Hi Guys
With a dominant single-pole filter is employed and hearing it makes a difference, it it because of what is objected about here, when Stanley Lipshitz and John Vanderkooy says that delta-sigma DACs should even be used for "High Quality Applications"?
http://sjeng.org/ftp/SACD.pdf
A bit deep and I don't confess to be a digital engineer, but noise is after all an analogue artefact and not a digital one, as indeed can easily be overlooked.
Sorting out noise always seems to improve digital playback.
Cheers, Joe
PS: Here is an excerpt.
Single-stage, 1-bit sigma-delta converters are in principle imperfectible. We prove this fact. The reason, simply
stated, is that, when properly dithered, they are in constant overload. Prevention of overload allows only partial
dithering to be performed. The consequence is that distortion, limit cycles, instability, and noise modulation can
never be totally avoided. We demonstrate these effects, and using coherent averaging techniques, are able to
display the consequent profusion of nonlinear artefacts which are usually hidden in the noise floor. Recording,
editing, storage, or conversion systems using single-stage, 1-bit sigma-delta modulators, are thus inimical to audio
of the highest quality. In contrast, multi-bit sigma-delta converters, which output linear PCM code, are in principle
infinitely perfectible. (Here, multi-bit refers to at least two bits in the converter.) They can be properly dithered so
as to guarantee the absence of all distortion, limit cycles, and noise modulation. The audio industry is misguided if
it adopts 1-bit sigma-delta conversion as the basis for any high-quality processing, archiving, or distribution format
to replace multi-bit, linear PCM.
So perhaps limiting the bandwidth, against common practice, is in some way limiting not only bandwidth, but noise properties?
.
With a dominant single-pole filter is employed and hearing it makes a difference, it it because of what is objected about here, when Stanley Lipshitz and John Vanderkooy says that delta-sigma DACs should even be used for "High Quality Applications"?
http://sjeng.org/ftp/SACD.pdf
A bit deep and I don't confess to be a digital engineer, but noise is after all an analogue artefact and not a digital one, as indeed can easily be overlooked.
Sorting out noise always seems to improve digital playback.
Cheers, Joe
PS: Here is an excerpt.
Single-stage, 1-bit sigma-delta converters are in principle imperfectible. We prove this fact. The reason, simply
stated, is that, when properly dithered, they are in constant overload. Prevention of overload allows only partial
dithering to be performed. The consequence is that distortion, limit cycles, instability, and noise modulation can
never be totally avoided. We demonstrate these effects, and using coherent averaging techniques, are able to
display the consequent profusion of nonlinear artefacts which are usually hidden in the noise floor. Recording,
editing, storage, or conversion systems using single-stage, 1-bit sigma-delta modulators, are thus inimical to audio
of the highest quality. In contrast, multi-bit sigma-delta converters, which output linear PCM code, are in principle
infinitely perfectible. (Here, multi-bit refers to at least two bits in the converter.) They can be properly dithered so
as to guarantee the absence of all distortion, limit cycles, and noise modulation. The audio industry is misguided if
it adopts 1-bit sigma-delta conversion as the basis for any high-quality processing, archiving, or distribution format
to replace multi-bit, linear PCM.
So perhaps limiting the bandwidth, against common practice, is in some way limiting not only bandwidth, but noise properties?
.
Last edited:
In my understanding Lipshitz is quite cool about multibit sigma-delta DACs (like ESS which uses 6bits) , he only says (and more to the point, shows) single bit (i.e. DSD) is fatally flawed. Where he falls down (in my estimation) is in using FFTs as evidence to claim noise modulation does not exist. It still does, even in the ESS implementations where they went to a lot of trouble in trying to weed it out.
Noise modulation is both a digital and an analog artifact.
You make some good points.
I know what you are saying and it's one of those things that depends from what angle you look at it. But noise is always noise, albeit it can be generated in a number of ways, but it is always analog - or *analogue.
I never had any doubts that the guys at ESS weren't up to scratch and indeed I see where Lipshitz is coming from, I think he is thinking more of DSD as an archival tool as wrong, as the is pure PWM (or PDM).
Cheers, Joe
* PS: Analog or analogue, artifact or artefact? The spell checker here is American and doesn't like the latter. Here in Oz-stralia we speak English, not Americana.
Got nuttin' against Gringos.
