• 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.

PCM61P DAC chip info requested from a Dennon DCD-660

gvl

Addicted to Fun and Learning
Joined
Mar 16, 2018
Messages
876
Likes
473
Location
SoCal
#41
The images follow a sinc (sin(x)/x) curve so signals in the upper midrange get folded around Nyquist (one-half the sampling frequency) and can appear above 20 kHz in "significant" amounts. The levels may indeed be "fairly low" but then again tweeters aren't designed to take a lot of power. I have an article on it, somewhere, but in any event have given up trying to convince anyone of anything in audio these days. More to the point, oversampled DACs (currently the vast majority of them) move the images well above the audio band where they are less likely to modulate a driver back into the audio band. Still likely to heat up the tweeter, of course.

This article touches on images: https://www.audiosciencereview.com/...ital-audio-converters-dacs-fundamentals.1927/
I think what saves the tweeters is the fact that HF content is relatively short-lived in nature. Just for kicks here is the peak spectrum of Pink Floyd's Money (in red), it would appear that the image levels should not exceed -40dB, hardly any danger to tweeters at normal listening levels.

2018-04-25 11_21_36-WaveSpectra - C__Temp_05 - Money.png
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,478
Likes
2,342
Location
Monument, CO
#42
@RayDunzl: NP Ray, you are correct in general and specifically for most conventional DAC architectures like R2R or similar; most R2R DACs are actually segmented with unary (unit-weight) MSBs and binary weighted (e.g. R2R) lsbs. I would never say you don't know what you are talking about. OK, unless you don't... :)

Something in the system will filter the steps, whether it's on the output of the DAC or the mechanical limits of the speaker drivers. I prefer to get rid of the trash earlier in the signal train before it starts to smell. :)

Analog proponents (remember I am an analog guy, have been for decades, in my day job) in the audio world (ok, so maybe not me) have thrown up those stair step signal pictures "forever" to show how "bad" digital looks. You have to remember those steps have very fast edges, well above the audio band, so they are filtered and never heard. Useful to spread FUD about digital but little else.

@gvl: Remember the entire band from DC to Nyquist is reflected about Nyquist (one-half the sampling rate) so a large LF tone has an image that is near the sampling frequency, albeit still reduced by the sinc function. A 100 Hz tone will be reflected about 22 kHz and appear at 43.9 kHz. Predicting the actual worst-case for music would mean taking your peak curve, imaging around Nyquist, and applying the sinc function to see where the worst case lies.

Me, I'll just stick with devices that explicitly incorporate the image filter.
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
6,257
Likes
2,311
Location
Riverview, FL
#43
I actually have a NOS/No Filter DAC here.

Unfortunately, I can't display the steps that I think I see because noise is interfering.

Here's as close as I could get, using 192khz sample rate to capture 44kHz analog with a cheap ADC and noisy analog cable:

upload_2018-4-25_14-38-41.png


I'm not 100% convinced I'm seeing steps, because of noise, but it does look a bit steppy...

(way zoomed in on this 70Hz signal)

Let me try this on a filtered DAC...

upload_2018-4-25_14-43-55.png


Noisy still, but doesn't have the more regular every sample jumpiness to it...
 

gvl

Addicted to Fun and Learning
Joined
Mar 16, 2018
Messages
876
Likes
473
Location
SoCal
#44
...
@gvl: Remember the entire band from DC to Nyquist is reflected about Nyquist (one-half the sampling rate) so a large LF tone has an image that is near the sampling frequency, albeit still reduced by the sinc function. A 100 Hz tone will be reflected about 22 kHz and appear at 43.9 kHz. Predicting the actual worst-case for music would mean taking your peak curve, imaging around Nyquist, and applying the sinc function to see where the worst case lies.

