+1
My DDDAC had it's guts spilled out over my workbench again last week. I had a orphan shunt lying around to upgrade the mainboard 5V regulator, another wonderfull addition. Seems to have made the SQ again darker but richer at the same time.
I've been perfectly happy with my DDDAC for a long time. But if I was to do it all again (which I'm not), I would do a lot different.
Last edited:
Hehe...! Everything are computerised these days, marce. Your little communication devices, TV, phone,...., instruments/electronic-tools/gadgets are nothing but a set of flip-flop switches. According to you, we might as well call everything on earth with a set of "Flip-Flop" or a "digital logic circuit" is a computer!! Your statement is too generalised. It goes by calling a BMW is nothing but motor-engine in a nice chassis. What're about the finer details, on going refinements those German's Enginneers have invested? Nonetheless, I must thank you for your self promotion enlightening. It has certainly remided me of my pitfalls many many years ago as a young graduated engineer with full of eager. Back than, I saw through a tiny view of an instrument like telescope, and arrogantly believed seeing the panoramic view of the Universe!
My apologies to the Opt and the rest, this needed get out off my chest! 😀
@Stijn - if you do not mind, please share with us your new development with DDDAC?! How else would you rather mod them differently? I am too very satisfied with my DDDAC via I2S direct from BBB. Keen to hear what other improvements can be made to it? 😉
Attachments
Last edited:
Hi Chan, On the dac side, I'm thinking potted transformers and chokes, a lot of nice and thick copper shielding and rubber dampening. Use my new osciloscope to get the PSU ripple down. Separate PSU for analogue and digital sides, like you, Stefan and others have done. Z-foils at Rload. Possibly even a tube I/V stage. Separate very low ppm clocking board to clock and reclock the isolated source.. That sort of stuff. I'm saving these for the next DSD adventure..
I to V
The quiescent current in the buffer is very sensitive to Vbe mismatch, the correct control of this can be improved by emitter degeneration resistors. These emitter series resistors will increase the impedance but also can make it more linear. A resistance value that is "26 mV / quiescent current" already results in a good linearity of the impedance and only doubles the impedance. This is most often the optimal choice. For a signal current sourced from a high impedance, this buffer impedance and its linearity are less important so higher values of the emitter series resistors can be chosen, but there is little reason to do so.
The signal current is replicated from the emitter to the collector minus the base current. The base current is the emitter current / (1 + hfe). The hfe is not constant and this results in a some distortion of the signal: the hfe changes with the current, with the collector-base voltage (Early effect) and with the junction temperature.
The change with current and collector-base voltage give non-linear distortion.
The junction temperature depends on the changes in dissipation, the thermal resistances and the thermal capacitances (thermal mass). So with a varying current signal the junction temperature is not constant, neither follows the signal instantaneously: there is a whole range of time constants involved ranging from the ten µs level to tens of minutes level. The result is an error that is not uncorrelated from the signal like noise, nor fully correlated like harmonic distortion and is called thermal distortion. This type of distortion will be less masked in our perception .
By adding the base current to the collector current, the variation of the hfe as source of (non-linear and thermal) distortion is eliminated. In the proposed schematic, this distortion is reduced by more than 40dB.
The change in junction temperature due to the dissipation variations also changes the Vbe (- 2.2 mV/°C). In the buffer circuit this will give a varying offset voltage signal and can also be a cause of thermal distortion. For a current sourced from a high impedance, the varying offset voltage will not result in any distortion current. (For lower source impedance a compensation is needed such as keeping the dissipation constant: e.g. a series resistor in the collector of specific value).
However, in the basic schematic the changes of the Vbe in the current mirror circuit will give thermal distortion. The Wilson current mirror also solves this issue.
The resistance values in the emitters of the Wilson current mirror can be chosen large (e.g. 1V drop) if the voltage swing does not need to be maximized v the supply voltage.
This basic schematic of a buffer and current mirror - the input and gain stage of a current feedback amplifier- has room for improvement.
The quiescent current in the buffer is very sensitive to Vbe mismatch, the correct control of this can be improved by emitter degeneration resistors. These emitter series resistors will increase the impedance but also can make it more linear. A resistance value that is "26 mV / quiescent current" already results in a good linearity of the impedance and only doubles the impedance. This is most often the optimal choice. For a signal current sourced from a high impedance, this buffer impedance and its linearity are less important so higher values of the emitter series resistors can be chosen, but there is little reason to do so.