I am going to play a little dumb here, but if it helps, it is worth it. But can you give a bit of a summary where modern DACs, both "voltage" and "current" type delta-sigma DACs, are they all low bit a la 6 bit? Or do they vary on this front?
Cirrus Logic, Wolfson, Burr-Brown, Analog Devices, AKM etc.
Cheers, Joe
Sounds like a question which might be worth a whole thread in itself
Without going back to the DSs, from memory they all nowadays use between 4 and 6 bits in the modulator.
The very first D-S design I encountered was Bob Adam's ADC he did whilst at dbx. This I think (I still have some of the parts in a kit) used 15 or 16 0.01% resistors in an array. That sets the lower bound for designs - equivalent to 4bits. It got about 106dB SNR I seem to recall.
dCS in their totally discrete implementation I think uses 24 levels, so that's closer to 5 bits. TI (PCM179x) talk about 64 levels (or is it 62, not sure?) so around the 6bit mark. Haven't seen anyone going higher. I think ADI is 5bits in AD1955 but my memory's fuzzy. As for AKM, Wolfson - haven't played sufficiently to know much at all.
Without going back to the DSs, from memory they all nowadays use between 4 and 6 bits in the modulator.
The very first D-S design I encountered was Bob Adam's ADC he did whilst at dbx. This I think (I still have some of the parts in a kit) used 15 or 16 0.01% resistors in an array. That sets the lower bound for designs - equivalent to 4bits. It got about 106dB SNR I seem to recall.
dCS in their totally discrete implementation I think uses 24 levels, so that's closer to 5 bits. TI (PCM179x) talk about 64 levels (or is it 62, not sure?) so around the 6bit mark. Haven't seen anyone going higher. I think ADI is 5bits in AD1955 but my memory's fuzzy. As for AKM, Wolfson - haven't played sufficiently to know much at all.
Thanks, this is what I expected. But I have here a particular behaviour that I don't have with full multi-bit DACs - and I am really thinking NOS-DACs in particular.
Indeed I have my own NOS-DAC and there is no doubt they have a particular sound all their own - often a more expansive midrange presentation and in my case a particularly solid bass. These are areas where they excel (not always in the bass, but that could also be because of the post-DAC circuit and in my NOS-DAC I didn't use anything other than a 1:1 quad-filar transformer).
But, when subjecting delta-sigma DACs to a single-pole filter that is anything but flat @ 20KHz, I then get some of that which I hear in NOS-DACs coming back to delta-sigma DACs. But again, the post-DAC stuff has to be well-sorted I suspect, to really get the benefit.
Get a termination Z for a current DAC of 3.3R (fit two small series resistors) and fit 0.68uF to 0.82uF across the output, before the 3.3R pair.
With "voltage" DAC, fit 15R to 47R and fit a somewhat smaller value cap across the phases after the added resistors, to get similar result.
Once determined and we get a consensus opinion, then maybe we could discuss what is really going on. The audible effect is similar to applying 'damping' to circuits that needs damping. Rolling off the top end and it doesn't get slower sounding, but actually faster sounding, something I associate with optimum damping in phono cartridges, recording microphones, loudspeakers etc. What I am saying is that this is not a frequency thing.
But the consistent point so far with the delta-sigma DACs tried so far, is that around -1.3dB and not lower, @ 20KHz gives that optimum point. So we are talking about a similar and consistent behaviour from the current batch of delta-sigma DACs encountered so far: Cirrus Logic, ESS Sabre, Burr-Brown and Analog Devices (AD1955).
Cheers, Joe
Do you FR-correct your NOS DAC? I mean fix up the sinc droop? If so that could be one similarity. Bruno Putzeys said somewhere that in his view the 'NOS sound' is the sound of the sinc droop and he was able to obtain the same sound from a non-NOS DAC just by emulating the NOS frequency response.
For myself I understand there's more to NOS sound than that, I prefer my NOS DACs flattened and they only gain in SQ terms. Its an interesting viewpoint nevertheless.
For myself I understand there's more to NOS sound than that, I prefer my NOS DACs flattened and they only gain in SQ terms. Its an interesting viewpoint nevertheless.
No, I don't.
I don't want to give the impression that I am trying to make a d-s DAC sound like a NOS-DAC, but rather the best SQ possible.
I was simply using the simplest multi-DAC around as an example and point out that this behaviour doesn't seem to affect those kinds of DACs, or genuine multi-bit, but rather that we are observing a consistent behaviour with d-s DACs - even from different manufacturers.
