THD measurement

[signals]

So, I recently purchased a Keithley 2015 THD multimeter. Wasn't that interested in the THD part, but it was a heckuva deal, and came with a calibration certificate, so I have one now. Up to this point, I've only played around with some basic breadboard amplifiers, intentionally introducing distortion, and watching the change on the THD meter. Pretty neat!

I realize this is my "first post," but I built a ton of Bottlehead gear back in the day, and had posted on the old forum. I guess I'm not much of an audiophile because I built a system that I love, and really haven't obsessed about gear much since. I obsess over music now. It's been the way it is for over 10 years now, I think. The Bottlehead parts are a Seduction, Foreplay III (original version, no upgrades), Paramount 300Bs (original version, no upgrades), and Straight 8s with the 3rd gen crossovers. Yes, I actually still use Straight 8 speakers...

Anyway, I thought it might be fun to take a full system measurement from the Foreplay to the speaker terminals and see what the THD meter had to say about it. I'm going to say right up front: I have no idea what the hell I'm doing with this thing, so any measurements I make with it should be taken with a boulder of salt. But, I'm hoping they're in the ballpark so to speak.

The way the meter works is that it has an audio frequency sine-generator in it, and then it compares the input with what its outputting and calculates THD, THD+N, or SINAD numbers. So, I hooked the sine generator into the input of the Foreplay III, put the probes on the speaker terminals and had a go at it. I made the assumption that the actual speaker load was the way to go, as opposed to a dummy load, since the actual speakers are the load that the amps see when I'm listening to music. Hope that wasn't a mistake.

Now, nobody buys a single ended DHT tube amp for the THD numbers, so I never really thought much about it. Just from what I know in general about SET amps, I was expecting to see numbers in the low to mid single digits, if I was lucky. At least that's what I would have expected just from the 300B amps. There was a little voice in the back of my head saying "Don't do it! You'll get 9% THD on the meter, then you'll think the rig sounds awful, even though it doesn't." But I did it anyway.

So, what kind of numbers did I get? Well, about 0.1% THD at 1Khz! It rose slightly with a THD+N number, but only to about 0.2%, and this is with the old un-upgraded amps.

I also took several other measurements at various frequencies and found that at 30Hz, I was looking at around 1.5% THD and at 15Khz I was seeing about 1%. And, remember this is at the speaker terminals, with the speakers in the circuit. So, a much more difficult to drive load than a resistive dummy load.

I have to say, I'm impressed! The guys who are designing these things at Bottlehead certainly seem to know what they're doing. Although, I already knew that. I've been listening to Bottlehead gear since the turn of the millennium, and haven't felt the need to run out and buy the latest and greatest thing at all during that time. Although... Now that I look over the products page here... I seem to be in need of some updating.

So, someone please correct me if I'm using the THD meter in some grossly incorrect manner, but even if I'm off by 100% the numbers still seem pretty good to me. A fully vacuum tube system with SET 300Bs, driving actual speakers that measures less than 1% THD seems pretty incredible.

[Doc B.]

You have to state the THD at a specific power level. If you use really low signal levels you will get really low THD readings. You also need to test individual pieces of gear - just the preamp, just the amp, just the speaker, to get a feel for what each part of the system contributes.

When we design a zero global negative feedback amp we generally crank the signal until we see about 5-10% THD. There we measure the power. For Paramount this would be around 8W output. For  a preamp you generally look at what the signal output voltage is at some chosen THD level. Foreplay will swing a ton of volts more than you need before THD gets ugly. I think we figured something like 18 volts RMS with the C4S upgrade installed for 10% THD.

[signals]

Exactly what I meant by "I don't know what the hell I'm doing with this thing." I measured at comfortable levels because I had to be in the same room with the speakers blaring a 1Khz sine signal.

And of course, the point of the exercise wasn't to measure the components on their own, nor was it to accurately measure the distortion for a spec sheet. I was really just curious to see much distortion is being caused by the all of the electronics under typical listening conditions. And... I also just wanted to be able to say I found a practical use for the THD part of my meter, which isn't something I'd normally need, but was basically a free bonus. 

But of course, you've completely explained away the "these seem like good numbers" part of my post. Yes, I suppose comparing the numbers I got from my Kindergarten-level experiment to the published numbers taken properly is a bit misguided.

But, I guess the point I was trying to make was, I can't believe how close the big signal coming out of the electronics is to the little signal going in. Don't let the naysayers hit you with 5%+ THD numbers, because you probably aren't seeing anything anywhere near that in normal listening conditions.

(Or does measuring the entire system cause additional problems, aside from just not being repeatable or meaningful to anyone who doesn't own the system?)
 

[Doc B.]

I think it is worthwhile to actually listen to the system when it is operating at 5% or 10% THD levels to understand what that sounds like. I'd guess most people would assume they will hear clipping and other extreme stuff but may be surprised to find that the sound is merely a little compressed and warmer/fatter sounding.

[Paul Birkeland]

If you're interested in doing more measurements, I'd buy an 8 Ohm 50 Watt dummy load resistor, as often times you'll need to put more signal through an amp than you can comfortably listen to in order to find its limits.