Building the ultimate NOS DAC using TDA1541A

Hi mattiasw,


I must say the only one specification of the PowerDAC I am a little bit worried about is the high output impedance of 1.7 Ohms.
Certainly it is perfect for a headphone amp but for a power amp output would a much lower impedance be desirable.
What I can imagine is that it will sound ultra transparent and clean but maybe offer less slam and drive.

Please, can you comment on this point.

The short answer,

1) Obtain critical damping with the open baffle speakers we also plan to offer.
2) Limit maximum power dissipation (maximum power dissipation occurs at top volume setting, volume is controlled by varying the supply voltage).
3) Obtain low bit errors, lowest resistor value is determined by MOSFET H bridge RDSon(flatness) & wiring.

Low output impedance is associated with high damping factor (load impedance to output impedance ratio) and the higher the damping, the better at least that's what we have been told.

However, all extra resistance in series with amplifier and speaker chassis is going to reduce -actual- damping factor to modest levels, even if the amplifier offers zero output impedance.

And then there is the speaker DC resistance (voice coil):

Butler Audio

In this example we get actual damping factor of only 1.33 with a power amplifier that offers infinite damping factor (zero Ohm output impedance).

What helps is connecting a shunt resistor to the speaker, this resistor will lower the total series resistance in the circuit, increasing actual damping factor.

Anyway, if a speaker has poor self damping properties then even a power amp with zero Ohms output impedance and huge copper interlinks with -very- low DC resistance won't keep it under control.

But how much damping do we need? for lowest distortion (best signal tracking) we need to obtain critical damping.

Here is a tube explaining critical damping. Magnet is fixated on a pivot. Solid aluminium block is placed at different distances from the magnet. This will vary Eddy current losses in the Aluminium block and related mechanical damping:

Critical Damping -- xmdemo 068 - YouTube


What I did was selecting speakers with good self damping properties and adjust the PowerDAC output impedance for acceptable power consumption and at the same time obtain critical damping.

Because we plan to offer our open baffle speakers with the Fractal PowerDAC, I left the output impedance at 1.7 Ohms for now.

But how do we get such low output impedance with amplifiers? by applying -massive- global feedback. After decades of testing and comparing I am not convinced that global feedback does no harm to the sound, especially when driving reactive loads like speakers. The degrading is not subtle, fatiguing dynamic distortion, smearing effects. With one single signal everything looks and measures very promising but as the signal becomes more complex and the speaker is throwing back voltage at the amplifiers, sound quality degrades.

The PowerDAC basically consists of switches and a resistors. There is no local / global feedback at all, speaker back emf is perfectly absorbed by resistors, nothing gets unstable and there is zero smearing effect.

IF lower output impedance is required there are several options. We plan to offer a Fractal PowerDAC bridge version that contains 4 Fractal Power D/A converters. These can be configured as follows:

- Two separate outputs / channel, 1.7 Ohms (comparable with bi amping).
- Two converters running in parallel, 0.85 Ohm output impedance.
- Two converters in bridge configuration, 3.4 Ohm output impedance.

There is also the possibility of using larger Fractal pattern size like 16 or 32 bits pattern:

8 bits -> 1.7 Ohm
16 bits -> 0.85 Ohms
32 bits -> 0.425 Ohms

But lower output impedance translates to higher power dissipation.

Fractal PowerDAC dissipation depends on the supply voltage (R remains constant). So the higher the supply voltage, the higher the volume and the higher the power dissipation.

After selecting desired volume level (supply voltage) power consumption remains constant as the PowerDAC is based on power division between both Fractal resistor array and speaker. The available amount of power (volume setting) remains constant.
 
Anyone had any difficulties driving their preamps / amps from The Fractal DACs 1.27V rms output?

It's a fair bit lower than the standard 2v RMS (although I understand the reasons why this is the case), and my TPA3255 amps have an input sensitivity of 2.7volts...

Is it a big deal, or given the input sensitivity of my amps should I wait for the PowerDAC?
 
Hi mattiasw,
The short answer,

1) Obtain critical damping with the open baffle speakers we also plan to offer.
2) Limit maximum power dissipation (maximum power dissipation occurs at top volume setting, volume is controlled by varying the supply voltage).
3) Obtain low bit errors, lowest resistor value is determined by MOSFET H bridge RDSon(flatness) & wiring.