Met Bruno Putzeys at ETF06 - he was the only Dutch guy there who didn't speak English with a somewhat US accent, but had a much softer Brit accent. Indeed most Europeans there spoke with a US accent, but I was the the other exception. Mine is definitely Aussie.
Cheers, Joe
Last edited:
Well, I did some more experiments about this filtering technique, and I think to make known my fresh results.
My observations confirm most of the Joe`s statements/findings.
This filter is the only one used in my DAC post processing circuit.
I repeat that my tests were done using a PCM1792 DAC (it were more convenient for me to work on this DAC now), but the principle here about the unusual filtering method it may apply to any delta/sigma DAC (post processing). Therefore, I do not find something wrong to take this subject in this Oppo mods dedicated thread.
Referring to my last/previous post, I did increased the value of the cap I used between the DAC`s phases, from 1nF to 15nF (and still also much lower than Joe used value on his ES9018 DAC).
I will say that this capacity increasing caused actually a really mess in how the signal, noises looked out on FFT, and how the sound could be heard. The noise floor have raised up dramatic, HF oscillations were present without signal, a heavy bass dominated the sound, and the sound stage it were really messed it up.
So, I have mounted some resistors trying to adapt the DAC`s output impedance to the used capacity. Still be all the bad... I have kept the resistors in place and got back to the 1nF. Still have big HF noise level, but the sound become well.
Got back to my 1nF, removed the resistors and all become just wonderful…
My conclusions so far:
This filtering technique is quite special, and have an exceptional positive impact to the outputted sound from a DAC system.
To achieve such result one may tweak the value of the filtering cap (as Joe state it) accordingly to the circuit this filter will be working on. To be more precise, to the impedance of the DAC output.
As using on my particular circuit, a 15nF value for the filtering cap it were much too high (or over the critical value). Starting in my case with the 100pF value it were positive quite by chance, when then it were confirmed that the cap value range it were for my circuit somewhere into 1nF range.
The critical value of the filtering cap it depends (I confirm here what Joe state) of many factors: DAC`s output impedance, the further configuration/circuits used (coupled) to the DAC output, and so on. The cap value it will be particular for the DAC chip and its post-processing configuration. As we can see it may vary quite much from one configuration or DAC chip used to another.
About the DAC output impedance, my opinion is different than Joe`s. I will say that one may use the impedance of the DAC chip as it is for best results, rather than correct/adapt it with additional resistors. Sometimes, the impedance of the DAC chip may be unknown, so one may tweak the cap value to adapt it. This is a hard work!
Now I have also a different opinion about the working procedure to adapt the cap value to the DAC system. I think it may be a better procedure to appreciate the critical value of the cap, observing when the noises start to increase in the system (connecting that cap between phases), rather than register the 1,3dB attenuation for a 1 kHz/20kHz signal out of the system. One may stop to increase the cap value, when the noise levels start to increase, and then chose that next lower (closest) value, which it may fit for the cap. Just under critical value, the result is only exceptional (as Joe has found out too).
I will not say that Joe`s procedure is wrong, but I think it may be more accurate to go the noises way, (noise increasing start before the outputted signal attenuation…) than choosing a cap value which it may affect/attenuate the useful signal.
I also agree that using ES9018 it may be easier to find the right cap value, as the output impedance of this DAC chip is well defined. Here it make also sense what Joe said, that the cap value reach quite high values (cause the relative big impedance on ES9018 output, when in voltage mode).
I can just confirm now that this filtering method is quite special and it give exceptional results when all fits all...
There are few days now I cannot stop listening my music outputted of such treated DAC system...
Much appreciated Joe`s contribution making known this filtering technique.
Again, I think it will be only benefit for all of us to find out more about this thing, if more people will take a closer look on it, and then come here with their thoughts/observations/findings…
My observations confirm most of the Joe`s statements/findings.
This filter is the only one used in my DAC post processing circuit.
I repeat that my tests were done using a PCM1792 DAC (it were more convenient for me to work on this DAC now), but the principle here about the unusual filtering method it may apply to any delta/sigma DAC (post processing). Therefore, I do not find something wrong to take this subject in this Oppo mods dedicated thread.
Referring to my last/previous post, I did increased the value of the cap I used between the DAC`s phases, from 1nF to 15nF (and still also much lower than Joe used value on his ES9018 DAC).
