Lampy cd player

[Nick]

Thanks for that Neal, don't take these questions as being directed at you, but you have the book

Yep, Principles of Digital Audio....in essence this is what the book is saying: the output low pass filter (correct name anti-aliasing filter) is needed to remove all frequencies above the half-sampling frequency, in just the same way as one is needed in the A to D process dictated by the Nyquist sampling theory.

But thats my point/question, is it in "just the same way", clear and understandable problems happen if the bandwidth of the signal into a A/D conversion contains components above the Nyquest frequency. I don't see how the "correct name" is the anti-aliasing filter as I can't see how aliasing can occur on the output of the process.

But the effect of the lack of the reconstruiction filter seems to be just

contains artifacts of the sampling process. The shifts in levels (steps) represent high frequency components not present in the original signal

And then goes on to say that bad things may happen, but I can't find a clear mathematical description of any bad things other than there being extra HF content. The Wiki page mentions low frequency components being introduced, but I can't find the mechanic that causes the geneation of these signals, and further if they do exist in the original signal from the DAC, how high the pass filter removes them.

[Richard]

I am trying to find references to just what the lack of reconstruction filter is meant to do (positive or negative), other than make the trace look better. References I found indicate its to remove low frequency images from the resultant sound. Anyone got a handbook on digital audio on hand?

Maybe Andrew L (or someone he works with) has some info on the pratical effect on the lack of reconstruiction filter?

Theres a good bit here of some relevance,
http://www.lavryengineering.com/lavry_f ... ?f=1&t=611

and an in-depth paper,
http://www.lavryengineering.com/documen ... Theory.pdf

[Neal]

Nick, I've corrected the name in my post, misread it, its an anti-imaging filter.

No mention of what these bad things are other than the HF components may modulate other equipment further down the chain. I wonder if things like wide bandwidth amplifiers (say solid state with loads of feedback) could have issues...our valve amps maybe immune...

And then goes on to say that bad things may happen, but I can't find a clear mathematical description of any bad things other than there being extra HF content. The Wiki page mentions low frequency components being introduced, but I can't find the mechanic that causes the geneation of these signals, and further if they do exist in the original signal from the DAC, how high the pass filter removes them.

I wonder if this is referring to the affects the low pass filter can have on the Audio band...phase shift, ringing etc which would appear in the audio band if the sampling Frequency wasn't shifted out...

[Richard]

Hello Richard et al, I've missed the boat here. What is 'NOS' and what is '4X'?

I've got a 'last year's model' lampucera board and intend to get digging with tiny soldering iron and then add a valve. I like the idea of what a 6N6P can give and don't need the welluy of SRPP output, so we shall see.

I bought another hongker DAC board last year and am very impressed mostly so when coupling it up to a DVD player that cost £20 - very impressive. But this board is stacked and crammed and far too complicated for me so the lampucera DAC is for me the place to begin 'sperryments.

I intend to use lampy with DVD so whatever player I am using I can use CD & DVD and, I hope, be bale to resolve DVD-Audio (I have this with a Pioneer deck anyway) and compare SACD output. I want better!

Hi Andrew, don't we all, and especially when £20 is involved

NOS = non over sampling, that is optical>digital decoder>dac and flat output with no deliberate further filtering but that is a question then as regards the merit of adding a passive low pass or if the system already includes it

4x is optical>digital decoder>4x upsampling filter>dac>output with reconstruction filter (low pass filter)

[Nick]

Yes, those are good reads, they talk about the Anti Imaging filter on the end of the process, but I can't find any definition of the imaging they talk about.

At the end of the based on that text, my hearing will function as a perfectly good sinc function.

Thats my point, I don't question that adding a filter will alter the resultant wave, what I am trying to find is why that filter seems to infer some audible product (image) is generated, such that the brickwall filter that is my ears will not remove.

Its going back to my question, I don't doubt that the sound with X4 and NOS is different, I am not questioning what you hear, but to me the interesting and usefull question is why? Normally you would say that X4 allows a higher filter corner, so it has less effect at audio frequencies, but in this case its can't be true as there is no filter.

