Harmonic distortion measurements

[troplin]

I totally understand that, Doc.

I was talking about my current setup without resistor/speaker. I don't understand why there is such high (odd harmonics) distortion in even an basically unloaded setup and I don't expect the distortion to be lower with more load. But we will see as soon as I have that resistor installed.

Strange enough, the distortion actually seems even higher at lower volumes (EDIT: which might just disprove what I've written above, I don't know).
Attached two FFT pictures, one at 1/2 volume (same as the measurements above) and one at 1/4 volume.
Same setup as above, at 200 Hz.

[Doc B.]

I think the problem here is you have no reliable point of reference yet. Do what PJ says, use the scope function to observe what is really happening to the wave form at the output of the SEX amp. Note the rms voltage at the input and the at output of the SEX amp, where a sine wave coming from the speaker output loaded into 8 ohms just starts to flatten. That should be about 5-10% THD. Maintain that same signal level and run your distortion plot. Verify that you measure somewhere roughly around 5-10% THD. If that is OK then back down to about 1/10 the signal level and do it again and compare the plots. That way you can start to see what is happening as the THD due to signal overload goes up. There's more than one possible cause for distortion and if your baseline measurement at a low input signal has distortion that seems excessive there may be something else you need to address in the test setup or the amp construction before you can get a useful measurement.

[troplin]

I think the problem here is you have no reliable point of reference yet. Do what PJ says, use the scope function to observe what is really happening to the wave form at the output of the SEX amp. Note the rms voltage at the input and the at output of the SEX amp, where a sine wave coming from the speaker output loaded into 8 ohms just starts to flatten. That should be about 5-10% THD. Maintain that same signal level and run your distortion plot. Verify that you measure somewhere roughly around 5-10% THD. If that is OK then back down to about 1/10 the signal level and do it again and compare the plots. That way you can start to see what is happening as the THD due to signal overload goes up.

Ok, I will try that.
I just fear that the input of the sound card will start clipping sooner, so I probably have to build a voltage divider from a few resistors or even directly use a pot.

There's more than one possible cause for distortion and if your baseline measurement at a low input signal has distortion that seems excessive there may be something else you need to address in the test setup or the amp construction before you can get a useful measurement.

Yes, that's what I'm actually interested in. I'm currently not as happy with my setup (S.E.X + Orcas) as I would have expected, but I'm not experienced enough to really tell what exactly I don't like about it. It could be an error in the construction, the room/placement, the missing bass (no sub), my personal taste or any combination of those. That's what I want to find out.
I hope it's something that is easy to fix, i.e. something in the amp.

[troplin]

Ok, there's definitely something severely wrong with the amp.
I've just listened to "Sand" by Nancy Sinatra and Lee Hazlewood and this strange thing happened:
I can barely hear his voice, while her voice is normally audible. So I switched to the internal speakers of the iMac and there, both voices are audible just fine. I already mentioned that I'm not quite satisfied with the sound, but I didn't hear such an extreme effect until now.

Now I'm totally unsure if the amp always sounded like that or if it got worse over time.

But I guess it means that I'm switching to the internal speakers now and don't use the S.E.X. until it's fixed. I hope I didn't cause any permanent damage by listening to it that often.

Besides the measurements that you already suggested, where should I start?
My voltages were all a bit high, but still in the tolerance.

[fullheadofnothing]

Are your speakers in phase?

Have you tried other speakers?

Have you listened to headphones on the SEX?

[Paul Birkeland]

I think I get this, but in my measurements above there's barely any load. The line-in has an input impedance of 27kΩ and a maximum input of 2dBV (~1.26Vrms). That means the means the maximum power taken from the amp without clipping in the ADC would be about 0.06 mW if my computations are correct.
So in that setup there should be no clipping, right?

You'll need to purchase an 8 Ohm L-pad to go with your test rig, or make up a reasonably high impedance 10:1 probe to give yourself some headroom.

[troplin]

Are your speakers in phase?

Yes

Have you tried other speakers?

No, I don't have any.

Have you listened to headphones on the SEX?

I've got no serious headphones but I just tried some old iPhone earbuds and it sounds absolutely horrible. Completely distorted. Much worse than with the speakers.

[troplin]

Ok, there we go.
I've made an L-pad with a 10Ω/4W pot and performed the measurements you suggested. I raised the volume until I got visible clipping on the oscilloscope.

Input voltage (after volume pot): 0.958VAC
Output voltage: 5.36VAC

Measurements are attached.
Is this a reasonable point of reference?

One question that was not answered yet and that makes me a bit nervous:
Is is normal that I can hear the sweep sound of the measurement from the amplifier itself?

[Doc B.]

Yes, that is starting to look like it should. You can see that the troughs of the waves are squashing a bit on the scope trace which implies you are at about 10%THD, and that correlates to your graph. 2nd harmonic dominates, followed by third, which also makes sense. And yes, your transformer will sing.

[troplin]

Ok, next measurement, 1/10 of the volume:

Input voltage: 0.095VAC
Output voltage: 0.58VAC

And two more measurements with input voltage 0.030VAC and 0.010VAC.
Those are starting to look like my initial measurements, i.e. odd harmonics are starting to dominate.

Is that still expected?

[Doc B.]

The higher harmonic stuff is looking the same no matter what the volume setting is in these lower output level sweeps. They may not necessarily be generated in the amp, but could be inherent in the test setup. You would need to start playing with the setup to see if you can isolate where the distortion is coming from.

[troplin]

The higher harmonic stuff is looking the same no matter what the volume setting is in these lower output level sweeps. They may not necessarily be generated in the amp, but could be inherent in the test setup. You would need to start playing with the setup to see if you can isolate where the distortion is coming from.

I doubt that it's something inherent in the test setup because if I take just the amplifier out of the loop, the picture looks completely different, i.e. almost no distortion (see my earlier post http://bottlehead.com/smf/index.php?topic=10046.msg93097#msg93097).
I don't know what else I could change to find out the source of the distortion.

As I've written earlier, if I use my old iPhone earbuds (TRRS with mic), some tracks sound awfully distorted. I'm not calling myself audiophile, but it's so obvious that even I can hear it clearly. Sure, the phones are bad, but there's no such distortion if I plug them directly into the DAC (which has an internal headphone amp).

Curiously, the effect is not as noticeable with a different pair of (even older) earbuds.

I'd be thankful for any pointer where to search next.

[Doc B.]

Is the distortion spectrum the same on both channels?

[troplin]

My first measurements (without load) were about the same on both channels. The latest measurements with 10Ohm load are only on the right channel, but since the picture is similar to the first measurement I'd expect that they are also the same on both channels.
But I can repeat them on the left side if you think that this will help.

The good news ist, that I've "solved" the massive distortion with the iPhone earbuds. It's not a problem with the amplifier but with the jack. TRRS jack does not seem to match well with the 6.3/3.5mm adapter. If I pull out the jack a little bit and find the right spot, it sounds ok. I can only guess, but I think that there's no proper ground connection and it connects to the mic instead.

But that also means that I'm back to just numbers and measurements. They still seem to be quite odd (haha the pun) to me, but I've no more actual audible distortions to back this up. I just haven't enough experience to hear such things.

What do you think, would a measurement between the amp stages be meaningful?

OTOH you say that those numbers are normal and to be expected I might also just give up searching and accept it how it is.