DIY IMD Analyzer

Status
Not open for further replies.
Jump to Latest
S

stellavox

Member
Joined 2007
#1
Back in 2004, AudioXpress published a DIY article by Richard Crawford on building an IMD Analyzer. As there has been some discussion here on this forum on building test equipment I thought you might like to know of it. The kindly folks at the magazine agreed to let it be published and my friend Mike Barney graciously agreed to put it up on his website so you could link to it:

https://docs.google.com/viewer?a=v&...pbnxtcGJjYXNmaWxlc3xneDo1M2M0YmY2N2YyNjAwOGJl

I liked the simple design so I built it and it works very nicely. For the display, I just used used a 0-100 microamp DC meter (with the appropriate resistor in series). Whole thing fit in a 7" X &" X 2" Bud box I had lying around. Crawford did make PC boards available for it but I doubt he even remembers the project now. Total cost plus the board was around $100. Can post an internal/external pic if anyone is interested

As I remember, there were a few "glitches" in the article like between the parts labeling and the material list but nothing major

Enjoy,

Charles
 
Last edited:
#3
much less hardware is needed if you start with a good soundcard

you can generate the two tone test signals in one channel each so that they aren't limited by the DAC or post filter/buffer IMD performance, then sum them

a filter to reject the test test frequencies and pass/amplify the IMD products is fairly relaxed in requirements, perfect rejection isn't required, only knocking down the test signal level sufficiently that the soundcard ADC circuitry's own IMD is invisible

multichannel soundcard or cheap DVD player could provide multi-tone signals

FFT math in many free software packages let you calculate any desired function of the data, the residual test signal is is eaily rejected in SW

I have used these techniques to get ~160 dB resolution for IMD - NP0 caps in my filters clearly showed ~ -110 dB IMD distortion which went away when relpaced with polystyrene

in principle the FFT data contains the info to tease out the separate "AM" and "FM" IMD components - so you could also directly measure "PIM" with the right math
 
Last edited:
#6
dchisholm said:
Are you aware of any follow-on work on triple-tone testing? Did any manufacturers standardize the method and adopt it for the published specs on their commercial gear?

Dale
Click to expand...

Hi Dale,

No I'm not aware of any significant use of the test by manufacturers, although I think one or two other people have published things where they used the frequency set that I defined (9kHz, 10.05kHz, 20kHz).

The MIM test was a good, inexpensively-implementable test back when spectrum analyzers were very expensive. SAs were needed to do a proper 19kHz + 20kHz CCIF IM test to capture the odd-order distortion products that lie at 1 kHz intervals away from the main tone-pair. Simple filtering of the 19kHz and 20kHz tones does not work adequately because of the unavoidable attenuation it introduces at the nearby distortion product frequencies.

Nowadays, things are much different with the availability of high-performance PC soundcards and software to do FFT. As jcx pointed out, on the generator side two channels from the soundcard producing the two tones and then adding them together externally in a low-distortion op amp is a good option. The proviso here is that there is no inadvertant intermodulation between the tones in the soundcard. If that is a problem, external analog oscillators can be used with a low-distortion summer to create an exceptionally good twin-tone source.

On the side receiving the signal from the DUT, the IM capability of the soundcard A2D comes into play. Some are very, very good, with a lot of dynamic range. For critical applications, however, it is desirable to reduce the incoming amplitudes of the main tones to make more of the dynamic range of the soundcard A2D useable. This must be done without changing the distortion products themselves. This is one reason why I built my distortion magnifier, described in Linear Audio Vol. 0 and available as a kit. It can reduce the main tones by 20 or 40dB, buying you 20 to 40 more dB of range on your spectrum analyzer. This is especially important when using an analog spectrum analyzer like the HP3580A, which advertizes only 80 dB dynamic range.

Finally, it is important to note that MIM, CCIF and others do not have the same readout sensitivity as THD-20. This is partly because their product of fast slew rate intervals and duty cycle of those high dv/dt intervals is usually smaller. There is correleation between the tests (see my MIM paper), but the numbers are different. For example, an amplifier with 0.01% THD-20 might read 0.002% on a CCIF or MIM test. In the article you saw where I think a sensitivity of 0.0001% was asserted, that is an impressive number, but is not reflective of an analyzer capability in the range of 0.0001% THD-20. It might only be roughly comparable to a THD analyzer that can show 0.001%.

Cheers,
Bob
 
#9
Bob Cordell said:
See also my 1981 JAES paper
Click to expand...

I referred to something of that time period to one of the folks in our choir, and she said "that was before the earth started cooling".

Then again, my XYL really f'd up a speaker at a major cancer research society meeting by saying that the stuff in his paper was published in 1973. I guess that's before the internet era in which all knowledge expanded forth unclothed from the mind of Zeus.
 
#10
Speedskater said:
Where can I find more information on "THD-20" ?
Click to expand...

Hi Speedskater,

THD-20 stands for 20 kHz Total Harmonic distortion. You run a 20 kHz sinewave through the amplifier under test and look at the output after you notch out the 20 kHz fundamental with a very sharp and deep filter. What's left over is the harmonics due to nonlinearities in the amplifier. Strictly speaking, THD-20 only measures the harmonics. In practice, it is usually THD+N that is being measured, meaning that noise in the amplifier is contributing to the reading.

Cheers,
Bob
 
#11
jackinnj said:
I referred to something of that time period to one of the folks in our choir, and she said "that was before the earth started cooling".

Then again, my XYL really f'd up a speaker at a major cancer research society meeting by saying that the stuff in his paper was published in 1973. I guess that's before the internet era in which all knowledge expanded forth unclothed from the mind of Zeus.
Click to expand...

That's funny. Sooner or later we all have to admit to being old farts - although the term is relative. I think I built my first tube amp in 1962 - before hand-held electronic calculators. The first transistors I worked with were the Germanium 2N107 and Raytheon CK722. The first ICs I worked with were RTL OR gates. Just this month I turned 64. Time flies when you are having fun - or behind schedule.

Cheers,
Bob
 
Status
Not open for further replies.
Top Bottom