IMD Testing - What am I seeing?

Hello, I’m back again, trying to understand the practical application of the QA403 (ver. 1.193) “AMP IMD (ITU and SMPTE)” automated tests.

Up to this point I have relied on THD, THD+N, and Frequency Response testing to characterize the performance of the amps, preamps and receivers I own or have been working on. As a fan of audio gear I have seen many many references to IM Distortion over the years and I have at least a decent understanding of the basic theory behind ITU and SMPTE, but I’m looking for some help with interpreting the actual results.

I’ve just finished up restoring a Marantz 2275 and I am really pleased with the sound, as well as with the THD and FreqResp test performance. So I figured I’d get off my butt and learn about running the IMD tests. Below are the ITU and the SMPTE results, running the 2275 at Max Gain and into 8 ohm dummy loads with -6dB external attenuation. I chose the -33 to -27 dBV range because that corresponds to roughly 1W to 4W per channel output when running 1kHz test signals in this setup.

And also just to muddy the waters, the old Marantz literature specifies the 2275 with “IM Distortion (S.M.P.T.E. method) = 0.25%” and I so far haven’t found an explanation of how that % would relate to dBC.

So again, if anyone cares to comment on these results - do they look good, bad, problematic, etc. - that would be greatly appreciated!! Thanks - Steve

Just finished doing an IMD measurement on an amp this evening and can relate. There is a paper on the QA wikipage called “QA40x & IMD measurements”. I have not had any luck with measuring it with the QA40x’s automated routine like what you show. Mostly I use the 19&20khz tones, and adjust them equally to give me 5watts/8ohms. I set the frequency range from 20hz-50khz and use dBr. I used the worst tones, usually at 1 & 17/18
khz, and the formula shown in the wikipage, which will give you a %. I have also seen the tones called out to be 70hz and 7khz, and use the worst tone(s) generated as well. The % is usually very small- the one I measured tonight was 0.095%. Here is the plot from that one:

hope this helps a little

1 Like

I’ve been finding that perceived quality seems to be tied more closely to IMD tests than THD. Nice to have this automated.

Thanks Scott, much appreciated. I will mess around with running the test manually, along the lines of what you showed. I’m not seeing a formula for a percentage result on that wiki page - is this is the correct page?

Also, ideally I’d like to be able to use the automated versions, but I still don’t have any idea what “good” or “bad” looks like, at least in terms of dBC. I will try running those tests on a late-70’s integrated amp today just to start getting some kind of qualitative feel for what the results should look like.

Another newbie question - do you run IMD tests on preamps as well? I will assume so.

Thanks again

The IMD page in your link is not the one that I downloaded. Here is a snippet of that document that shows the equation:

There is also CEA standard CEA-490-A-R2008:

I actually built a spreadsheet so I can export the trace data (I just use the left channel since both are going to be pretty much the same if the amp is working ok), and it will convert the dBV to volts rms and calculate the IM distortion as a percent- they have been very, very small numbers on the few I have measured that way. I don’t think I have measured IMD on a preamp yet, but it would be done the same way though the signal levels are going to be much smaller than an amp putting out 5w.

1 Like

Wouldn’t this “converter” be useful?

1 Like

@vidax: Yes, that converter seems useful, thank you! Or at worst the theory described on that page seems to show the way. I will need a little time to digest what @VAR describes doing, assuming I can, and I suspect that will agree with that page.

@VAR: Thank you for the additional details and source references, and wow, the spreadsheet sounds awesome. I will go through the exercise of building my own, since that is the only way this hard head ever seems to I really learn something. Also, I did run some manual IMD test setups on the preamp of the 2275 and the plots look in line with yours. Seems to make sense that the IMD comes from the preamp as much as the power amp. And IMD is working with ratios of distortion products to primary signals, so the absolute signal levels are basically irrelevant as long as they’re above the noise floor/resolution of the measuring device.

I would be more than happy to send you the spreadsheet- if you email me at :

Thanks for the offer Scott, you’re very generous. I may take you up on it depending on how the exercise goes for me. -Steve