the WM8805 is 2,96 USD Mouser WM8805GEDS/V Wolfson Microelectronics Encoders, Decoders, Multiplexers & Demultiplexers
The Sabre both upsamples and oversamples.
Aliasing products in non OS DACs will not clip later stages, nor have I ever heard of them burning out anyone's tweeters. I don't have any prejudices for or against oversampling, by the way.
Also at WOLFSON MICROELECTRONICS|WM8805GEDS/RV|TRANSCEIVER, S/PDIF, 8:1, 28SSOP | Farnell Nederland Farnell.
A switchable input is very convinient in allday use.
Gr GJ
Hi,
I am sorry to say, but in this case you really did not look very far, did you?
Really? You think so? Interesting.
Do you actually speak from real experience with such devices, or are we just talking about what "theoretically will happen"?
Ciao T
I am sorry to say, but in this case you really did not look very far, did you?
Really? You think so? Interesting.
Do you actually speak from real experience with such devices, or are we just talking about what "theoretically will happen"?
Ciao T
This looks like you have sources for components that other people don't have.
As to my limited understanding of the subject, semiconductors have a transfer function that is not exactly:
y = ax + b
No I have not.
Nope, we just use industrial grade components with the specs we find relevant, which excludes racer audio components.
But also the design of the analog stage is quite unique and very well sounding, when supplied from low noise and low impedance sources.
Actually the analog stage is very powerfull and realistic in transients, as well as it is capable of reproducing very low level acoustic information. And contrary to tube gear it has full and uncompressed low end reproduction.
When supplied from shunts, it can even reproduce ambience realisticly.
OK,
TDA1545...what a mess! It´s easy to get good sound out of a 1543. The 1545 isn´t that easy.
You better put your energy to a 1541. But that´s only MY point of view! YEMV!!!
OK, I´ll make a short hint: Take a look @ TDA1387 = a better I2S version of the TDA1545A. With better implemented charge-pump. Input-format = like TDA1543!
It´s easy to source this chip: ~1,-Eu @ iBuy for an old isa soundblaster-soundcard. Go on .... but remember: my advise is 1541, nothing else ....
Carsten
TDA1545...what a mess! It´s easy to get good sound out of a 1543. The 1545 isn´t that easy.
You better put your energy to a 1541. But that´s only MY point of view! YEMV!!!
OK, I´ll make a short hint: Take a look @ TDA1387 = a better I2S version of the TDA1545A. With better implemented charge-pump. Input-format = like TDA1543!
It´s easy to source this chip: ~1,-Eu @ iBuy for an old isa soundblaster-soundcard. Go on .... but remember: my advise is 1541, nothing else ....
Carsten
Hi,
As said, you did not look far. All my commercial stuff uses shunt regulators where appropriate and lower noise and more optimised ones than those you copied from LC Audio.
So, you are speaking actually from pure theory.
My Digital Products allow the selection between various up and oversampling modes plus they allow to bypass all digital filtering and to select to use an analogue filter or not. From practice with those, up till now with quite substantial numbers no-one has reported clipping amplifiers, blowing amplifiers or blown up tweeters.
And that is of course the way how it should be, if one does REALLY apply oneself to the theory and attempts to understand what is really going on, instead of just blasting off randomly. You arguments remind of some that where had on german discussion boards over ten years ago on the subject of non-oversampling dac's.
Then I did take the liberty of actually providing a full expose of what really happens and prooved why all these nay saying supposed Electronic Engineers and Sound Engineers did not know what they where talking, today I do not have the time nor inclination, plus the sheer fact that there must by now thousands upon thousands non-oversampling DAC's and not ONE report of gear or speakers blown up as a result should suffice.
Hmm, if your Amp cannot handle an input signal of it's own "full power" at any given frequency without shutting down it would appear the design is faulty somewhere. If you would mind posting the schematic and other details I could help you diagnosing the problem.
Why not? If an Amplifier has a power bandwidth of (say) 10Hz to 400KHz why should there be any issue from giving it a full power signal at 400KHz?
I don't know. From where I stand simply because the resulting sound quality is so much more in line with real live music that it makes most oversampling DAC's sound broken, especially high order Delta Sigma types? And this seems not only my view, but that of anyone who ever bought one of the digital products I helped design.
I would add that some of the most recent Chip Designs (including some ESS, AKM and WM ones, but not a single AD, BB and CS one that I tested) using some of the less "textbook" digital filters come very close to the "a-live-ness" of a good non-oversampling DAC.
But as said, that is only my view. And you do not need to take them into account. I do however take exception when you spread false rumors...
Ciao T
As said, you did not look far. All my commercial stuff uses shunt regulators where appropriate and lower noise and more optimised ones than those you copied from LC Audio.
So, you are speaking actually from pure theory.
My Digital Products allow the selection between various up and oversampling modes plus they allow to bypass all digital filtering and to select to use an analogue filter or not. From practice with those, up till now with quite substantial numbers no-one has reported clipping amplifiers, blowing amplifiers or blown up tweeters.