I wonder if the image in question is similar to SSB radio or FM Stereo modulation, where there are duplicates of the original signal centered around 44k, 88k and so on.

[Neal]

I wonder if the image in question is similar to SSB radio or FM Stereo modulation, where there are duplicates of the original signal centered around 44k, 88k and so on.

Maybe Nick:

"viewing the output filtering process from a more mathematical view point we can observe how sampling creates the need for filtering. Sampling multiplies the time domain audio signal with the time domain sampling pulse signal. In terms of the spectrum of these two sampled signals, this convolution produces a new sampled spectrum identical to the original unsampled spectrum. However, additional spectra are infinitely repeated across the frequency domain at multiples of the sample frequency IE 1Khz sampled at 44Khz creates components at 43, 45, 87 and 89Khz. Although the sample and hold process reduces the amplitude of these extra frequencies, significant components still remain, particularly in the region near the Audio band"

[Richard]

Hi Neal and Nick,

Yes, I liked this Q and A which turns it around somewhat,

http://stason.org/TULARC/entertainment/ ... ecord.html

My gut feeling at the mo is that we may be accepting upsampling as a benefit without questioning it. It was used by Philips to make their 14 bit chips work with 16 bit. They don't miss a trick though and although the later 1541 chip didn't actually need it they sold it as a benefit. Then reversed track and took it to the ultimate level applying it to make a cheap 1 bit chip work over 16 bits, originally for cheap players but then stuffing the hifi market with it too as bitstream.

Nyquest theorem says sample at twice the max analogue f then filter off above the max f and we won't get aliasing. The problem is a suitable low pass filter which is in the analogue domain and may take some of the audible spectrum with it and oversampling was introduced to allow filtering higher up. This brings us back to Nick's point which is the equipment and his ears so no filter needed. Upsampling means a complex filter in the analogue domain which may shifts phase and mess the sound up. So upsampling was invented through need but sold as a benefit.

[Nick]

Well, I don't think oversampling means the added image filter any more than NOS. I believe they are two different things, Originally oversampling was done by inserting blank samples, so 4X was one sample period of the signal followed by three sample periods of digital zero.

I think upsampling is a different process, where digital interpolation between samples is introduced. that interpolation may be viewed as a filter in the digital domain, but still seperate from the analog filter at the end of the line.

But in this case if the comparison is NOS + NO filter against x4 + filter, then its not as I thought, so I take most of it back

[Richard]

Sorry, yep used the wrong word intead of OS. Should read oversampling not upsampling in my last post, thinking of what I was trying to say not what I was actually saying.

[andrew Ivimey]

Yep that's clear, Ta, as clear as an azure sky of deepest summer - or not but thanks. I have a problem with - its what they say in ebay adverts when they think they can get away with the valve for sale not being tested.

£20 is a lot of money for a DVD/CD player! oh alright its cheap and I want to be cheerful - probably asking too much.

[Richard]

Yep that's clear, Ta, as clear as an azure sky of deepest summer - or not but thanks. I have a problem with - its what they say in ebay adverts when they think they can get away with the valve for sale not being tested.

£20 is a lot of money for a DVD/CD player! oh alright its cheap and I want to be cheerful - probably asking too much.

Ah yes NOS new old stock or new old socks depending on the seller or non over sampooling...

Sorry I also got upsampling and oversampling mixed in my reply.

...and it was not clear in my post that the NOS chain I posted is how mine is run at the present and that my own conundrum was whether it needed a reconstruction (low pass) filter or not to work correctly. I have this morning answered that and will post forthwith

So not a very good reply at all really...

[Nick]

So not a very good reply at all really...

But we do have standards here that you are expected to live down to

[pre65]

So not a very good reply at all really...

But we do have standards here that you are expected to live down to

As in rule, hang on, there are no rules.

NURSE, where are the rules.

[Nick]

Who mentioned rules, I am talking about standards

[pre65]

Who mentioned rules,

Jethro Gibbs.

Just to say I'm finding this thread most interesting. I wonder why no Chinese hi-fi entrepreneurial person has produced a DAC like wot Andrew and I have, but with a valve output stage on board.

I have asked Lawrence if one is in the pipeline.