A cost no object ideal Analog Reconstruction Filter - is it a thing?

[restorer-john]

My feeling was that it was worthwhile but that was very early in the CD era and the issue of filtering was fairly simplistic.

Just curious, exactly how was the understanding of filtering theory simplistic in the early CD era? If you are talking about over-sampling and linear interpolation used by Philips, that would be simplistic by modern standards.

As for the analogue reconstruction LPF on D/A converter outputs, there was nothing simplistic about filter theory- it was completely understood wouldn't you say?

 

[DonH56]

Speaking for myself, the DSP theory and HW was not fully developed in the late 1970's/early 1980's, so the power to implement the interpolation filters wasn't really there. Also, speaking from a foggy memory of my 1980's-vintage digital theory grad classes, many of the early digital implementations mimic'd analog filter designs and did not take advantage of advances in digital filter theory (some made sense then). There have been major advances in delta-sigma theory and technology since then as well. I was very lucky to have Henry Samueli as my advisor back then, though I was (and am) an analog guy, and took a couple of classes with Gabor Temes (really nice guy!) I have worked on a few delta-sigma designs and they are very challenging beasts, albeit mine targeted RF ADCs and DACs.

As for the analog filters, true that the theory was (and is) there, although for commercial applications some of the things like state-space and Kalman filtering was rarely used IIRC. Audio filters tended to be Butterworth and other filters like Bessel, Chebyshev, and elliptical designs tended to be found more in the RF world so part of the issue may have simply been unfamiliarity on the part of the designers. I do not know that, however, as at that time I was working in a design world relatively far from audio. I tend to think a variety of issues were very challenging, including dealing with the (relatively) low sampling rate so image filters had to be very high-order, op-amps that sometimes struggled to provide the gain-bandwidth required and ability to recover quickly and cleanly from overload (since glitches from conventional DACs are very large, very fast, and just generally obnoxious), and limitations of the filter designs themselves (high-order analog filters are touchy IME). Finally, there were commercial realities, as the players were expensive and there was naturally pressure to reduce component counts and reduce cost.

Didn't Philips end up having to depend upon Sony for much of their early players? Been a while, maybe I am mis-remembering.

All IMO - Don

 

[gvl]

And spew all sorts of toxic substances into the air as well as producing excessive and undesirable noise- that was my analogy.

I'm not advocating filterless NOS nor do I listen to it on a regular basis, but I have to admit some of the more involving digital audio experiences I had happenned when I listened to a filterless NOS DAC. Definitely not all types of music, but certain albums for sure. I suppose it depends on the frequency content of the recording, acoustic and vocal material typically yields better results through NOS than electronic. Call it psychoacoustics or whatever but it is not about all that noise, that gets filtered on the way to the brain anyway. Sure, it may upset the gear downstream of the DAC, that's probably the main downside.

 

[Wombat]

Speaking for myself, the DSP theory and HW was not fully developed in the late 1970's/early 1980's, so the power to implement the interpolation filters wasn't really there. Also, speaking from a foggy memory of my 1980's-vintage digital theory grad classes, many of the early digital implementations mimic'd analog filter designs and did not take advantage of advances in digital filter theory (some made sense then). There have been major advances in delta-sigma theory and technology since then as well. I was very lucky to have Henry Samueli as my advisor back then, though I was (and am) an analog guy, and took a couple of classes with Gabor Temes (really nice guy!) I have worked on a few delta-sigma designs and they are very challenging beasts, albeit mine targeted RF ADCs and DACs.

As for the analog filters, true that the theory was (and is) there, although for commercial applications some of the things like state-space and Kalman filtering was rarely used IIRC. Audio filters tended to be Butterworth and other filters like Bessel, Chebyshev, and elliptical designs tended to be found more in the RF world so part of the issue may have simply been unfamiliarity on the part of the designers. I do not know that, however, as at that time I was working in a design world relatively far from audio. I tend to think a variety of issues were very challenging, including dealing with the (relatively) low sampling rate so image filters had to be very high-order, op-amps that sometimes struggled to provide the gain-bandwidth required and ability to recover quickly and cleanly from overload (since glitches from conventional DACs are very large, very fast, and just generally obnoxious), and limitations of the filter designs themselves (high-order analog filters are touchy IME). Finally, there were commercial realities, as the players were expensive and there was naturally pressure to reduce component counts and reduce cost.

Didn't Philips end up having to depend upon Sony for much of their early players? Been a while, maybe I am mis-remembering.

All IMO - Don

FYI: http://www.turing-machines.com/pdf/beethoven.htm

 

[restorer-john]

FYI: http://www.turing-machines.com/pdf/beethoven.htm

Yes, that often linked article is useful, albeit a biased and skewed account by the 'inventor' of EFM- something I think he feels deserves more credit than all the rest of the format inventor engineers together...

Clearly, there was little love lost between Sony and Philips. Simply put, Philips couldn't keep their end up at all- they couldn't get a 16bit D/A to market or functional machines ready for the release, Sony had to delay six months and release in the home market only in October 1982 and wait for Philips until March 1983 for a worldwide release. What Philips did release even in 1983 were cobbled together machines with all sorts of production modifications to make them work.

On the other hand, the Sony CDP-101 was an accomplished machine with a wireless IR remote and flawless aesthetics and functionality. While all the other manufacturers were still messing around with ugly 1st generation vertical spinners, Sony had secretly developed a drawer loading mechanism enabling the unit to be placed in a rack like all other components. They even made a lovely rosewood outer case to make the machine a little larger, more deluxe, and provide storage for the (then) novel 'remote commander'.

If you ever saw the first home market CDP-101s, they were sold multivoltage in duty free with a special cardboard carton that the top flaps enclosed a handle to carry through customs- they knew well-heeled international travellers would snap them up in droves.

I'll dig out one I have here in it's original box and post a pic.

If you are interested in a little more balance and a lot more detail, get hold of 'Digital Audio Technology' by the Heitaro Nakajima, Toshitada Doi, Jyoji Fukuda and Akira Iga, the engineers from Sony. First English Language edition, 1983. 312 pages.

Heitaro Nakajima only died late last year and received Japan's highest civillian honour for the invention and commercialisation of the Compact Disc. A truly amazing engineer and responsible for the entire Audio Division of Sony through the 1970s. The Esprit line can be attributed to him and his team.

 

[restorer-john]

Here's the pics of the CDP-101 box. Note the 'baggage' tags via Cathay Pacific.

The plastic carry handle fit here, so the super-hip traveller could breeze through the airport with the latest digital laser disc player for all the world to see.

They even proclaimed its attributes on the side.

Not trying to derail the thread, but we are discussing early analogue LPF filtering in CD players and some background on the first machines I think is helpful.