• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

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

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
1,289
Likes
858
Location
SoCal
#1
The endless agonizing about oversampling digital interpolation filters and how they can badly affect the sound leaves one wondering if it is possible to create an ideal analog filter that would make a digital filter unnecessary. The common knowledge tells us that digital filters were introduced to simplify the design of the analog anti-imaging filters, i.e. reduce costs. What if one doesn't care about the cost? I'm pretty sure it is possible to create a heck of an analog filter for the amount of money some makers charge for their DACs, some of which, ironically, have no filters at all.
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,534
Likes
4,175
Location
Gold Coast, Queensland, Australia
#2
The common knowledge tells us that digital filters were introduced to simplify the design of the analog anti-imaging filters, i.e. reduce costs. What if one doesn't care about the cost? I'm pretty sure it is possible to create a heck of an analog filter
Digital filters in the case of Compact Disc were created due to the fact that Philips had already committed to silicon 14 bit D/A converters (as had Toshiba) when the working group decreed that 16 bit was the standard. Oversampling gave them almost 16bits of resolution using 14 bit D/As and precious extra time to get true 16bit D/As fabricated. (the dual channel TDA-1541)

It was a welcome byproduct of oversampling (4 times in the case of the first TDA-1540D, single channel 14 bit D/As released in their CD players in March 1983), that the analogue reconstruction filters could be less complicated and less costly.

The analogue filters used in the very first generation CD players were expensive to buy and align. 13th order Chebyschev were used in some early machines IIRC.

This is single sample, full 0dBFS impulse as produced by the very first commercially released CD player- the Sony CDP-101 (I have several in my collection). Note, due to oversampling not appearing on this unit, there is absolute zero pre-ringing. There is considerable post-ringing however.

RIGOL Print Screen23-02-2018 12_15_20 PM.842.jpeg


Here's a 100Hz square wave, again on the CDP-101

RIGOL Print Screen20-01-2018 5_28_39 PM.368.jpeg


Here's the same single sample pulse on a 24/196 CD player with OS (A Marantz PMD-325 professional machine) Horizontal Scales are not the same however- it is not really that clean

RIGOL Print Screen23-02-2018 10_49_37 AM.948.jpeg


RIGOL Print Screen23-02-2018 10_49_58 AM.403.jpeg


Zoomed in to 100uS/Div to see the pre-ringing of the oversampling DF.

The same square wave for comparison

RIGOL Print Screen23-02-2018 10_47_23 AM.342.jpeg


Personally, I have absolutely no problem listening to well implemented 16/44 D/As, using analogue (brickwall) filters.

But, my favourite machines are Sonys, their flagship machines from 1989-91 which used BB PCM-58P K in 8x OS. To this day, their CD (measured) performance is truly beyond any rational criticism. It makes me smile to see standalone D/As of today, that can't approach what those machines achieved on Red Book nearly 30 years ago.
 
Last edited:

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
1,289
Likes
858
Location
SoCal
#3
Thanks, I suppose my question is can you do better than the filters in those early players.
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,534
Likes
4,175
Location
Gold Coast, Queensland, Australia
#4
I suppose my question is can you do better than the filters in those early players.
I would say, that at 16/44 NOS, no, you couldn't do any better than the analogue filters they used.

The choice of filters was based solely on performance, and LPF filter price was not a an issue in the early generation machines as they were loss making for several years in any case. Murata made most of the filters and once economies of scale kicked in, their price dropped, but the bit race was on and OS was the rage. There were only a few years where ultra high grade analogue filters were in use. Basically, I would say from 1983/4- 1986.
 

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
1,289
Likes
858
Location
SoCal
#5
That impulse response looks not much different than that from the minimum phase filter that many seem to like these days .
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,013
Likes
1,856
Location
UK
#6
The endless agonizing about oversampling digital interpolation filters and how they can badly affect the sound....
They may agonize over it, but it doesn't mean that oversampling filters do badly affect the sound - the opposite is more likely because oversampling filters are closer to the theoretically perfect filter, as designed by humans using mathematics, just like the rest of the digital audio system.

Audiophiles need something to agonize over, and in the case of digital audio they're struggling because it is so close to perfection in a $1 integrated circuit. That cannot be allowed to happen or else they would be left with nothing but music to spend their time listening to.
 

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,137
#7
They may agonize over it, but it doesn't mean that oversampling filters do badly affect the sound - the opposite is more likely because oversampling filters are closer to the theoretically perfect filter, as designed by humans using mathematics, just like the rest of the digital audio system.

Audiophiles need something to agonize over, and in the case of digital audio they're struggling because it is so close to perfection in a $1 integrated circuit. That cannot be allowed to happen or else they would be left with nothing but music to spend their time listening to.
True, the accuracy problem in a $100 digital device is a fraction of the accuracy problem in, say a $100k analog turntable system.

Why is it that analog inaccuracy is benign while digital inaccuracy is malign?
 

Wombat

Major Contributor
Joined
Nov 5, 2017
Messages
4,047
Likes
2,358
Location
Australia
#8
True, the accuracy problem in a $100 digital device is a fraction of the accuracy problem in, say a $100k analog turntable system.

Why is it that analog inaccuracy is benign while digital inaccuracy is malign?
It is to do with which crowd and which neutral.
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,534
Likes
4,175
Location
Gold Coast, Queensland, Australia
#9
That impulse response looks not much different than that from the minimum phase filter that many seem to like these days .
Ironic isn't it?

