DSD comparison to PCM.

[LynnOlson]

Still coming down. Looks pretty out there; but I'm staying indoors today unless we need to use the snow thrower again. Plenty of food and fortunately, very reliable utilities here in Colorado.

I also want to thank Bud Purvine, who got me out the feeling-sorry-for-myself doldrums after my accident in the snow in the Colorado blizzards of 2006. The recovery from the broken leg took 6 months before I was allowed to touch the floor with the toes, and another 6 months to learn to walk again.

I was feeling pretty gloomy, and Bud suggested I get moving on one of my deferred projects - the high-efficiency speaker. Thus began the "Beyond the Ariel" postings over in the DIY forum. Down the rabbit hole on horns, but it got me going again, and got me in touch with some really talented people around the world, for which I am very grateful.

But I'm still pretty wary about snow, which is why I bought the Honda snow thrower. Shovelling snow that's 6" or more deep is something to be careful about. I'm trying to persuade Karna that maybe we should get an electrically heated driveway, but she's not going for it.

 

[Steve Williams]

Ouch.

73 right now where I live

 

[garylkoh]

Well, as of a few minutes ago I would have to add another tube amp to my very short list. Gary Koh loaned me a pair of his personal designs for unobtainium mono blocks, from out of his storied and smokey history.

Incidentally Gary, placing the notch filter after the speaker and placing a Ground Control in between appears to be the answer. For the rest of you, I have been trying to keep a notch filter applied to my EnABL'd Lowther's from absolutely killing all life from those drivers. So, two delights in one evening... no chocolate required either.

Bud

Bud,

glad that one worked. I'll have to try the Ground Control between the speaker and the notch filter too. Did you place it on the return or +ve?

You should also declare your industry affiliation like Lynn did. I think that your history would also be extremely interesting.

 

[DonH50]

Whereabouts in CO, Lynn?

23 degF and still snowing here, but I can see the neighbor's houses (could not just a little while ago) so wind and snow rate falling off. About 7-8" so far, on top of the 8" just a few days ago. Blizzard warning goes until 11 pm tonight; calling for 9-13" today and another 1-2" tonight. If goes like normal it will end this afternoon but who knows. Lows in the single digits next few nights, highs in the 20's, another chance of snow Tuesday. Getting to work tomorrow morning could be interesting.

We really need the moisture, though!

And one can never have enough chocolate...

 

[LynnOlson]

About 15 miles east of Boulder, and 2 miles west of I-25. On a clear day, there's a great view of the Rocky Mountains, including Long's Peak. Here's a photo I took a few years ago, looking west, as the first rays of sunrise illuminate the Rockies. (Yes, photography is one of my other passions, going back to my first camera, a Pentax Spotmatic that I bought when it was brand new in 1964. These top photo was taken with a Olympus OM-D, and the photo below with a Pentax K10D.)

 

[DonH50]

Nice! We can see Pike's Peak but we're higher and in the trees so don't have that broad a view. There are other benefits, and of course a few drawbacks living near the top of the Palmer Divide...

Seem to be a lot of audiophiles in Colorado.

Sorry for going OT; back to debating DSD vs. PCM (I have no dog in that hunt).

 

[LynnOlson]

Back on topic: most people, well OK, most audiophiles have never heard the full potential of DSD. There's a studio in Boulder that has a full Sonoma editing suite, which where I first heard pro-level DSD. I've also heard David Robinson's setup in Portland, but I had to make mental adjustments for the sound of an all-MBL system. The Playback Designs sounded mighty good, but again, I was mentally compensating for all the solid-state. I have no idea how the Playback Designs would sound in my all-triode, zero-feedback setup, which is about as far from MBL as you can get.

Even more so, most audiophiles have not heard the full potential of PCM, which is severely damaged by active I/V conversion, unless the converter is extremely exotic ... more exotic than anything I've seen, and I've worked with some of the world's best microwave engineers when I was at Tek. The "low-bandwidth" spectrum analyzer I had on my desk only went up to 1.8 GHz; the full-bandwidth model went up to 18 GHz without external conversion. The analyzers could easily resolve 50 Hz of phase noise (jitter) at 1 GHz, for example, and would be superb tools for looking at clock noise today.

I am not aware of any DSD DACs that use passive I/V conversion followed by wideband vacuum-triode analog amplification. Maybe they're out there, but I haven't seen them. If such a thing existed, it might be very interesting to see how it compares to a TDA 1541 or Brown PCM-63, 1702, or 1704 ladder converter with the same type of I/V conversion and amplification.

Short of that, though, a lot of what we're hearing is the result of 1~50 MHz RFI crossmodulating with audio, and all bets are off. Rather than reducing 0.0008% audio-band distortion down to 0.00008% audio-band distortion, we should get rid of the grossly nonlinear RFI crossmodulation first. Then we can finally hear the difference between the two modulation systems. Based on my limited listening tests, I think they might be closer than you'd expect.

