The Sound of the Machine The Hidden Harmonics behind THD

[Tom WA3KLR]

Anatoly,

Did you ever build and test your high-powered source follower amp design?

[WBear2GCR]

Jim,

No doubt power supplies or lack of regulated supplies in the amplifier is a source of many problems and complaints, transient phenomenon.  For many years when I hear of the line cords, penny in fusebox etc. improvements I said to myself well maybe there is power supply problems.

My receiver/amp here is a Harmon Kardon 630 which is 38 years old now.  Many years ago after noticing the significant hum level (saw-toothed also) at the speaker terminals, I investigated and took out the simple series pass transistor/zener diode regulator for the +22 Volt supply and replaced it with a LM723 circuit I think it was.  Now pure dc.  Also the main speaker amp supplies had bad ground loops involving the filter capacitor currents.  I put all of the raw filter stuff on isolated buss wire which came to a single point ground on the chassis.  I moved the speaker ground return points.   Afterwards the hum was in the noise level and sinusoidal.

How many high end amps use a linear regulated supply ahead of the output amps?  Does anyone at all?

Time to update ur amplification... many reasons.

But anyhow, many amps and preamps use regulation in the power supply...

There are issues with "regulation" it's not a simple cure-all. You can look at regulation as a type of amplification, in as much as it uses feedback to control the "linearity" of the output (nominally DC). What it does steady state is not the same as it does in dynamic conditions, just like the amplification stage it is feeding.

To quote Bill Witlock (Jensen Transformers, and AES Fellow):
"A variation of your method was presented in a paper by the late Deane Jensen and Gary Sokolich, "Spectral Contamination Measurement", at the AES 85th Convention, 1988 (preprint #2725). In their (and my) opinion, much of what audiophiles describe as "the veil" consists of low-level, non-harmonic distortion products. I believe a major source of these is due to widespread use of op-amps under heavy negative feedback. As frequency increases (at ultrasonic frequencies, for example), the op-amps essentially run out of open-loop gain and start approaching open-loop conditions. This allows their inherent non-linearity to intermodulate distortion products from previous stages and the result is non-harmonic distortion products. It is one very good reason for passive bandlimiting at the input of every active amplifier stage. Otherwise, the effect is cumulative - and demonstrably ugly sounding. The test Jensen and Sokolich proposed is very similar to the AP (Cabot) multi-tone test.

Bill Whitlock
President & Chief Engineer
Jensen Transformers, Inc.
(and recent AES Fellow)"

Of course that's his approach to the "cure" and CD is bandwidth limited already...

             _-_-bear

[WD5JKO]

  I have another story where bad power worked against me. I had a RCA AR-88 receiver and after a lot of work I got it going. At one point I tried to beef up the audio output that I think was a single ended 6K6 good for 2-3 watts. I did the usual coupling cap change and added some NFB. The efforts were not responding well, and any attempt to reproduce the lower frequencies became futile. I was suspecting the audio output transformer, but before I got too far I hooked up my scope and snooped around. Much to my surprise I had over 100v P-P of B+ ripple to the 6K6 transformer whenever I cranked the audio while reproducing a male broadcast voice. The B+ was about 250v without audio, and it was all over with audio. The frequency was around 80 Hz, and somewhat sinusoidal. It turns out the the HV power supply used a 5Y3 and a dual pi C-L-C-L-C filter. The cap values were very small and the L values were very high. I'm pretty sure I had a filter resonance condition working against me. When I fixed the power supply (another story) the 6K6 behaved exactly how I had intended.

Don't overlook the power supply folks...

Jim
Wd5JKO 

[Tom WA3KLR]

"Time to update ur amplification... many reasons."

Update to tubes and transformers Bear?

"I believe a major source of these is due to widespread use of op-amps under heavy negative feedback. As frequency increases (at ultrasonic frequencies, for example), the op-amps essentially run out of open-loop gain and start approaching open-loop conditions. This allows their inherent non-linearity to intermodulate distortion products from previous stages and the result is non-harmonic distortion products. It is one very good reason for passive bandlimiting at the input of every active amplifier stage. Otherwise, the effect is cumulative - and demonstrably ugly sounding."

Bear, what breakpoint frequency would you choose "for passive bandlimiting" to filter out the offending audible harmonics yet allow the overall response to go to 20 kHz.?

Jim, yep, never forget that supply LC filtering are resonant L-C circuits.  O.k. with steady loads but can swing greatly with varying loads.  This is why the SSB linear amp B+ supplies are just capacitors; very stiff.

[WBear2GCR]

Tom,

Please don't assume?

I said a few things that seem to have gotten lost in there...

Whitlock's cure is his, not mine. He sells transformers?

There have been substantial improvements in solid state amplifier topology, with attendent improvements in steady state and transient performance since your old HK was designed and built. A newer design potentially would both measure better and sound better - potentially.

Tubes or solid state depends on things that are unclear from here, and more or less up to you. I'd bet that your old Utah speakers would sound a lot "nicer" on a modern tube amp of high quality design. They might even perform better in terms of measurement too...

