I got my sample of Andreas TWSDAC LT running!
The power supply for the dac is made with Sparkos Labs regulators and mosfet rectifiers.
It plays very well, very detailed, and dynamic. It is good, very good! I bet you will not find anything better for up to double the price.
But is it better than my current reference dac with the ancient TDA1541A in balanced mode? No, it is not. Voices sound so extremely natural on the TDA, and has a musical flow (an analog sound) that is very satisfying to listen to for hours and hours. I have never heard this quality from any digital source.
But I want the TWSDAC to perform better than my TDA dac, I need improvement😫
So now the experimenting for further improvement starts.
I am experimenting if the TWSDAC can be improved with changes in the power supply, l just lowered the VREF voltage from 5 to 4V to see if this changes the sound balance.
Another spot of improvement is maybe the power supply of the Raspberry Pi? I use an Allo Shanti now. And maybe adding some extra galvanic isolation will have a positive influence?
Has anyone experimented with this dac yet? I am interested in your findings.
The power supply for the dac is made with Sparkos Labs regulators and mosfet rectifiers.
It plays very well, very detailed, and dynamic. It is good, very good! I bet you will not find anything better for up to double the price.
But is it better than my current reference dac with the ancient TDA1541A in balanced mode? No, it is not. Voices sound so extremely natural on the TDA, and has a musical flow (an analog sound) that is very satisfying to listen to for hours and hours. I have never heard this quality from any digital source.
But I want the TWSDAC to perform better than my TDA dac, I need improvement😫
So now the experimenting for further improvement starts.
I am experimenting if the TWSDAC can be improved with changes in the power supply, l just lowered the VREF voltage from 5 to 4V to see if this changes the sound balance.
Another spot of improvement is maybe the power supply of the Raspberry Pi? I use an Allo Shanti now. And maybe adding some extra galvanic isolation will have a positive influence?
Has anyone experimented with this dac yet? I am interested in your findings.
Attachments
Yes, I do...I have built everything with Salas ultrabib...
I found the quality of the drixo clocks PSU very important...just finisHed the battery PSU but have not had any time to listen and try.
I spend lots of time to see what effect the source has...which is your question here as well.
It is audible, even though many love to argue about this. I just can give you the hint to try a small x86 based machine vs. Rasperry.
After many tries I personally and completely subjectively favour a Pink Faun like build with a Asus crosshair vii and Amd 5700x using USB out with a debian system tickless and isolated cores etc...see my threat.
The Rp plays really nice. You can be completely happy with it.
The x86 adds a level of realismn which is hard to describe...subtile..but there seem simply to be more information, less rounded transients, finer details, more air...but we talk here about moving from 95% to 98%. So dont expect wonders...But i went back and for now for months, invited others which did blind tests not to fool myself etc...and that is where I stand with My 5ct.
I found the quality of the drixo clocks PSU very important...just finisHed the battery PSU but have not had any time to listen and try.
I spend lots of time to see what effect the source has...which is your question here as well.
It is audible, even though many love to argue about this. I just can give you the hint to try a small x86 based machine vs. Rasperry.
After many tries I personally and completely subjectively favour a Pink Faun like build with a Asus crosshair vii and Amd 5700x using USB out with a debian system tickless and isolated cores etc...see my threat.
The Rp plays really nice. You can be completely happy with it.
The x86 adds a level of realismn which is hard to describe...subtile..but there seem simply to be more information, less rounded transients, finer details, more air...but we talk here about moving from 95% to 98%. So dont expect wonders...But i went back and for now for months, invited others which did blind tests not to fool myself etc...and that is where I stand with My 5ct.
I have experimented also with different sources and software. I also used an x86 Alix board but found a USBridge Signature the best sounding source. Also the software matters.
But these differences are subtle as you also have noted. I am looking more into other refinements that change the sound balance and musical flow of the dac.
For the clocks I have tried different PS also LifePo4 batteries and found the Sparkos to be the best sounding.
I also use Ultrabib and Reflector PS on my TDA dac. I am only wondering if I should take the effort because this dac needs 4 separate PS and 2 for the Fifo and tow more for clock and RP.
But these differences are subtle as you also have noted. I am looking more into other refinements that change the sound balance and musical flow of the dac.
For the clocks I have tried different PS also LifePo4 batteries and found the Sparkos to be the best sounding.
I also use Ultrabib and Reflector PS on my TDA dac. I am only wondering if I should take the effort because this dac needs 4 separate PS and 2 for the Fifo and tow more for clock and RP.
Supersurfer, interesting comparison and from the pics it seems you have a serious TDA1541 setup (if I am not wrong you are using the D3 boards).
Andrea mentioned in his forum that he noticed no difference in sound quality in the range of 4-4.7V for Vref, but that there are some who have preferred the sound using shunt regs there.
Are you taking the I2s signal directly form the Rpi? Have you tried going trough Andrea's HDMItoI2S or with the JLsounds I2SoverUSB cards? Have you used the same source and connections for the comparison of both DACS?
Finally, I see you have higher freq. clocks on the LT than on the TDA1541. I have no way to compare, but from measurements alone the lower freq. ones are better and Doede Douma (DDDAC) reported hearable differences too. Perhaps you could swap them form some time and check if the LT improves, most probably you already have.
Andrea mentioned in his forum that he noticed no difference in sound quality in the range of 4-4.7V for Vref, but that there are some who have preferred the sound using shunt regs there.
Are you taking the I2s signal directly form the Rpi? Have you tried going trough Andrea's HDMItoI2S or with the JLsounds I2SoverUSB cards? Have you used the same source and connections for the comparison of both DACS?