The signal current is replicated from the emitter to the collector minus the base current. The base current is the emitter current / (1 + hfe). The hfe is not constant and this results in a some distortion of the signal: the hfe changes with the current, with the collector-base voltage (Early effect) and with the junction temperature.
The change with current and collector-base voltage give non-linear distortion.
The junction temperature depends on the changes in dissipation, the thermal resistances and the thermal capacitances (thermal mass). So with a varying current signal the junction temperature is not constant, neither follows the signal instantaneously: there is a whole range of time constants involved ranging from the ten µs level to tens of minutes level. The result is an error that is not uncorrelated from the signal like noise, nor fully correlated like harmonic distortion and is called thermal distortion. This type of distortion will be less masked in our perception .
By adding the base current to the collector current, the variation of the hfe as source of (non-linear and thermal) distortion is eliminated. In the proposed schematic, this distortion is reduced by more than 40dB.
The change in junction temperature due to the dissipation variations also changes the Vbe (- 2.2 mV/°C). In the buffer circuit this will give a varying offset voltage signal and can also be a cause of thermal distortion. For a current sourced from a high impedance, the varying offset voltage will not result in any distortion current. (For lower source impedance a compensation is needed such as keeping the dissipation constant: e.g. a series resistor in the collector of specific value).
However, in the basic schematic the changes of the Vbe in the current mirror circuit will give thermal distortion. The Wilson current mirror also solves this issue.
The resistance values in the emitters of the Wilson current mirror can be chosen large (e.g. 1V drop) if the voltage swing does not need to be maximized v the supply voltage.
Attachments
Many thanks Stijn! Personally, I found having separate ps for digital and analog brought noticeable improvements. [emoji106]
These days, I found DSD are more and more engaging! Still waiting for Doede with his first prototype. For the interim, I have starting to play with JLSounds AK4490 and like it very much. Would love to have a full discrete output stage, one day. 🙂
I think David, Juancho, is visiting Perth in a fortnight. Hopefully he has spare time to listen to my assemble from Doede's creation. 😉
Sorry for going OT, again (at least this is dac related 😱)
Chanh, have you found the thread below.?This is a discrete way of doing it.
I think the JLSounds products are great and well documented.
http://www.diyaudio.com/forums/digital-line-level/273474-best-dac-no-dac.html
Chanh, have you found the thread below.?This is a discrete way of doing it.
I think the JLSounds products are great and well documented.
http://www.diyaudio.com/forums/digital-line-level/273474-best-dac-no-dac.html
I think David, Juancho, is visiting Perth in a fortnight. Hopefully he has spare time to listen to my assemble from Doede's creation. 😉
I am indeed and will report back. I'm really interested in how it compares with my fully-pimped up single decker!
David
Thanks Stijn, Ray was going to send me a board when it is made available. Unfortunately, with the fabrication error(s), it will test our patience a little longer! 🙂
Chanh.
a little off topic but curious
Hi Chanh
I have followed the development and innovations of this thread with much interest. I am not a DDDAC user but have applied many of the innovations in this thread with good results to a 1541a implementation.. including chokes in PS and bbb to S03.
First point. Ignore Marce.. incorrigible....
But out of curiousity. Like him and many others I expected computer source to have no impact on sound. How wrong in practice. Though we'd get there with async usb with good isolation on the WaveIO and similar. But no! PS to computer and software, ssd and all sorts of items still made big impact. But finally, eliminating usb, and having good isolation using S03 in my case and now tp rig in yours, I finally think the bbb and its software settings are finally making little or no impact on sound. I think I am now listening to the S03 without impact of upstream. I have cleaned up S03 PS with advanced regs and BlackGate NX HiQ caps to good effect. But all the settings I used to need to fuss with in the BBB now seem to have little effect... finally as one would hope. Finally isolated. Just interested in your thoughts on this topic as you and the other colleagues on in the DDDAC crowd seem to be on the same page. Thanks for your thoughts
Hi Chanh
I have followed the development and innovations of this thread with much interest. I am not a DDDAC user but have applied many of the innovations in this thread with good results to a 1541a implementation.. including chokes in PS and bbb to S03.
First point. Ignore Marce.. incorrigible....