Me, I'll just stick with devices that explicitly incorporate the image filter.
No, I get it. It is just the 100 Hz tone image will be significantly reduced by sinc.

ai_sinc_fresp_bas459.gif
 
Last edited:

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,478
Likes
2,342
Location
Monument, CO
#45
I actually have a NOS/No Filter DAC here.

Unfortunately, I can't display the steps that I think I see because noise is interfering.

Here's as close as I could get, using 192khz sample rate to capture 44kHz analog with a cheap ADC and noisy analog cable:

I'm not 100% convinced I'm seeing steps, because of noise, but it does look a bit steppy...

(way zoomed in on this 70Hz signal)

Let me try this on a filtered DAC...
The marketing blurb says no digital filter, I would guess there is still an image filter on the output though could well be wrong. At low frequencies it is easier to see the steps because the edges are slow enough, despite HF filtering, to show the stair steps.

You can still see steps at the output, there is finite resolution whether delta-sigma or not, but it depends upon the filtering performed (digital and analog).
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,478
Likes
2,342
Location
Monument, CO
#46
No, I get it. It is just 100 Hz tone image will be significantly reduced by sinc.

View attachment 12336
OK, sorry for misunderstanding. The chart is very useful for illustrating the issue; LF tones are usually much bigger, but the images are much closer to the nulls of the sinc function so contribute less. Looking at the picture, if you have an 20 kHz tone sampled at 40 kS/s the image lands near the first peak of the sinc function and would be about 30 dB down. Near DC, the image is near the null of the sinc function and so greatly attenuated. A 10 kHz signal has an image at 30 kHz and is only about 15 dB down, but the 10 kHz signal is as you say most likely very small.

Some RF DACs use short pulses instead of 50% duty cycle to take advantage of the images to generate higher-frequency signals. The short pulses have less energy to begin with but the sinc function no longer applies (or rather applies but the response is much flatter with less HF attenuation).
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
6,257
Likes
2,311
Location
Riverview, FL
#47
I would guess there is still an image filter on the output
The claim is the DAC chips directly drive the cable ouputs... there's not a lot in the box:

upload_2018-4-25_14-56-1.png


The DAC chips are TDA1543T.

Maybe it has some filter inside it. You tell me.
 

gvl

Addicted to Fun and Learning
Joined
Mar 16, 2018
Messages
876
Likes
473
Location
SoCal
#48
The claim is the DAC chips directly drive the cable ouputs... there's not a lot in the box:
The DAC chips are TDA1543T.
Maybe it has some filter inside it. You tell me.
Interesting the specs say 4xTDA1543, I see 2, are the other 2 on the other side of the PCB? Ahd what the heck is "differential" mode? The trend has been to just parallel these chips together, there is nothing differential about it.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
6,257
Likes
2,311
Location
Riverview, FL
#49
I see 2, are the other 2 on the other side of the PCB?
Don't know, just grabbed a picture.

---


Uh-oh, the picture is not of the same device.

Doh....

Here:

4 DACs under the heat sink, Cirrus something in the center, nothing underneath.

upload_2018-4-25_15-57-19.png
 
Last edited:

Guermantes

Active Member
Joined
Feb 19, 2018
Messages
286
Likes
212
Location
Brisbane, Australia
#51
The other gotcha for me was always that the sinc function will cause HF roll-off in the audible band in NOS DACs. So the frequency response of NOS DACs operating at 44.1 kHz is always going to be non-flat unless pre-emphasis or similar is used.