Low output impedance is associated with high damping factor (load impedance to output impedance ratio) and the higher the damping, the better at least that's what we have been told.

However, all extra resistance in series with amplifier and speaker chassis is going to reduce -actual- damping factor to modest levels, even if the amplifier offers zero output impedance.

And then there is the speaker DC resistance (voice coil):

Butler Audio

In this example we get actual damping factor of only 1.33 with a power amplifier that offers infinite damping factor (zero Ohm output impedance).

What helps is connecting a shunt resistor to the speaker, this resistor will lower the total series resistance in the circuit, increasing actual damping factor.

Anyway, if a speaker has poor self damping properties then even a power amp with zero Ohms output impedance and huge copper interlinks with -very- low DC resistance won't keep it under control.

But how much damping do we need? for lowest distortion (best signal tracking) we need to obtain critical damping.

Here is a tube explaining critical damping. Magnet is fixated on a pivot. Solid aluminium block is placed at different distances from the magnet. This will vary Eddy current losses in the Aluminium block and related mechanical damping:

Critical Damping -- xmdemo 068 - YouTube


What I did was selecting speakers with good self damping properties and adjust the PowerDAC output impedance for acceptable power consumption and at the same time obtain critical damping.

Because we plan to offer our open baffle speakers with the Fractal PowerDAC, I left the output impedance at 1.7 Ohms for now.

But how do we get such low output impedance with amplifiers? by applying -massive- global feedback. After decades of testing and comparing I am not convinced that global feedback does no harm to the sound, especially when driving reactive loads like speakers. The degrading is not subtle, fatiguing dynamic distortion, smearing effects. With one single signal everything looks and measures very promising but as the signal becomes more complex and the speaker is throwing back voltage at the amplifiers, sound quality degrades.

The PowerDAC basically consists of switches and a resistors. There is no local / global feedback at all, speaker back emf is perfectly absorbed by resistors, nothing gets unstable and there is zero smearing effect.

IF lower output impedance is required there are several options. We plan to offer a Fractal PowerDAC bridge version that contains 4 Fractal Power D/A converters. These can be configured as follows:

- Two separate outputs / channel, 1.7 Ohms (comparable with bi amping).
- Two converters running in parallel, 0.85 Ohm output impedance.
- Two converters in bridge configuration, 3.4 Ohm output impedance.

There is also the possibility of using larger Fractal pattern size like 16 or 32 bits pattern:

8 bits -> 1.7 Ohm
16 bits -> 0.85 Ohms
32 bits -> 0.425 Ohms

But lower output impedance translates to higher power dissipation.

Fractal PowerDAC dissipation depends on the supply voltage (R remains constant). So the higher the supply voltage, the higher the volume and the higher the power dissipation.

After selecting desired volume level (supply voltage) power consumption remains constant as the PowerDAC is based on power division between both Fractal resistor array and speaker. The available amount of power (volume setting) remains constant.

Hi ecdesigns,

thank you very much for your explanation though I must say the speaker cable resistance in my case is 0,05 Ohms only.
I really hope the PowerDAC will sound superior to my existing set-up.

Matt
 
So we plan to use one universal DAPI receiver for future projects

Hi ecdesigns,

am I right that both UPL and U192 will be replaced with DAPI versions of both devices when DAPI is ready?

Please, can you shed some light on how DAPI works?

Would it make sense to launch a one-box-solution (DAPI and FractalDAC in one case)?

Thank you

Matt
 
Last edited:
I may be posting this in the wrong place - forgive me if that's the case. I'm a noob to this forum.

Once covid is over I plan to take an electronics course and build a DAC. Until then I've bought a CD player with a TDA1541A chip.

I'm intending to buy a "finished" Chinese DAC to use the chip. I can find three alternatives that are possibilities but I wondered if anyone could advise on which of these three looks the most promising for immediate use and future modding.

Many thanks in advance!
 

