Distortion - which is more significant - IMD or THD?

[w5jag]

Good Morning,

What is the best predictor of sound quality, IMD or THD?

I want to do some basic measurements for characterizing audio amps, high and low level. The available space in my radio room is slight, at best, for additional test equipment.

I lean toward IMD as the most significant.

Win W5JAG

[tubelab.com]

It is possible to make an amp with a THD of 1% at low power and 5 to 10% at near max output that sounds very good. It is also possible to make an amp that sounds rather sterile and boring with .002% distortion.

I don't think it is possible to make an amp with 5% IMD that sounds anything but ugly with complex music. The best tool for measuring and observing both effects is an ols PC with a 24/96 sound card and some software.

There are other factors that contribute to sound quality that are not easilly measured. Time smear and PIM (phase intermodulation distortion) come to mind. Both can be caused by devices that change characteristics with the applied signal level. All semiconductors exhibit voltage varying capacitance effects. This is known. Vacuum tubes do too. This is a much lesser effect but as the space charge changes with bias voltage, the interelectrode capacitances are modulated.

[jcx]

they are NOT "either/or" quantities

harmonic distortion is evidence of nonlinearity which will give IMD with 2 or more frequencies

it is impossible to build amps with harmonic distortion that don't produce IMD when fed a complex signal
it is only "very hard" to build circuits with low harmonic distortion and high IMD, impractical without intent on the designers part

even a “benign” 2nd harmonic distortion only tube or fet circuit will give sum and difference frequency IMD with multiple tones, 1 KHz diff should be “clearly audible” with CCIF 19 kHz + 20 kHz test tones

[jcx]

for testing at audio frequencies better PC "prosumer" add in sound cards are quite good today

I wouldn't buy hardware low distortion audio osc, audio spectrum analyzers today unless collecting test equipement was your actual hobby

[w5jag]

Well, I already have some HP oscillators, and some scopes, but I was asking because hamfest season is coming up so I wanted to keep my eyes out for test equipment.

Do the sound card solutions have the tone generators built in?

Win W5JAG

[jcx]

I certainly recommend having basic benchtop function generator, audio up to 10 MHz for gross debugging on the 'scope - even 1% distortion isn't a problem


the soundcard can give much higher resolution, lower distortion, but requires more effort than purpose designed hardware

various "PC oscilloscope" software exists, spectrum analysis, test tone generation, loudspeaker designers have extensive "real time analyzer" SW that runs the soundcard

http://audio.rightmark.org/products/rmaa.shtml RMAA free version is a standard - add RMAA to a search of any make soundcard for loopback test results

diy audio forum google search can be your friend http://www.diyaudio.com/forums/gtsea...ware&sa=Search


Test & Measurement interface for Soundcard

[abraxalito]

IMD I'd say is the better predictor of sound quality - but IMD with out of band (e.g. RF and ultrasonic signals). Check out Deane Jensen's work on 'Spectral contamination'.

[tubelab.com]

Do the sound card solutions have the tone generators built in?

Many do, but one common test of audio amps involves a 10KHz square wave. Most sound cards will not do a good job with this unless it has a 192 KHz sampling rate. A basic function generator that can produce a 10 KHz square wave is a good and relatively cheap thing to look for at a hamfest.

I have an old HP8903A audio analyzer. It measures THD, audio level / power and contains a very low distortion generator. The level meter can do absolute and relative measurements in linear or DB, so frequency response measurements are easy. Of course all of this can be done with a PC too. I usually use the generator in the 8903A to drive an amplifier under test while the output of the amp goes to both and the scope. I can look at power or THD on the 8903A while observing the output spectrum on the FFT in the PC. Makes tweaking a bit easier.

I got my 8903A at the now defunct Miami hamfest for $75 about 10 years ago. It was broken and I fixed it.

Old HP 331A distortion analyzers are becoming cheap. One will measure THD, and two can be connected together to measure IMD. Most budget audio oscillators have too much distortion to be useful. An exception is the HP204C. I have 2 of them. They aren't too common yet on the surplus market. You need two good oscillators to measure IMD. I can run two instances of the audio generator program on my PC and get two tones out of the sound card.

Hamfests???? I will probably go to Dayton and Orlando. Others are maybe's.

[w5jag]

I used to have a function generator, but I haven't seen it for a few years - might be in my warehouse of stuff.

I think a dedicated hardware audio analyzer or two is really what I am looking for, but may try the sound card / software route in the interim.

I already have a couple of decent HP audio oscillators, the model number escapes me, but they look like an overgrown lunch pail and have a pair of 6CW5's for output. I also have a depot reconditioned mil surplus audio level / power meter that I got from the same guy that had the depot overhauled TV-7D's ( for $50 !) about fifteen years ago.

The small hamfests start around here in about a month. I have my schedule blocked for Dayton and Ham Com, and should make both this year.

Win W5JAG

[w5jag]

Came across a Sound Technologies ST 1700A this weekend. It looked good, but I didn't buy it. The seller is Local to me, so I didn't see any urgency.

It looked good, asking price was $250, doesn't appear to be able to do IMD, just THD. Anyone have any experience with these?

Win W5JAG