 

[DonH50]

Isn't an output buffer and filter of some sort needed whether the DAC's input is PCM or DSD? Maybe I am not understanding the actual DSD protocol...

Having done a lot of mW design, I am not sure I would stack it against the best audio designers; HF analog requires very few transistors, little feedback, etc. I can do things in baseband, let alone at DC (which means most of audio to an RF guy), totally impractical at X band or even 1 GHz. That said, I agree choosing opamps without enough bandwidth and linearity to handle even an audio DAC's output is a problem, not to mention (so I shall) the travesty most audio designs use for an output image filter. Or they trade an ugly filter for DS DACs; simple analog filter, way more other issues...

 

[BudP]

and right back off topic. Since both Steve and Gary have asked, here is what I wrote Steve on the question.
"Not entirely certain what constitutes an industry affiliation. I am personal friends with people like Gary Koh, Rene Jaeger, Gary Pimm, Lynn Olson, Vic Tiscareno etc. and on occasion, as with Vic at Audio Prism, I have sold them transformers, for their commercial products Vic is the only one though. I do support the DIY community on a number of fronts and did design the Ground Control product for AP Labs Byron Collet. Does this qualify?"

Steve said I was probably clear to go on. As for history, mostly just digging around in my own pea patch, until a friend asks what do I think I am doing and exactly what is that weird looking thing right there? At which point, I mumble something about the cracked physics I think I am fiddling with and everyone laughs. Seriously, that is my history in a nutshell.

The notch filter is on the return leg, after your suggestion, right where the Ground Control needs to be Gary.

 

[LynnOlson]

Don't sell yourself short, DonH50. Microwave jocks make some of the best audio engineers ever. Linkwitz used to work at HP in the microwave group, A.D. Blumlein with radar and television, and it's rather convenient that microwaves are about the same dimensions as soundwaves (that old million-to-one difference between the speed of light and sound).

You guys know RFI suppression frontwards and backwards, and actually have RF spectrum analyzers sitting on the test bench. I'm embarrassed to say this, but I have *never* seen a HP or Tek RF analyzer in a high-end audio company. If you can't see what's going on at 50 MHz - say, a spur that's 20 dB down from the main signal, so it won't trigger a scope - how are you going to get rid of it? The most famous and expensive equipment on the market (no, I'm not naming names) is notorious for RF oscillations in the regulator sections, which is why XYZ brand has a "bright" house sound. Uh, yes, an uncontrolled, low-level oscillation in the 10 to 50 MHz region will certainly brighten up the sound. Swap the internal wiring for $1000-a-meter audiophile wiring, and guess what, it sounds different!

If a DAC designer wants to avoid tubes, OK, I sympathize. Not everyone wants devices that need replacing every five years or so, and 120V to 250V B+ supplies are kind of tedious. Not a solid-state expert, but I'm guessing a well-designed cascode that's well-behaved up to 100 to 200 MHz would be a good start, combined with the previously mentioned passive I/V conversion with lab-grade resistors, and a well-chosen cap array for 1st-order lowpass filtering. That should provide an adequate guard band between the abundant RFI coming from the raw converter and the more delicate, low-speed analog circuits in the rest of the system.

Opamps sound a lot better if you don't blast them with RFI, which isn't surprising when you consider the dominant pole is usually at 10 Hz or so. Not much feedback left at 50 MHz - and input diff-pairs don't like RFI either. The SPICE jockeys forget that distortion in the input-summing node *cannot* be corrected by feedback; it's there to stay, and you'll have loads of distortion above 1 MHz.

 

[DonH50]

Thanks for the kind words, "man's got to know his limitations".

That is very interesting about RFI. I have seen (and heard) some cases where amps went unstable and preamps suffered from RFI, but I would have assumed we were past all that. And, I agree, especially with digital processors and clocks flying around, that I would have expected at least a fairly inexpensive 3 GHz field spectrum analyzer would be in any audio lab these days. It's no longer just the AM band and 10.7 MHz FM LO that modern circuits have to deal with...

Years ago I made significant changes by adding RFI filters to some top-notch preamps, tube and SS. Audio circuits have very high gain so a little RF can wreak havoc if not killed before it gets mixed/rectified/amplified.

BTW, welcome to you and to Bud! - Don

 

[LynnOlson]

Nope, you'd be shocked at what you'd see at high-end manufacturers selling products in the $5,000 to $100,000 range. A front office full of marketing clowns schmoozing with reviewers, and a harried tech in the back room that's actually doing the real work. One real engineer who is well-hidden from the public is the norm. The multitude of design errors that slip through is why "modding" high-end works so well.