Attachments

  • rps20201224_020617.jpg
    rps20201224_020617.jpg
    58 KB · Views: 880
  • rps20201224_020546.jpg
    rps20201224_020546.jpg
    277.2 KB · Views: 855
  • rps20201224_020502.jpg
    rps20201224_020502.jpg
    181 KB · Views: 883
  • rps20201224_020434.jpg
    rps20201224_020434.jpg
    135.5 KB · Views: 844
  • rps20201224_020406.jpg
    rps20201224_020406.jpg
    41.6 KB · Views: 769
  • A+ Gallery_10.jpg
    A+ Gallery_10.jpg
    193.2 KB · Views: 233
  • A+ Gallery_9.jpg
    A+ Gallery_9.jpg
    167.1 KB · Views: 234
  • rps20201224_022943.jpg
    rps20201224_022943.jpg
    236.3 KB · Views: 207
Hi matthiasw,

am I right that both UPL and U192 will be replaced with DAPI versions of both devices when DAPI is ready?

The S/PDIF DAPI receiver does not replace UPL96 or U192.


Please, can you shed some light on how DAPI works?

With DAPI we have 32 ... 48 parallel data bits (L+R data), these change simultaneously. After all bits have changed, one single LE (Latch Enable) signal is used to latch the outputs and output a new sample.

MCK / BCK clock signals are no longer needed as we use parallel data interface.


This DAPI interface sits between both, the digital audio receiver and Fractal D/A converter.

The receiver part is now based on a single chip (microcontroller), 4-input, 192 KHz S/PDIF Toslink receiver with independent low jitter LE sample timing signal (not recovered from S/PDIF).

It converts S/PDIF directly to DAPI. My brother developed the required software for this micro controller, one could say this is a complete digital (software) based S/PDIF receiver.

Because the LE timing signal is no longer recovered from the jittery S/PDIF input signal, both, source & interlink jitter dependency are greatly reduced.

Now we finally have all required ingredients:

- Perfect galvanic insulation (Toslink / ElectroTOS).
- Minimum data bandwidth (18 MHz, Toslink ElectroTOS).
- Effective source & interlink jitter blocking (independent low jitter timing signal).
- Silent receiver -> DAC interface (DAPI).
- Accurate Fractal D/A converter with "passive" low impedance output, low enough to directly drive headphones & speakers, completely removing audio interlink and (pre) amp related distortion & degrading from the signal path.


Would it make sense to launch a one-box-solution (DAPI and FractalDAC in one case)?

Yes, we plan to offer an affordable medium power entry level version that is primarily intended to be used with (pre) amps but is powerful enough to directly drive 32 ... 600 Ohm headphones. Prototype is already up and running and is being thoroughly tested.
 
A question - can volume of the Fractal DAC be attenuated via software (HQPlayer in my case) on my music server via USB to the U192 -> Electrotos -> fractal DAC?

I’m still waiting for my preamp to arrive, but somebody reported that this works. I just don’t want to blow my speakers, historically I had the impression that optical solutions like spdif/toslink didn’t pass volume control.

Thanks
 
A question - can volume of the Fractal DAC be attenuated via software (HQPlayer in my case) on my music server via USB to the U192 -> Electrotos -> fractal DAC?

I’m still waiting for my preamp to arrive, but somebody reported that this works. I just don’t want to blow my speakers, historically I had the impression that optical solutions like spdif/toslink didn’t pass volume control.

Thanks

In answer to my own question - ECDESIGNS U192ETL and factual DAC work fine with software control.
 
Hi stretchneck,

Will the new Power DAC do 192/24, or even higher PCM rates?

The S/PDIF DAPI receiver for the Fractal PowerDAC SE offers 4 selectable Toslink inputs and supports 44.1/16 ... 192/24 (limit for S/PDIF 24 bit stereo).

The S/PDIF receiver for the small and affordable Fractal PowerDAC E offers one Toslink input and supports 44.1/16 ... 192/24.
 
@ecdesigns

"The S/PDIF DAPI receiver for the Fractal PowerDAC SE offers 4 selectable Toslink inputs and supports 44.1/16 ... 192/24 (limit for S/PDIF 24 bit stereo)."

The prospective buyer of the PowerDAC SE will most probably have an USB source, maybe a HQ server or something like a MacBook Pro.

I can not imagine that he/she wants to fiddle around with a separate box like the U192 and a Toslink connection to the PowerDAC.