I will say that this capacity increasing caused actually a really mess in how the signal, noises looked out on FFT, and how the sound could be heard. The noise floor have raised up dramatic, HF oscillations were present without signal, a heavy bass dominated the sound, and the sound stage it were really messed it up.
So, I have mounted some resistors trying to adapt the DAC`s output impedance to the used capacity. Still be all the bad... I have kept the resistors in place and got back to the 1nF. Still have big HF noise level, but the sound become well.
Got back to my 1nF, removed the resistors and all become just wonderful…
My conclusions so far:
This filtering technique is quite special, and have an exceptional positive impact to the outputted sound from a DAC system.
To achieve such result one may tweak the value of the filtering cap (as Joe state it) accordingly to the circuit this filter will be working on. To be more precise, to the impedance of the DAC output.
As using on my particular circuit, a 15nF value for the filtering cap it were much too high (or over the critical value). Starting in my case with the 100pF value it were positive quite by chance, when then it were confirmed that the cap value range it were for my circuit somewhere into 1nF range.
The critical value of the filtering cap it depends (I confirm here what Joe state) of many factors: DAC`s output impedance, the further configuration/circuits used (coupled) to the DAC output, and so on. The cap value it will be particular for the DAC chip and its post-processing configuration. As we can see it may vary quite much from one configuration or DAC chip used to another.
About the DAC output impedance, my opinion is different than Joe`s. I will say that one may use the impedance of the DAC chip as it is for best results, rather than correct/adapt it with additional resistors. Sometimes, the impedance of the DAC chip may be unknown, so one may tweak the cap value to adapt it. This is a hard work!
Now I have also a different opinion about the working procedure to adapt the cap value to the DAC system. I think it may be a better procedure to appreciate the critical value of the cap, observing when the noises start to increase in the system (connecting that cap between phases), rather than register the 1,3dB attenuation for a 1 kHz/20kHz signal out of the system. One may stop to increase the cap value, when the noise levels start to increase, and then chose that next lower (closest) value, which it may fit for the cap. Just under critical value, the result is only exceptional (as Joe has found out too).
I will not say that Joe`s procedure is wrong, but I think it may be more accurate to go the noises way, (noise increasing start before the outputted signal attenuation…) than choosing a cap value which it may affect/attenuate the useful signal.
I also agree that using ES9018 it may be easier to find the right cap value, as the output impedance of this DAC chip is well defined. Here it make also sense what Joe said, that the cap value reach quite high values (cause the relative big impedance on ES9018 output, when in voltage mode).
I can just confirm now that this filtering method is quite special and it give exceptional results when all fits all...
There are few days now I cannot stop listening my music outputted of such treated DAC system...
Much appreciated Joe`s contribution making known this filtering technique.
Again, I think it will be only benefit for all of us to find out more about this thing, if more people will take a closer look on it, and then come here with their thoughts/observations/findings…
Last edited:
Coris,
I would be interested in knowing more about your PCM1792 DAC set up. It could be here or in a separate thread.
In any case, in what ways do you find this sounds better than the OPPO?
How does it compare before the most recent I-V output changes?
Everyone,
I think that Joe's proposed suggestions for the next OPPO are very interesting and productive. However, I wonder if they are not ambitious enough. Perhaps we should ask for DC coupling, better power supply regulation, better clocks, more DAC output lines summed together, and a more elaborate I-V converter perhaps even discrete. We know that OPPO likes to be very ambitious.
Looking at their recent headphone amplifier, it seems that they are pursuing many of these themes already.
Eric
I would be interested in knowing more about your PCM1792 DAC set up. It could be here or in a separate thread.
In any case, in what ways do you find this sounds better than the OPPO?
How does it compare before the most recent I-V output changes?
Everyone,
I think that Joe's proposed suggestions for the next OPPO are very interesting and productive. However, I wonder if they are not ambitious enough. Perhaps we should ask for DC coupling, better power supply regulation, better clocks, more DAC output lines summed together, and a more elaborate I-V converter perhaps even discrete. We know that OPPO likes to be very ambitious.
Looking at their recent headphone amplifier, it seems that they are pursuing many of these themes already.
Eric
Last edited:
OK, I have to precise something which it may be quite important about my previous statement that PCM1792, in my set up it sounds better, than the sound I could get from ES9018 in a Oppo modified set up (system).
