HP 3580A Spectrum Analyzer and Harmonic Distortion on the McIntosh MC275.MOV 

Yes, tubes create odd and even order harmonics. I don't understand the idea that that all even order harmonics are good and all odd order harmonics are bad. It's just nonsense. If one compared high-end solid state to high end-tube amplifiers, the solid state will win every time - but - we humans and our ears like the sound and warmth (and distortion) of the vacuum tube amplifier. It is very subjective. I listen to both types and like both of them - but they are slightly different - to me.

I did not know that about the summing of harmonics. It will complicate any video trying to make comparisons. Good to hear you have your HP1740A running. A perfect mate for the 3580A. I am sure you will love your AA501 and SG505. You mentioned attaching a heat sink to one of your instruments IC's. I did a similar thing by attaching a heat sink to one of the IC's in a Tek 2465 scope I had. Can't remember the U# - horizontal output, I think. Ultimately went bad anyway. Always good talking with you.

In hifi stereo equipment, the clear objective is to have the amplifier create no harmonics as all - push pull amplifiers suppress the second harmonic by their very design and this is desirable. On the other hand, musical creation amplifiers along with everything else in the creation process, intentionally create and enhance harmonics along with clipping of the signal for "overdrive" sound. Odd and even order harmonics are an every day fact of life and both are desirable (or not) in some mixture.

I've been using an old FFT spectrum analyzer to look at audio amps. Its advantageous as it has a 120 dB log function on the vertical scale. Thus you don't have to drive the amplifier in to clipping and have the amplitude of the center frequency spectrum go way off scale to observe very low level haromonics. It can also measure the amplitude of each harmonic down from the peak of the CF eliminating the need for a distortion meter as the percentage at any power level can be easily calculated.

@MarkPalmer1000 The HP 3580A actually does have a log amplitude mode, not used in this video, but can show everything down into the noise floor. For instance, the SG502 oscillator is specked at .035% THD, and my HP 3580A shows 2nd harmonic around -74dB, right where it should be. I believe the purpose of this video is to illustrate how tube amplifier distortion looks under overdriven conditions. A nice comparison with be transistor vs tube, showing the differences.

@essoterric THD=V1^2+V2^2+V3^2 etc / V1^2. Reminds me of the resistance calculation for multiple parallel resistors. My 1740 developed a weird problem last night that would affect display intensity and make it bounce. The +5 supply was failing. It was the 723 regulator, somehow common mode conducting and feeding back current from the +9 raw back into the +15 via the 723 internals which were getting hot. 85C on my Tektronix temp probe. Replaced U2 and Q5 and all is well. Have some mods for 3580A.

@essoterric The sum of the harmonics is, if I recall correctly, similar to the calculation for the sum of parallel resistors. Where it gets tricky is weighting the various harmonic intensities into the final THD %. UPS has my AA501 and SG505 scheduled for end of day delivery today, so I'll be playing with them tonight. I just fixed my HP 1740A... U2 and Q5 in the LVPS were acting up. Did a calibration and balanced the FETs in the vertical amps. Runs nice now. Will follow up when the gear arriv

@essoterric It takes a little getting used to, but Log is really a better way to see everything that's going on with signals at a glance. It lets you see things down to about -90dB, and, assuming there's no ground loops and the mixer is balanced out nicely, the display should not contain any anomalies. Linear and Log 1dB have their uses, but I find Log 10dB to be the most useful for testing amplifiers. Recently, I found harmonics in a Phase Linear D500 that were traced to ultrasonic oscillation.

@essoterric Two things I shifted away from as I got used to the 3580A: using the center freq, and using Linear amplitude. Nowadays I just set zero, or sometimes 200cps to chop off the zero response peak, and use Start. Then I put the Freq/span to 2Kc/div for most audio measurements. That gives me 20Kc bandwidth at a glance, good for seeing the overall spectrum. Sometimes I set Sweep Mode to Log when I need to see the entire range. SG502A outperforms its specs when output pot maxed out only.