3853Fig03.gif


This figure is from here: https://www.maximintegrated.com/en/app-notes/index.mvp/id/3853
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
1,571
Likes
2,066
Location
Gold Coast, Queensland, Australia
#53
...Analog proponents...have thrown up those stair step signal pictures "forever" to show how "bad" digital looks. You have to remember those steps have very fast edges, well above the audio band, so they are filtered and never heard. Useful to spread FUD about digital but little else...
And, especially for you Don, I present exhibit 'A' and 'B' below- complete with official 'stairsteps' from the (co)inventors themselves. :)

The first release sales documentation from Philips in 1983. I remember picking these sheets up at the HiFi show where Compact Disc was being shown. There were long queues at the Sony stand where they had 5x CDP-101s sitting on pedestals with some MDR-xxx headphones to listen. You had choice of Jazz, Classical or Popular. I listened to Jean Michel Jarre's Oxygene on an early Polygram disc- it was the go-to disc to demonstrate CD that year, all the manufacturers who had machines seemed to be playing it.

scan199.jpg


scan200.jpg

We ended up buying an Akai CD-D1 (Kyocera DA-01 etc) which incidentally used the Philips chipset with 4x OS and dual TDA-1540D ceramic pack D/As in a vertical spinning design. Very pretty machine, but drove my dad mad with random 'blips'- something they could never fix and ended up replacing it with the next years model.
 
Last edited:

Wombat

Major Contributor
Joined
Nov 5, 2017
Messages
3,123
Likes
1,386
Location
Australia
#54
And, especially for you Don, I present exhibit 'A' and 'B' below- complete with official 'stairsteps' from the (co)inventors themselves. :)

The first release sales documentation from Philips in 1983. I remember picking these sheets up at the HiFi show where Compact Disc was being shown. There were long queues at the Sony stand where they had 5x CDP-101s sitting on pedestals with some MDR-xxx headphones to listen. You had choice of Jazz, Classical or Popular. I listened to Jean Michel Jarre's Oxygene on an early Polygram disc- it was the go-to disc to demonstrate CD that year, all the manufacturers who had machines seemed to be playing it.

View attachment 12344

View attachment 12345
We ended up buying an Akai CD-D1 (Kyocera DA-01 etc) which incidentally used the Philips chipset with 4x OS and dual TDA-1540D ceramic pack D/As in a vertical spinning design. Very pretty machine, but drove my dad mad with random 'blips'- something they could never fix and ended up replacing it with the next years model.

The absence of superlatives in the second document is noted.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
6,257
Likes
2,311
Location
Riverview, FL
#55
The other gotcha for me was always that the sinc function will cause HF roll-off in the audible band in NOS DACs. So the frequency response of NOS DACs operating at 44.1 kHz is always going to be non-flat unless pre-emphasis or similar is used.



This figure is from here: https://www.maximintegrated.com/en/app-notes/index.mvp/id/3853
Here's a 0~23999Hz log sweep, sampled at 192khz, though the Pro-ject box mentioned above, using my Cheesetools:

upload_2018-4-25_19-40-27.png


Linear Scale:

upload_2018-4-25_19-40-46.png
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,478
Likes
2,342
Location
Monument, CO
#57
The other gotcha for me was always that the sinc function will cause HF roll-off in the audible band in NOS DACs. So the frequency response of NOS DACs operating at 44.1 kHz is always going to be non-flat unless pre-emphasis or similar is used.

View attachment 12341

This figure is from here: https://www.maximintegrated.com/en/app-notes/index.mvp/id/3853
Yup. The sinc function is down about 3.92 dB at Nyquist (22.05 kHz at CD (44.1 kS/s) rate). However, most everyone these days uses a peaking filter*, now usually part of the digital filter before the DAC's output, to boost the response so there is no rolloff at the output. As evidence I refer you to the innumerable frequency response plots Amir has made which show essentially no rolloff until the image filter kicks in (with typically very steep rolloff). IIRC, I used a simple analog zero circuit at the output of the first RF DAC I designed to compensate the sinc rolloff.

At this point in my life, -4 dB at 20 kHz no longer matters... :(

Some of the early image filters modified the response well down into the audio band. One of the things to note about the Philips advert is that it highlights something I keep saying: you can oversample any DAC, delta-sigma or not, to decrease in-band quantization noise and simplify the output filter. You move the required filter corner up an octave, and gain 0.5 bit (~3 dB) in SNR (from quantization noise), for every doubling of the clock (sampling) rate assuming you keep the signal bandwidth the same.