One may know first that in the back of my PCM1792 it is a whole computer system (my 1792 and its analogue post processing it runs on a sound card). The last processor type on one of the last motherboards, which run win 8 (tweaked for better/best audio performances). A such hardware did not run in any of Oppo devices... This fact it may due to an unfair competition between these two DACs set up, when about sound quality, in my previous comment...
So, to right compare these two DACs one may use the same "source/hardware/software".
This is of course possible, but mean also a lot of work and enough time to be accorded to this task.
In fact ES9018 it have to sounds better, as it has implemented another technology, is more sophisticated engineered/designed, it is anyway a step forward for a DAC system.
I did my best to improve what Oppo has designed in theirs two last types player. I did the same with my sound card (Asus STX - DAC and analogue stage). I have to admit (and I`m not glad for that...) that the sound out of my own designed 1792 stage it is much better than what I could get from a modified Oppo (at the present time...).
My opinion is that the analogue post processing of the signal is quite crucial for the sound quality. There is of course the power system, the clock system, and some other things, but I think the way one treat the analogue signal out of the DAC it may be the most important.
My plan is to implement quite soon my findings to the ES9018 DAC, and then comparing these two systems it may be more fair, I suppose... But first I have to put all on a right designed PCB, and then having a functional module/set up, it will be more professional to experiment on different types DACs...
Anyway this way of filtering (we have last discussed here), it gave me huge improvements, when I have implemented it first to this (I was working on at the time) PCM1792 DAC.
One may know first that in the back of my PCM1792 it is a whole computer system (my 1792 and its analogue post processing it runs on a sound card). The last processor type on one of the last motherboards, which run win 8 (tweaked for better/best audio performances). A such hardware did not run in any of Oppo devices... This fact it may due to an unfair competition between these two DACs set up, when about sound quality, in my previous comment...
So, to right compare these two DACs one may use the same "source/hardware/software".
This is of course possible, but mean also a lot of work and enough time to be accorded to this task.
In fact ES9018 it have to sounds better, as it has implemented another technology, is more sophisticated engineered/designed, it is anyway a step forward for a DAC system.
I did my best to improve what Oppo has designed in theirs two last types player. I did the same with my sound card (Asus STX - DAC and analogue stage). I have to admit (and I`m not glad for that...) that the sound out of my own designed 1792 stage it is much better than what I could get from a modified Oppo (at the present time...).
My opinion is that the analogue post processing of the signal is quite crucial for the sound quality. There is of course the power system, the clock system, and some other things, but I think the way one treat the analogue signal out of the DAC it may be the most important.
My plan is to implement quite soon my findings to the ES9018 DAC, and then comparing these two systems it may be more fair, I suppose... But first I have to put all on a right designed PCB, and then having a functional module/set up, it will be more professional to experiment on different types DACs...
Anyway this way of filtering (we have last discussed here), it gave me huge improvements, when I have implemented it first to this (I was working on at the time) PCM1792 DAC.
Last edited:
Coris,
Well, you should also specify whether you are comparing CDs or high-resolution formats. You can make the Comparison more even by playing music from USB thumb drive. You could also use the computer to drive the rear USB input. Both of these would be easy to try. In this case the transport is not being used. Of course the OPPO is at its best in high-resolution formats…
Also, did you compare the two setups before this new trick?
Also, I think some of us might be interested in how you built your PCM 1792 DAC. I think the modern part is now a PCM 1792A
Eric
To answer more precise to one of Eric question, I will say that referring to sound improvements, I mean improvements in the sound stage, and not exactly the fidelity of the sound. At least a better sound stage mean in fact a higher reproduction fidelity... Not very easy to describe such things.
I can make here some personal considerations. High fidelity mean (in my opinion) to reproduce the recorded sound as accurate as possible, when about frequency range, dynamics, noise levels, and so on. To have improvements in the sound stage of a stereo recording, means to have the high/highest fidelity in reproducing the relative phase changes of the different recorded sounds, which a sound stage contain. By hearing the sounds phase changing is the way the human being can only localize the sounds sources. This (phase) information is crucial for both recording and reproducing (playback), and this is actually that fidelity one may wish to have: the exact sound sources disposal from recordings played it back to the listener (in theirs room). To have such delicate information out of a recording, one need first the best possible playback integrity of the signal. To reproduce such microscopic phase changing between sounds, to permit to a listener to localize the sound sources (in a stereo sound set up) is quite a challenge for an audio system.