And that is of course the way how it should be, if one does REALLY apply oneself to the theory and attempts to understand what is really going on, instead of just blasting off randomly. You arguments remind of some that where had on german discussion boards over ten years ago on the subject of non-oversampling dac's.
Then I did take the liberty of actually providing a full expose of what really happens and prooved why all these nay saying supposed Electronic Engineers and Sound Engineers did not know what they where talking, today I do not have the time nor inclination, plus the sheer fact that there must by now thousands upon thousands non-oversampling DAC's and not ONE report of gear or speakers blown up as a result should suffice.
Hmm, if your Amp cannot handle an input signal of it's own "full power" at any given frequency without shutting down it would appear the design is faulty somewhere. If you would mind posting the schematic and other details I could help you diagnosing the problem.
Why not? If an Amplifier has a power bandwidth of (say) 10Hz to 400KHz why should there be any issue from giving it a full power signal at 400KHz?
I don't know. From where I stand simply because the resulting sound quality is so much more in line with real live music that it makes most oversampling DAC's sound broken, especially high order Delta Sigma types? And this seems not only my view, but that of anyone who ever bought one of the digital products I helped design.
I would add that some of the most recent Chip Designs (including some ESS, AKM and WM ones, but not a single AD, BB and CS one that I tested) using some of the less "textbook" digital filters come very close to the "a-live-ness" of a good non-oversampling DAC.
But as said, that is only my view. And you do not need to take them into account. I do however take exception when you spread false rumors...
Ciao T
Some design suggestions/my $0.02
Whew this is a long thread! But a good project, I too am getting tired of the cheapo kits with poor designs. Here are some suggestions
=======================================================
C504, C507, C512, C515, C519, C522 should be at least 10uF for increased ripple rejection, or take them out entirely.
I would put 33-47 ohm resistors in the DAC digital inputs (pin 3,4,5,6) to prevent negative undershoot from getting in.
1000uF filter caps is a bit small, how about sizing for 2200uF's?
Clamp diodes on each regulator's output (in case of weird power up/down sequencing).
R111 and C218 aren't needed. I'm not sure about using three 3.3V regulators, VD1 and VD2 could be combined...
Some ferrite beads between the analog and digital grounds. Ferrite beads on the chip power pins (pre-decoupling caps). A power on LED and maybe testpoints/headers to connect to the internals would be nice.
Is this a 2 layer or 4 layer PCB? How much SMT?
I am willing to contribute to the project if you need help
Whew this is a long thread! But a good project, I too am getting tired of the cheapo kits with poor designs. Here are some suggestions
=======================================================
C504, C507, C512, C515, C519, C522 should be at least 10uF for increased ripple rejection, or take them out entirely.
I would put 33-47 ohm resistors in the DAC digital inputs (pin 3,4,5,6) to prevent negative undershoot from getting in.
1000uF filter caps is a bit small, how about sizing for 2200uF's?
Clamp diodes on each regulator's output (in case of weird power up/down sequencing).
R111 and C218 aren't needed. I'm not sure about using three 3.3V regulators, VD1 and VD2 could be combined...
Some ferrite beads between the analog and digital grounds. Ferrite beads on the chip power pins (pre-decoupling caps). A power on LED and maybe testpoints/headers to connect to the internals would be nice.
Is this a 2 layer or 4 layer PCB? How much SMT?
I am willing to contribute to the project if you need help
Whew this is a long thread! But a good project, I too am getting tired of the cheapo kits with poor designs. Here are some suggestions
=======================================================
C504, C507, C512, C515, C519, C522 should be at least 10uF for increased ripple rejection, or take them out entirely.
I would put 33-47 ohm resistors in the DAC digital inputs (pin 3,4,5,6) to prevent negative undershoot from getting in.
1000uF filter caps is a bit small, how about sizing for 2200uF's?
Clamp diodes on each regulator's output (in case of weird power up/down sequencing).
R111 and C218 aren't needed. I'm not sure about using three 3.3V regulators, VD1 and VD2 could be combined...
Some ferrite beads between the analog and digital grounds. Ferrite beads on the chip power pins (pre-decoupling caps). A power on LED and maybe testpoints/headers to connect to the internals would be nice.
Is this a 2 layer or 4 layer PCB? How much SMT?
I am willing to contribute to the project if you need help
Thanks!!
About the capacitors C504, C507, C512, C515, C519, C522, it may have a positive effect to increase these to 10uF. In our Reference DAC, we did start out with a parallel of a 47uF/100nF, but ended up using only the 100nF, even though in theory it's not ideal. This choice was based purely on listening tests. But I agree with you, that it may be an idea to let the user choose this. Therefor, I will add an 10uF in parallel. We normally prefer Panasonic FM types over all other (Including Blackgate, OS-CON). This may lead to the need of clamping diodes.
A resistor in series with the data-lines could also be an idea. We have not seens any bad effect of not having these in our Reference DAC, but I will take it into consideration.
1000uF may be to little. Normally we use 2 x 1800uF Panasonic FM i parallel.