Ray's plots are a bit surprising to me, with significant high- and low-frequency rolloff. The first null at 24 kHz is the (digital, I assume) image filter; the second null at 48 kHz is the first sinc null.

Onwards - Don

* Peaking filter = pre-emphasis, though in my present world of DFEs and such that term has somewhat different meaning than it did to me a few years ago.
 
Last edited:

sonci

Active Member
Joined
Feb 18, 2018
Messages
167
Likes
53
#58
From what I gather measurably multibit has poor linearity at very low levels which are usually inaudible in music listening. I think this is due to the challenge of having accurate resistors in the circuit. Other than that, I'm not aware of differences that may exist. I've heard talk about "faster impulse" and "better timing" but I don't have a clue if that's just talk or an actual meaningful thing.

Having said that, ...once again completely subjective, prone to error and bias, impressions. No DBT tests done at all, not making absolute claims here etc. Anyway, these are my listening impressions to follow up to your question:

I don't have a lot of experience listening to multibit dacs. I've only listened to this Denon CD player and a Schiit Gungnir Multibit. I thought that they did sound different than my DC-1 DAC which was more or less the best sounding unit I had heard. Right now the Denon is my go to CD player for classical music. I hear as if my delta sigma DC-1 to sound warmer (a sense of more lower midrange) and have more depth, and subtle detail., But it has a slight hazey coloration to it. For some reason this Denon excites my auditory senses in a different way, namely around timing, dynamics, and an ease in recognizing an instruments position and melody patterns in a song. However that sense of 3 dimensions is not quite as good as the DC-1.
I've had enough experience with DACs where I won't come out and say multibit is the only way to go. I've heard several good units that I thought sounded great. However, at this point, the Schiit Ygdrassil is going to be hopefully my final upgrade to my DAC. Unless it's a Holosprings unit! :D But only listening will tell if they stay or not.
Do you still have the Denon?
I totally agree with your subjective review. It’s really great and eventhough I have other cd players which have more resolution or detail, there’s smth special about this, but until I found this thread I didn’t know this was no delta sigma dac.
Anyway, not the best place to make this discussion. ASR is where all cd players and Dacs sound the same, which is one of the reasons I come here often..
 

solderdude

Addicted to Fun and Learning
Joined
Jul 21, 2018
Messages
956
Likes
1,476
Location
The Neverlands
#59
I feel I have to correct something here.
ASR is where DAC's and players are measured and analyzed and where one deducts from the results whether measurements indicate something may be audible different or not.
We do not say all DACs sound the same. In fact we know this is not the case and is either because of rol-off in the audible band. Incorrect reconstruction filtering (NOS 44.1kHz) or distortion levels etc. being different.

DAC's that measure closely the same and differ only on persumed inaudible levels are asuumed to sound the same though.

CD players usually have proper reconstruction filters because they only need to convert 44.1 or upsample to fixed frequency.
 
Last edited:

sonci

Active Member
Joined
Feb 18, 2018
Messages
167
Likes
53
#60
I feel I have to correct something here.
ASR is where DAC's and players are measured and analyzed and where one deducts from the results whether measurements indicate something may be audible different or not.
We do not say all DACs sound the same. In fact we know this is not the case and is either because of rol-off in the audible band. Incorrect reconstruction filtering (NOS 44.1kHz) or distortion levels etc. being different.

DAC's that measure closely the same and differ only on persumed inaudible levels are asuumed to sound the same though.

CD players usually have proper reconstruction filters because they only need to convert 44.1 or upsample to fixed frequency.
Pardon me but there are only 2 or 3 dacs that measure bad enough to affirm they sound different from the others.
The only dac I owned that was measured here was Topping D30, and when compared to the Denon, it was like night and day difference.
I don’t think anyone subjectively can prefer the Topping to the Denon.
 
Top Bottom