setup provides the most realistic rendering of acoustic instruments. I thought that a mono recording might help the listener concentrate on the tone. After listening, I think it does. It's less distracting, especially on piano, where stereo or other multi-mic recording setups tend to splay out the notes across the stereo field.

I made two recordings for the test and will place links below so that the audio can be downloaded. I won't at this point give the make and model of the power amps involved, but this is the method used:

Method

1. I created an audio file with white noise at -10dB RMS and put the file on a Logitech Media Server so that I could play it on my stereo using a Raspberry Pi 3 with Audio Injector Pro card and RCA interface (192kHz 24bit DAC).

2. I then put on an LP on a Pro-ject 1.2 and set the volume to my usual listening level on a Quad 34 preamp. Following this, I then played the white noise and used a decibel meter, positioned next to the mic, to measure the level. It measured 67.3 dB.

3. Still playing the noise, I set the record level on a portable Tascam digital recorder arbitrarily to somewhere above -15dB. The microphone used was a large diaphragm valve condenser mic. The Tascam was set to record at 192kHz 24bit.

4. I then recorded the first track of the LP on the Tascam.

5. After that, I wired up the other amp configuration. I played the white noise and adjusted the volume of the preamp such that the decibel meter again measured 67.3dB at the position of the mic. The volume control on the Quad 34 is stepped, so I was lucky it matched!

6. I then recorded the same track on the LP as before, leaving the Tascam record levels unchanged.

7. I tidied the two recordings in Ardour (trimming start and finish only) and exported each as a 192kHz 24bit Flac file. I did not adjust the gain on either recording.

8. I listened to the recordings on the computer with a pair of AKG K501 headphones and Focusrite Scarlett interface.

Results

At first, I could distinguish a marked difference between the two. But now, I'm uncertain of the first qualitative difference that I'd noticed but I have noticed other more subtle differences (for the moment anyway). And that's why I'm here!

It would be wonderful if some people here could listen to the recordings and say which recording produces the most realistic rendering of the three instruments therein, and why. The instruments being piano, drums and string bass.

I've given the two files nondescript names: e.flac and t.flac. If anyone needs a different format or for me to down-sample, please let me know.

Finally, here are the files:

https://escuta.org/webtmp/e.flac

https://escuta.org/webtmp/t.flac

Cheers,

128x128surdo

Showing 5 responses by erik_squires

erik_squires
15,772 posts
 

OP:

It’s your experiment, but as someone who worked quite a bit with microphones and speakers and electronics, the advice I gave you is solid.

Measuring the SPL of quasi random white or pink noise is hard to get precisely accurate, it’s even hard with a multi-meter, which is why level matching should be done with a multimeter and single signal.  60 Hz is a frequency which is convenient as any $10 multimeter will read it, and all are more relatively accurate IMHO than an audio meter, given they are not subject to random room noise or the periodic variability of white/pink noise sources. Level matched experiments comparing DAC’s for instance should be level matched this way and with a decent meter will give an accuracy well better than 0.1 dB.

Now, once we agree to use a multi-meter, and a standard sine wave, whether they measure RMS voltage accurately, peak or peak to peak is irrelevant so long as they are absolutely RELATIVELY accurate. That is, you can trust that 3V now will be 3V in an hour after you switch amps.

Also, with a 60 Hz sine wave signal, again, any $10 meter will measure the voltage pretty accurately and almost perfectly relatively accurately. A nice meter will give you more zeroes.

erik_squires
15,772 posts
 

BTW, I don't believe in static tests very much.

That is, as a builder I use static tests such as frequency sweeps, voltage, etc. to guide my work, but I also know that these types of static tests may not cover all possible dynamic situations.  By this I mean frequency response, output, distortion.

We use these static tests and basic models because they are damn convenient and accurate for what they are, but the testing I'd want to see done, which I never see done, is to compare the output at the speaker or across the speaker cable with actual music.  With demanding speakers or loads we may very well see and measure dynamic behavior which is more complicated and unexpected based on the traditional static measurements say taken by Stereophile. 

erik_squires
15,772 posts
 

OP:  Should be about the same.  Another test is to connect your line inputs to either side of the HOT cable.  This will record any and all variations caused by the cable alone. :)

erik_squires
15,772 posts
 

OP:

Yes, use the speaker terminals to create a line-level input, which you feed to your PC's inputs.  It's a little dangerous because your amp CAN fry the inputs.

2.83V is ~ 1 Watt if fed to 8 Ohm speakers. 

For your listening tests, it doesn't matter what reference level you set, as the multimeter will take several hundred volts, but whatever you pick, use the meter to get you to near millivolt accuracy in terms of level matching.

If you attempt to record off the speaker terminals though you have to be sure you keep your voltages low, probably 2V or less at all times.

erik_squires
15,772 posts
 

OP:

I see where you are going with this and I have a couple of suggestions:

 

1. Match level by multimeter instead of SPL. Get a 60 Hz signal and check the output on either speaker. You can get really accurate this way.

2. Consider recording AT the speaker connections. You have to be careful to keep the peak voltages ~ 1V but this would let you get a noise-free recording of what the amp is actually offering the speakers. Then you can do some very interesting analysis such as output by frequency, compression by frequency, etc.

This option also makes it easier to listen to any differences directly as opposed to via another microphone, though the max signal to noise will be different.

 

More to discover