The issue however is many of the early machines using NOS also had a single multiplexed D/A resulting in a phase error (interchannel delay) that increased as the frequency went up. That said, in a normal 2 channel stereo I have never been able to hear it. It only becomes remotely significant at high frequencies.

Here's a 3.15KHz example of the CDP-101, showing the interchannel phase delay.

RIGOL Print Screen20-01-2018 5_01_14 PM.846.jpeg
 

Theo

Active Member
Joined
Mar 31, 2018
Messages
257
Likes
142
#10
If this is just a delay between channel, moving your armchair a few inches should compensate that, shouldn't it?
 

Kal Rubinson

Major Contributor
Industry Insider
Joined
Mar 23, 2016
Messages
1,109
Likes
1,018
Location
NYC/CT
#11
The endless agonizing about oversampling digital interpolation filters and how they can badly affect the sound leaves one wondering if it is possible to create an ideal analog filter that would make a digital filter unnecessary. The common knowledge tells us that digital filters were introduced to simplify the design of the analog anti-imaging filters, i.e. reduce costs. What if one doesn't care about the cost? I'm pretty sure it is possible to create a heck of an analog filter for the amount of money some makers charge for their DACs, some of which, ironically, have no filters at all.
I published an article in The Audio Amateur on doing that way back in the last century.

MY ERROR: I was referring to the reconstruction filters on the output of the DAC.
 
Last edited:

Kal Rubinson

Major Contributor
Industry Insider
Joined
Mar 23, 2016
Messages
1,109
Likes
1,018
Location
NYC/CT
#13
What was the verdict?
My feeling was that it was worthwhile but that was very early in the CD era and the issue of filtering was fairly simplistic.

EDIT: MY ERROR: I was referring to the reconstruction filters on the output of the DAC.
 
Last edited:

DonH56

Technical Expert
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
3,166
Likes
3,866
Location
Monument, CO
#14
The interpolation filter at the input of DACs is usually a digital thing, not sure why you'd want to insert an analog network into that path. Interpolation is used to assist in oversampled designs to create an approximation of the signal generated by a higher sampling rate. The higher sampling rate allows use of a simpler analog output anti-imaging (some just say image) filter since the images are reflected about Nyquist (one-half the sampling rate) so higher sampling rate means images are higher and easier to filter. Digital interpolation filters can be much higher order, with desirable characteristics such as linear phase (for better time-domain response), and are stable over process, voltage, and temperature (PVT) variations.

The output filter is normally analog, and numerous types have been used, with varying impact on the signal. Some of the best for roll-off (image suppression) have more in-band ripple, and some of the best for time response have slow roll-off so very high orders (complex filters) are needed. Always trades (digital or analog). In-band artifacts are greatly reduced when the filter's transition band (corner frequency) can be moved well above the audio band. Early filters introduced significant phase shift well into the audio band and often caused HF ringing. Later filters were based on more modern filter theory and exhibited fewer problems.

FWIFM - Don

Edit: Was typing so missed Kal's response... @Kal Rubinson , is the article on-line?
 

Kal Rubinson

Major Contributor
Industry Insider
Joined
Mar 23, 2016
Messages
1,109
Likes
1,018
Location
NYC/CT
#15
The interpolation filter at the input of DACs is usually a digital thing, not sure why you'd want to insert an analog network into that path. .....................

The output filter is normally analog, and numerous types have been used, with varying impact on the signal. Some of the best for roll-off (image suppression) have more in-band ripple, and some of the best for time response have slow roll-off so very high orders (complex filters) are needed. Always trades (digital or analog). In-band artifacts are greatly reduced when the filter's transition band (corner frequency) can be moved well above the audio band. Early filters introduced significant phase shift well into the audio band and often caused HF ringing. Later filters were based on more modern filter theory and exhibited fewer problems.

FWIFM - Don

Edit: Was typing so missed Kal's response... @Kal Rubinson , is the article on-line?
1. MY ERROR: I was referring to the reconstruction filters on the output of the DAC.
2. I doubt it. I have been trying to find it but all I can find is this reference: "Passive Filters for Digital Audio", Kalman Rubinson, TAA 4/94, p. 30
3. Cited here: http://info.linuxoid.in/radio/audio/cd1/AMPLIFIERS/39/JT DAC No_ 3.htm
I am sure that I have a print copy at home.
 

DonH56

Technical Expert
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
3,166
Likes
3,866
Location
Monument, CO
#16

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
1,289
Likes
858
Location
SoCal
#17
Sounds like there are missed opportunities in the modern audiophile business. Given a decent analog anti-imaging filter they can slap the mysterious NOS stamp on a DAC (NOS sounds good, right?) and have this DAC measure reasonably well and probably sound better than filterless NOS. Who knows, may be it is the next big thing.
 
Last edited:

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,534
Likes
4,175
Location
Gold Coast, Queensland, Australia
#18
...and probably sound better than filterless NOS...
Filterless NOS is just utterly ridiculous. No serious manufacturers would allow a product out the door with that implementation. It is essentially a faulty product, not fit for purpose.

It's a bit like removing the catalytic converter and muffler from a car because they 'stifle' the performance when they do the complete opposite.
 

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
1,289
Likes
858
Location
SoCal
#19
Objectively, Dyno tests confirm removing the cat, and muffler is a way to increase engine power :)
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,534
Likes
4,175
Location
Gold Coast, Queensland, Australia
#20
Objectively, Dyno tests confirm removing the cat, and muffler is a way to increase engine power
And spew all sorts of toxic substances into the air as well as producing excessive and undesirable noise- that was my analogy.
 
Top Bottom