One of my good friends, with an RF background, went through a famous-name tube preamp with glowing reviews from all the magazines, and found that *all* the regulators were oscillating in the 10 MHz region. Not synchronously, mind you, but each singing their own little tune, and modulated by the audio signal, too, as the current demand through the regulator changed. Really pretty front panel, though. He told me that oscillating regulators is the trademark of that company, and the reason for the "bright sound" that reviewers love so much.

One of the nasty things to look out for in any feedback amplifier, particularly ones with tricky SPICE-optimized feedforward elements, is intermittent oscillations, or put another way, not quite enough phase margin. Designers forget that phase margin is affected by the reactance of the load, which in turn affects settling time. If an interior element sees a reactive load that is dynamic - like the gate capacitance of MOSFETs - the preceding stage can momentarily lose stability. This kind of thing is very difficult to see on a scope - at best, just a whiff of blur, and you won't see it on a digital display at all. The designer needs to use a wideband spectrum analyzer that can peer into the noise, and look for spurs where they don't belong.

Another way to winkle out incipient instability is remove all the lowpass filtering from the input section of a linestage or power amplifier and send an ultrafast pulse into the thing, just short of clipping. Poorly-designed amps will blow up, of course. Too bad for them. Assuming they survive - and they should - you then look very carefully at the rising edge of the pulse, especially towards the top, at every stage of amplification inside the amplifier. That, and the settling time of the pulse, will tell you a lot about how stable the amplifier really is. Repeat the pulse test with reactive loads that simulate fairly long speaker cables and complex crossovers, and you'll find which amps sound good and which don't.

 

[DonH50]

Well, you can catch glitches on a good wideband scope with deep memory, we do it all the time. But, you have to know how to do it and what to look for. A sniffer probe on a SA is much easier...

Pulse-testing can yield some interesting results. I think time-domain analysis is something that would really benefit a lot of audio (and other) companies, but it seems to be somewhat a lost art. The emphasis on filter pulse response at DAC outputs is bringing some of it back, thankfully.

Snow picking up again, suspect tomorrow will be a snow day...

 

[hvbias]

Lynn really great to see you here, along with Gary Pimm, your DIY pages got me interested in hifi many years ago

I am a big fan of the few modified PCM1704 converters I've heard as well (though using opamps for I/V). I also agree with you that I've never liked any delta-sigma DAC I've heard besides the ES9018 I'm using now, which is very similar to the ESS reference design.

The latest MSB digital is very good for PCM, I would love to hear their converters with DSD.

 

[Ragnar]

I can contact both Lynn and Rene Jaeger (Berkley designer) and see if they will post here. Lynn likely will, Rene', not so sure, but I will prod him tomorrow.

Bud

Thanks for the invite, Bud, now that you've dragged me kicking and screaming into this athabascan medley.

Here I see discourses on archaic technologies, like multi-bit dacs and vacuum tube amplifiers, on cognition and premeditation, even some on what makes DSD and PCM different. Post 155 cites an very good entry level paper which all should have read. Post 251 points to an excellent interview with Bruno Putzeys, who knows what he's talking about. Those two should have laid things to rest

A brief disclaimer, and then I will attempt to satisfy some of the curious about the technical aspects of converters from the past, PM in particular.

I have worked as an engineer for a number of manufacturers of digital audio equipment since 1979 and during that time have spent time developing or examining all parts of the audio electronics chain. I've dabbled in loudspeakers and microphones as well but do not profess any significant expertise in these areas. As a hobby I work on some of the more intractable aspects of audio reproduction such as: why does every piece of electronics gear have a different sound?

Most of my audio development career has been directed at designing equipment for profitable sale. As such, it often requires a deep knowledge of how things work and an understanding of how humans hear. I have worked in pro audio, consumer audio, broadcast audio and video and the musical instrument business.
I am still employed in the industry, live by my wits and love music (live or well reproduced, that is.)

Rene

 

[Bruce B]

Thanks for the invite, Bud, now that you've dragged me kicking and screaming into this athabascan medley.

Rene

Welcome Rene.... glad you could make it!

 

[LynnOlson]

The not-so-great aspect of living in Colorado.

 

[NorthStar]

-- Is Colorado part of Canada?

_______________

* Rene, Welcome! :b

Bob

 

[garylkoh]

Thanks for the invite, Bud, now that you've dragged me kicking and screaming into this athabascan medley.

Rene

Ha! Three for the price of one

Welcome, Rene!

 

[Ragnar]

Welcome Rene.... glad you could make it!

I guess it's time to stick my neck out! I've been dancing around this topic for years, ever since Sony threw down the DSD gauntlet against the DVDA consortium. Actually since the original shootout between analog tape, DSD and HDCD at Sony Studios in NYC with David Smith.