However, this is very possible and it works actually. I`m really amazed to find out that such extremely fine piece of information it is transmitted from the recording microphones, through all of digitalizing, coding, files and analogue processing out to the reproducing system/speakers.
I think this, the sound stage it may be the goal of a sound system, and this mean actually that in fact high fidelity of a gear. When a listener is capable to localize the instruments in a band sitting in front of two speakers. Then the goal is reach it for the all involved factors in an audio chain, system, device, etc.
To get back to Eric`s question, in this area of a sound stage, I found improvements in my last experiences (using quite by chance an 1792 DAC). I`m sure that it may be even more improvements when using an ES9018 DAC, but only I personally come not yet at that point...
I can make here some personal considerations. High fidelity mean (in my opinion) to reproduce the recorded sound as accurate as possible, when about frequency range, dynamics, noise levels, and so on. To have improvements in the sound stage of a stereo recording, means to have the high/highest fidelity in reproducing the relative phase changes of the different recorded sounds, which a sound stage contain. By hearing the sounds phase changing is the way the human being can only localize the sounds sources. This (phase) information is crucial for both recording and reproducing (playback), and this is actually that fidelity one may wish to have: the exact sound sources disposal from recordings played it back to the listener (in theirs room). To have such delicate information out of a recording, one need first the best possible playback integrity of the signal. To reproduce such microscopic phase changing between sounds, to permit to a listener to localize the sound sources (in a stereo sound set up) is quite a challenge for an audio system.
However, this is very possible and it works actually. I`m really amazed to find out that such extremely fine piece of information it is transmitted from the recording microphones, through all of digitalizing, coding, files and analogue processing out to the reproducing system/speakers.
I think this, the sound stage it may be the goal of a sound system, and this mean actually that in fact high fidelity of a gear. When a listener is capable to localize the instruments in a band sitting in front of two speakers. Then the goal is reach it for the all involved factors in an audio chain, system, device, etc.
To get back to Eric`s question, in this area of a sound stage, I found improvements in my last experiences (using quite by chance an 1792 DAC). I`m sure that it may be even more improvements when using an ES9018 DAC, but only I personally come not yet at that point...
Last edited:
I have compared these two DACs systems before. It were off course better the ES9018 in Oppo.
Well, a computer system it may reproduce also DSD signals, BluRay, and so on. High resolution files. For some formats there is a conversion on the fly, but anyway...
Yes I use the last chip PCM1792A. At least that another one type (without A) I think it become obsolete... Anyway, I use what is to be found now on market.
I know, of course that is not difficult to connect the computer to the Oppo player... But is not this the point here. The clue is to eliminate most of the "inter stages"... The best is to have I2S interface... So, here is a little problem when to connect the computer to Oppo....
I really think that the 1792 it may be where it is now (connected to the sound card ), and the ES9018 in its Oppo player. With the same (in principle) set up, playing hi res digital files, it may be a fair competition... We will see..
I`m still working to put together all the things in a professional way, then I will show you the result...
Last edited:
Next Generation
I sent the following suggestions to OPPO for the next generation of players. I incorporated both my ideas and those in this forum.
"I wanted to make some suggestions for the next generation of players.
The current model has a superb headphone output (into my Sennheiser HD-650 headphones), and also excellent balanced outputs. However, the unbalanced outputs are not quite as strong. This has been observed by many professional reviewers as well.
Therefore, my request would be for improved audio performance from the left and right 2 channel unbalanced and balanced outputs.
I'm sure you have your own ideas, but I suggest the following:
The output coupling capacitors can be removed and the unit should be DC coupled. (There is already zero DC output, and if necessary the output stage can have better heat sinking through a power pad to protect it).
The power supplies should have better regulation, at the minimum some larger capacitors, and better bypassing. Currently the unit sounds better when the transport is not running, and in pure audio mode.
Additional local bypassing should be applied liberally. For example 10 µF smd ceramic capacitors should be added to the DAC power lines as suggested by Joe Rasmussen. This is inexpensive and very effective.
Better clocks should be used. Even slightly better clocks appear to have a beneficial effect. Perhaps synchronized clocks such as 54 MHz and 27 MHz could be used as suggested by Coris.