R111 is not needed
My bad.. However, C218 is...We spend about a year tweaking the power supply in our Reference DAC, and discovered a great effect by using individual regulators for digital and analog supply on both receiver and DAC chip. It was one of the things we discussed prior to starting this project, in order to keep cost down. But both KvK and me agreed, that this was to important to the sonic performance, to not have even in this 199USD design.
Ferrites may be an idea... Again... We did spend a long time tweaking the supply. But it is taking into consideration.
And naturally we need a power ON LED... Forgot that
The PCB will be 2 layers, which is easy to source for everyone. We expect to be able to fit the whole design including power transformers, into a "Euroboard" size PCB = 100x160mm.
Once again, thanks for some positive and constructive ideas!!
If you are interested in joining the "design team", you are most welcome. I am considering using a Wolfson Micro SPDIF reciever, WM8805, since many people seem to love this chip.
We may need someone to program a small PIC, to set up the registers in the WM8805. Would that be of any interest to you??
I did look pretty far, but none in sight so far.
Tell me where to look.
I am not speaking of theory only, I´ve listened to NOS stuff, and with absolutely no go.
Frankly spoken, they are of no interest at all to me.
My power amp just excactly has a power bandwith of 3-400KHz, and that is the reason for the protection circuits.
If you look at a speaker, it is designed to reproduce the audioband + a little for headroom. The performance of any speaker degrades heavily with SPL, talking about the top octaves in music, you´ll find, that information above lets say 6-7 KHz is present only at very low levels.
Having a 400KHz crazy noise signal screaming into your speaker at FS, will not improve its performance, both due to heat and harmonics, on top any amp would perform better without that signal.
I must say, that I completely disagree about this.
I´ve never heard any NOS DAC not sounding dull and dispersed, and I do not really want to discus that subject, because I would never participate in a project like that anyway due to sound quality.
So as long as I´m part of this project, it makes no sense at to promote NOS DACs and analog brickwall filters or unfiltered noise radiators.
Maybe you can talk Hurtig into designing a thing like that some day, but I doubt it.
For me the NOS era is completely idiosyncratic, but still your input is important to us, since you did inspire about the reciever.
My choise would be CS8416, CS 8406, TI DIR or maybe ESS sabre.
But anyways - may the force be with you.
Last edited:
Hi,
Which is fine, note I did not suggest adopting Non-Os in your project, as much as I tend to prefer this option. As to what Non-Os DAC's you have heard or not I cannot comment. Many do indeed sound like you describe, but in my experience not all of them.
So it should have no issues and should not shut down with any signal containing supersonic noise at -20dB...
I see ton's of inductance in all the cabling and so on, hence the issue of high frequency noise is not even academic. At any extent, I design my speakers just like amplifiers, that means with a full power bandwidth, so you can sweep my speakers with a sinewave at rated RMS power across the rated bandwidth with zero issues.
It is one of getting certain results.
Must then use a secondary PLL and several words worth of FIFO buffer to reduce the jitter to acceptable levels.
Even though you are seeming to tend towards the down side of it...
Ciao T
Which is fine, note I did not suggest adopting Non-Os in your project, as much as I tend to prefer this option. As to what Non-Os DAC's you have heard or not I cannot comment. Many do indeed sound like you describe, but in my experience not all of them.
So it should have no issues and should not shut down with any signal containing supersonic noise at -20dB...
I see ton's of inductance in all the cabling and so on, hence the issue of high frequency noise is not even academic. At any extent, I design my speakers just like amplifiers, that means with a full power bandwidth, so you can sweep my speakers with a sinewave at rated RMS power across the rated bandwidth with zero issues.
It is one of getting certain results.
Must then use a secondary PLL and several words worth of FIFO buffer to reduce the jitter to acceptable levels.
Even though you are seeming to tend towards the down side of it...
An externally hosted image should be here but it was not working when we last tested it.
Ciao T
dispersed = no analog filter
dull = no sinx/x compensation
harsh & metallic = digital filter + multibit DAC
hazy & sandy = digital filter + deltasigma DAC
One of the reasons why digital filters are worse than non os is that for 16 bit CD, you should have at least 20 or 24 bit real resolution while you apply filter algorithms to the signal and go back to 16 bit after.
But what you have in a digital filter is faked interpolated data.
It just can not work without loss of "resolution".
It must be worse, there is no way around.
But what you have in a digital filter is faked interpolated data.
It just can not work without loss of "resolution".
It must be worse, there is no way around.
Well I do not know where you got this idea, but it surely is not compliant to the Nyquist theorem.
In oversampling the data converted is the original 16 bit data, but with estimated data interpolated between each original sample.
Of course this is a resolution enhancement, but actually only at very low levels.
NOS DACs need analog filtering, which completely ruins impulse response and phaselinearity at any level.
To me the choise is obvious, but I would actually prefer upsampling, which is the real jitter killer.
But anyways, discussing NOS technology in this thread is of topic, there will not be introduced such things in this project.
So no matter how much someone might like NOS DACs, it is just a waist of beautiful time to promote this here in this thread.'
NOS DAC discussions belong in NOS DAC related threads.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.