But out of curiousity. Like him and many others I expected computer source to have no impact on sound. How wrong in practice. Though we'd get there with async usb with good isolation on the WaveIO and similar. But no! PS to computer and software, ssd and all sorts of items still made big impact. But finally, eliminating usb, and having good isolation using S03 in my case and now tp rig in yours, I finally think the bbb and its software settings are finally making little or no impact on sound. I think I am now listening to the S03 without impact of upstream. I have cleaned up S03 PS with advanced regs and BlackGate NX HiQ caps to good effect. But all the settings I used to need to fuss with in the BBB now seem to have little effect... finally as one would hope. Finally isolated. Just interested in your thoughts on this topic as you and the other colleagues on in the DDDAC crowd seem to be on the same page. Thanks for your thoughts
Last edited:
You are an ignorant piece of work mate, I I do not appreciate being called a jerk OK.
Please note moderators that this person changes his reply after calling me a jerk on a public forum.
No need top resort to name calling just because you and your little friends cant handle the truth. All I have done is provide some facts and if you bother to open your tiny little mind and read my posts you will see that there is always the same commonality about interfaces and how to combat the perceived problems by using engineering not magic.
Last edited:
I think somewhere in this thread, I did mention a slightly advantage in SQ when using Acko S03 than TPA Herman/Cronus. I have chosen the TPA over Acko S03 as it is more completeness. All seem to integrate well, including switch/reset and powering BBB direct from Herman board with a Lipo battery as backup feature at that given time. With S03, one must have two set of co-axial cables and three seperate shunt regulators. Most of all, I was trying to have a shortest I2S path as possible. One can appreciate the length from DDDAC mainboard to level 11 DAC chips is somewhat not ideal. 🙂
Edit: If you are using 45/49.xx, I encourage you to try the exact speed setting on your uEnv.txt. Reroute R4 to R2, from memory. Somewhere in Acko S03 we did have an extensive discussion on this topic. I would be interesting to hear your findings.
Chanh.
Last edited:
Thanks for reminding me of this Chanh. I recall Acko referencing your experience with improved sound stage with this setting. Makes sense. As good as they are one less Potato in the mix. I'll do a little digging to ensure it will not mess up downstream Ian I2S to PCM board and if ok give it a go. Could be a little while before I report back... thanks again for the tip.
Last edited:
Hi Guys,
I have the privilege having David here in my shed last night with fews local audiophiles. It was most enjoyable having him here.
Unfortunately I still have a fridge full of beverages (Alcohol). Hopefully, they'll all be consumed next gtg with our local. [emoji6]
In the Shed, we did;
1. DDDAC vs JLSound AK4490 for PCM
Materials,
2. DSD vs PCM on JLSounds AK4490 only.
3. Ethernet cable variations on DDDAC and including JD's SOtM LAN isolator unit.
4. HQA DAC vs JLSounds AK4490 for DSD materials
5. Banana Pi vs Tassopodes' with special build Linux OS on HQA DAC.
6. HQA DAC vs JLSounds AK4490 with DSD material transporting by Tassopodes.
IMO: Was not a good night for HQA DAC, unfortunate!
At Tuyen's setup, we did only one comparison. His optimised Server vs my Banana Pi all in one server.
Findings: it was obvious better with BPi, fuller sound and more engaging with better level of organic sounding.
Hopefully David can post his neutral observations on my "Dubai Tower" DDDAC and the tests we carried out at his next opportunity? .
Chanh.
Hi Guys,
I have the privilege having David here in my shed last night with fews local audiophiles. It was most enjoyable having him here.
Unfortunately I still have a fridge full of beverages (Alcohol). Hopefully, they'll all be consumed next gtg with our local. [emoji6]
In the Shed, we did;
1. DDDAC vs JLSound AK4490 for PCM
Materials,
2. DSD vs PCM on JLSounds AK4490 only.
3. Ethernet cable variations on DDDAC and including JD's SOtM LAN isolator unit.
4. HQA DAC vs JLSounds AK4490 for DSD materials
5. Banana Pi vs Tassopodes' with special build Linux OS on HQA DAC.
6. HQA DAC vs JLSounds AK4490 with DSD material transporting by Tassopodes.
IMO: Was not a good night for HQA DAC, unfortunate!
At Tuyen's setup, we did only one comparison. His optimised Server vs my Banana Pi all in one server.
Findings: it was obvious better with BPi, fuller sound and more engaging with better level of organic sounding.
Hopefully David can post his neutral observations on my "Dubai Tower" DDDAC and the tests we carried out at his next opportunity? .
Chanh.
Thanks to Chanh for hosting a memorable evening! Sorry I couldn't drink more beer but with jet lag I would have slept through all the comparisons! It was great to meet others with like-minded philosophy audiophiles! I will write up my impressions once my Conference (the other reason for being in Perth) is over!
Has anyone else done this? I consider this one of the most significant developments in ages, yet I haven't seen many giving it a go.