You should return to the system of using four pairs of DAC output lines summed together. This is clearly beneficial. It should be possible to extract the signal from after the IV converters without picking up extra noise. Currently, the headphone output seems to benefit compared to the others in part by having more DAC output lines.
Currently the IV converter for the unbalanced stage is not symmetric. This may be part of the reason this output is less good than the others. In particular, on the audio board, Resistors R34, R101, R30, R33 should all have the same value. Making this stage symmetric will lead to proper cancellation of common mode noise out of the DAC.
Most of these ideas would also improve or could be applied to the 8 channel outputs.
More ambitious suggestions could include:
A more elaborate I-V converter, perhaps even a discrete one. If you want to stick with SOIC op amps, the OPA1632 has been suggested by Ric Schultz for example. This op amp also comes in a power pad version for better thermal dissipation.
Synchronous clocks could be chosen to match the source material better. Also one centralized clock should reduce jitter between the digital and analog sections. Some people report good results with Crystek 100 MHz DAC clock.
Higher quality parts could potentially be used in targeted locations such as key resistors in the I-V stage and in the signal path. Oscon capacitors could be used for digital boards. Fast recovery diodes could be used in the power supplies.
It may also be possible to improve the standalone usage of the unit without an attached video display.
Better vibration isolation of the transport mechanism. Better shielding of the audio board which is currently on top of the digital board. More attention to power supply noise, and to high-frequency shielding in general.
Many of these ideas have been discussed on the DIY forum."
I sent the following suggestions to OPPO for the next generation of players. I incorporated both my ideas and those in this forum.
"I wanted to make some suggestions for the next generation of players.
The current model has a superb headphone output (into my Sennheiser HD-650 headphones), and also excellent balanced outputs. However, the unbalanced outputs are not quite as strong. This has been observed by many professional reviewers as well.
Therefore, my request would be for improved audio performance from the left and right 2 channel unbalanced and balanced outputs.
I'm sure you have your own ideas, but I suggest the following:
The output coupling capacitors can be removed and the unit should be DC coupled. (There is already zero DC output, and if necessary the output stage can have better heat sinking through a power pad to protect it).
The power supplies should have better regulation, at the minimum some larger capacitors, and better bypassing. Currently the unit sounds better when the transport is not running, and in pure audio mode.
Additional local bypassing should be applied liberally. For example 10 µF smd ceramic capacitors should be added to the DAC power lines as suggested by Joe Rasmussen. This is inexpensive and very effective.
Better clocks should be used. Even slightly better clocks appear to have a beneficial effect. Perhaps synchronized clocks such as 54 MHz and 27 MHz could be used as suggested by Coris.
You should return to the system of using four pairs of DAC output lines summed together. This is clearly beneficial. It should be possible to extract the signal from after the IV converters without picking up extra noise. Currently, the headphone output seems to benefit compared to the others in part by having more DAC output lines.
Currently the IV converter for the unbalanced stage is not symmetric. This may be part of the reason this output is less good than the others. In particular, on the audio board, Resistors R34, R101, R30, R33 should all have the same value. Making this stage symmetric will lead to proper cancellation of common mode noise out of the DAC.
Most of these ideas would also improve or could be applied to the 8 channel outputs.
More ambitious suggestions could include:
A more elaborate I-V converter, perhaps even a discrete one. If you want to stick with SOIC op amps, the OPA1632 has been suggested by Ric Schultz for example. This op amp also comes in a power pad version for better thermal dissipation.
Synchronous clocks could be chosen to match the source material better. Also one centralized clock should reduce jitter between the digital and analog sections. Some people report good results with Crystek 100 MHz DAC clock.
Higher quality parts could potentially be used in targeted locations such as key resistors in the I-V stage and in the signal path. Oscon capacitors could be used for digital boards. Fast recovery diodes could be used in the power supplies.
It may also be possible to improve the standalone usage of the unit without an attached video display.
Better vibration isolation of the transport mechanism. Better shielding of the audio board which is currently on top of the digital board. More attention to power supply noise, and to high-frequency shielding in general.
Many of these ideas have been discussed on the DIY forum."
Coris, what is your I-V arrangement?
From memory you stated you were using differential opamp, ie; virtual
ground? If this is the case the caps across DAC OP's work in a completely
different way and yes you do increase noise gain, distortion etc.
You also change the compensation. There is a correct way to do this and
I believe a few white papers written on it.
T