It would be nice to have a video on using a spectrum analyzer to MEASURE THD. Also, a video on using the S. A. to detect TIMD. If you feed a square wave into an amp (proper termination, please), you know, from the Fourier series, what harmonics are supposed to be there, and their amplitudes. Anything else is garbage, and probably TIMD. And you could do the test with op-amps, of various slew-rates. A comparison of those whiz-bang DSOs, with FFT, and a REAL spectrum analyzer, would be nice. I have a good scope (Tek 475A), but have considered getting a DSO, because of the FFT feature. I've used a REAL spectrum analyzer, and from what I can determine from the reviews of the DSOs, I'm not impressed. Actually, most of the reviewers of those scopes, don't seem to know how to evaluate the quality of a freq. domain display. Maybe, I'm just really picky.

No doubt about the Log scales. I haven't found much use for the Log 1 dB scale either. It would be nice if one of us made a video to correlate the actual sum of the (most significant) harmonics in the HP 3580A display to the THD of the Tektronix AA501 and posted it. Making sure we make a point of voltage gain/loss vs power gain/loss, etc. I know a few out there would enjoy it.

Wow... I can see you have taken the AA501 to another order of magnitude of usage. I have taken the lazy approach and run mine in AUTO everything. I can see that just pushing in the 400 Hz HPF reduces the noise and THD display. However in my case I use it almost exclusively with a low impedance input, i.e., across 4 or 8 ohms and in that case seems to confirm what one would expect from accepted ratings of amplifiers, etc. I can see that if I do preamps (high Z), I will have to be more careful.

@basspig I agree with using the log scale but for some reason I just seem to always pick the linear scale. I do have to be careful not to overdrive the front end and watch the little red over-drive light. I just performed the same measurement as you did on the SG502 and as best I can read it, I get about -72dB to -74 dBfor the second harmonic also. You are the second one that suggests I use the log scale. I am going to take that to heart and start using both in my demonstrations. Thanks

Yes, for audio measurements the HP is really nice. It's shortcoming is that it only goes to 50 KHz so, if you have a fundamental of 20 KHz the only harmonic you can see is the second at 40 KHz. Anything past that is off scale. I use a Tektronix 7L5 for higher frequency measurements but haven't found it necessary to evaluate audio amps yet. Thanks for your comments.

@waldo041 I have posted a video on the MC 2100 solid state power amplifier under the title of "McIntosh MC2100 Under THD and Output Power Tests". Hope you enjoy.

It is so interesting you mention using Start instead of CTR. I was working with the 3580A last night and came to the conclusion that was the easiest way to get rid of the zero response peak (by shifting it off to the left) and your suggestion of 2Kc/Div is just beautiful for a 1KHz square wave and its harmonics all the way out to the 10th in one display sweep. I am getting a bit more prone to using Log scale instead of Linear but I like the sharp display that Linear gives me. Thanks.

I actually have been thinking about doing a video on dummy loads. I do hope to get around to it soon. Thanks for your comments.

@essoterric I have pretty much concluded my experiments to modify the 3580A to reduce the carrier feedthrough (zero response peak). My conclusion is that bypassing the unused carrier input and increasing the size of the capacitor that feeds that carrier to the double-balanced modulator is about the only mod that will help and not hurt performance. I've gotten the zero beat peak down to about -58dB. Also, found that soldering the pins of that chip to the PCB helps to make it more stable over time

Nice ...

@essoterric I've been using my AA501 today, and one thing I realized is that a VERY small amount of noise can totally invalidate readings on systems with very low distortion i the .00X range. Even 60Hz hum @-100dB can wreak havoc. The 400Hz HPF switched in, shows MUCH lower distortion. Just moving my hand near the cable affects readings, unless source is low impedance. I made shielded cable for this reason. Just modded my dummy load with 10:1 voltage divider/sample out for plugging to AA501.

Great Video and thanks for sharing. I am just getting my shop set up for diy audio work and was hoping you might share your dummy load design so that I can make a similar one. Cheers

So if you crank the signal up to get a square wave, you're going to have all odd harmonics present?

That was a great video! And the old school analog electronics just make it better...

GREAT VIDEO.....You have made sense to me by visually showing me how harmonics work.

Very helpful , thank you but I have so much to learn.

I'd like to find one of these in my X-mas stocking. :)

Very well explained.

That 485 is sweet! Can't beat em

Thanks ! -