So my questions:

Why should the PowerDAC SE have four Toslink inputs?
With multiple inputs you need a switch and every switch degrades SQ.

Why does it have no USB input?
You could easily put an U192 inside the PowerDAC and connect it internally via Toslink to the DAPI. So the internal Toslink connection would be much better protected and there would be no fiddling around with these connectors.

In a former post you mentioned that the PowerDAC will have a direct USB input to reduce the box count that means a nice simple system:
Server > PowerDAC > Speakers.

Please can you shed some light on your reasoning?

Thank you

Matt
 
Last edited:
Hi matthiasw,

The prospective buyer of the PowerDAC SE will most probably have an USB source, maybe a HQ server or something like a MacBook Pro.

I can not imagine that he/she wants to fiddle around with a separate box like the U192 and a Toslink connection to the PowerDAC.
Why should the PowerDAC SE have four Toslink inputs?
With multiple inputs you need a switch and every switch degrades SQ.

We already designed, built and tested a USB DAPI receiver and I reported about this. Further research showed that this was not the most optimal solution. The same DAPI receiver (everything else being the same) with S/PDIF Toslink input exceeded our expectations.

The S/PDIF DAPI receiver does -not- recover the S/PDIF clocks for D/A converter timing like most other S/PDIF receivers. It uses an -independent-, local low jitter clock for D/A converter timing. So S/PDIF jitter is no longer relevant.
Low Toslink bandwidth and very low DAPI parallel interface bandwidth effectively attenuate S/PDIF -noise-
So we are now free to choose whatever S/PDIF Toslink source we want without worrying about S/PDIF signal quality or S/PDIF input switch related degrading.

The 4 S/PDIF inputs on the new experimental S/PDIF DAPI receiver are I/O pins on the micro controller. Each pin offers the same properties (signal quality) anyway.


-> Macbook pro has 3.5mm Toslink output and can be directly connected to the PowerDAC series, even with 5 meter Toslink fibre and a 3.5mm adapter plug if desired. I am using this for driving my Fractal PowerDAC prototype.
-> Most streamers have Toslink output and can also be directly connected to the Fractal PowerDAC series.

IF one -must- use USB no matter what, any available USB to Toslink converter will do. We offer a tiny (2cm x 5cm x 10cm) USB Toslink converter module that is USB bus powered. The big difference here is that we no longer need a low jitter source as we fixed this problem in the S/PDIF DAPI receiver. This greatly simplifies converter design.


Why on earth provide Toslink only? Why not add each and every existing digital audio interface on the DAC for maximum flexibility like all others do?


The short answer: Unwanted large bandwidth noise injection and ground loop noise injection (poor galvanic insulation). This is the reason we had to abandon the USB DAPI receiver project that we already completed and tested in favour of the S/PDIF DAPI project.




I added a graph illustrating USB vs Toslink vs DAPI data bandwidth.

Larger data bandwidth -> more high frequency noise enters the DAC (fused to the data signal, must enter the DAC to retrieve contained data) and the easier this noise spreads across the DAC (stray capacitances and parasitics).

We can also see from the graph that lower sample rate support also reduces required data bandwidth and related noise injection and spreading.



The ideal digital audio interface foe connecting a source to an external DAC would be a DAPI interface based on 48 separate optical fibres (one for every bit) plus one fibre for sample timing (latch). This would reduce data bandwidth to 100 ... 200KHz, 20,000 times lower bandwidth compared to USB!

When providing multiple different digital audio interfaces on a DAC, one interface can inject large bandwidth noise, degrading all others (noise, galvanic insulation bypass).

With Toslink we don't have this problem (lowest noise injection + perfect galvanic insulation).



More on this subject in my next post.
 

Attachments

  • graph1.png
    graph1.png
    31.4 KB · Views: 783
Like most I find the PowerDAC an interesting product. I hope to hear reviews of its sonic performance and wish ecdesigns every success with its commercial launch.

OTOH, is it time to start a thread devoted to this commercial product in the vendor area? I have valued this thread as a resource on TDA1541a NOS DAC innovation. It will be hard to find that information if this becomes simply the ecdesigns product page. I fear we'll soon be discussing active speakers and transports that have nothing to do with 1541a.