The ironic thing is that ur an AMer? You run tubes?
It would be easy, no simple to build a really excellent P-P tube amp for you using modern designs and parts... I suspect you'd wonder why you ever used that HK when you were done, truly!

FYI, I'd suggest a pair of triodes run in AB2, using a DC coupled cathode driver - hey wait a sec, that sounds a lot like a modulator!!     The thing is that you don't want to push for max power at all, so you can keep the B+ modest... all you need is excellent output iron... high quality parts (resistors, caps etc...) and if you insist separate supplies for each stage (not a bad idea at all).

For a killer result consider some 845s, 211s in class A... etc... these are not your grandfather's tubes, using modern parts and power supplies!

If ur interested in modern solid state, take a look at the kit offerings from http://Borbelyaudio.com they are first rate when done in a first rate implementation. Just an example of excellent modern solid state design...

If you want to see what a very leading edge design for SS looks like, just one person's project, go here: http://www.synaesthesia.ca/PGP.html  follow it through, it is a fairly amazing design... dunno how it sounds, but the circuitry is outstanding. There are some others that are equally interesting out there...

                _-_-bear

[Tom WA3KLR]

Bear,

More on my Utah HS4-B 3-way bass-reflex cabinets with 12" woofer.  I bought them in the early 1970's.  I did a major overhaul on them in 1986 for many reasons.  The original mid-range drivers were horns.  They were replaced with Peerless K040MRF closed-back midrange cone drivers.  The spec on them is 91.5 dB SPL. 1 W., 1 M.  The loss through the cross-over is negligible, in fact a slight step-up in voltage at 1 kHz., just measured.  So I did not do my own SPL measurement at 1 W, just going with the mfr. spec.

The original “cross-over network” was a joke.  Got rid of rotting front foam cover (no kids here to poke holes in woofer).  Incorporated a new much more complex cross-over design and tweaked it.  Changed driver positions in cabinets (one cabinet became mirror image of other).   Reinforced cabinet corner seams and joints from inside with wood strips, nails and glue, and nailed in finishing nails from outside of cabinet.  This eliminated a “bark” I would get from the cabinet at loud peaks sometimes (loose sidewall).  Yes, one transient distortion source identified and fixed!

[WBear2GCR]

Tom,

Sounds like ur poised to make a breakthrough here! 

You might want to try downloading "sound workshop" - it's a freeware FFT type program that runs on the PC... use it an ur ratshack spl meter for a microphone and you can get a really good idea of what ur freq reponse looks like. I'd measure the system at <3ft on axis to the tweeter, and on axis to each driver, then measure it up close, like 4-5" in front of the cabinet on axis to each driver to see what the individual driver responses look like... that will tell you one heck of a lot more than you probably want to know about what they are doing or can do.

You regular everyday onboard soundcard will work fine for this.

No matter what if you have any electrolytics in the xover, at least bypass them with the largest value film caps you can find, and if at all possible don't use any electrolytics at all, use all film, and better still use polypropylene only. At least use polypropylene for the tweeter.

I think that you will be surprised what that does, changing nothing else.

                       _-_-bear

[Tom WA3KLR]

Bear, there are many many much higher priority projects here. 

When I did the overhaul work I used a stand-alone 20 band/half-octave filter hand-held spectrum analyzer with pink noise generator.  Each driver was individually connected to the amplifier output and tested for frequency response.  I used 3 snapshots at on-axis and 2 off-axis readings and averaged them together for the frequency response.  This is how the original midrange horns were kicked out of the design.

The two new cross-over frequencies were then chosen to be in the middle of the drivers’ overlap regions.  The cross-overs are a mixture of film and NP electrolytics.  They stand for now.

After all the work, shortly after I moved into this house in 1990 one original 1” soft dome tweeter died.  I just stuck another different tweeter on hand in it’s place and haven’t bothered to mess with that either.

[WBear2GCR]

Ok, if you want to have some fun, download that Speaker Workshop program and have a go at it with almost any mic you have on hand - if the mic from that spectral analyser can be tapped, that would work. About a half hour or so dorking with the program would open up a whole lot of insight that the bar graph type display simply glossed over. The fact that the mic you pick might be non-flat won't change the value of taking a look at it. Of course the Panansonic $1 electret mic element is flat enough until out to ~15khz so as to be virtually perfect for this application (looking at the real response through the xover areas...)

But let's leave it here, since we're trying to bridge two points of reference that are very far apart, making it hard to accomplish.

I can try to use an analogy to explain, and hope that it doesn't insult you or anyone, I'm simply trying to make it clear and easy to understand. In effect we're comparing a 3x5 snapshot (of the same subject, for example, corresponding to the same recording) to a professional large format photograph blown up to poster size and printed on Cibachrome, with color controls for hue and saturation... Of course if you never ever saw the big pro print, you might not ever realize what was possible, or how much attention to detail is required to produce it!

           _-_-bear