Finally, I see you have higher freq. clocks on the LT than on the TDA1541. I have no way to compare, but from measurements alone the lower freq. ones are better and Doede Douma (DDDAC) reported hearable differences too. Perhaps you could swap them form some time and check if the LT improves, most probably you already have.
Where do you guys source Salas ultrabib regulators? Is there any way to get them fully build?
I wonder if anyone has tried Sjostrom psu, seem a very high quality option as well. Maybe Sellarz psus are another option.
In the transport, I want to try the x86 way as well. I have an UP2 board connected (not yet built) to a coreaudio P24 pcie-usb card. This card has an mmcx socket to feed an external 24 Mhz clock. Just for fun, this clock will be supplied by a TWTMC-STS-FSDO board which in turn will be connected to DRIXO (24Mhz) and a ECC88 tube clock from Abbas. PSU for the pack (excepting UP2) will be Ian´s PurePi.
Sjostrom
SSR03 --> https://sjostromaudio.net/shop/buil...utput_voltage-15_v/42-voltage_reference-lm431
SSR04 --> https://sjostromaudio.net/shop/buil...tage_reference-lm431/47-mains_voltage-230_vac
I am not sure if both can be configured for dual positive output. I would use it for Clocks and VREF in DAC.
Sellarz
ULN-PS2 --> https://sellarz.com/goods/view?no=57
I have heard a TDA1541 implemented with this PSU performing REALLY good!
SSR03 --> https://sjostromaudio.net/shop/buil...utput_voltage-15_v/42-voltage_reference-lm431
SSR04 --> https://sjostromaudio.net/shop/buil...tage_reference-lm431/47-mains_voltage-230_vac
I am not sure if both can be configured for dual positive output. I would use it for Clocks and VREF in DAC.
Sellarz
ULN-PS2 --> https://sellarz.com/goods/view?no=57
I have heard a TDA1541 implemented with this PSU performing REALLY good!
https://www.diyaudio.com/community/threads/gb-for-salas-sslv1-3-ultra-bib.322596/
Those you mentioned are expensive!
A super regulator and a Salas type shunt regulator are two different things. For some applications they may sound rather different. Vref would probably show a clear difference. Andrea reports that shunt regulators sound best for DAC_Lite Vref supplies. It may matter which particular shunt design though.
About super regulators:
https://linearaudio.nl/sites/linearaudio.net/files/superreg V2.3.pdf
Salas Shunt:
https://www.diyaudio.com/community/threads/salas-sslv1-3-ultrabib-shunt-regulator.322411/
According to a patent by Martin Mallinson of ESS:
"...This abrupt change in incremental resistance allows the shunt regulator 200 to provide a stable output voltage for a wide range of load conditions at the same regulated or zener voltage. In addition, compared to a series regulator in which the output impedance increases with frequency, a shunt regulator has a lower output impedance as frequency increases and thus may work better in suppressing jitter."
https://patents.google.com/patent/US9383762B2/en
My guess would be that shunts are often deemed to sound better for sensitive analog audio circuits for reasons such as those described by Mallinson.
https://linearaudio.nl/sites/linearaudio.net/files/superreg V2.3.pdf
Salas Shunt:
https://www.diyaudio.com/community/threads/salas-sslv1-3-ultrabib-shunt-regulator.322411/
According to a patent by Martin Mallinson of ESS:
"...This abrupt change in incremental resistance allows the shunt regulator 200 to provide a stable output voltage for a wide range of load conditions at the same regulated or zener voltage. In addition, compared to a series regulator in which the output impedance increases with frequency, a shunt regulator has a lower output impedance as frequency increases and thus may work better in suppressing jitter."
https://patents.google.com/patent/US9383762B2/en
My guess would be that shunts are often deemed to sound better for sensitive analog audio circuits for reasons such as those described by Mallinson.
The TDA1541A is indeed the D3 boards in balanced mode with the I2S-SIM board from Ryanj. I have planned to build 2 extra boards to add them in parallel. This dac is fed by an Allo USBridge sig with fifopi Q3 and reclocker, power supply of the streamer with supercaps and the dac with ultrabib and reflector shunts.
I have the 5 and 6mhz clocks from Andrea on the TDA and currently the 22-24mhz clocks on the TWSDAC LT, will switch them later to evaluate but I have already extensive experience with these on the TDA dac; the 5-6mhz are indeed better sounding but in a different way as the differences between both dacs I am hearing.
subjectively the 4V on VREF sounds a bit better than 5V, but as I have also changed the regulators from 5V sparkos to 4V new class-d there are more differences than only the voltage. Looking at the specifications the Sparkos would perform better though.
I had an Allo Isolator in my parts cabinet and just installed it to see what kind of effect this has.
Attachments
You mean to put the phase on the other pin for all transformers? Just like plugging the ps cable reversed in the wall socket (you can do this in Europe you know)
I think I am going to try the ultrabib for vref if the gb is available....
Lt3045 is enough for digital and clock sections?
Supersurfer, I will definitely be following your experiments with the FIFO+LT, I should be getting mine soon.
I believe the source (plus source PSU) and clock should be the same for a closer comparison. While the FIFO (+ iso) should decrease the impact of the source, I don't trust much the direct I2s out of the RPi.
What kind of regulation does that newclass-d have?
Even if you find at the end that the TDA1541 DAC suits you better, you can compare Andreas FIFO to your other one, who knows if it will still make it better for you.
I believe the source (plus source PSU) and clock should be the same for a closer comparison. While the FIFO (+ iso) should decrease the impact of the source, I don't trust much the direct I2s out of the RPi.
What kind of regulation does that newclass-d have?
Even if you find at the end that the TDA1541 DAC suits you better, you can compare Andreas FIFO to your other one, who knows if it will still make it better for you.
Last edited: