R&D - PowerDAC
Blog about R&D of our PowerDACs
The Fractal PowerDAC was developed in order to remove active linear circuits like (discrete) OP-amps, buffers and similar analogue circuits and all related degrading! from the signal path. The idea is sound quality improvement by elimination of imperfect analogue circuits and interlinks.
So the load (power amplifier, headphones, speakers) is directly driven by the D/A converter raw resistor array output. So all active linear circuits we would typically find in conventional DACs and (pre) amplifiers are no longer present in the PowerDAC.
There are two versions:
PowerDAC - R
PowerDAC - R (R stands for RCA output). This PowerDAC is intended to directly drive most headphones and it can be used to directly drive power amps (no pre amp required). This is a relatively small and affordable high performance DAC. The PowerDAC - R is approx. 17cm high (including foot), 20cm wide, 6cm deep, round base of 16cm.
Toslink input (44.1/16 ... 192/24).
Compatible with ElectroTOS with standard S/PDIF protocol.
Integrated lossless volume control + muting (10 steps of 3dB, 30dB attenuation range) push buttons for vol - and vol +
7 segment white LED display for volume, muting, lock indicator
Jumper setting for fixed max. volume output.
125mVpp ... 4 Vpp (44mV ... 1.4 V rms) output voltage or fixed 4Vpp (1.4V rms) output voltage (jumper setting).
30 Ohms resistor array output impedance.
2 x RCA output, headphones can be connected through RCA to jack adapter cable.
External 5V / 250mA power supply (can be tweaked for optimal sound quality).
Power consumption approx. 1.25 watts.
Firmware updates through USB socket that also serves as 5V power supply socket (the PowerDAC - R does not support USB audio, see Connectivity section below).
PowerDAC - S
PowerDAC - S (S stands for Speaker output). This PowerDAC is intended to directly drive efficient (90dB and up) speakers and all dynamic headphones, it outputs up to 6W rms. If more power is required for inefficient speakers, it is possible to add external, low gain (1x ... 5x) single semiconductor / single tube buffers. Such buffers only need few parts in the signal path like one coupling cap or insulation transformer, power tube + OT or one coupling cap or insulation transformer, power MOSFET + CCS or OT so the damage to the analogue signal (loss of resolution) is small compared to a conventional (pre) amp. The PowerDAC - S is about the size of a vintage CD player and more expensive compared to the PowerDAC - R.
4 selectable Toslink inputs (44.1/16 ... 192/24), push buttons + IR remote control for input selection.
Compatible with ElectroTOS with standard S/PDIF protocol.
7 segment yellow LED display for input selection indicator.
Integrated lossless volume control + muting (13 steps of 3dB, 39dB attenuation range) push buttons + IR remote control for vol - and vol +
dual 7 segment white LED display for volume, muting, lock indicator.
125 mVpp ... 18 Vpp (44mV ... 6.36 V rms) output voltage.
1.7 Ohms resistor array output impedance.
2 x 6W rms output power.
4mm speaker terminals, headphones can be connected through banana plug to jack adapter cable.
External 5V / 250mA power supply (can be tweaked for optimal sound quality) for all circuits except the Power D/A converters.
115 or 230V mains power input, integrated power supply for power D/A converters only.
Power consumption (depends on volume setting) between 30 watts (lowest volume setting) and 120 watts (highest volume setting).
Firmware updates through USB socket (the PowerDAC - S does not support USB audio, see Connectivity section below)
The digital audio receiver:
S/PDIF Toslink was chosen like with the DA96ETF because it offers lowest data bandwidth and related lowest noise injection. Following graph shows the huge reduction in data bandwidth (and related noise injection) for USB and Toslink.
Noise injection doesn't stop at the DAC digital input. Therefore we developed DAPI (Digital Audio Parallel Interface). It further reduces data bandwidth and related noise injection into the D/A converter (graph). DAPI is an internal digital audio interface that can only be used with discrete D/A converters that have parallel data input.
But Toslink low data bandwidth plus the optical interlinks in combination with biphase mark encoded S/PDIF results in high jitter. Even when using a zero jitter Toslink source, jitter at the Toslink input on the DAC will be relatively high.
Conventional S/PDIF receivers (used in most DACs) recover the clocks from the incoming S/PDIF signal, so if there is jitter on the input, there will be some jitter on the recovered timing signals as well. Therefore it is important to minimise S/PDIF source & interlink jitter when using conventional S/PDIF receivers. The larger the data bandwidth, the lower the jitter (biphase mark encoding). So we have a paradox, we need low data bandwidth for minimum noise injection but this will also lead to increased jitter.
We fixed this problem by scraping only the data from S/PDIF using a custom software-based S/PDIF receiver. The D/A converter timing signal is derived from a local, independent low jitter clock. This way source jitter is no longer a problem either.
But now both, source and D/A converter clocks are no longer in perfect sync all of the time as we use two different free running clocks (two different time domains). This problem is fixed by smart software algorithms and a large RAM buffer. The RAM buffer temporarily stores audio data until it is needed.
We now have a digital S/PDIF audio receiver that effectively attenuates source noise and jitter, eliminates ground loops (Toslink optical). We now have the freedom to choose whatever Toslink source we like.
S/PDIF input selection is no longer a problem either, because S/PDIF signal quality only has to be good enough to prevent data errors.
The PowerDACs work with all standard S/PDIF sources including ElectroTOS configured for S/PDIF (U192ETL / UPL96ETL) . Other sources like USB and coax can be connected through converters, one can choose one of many existing converters or use our USB to Toslink or coax to Toslink converter.
Why not integrate all digital audio interfaces in the PowerDAC?
We need minimum noise injection into the DAC circuits, this is the only way to maintain reasonably low noise and EMI levels inside the DAC. Unused USB input (still connected to the source) would bypass the perfect galvanic insulation of Toslink, creating a ground loop and injecting ground loop noise. The connection would also inject large bandwidth noise into the DAC circuits. Same goes for coax (large bandwidth) and I2S (large bandwidth). So in order to make absolutely sure no such degrading could ever occur we use Toslink input only. When using converters it is advised to keep these away from the PowerDAC. If we would place such converter on top of the DAC we would bypass the perfect galvanic insulation (coupling capacitance) and still end up with noise related degrading.
The master clock:
We use independent low jitter master clocks for generating the critical PowerDAC sample timing signal (DAPI only requires one single clock signal because it s a parallel data interface).
The Fractal Power D/A converter:
The PowerDACs use a similar discrete D/A converter and same DAPI interface as the DA96ETF. The D/A converter is just scaled up, higher supply voltage, higher bit currents, lower resistor values, lower output impedance.
These converters basically attenuate the supply voltage by selection of resistor combinations. Imagine a battery connected to a stepped volume attenuator, when rotating the switch (changing resistor values) we can accurately attenuate the battery voltage between say 1 micro volt and 6 volts. The practical version uses high speed electronic switches instead of a rotary switch and we can now reproduce audio signals by changing the electronic switch setting with every new sample. So for CD quality we change the switch settings 44,100 times / second. The switches and resistors are arranged such way that output impedance remains constant regardless of output voltage. This way we no longer need active linear circuits like (discrete) OP-amps, I/V converters, buffers, FETs, transistors, tubes and so on. We only need a balanced power supply, electronic switches and resistors, that's all.
If we want to drive headphones or speakers we need certain output impedance and this has to be realised using resistors only. This translates to low resistor values, high currents and related high power dissipation in the resistors. The efficiency is lower compared to conventional analogue circuits, only 10% of the power can be converted into an output signal, the rest has to be dissipated in the resistors. For 6W output we have to dissipate at least 60watts in the resistors for each channel. Therefore the PowerDAC concept is most suitable for modest output power.
So we have following:
Digital audio source -> Toslink -> DAPI receiver -> Power D/A converter (switches and resistors only) -> headphones / speaker.
The volume control:
RCA / XLR interlinks and related unavoidable degrading are gone, so how do we control the volume? We use a combination of supply voltage programming (3dB steps) and bit shifting (6dB steps). This way we can realise lossless volume control with 30dB range (PowerDAC - R) or 39dB range (PowerDAC - S). This integrated volume control enables us to directly driver headphones, speakers and power amps / buffers without the need for a pre amp. Keep in mind that perfect pre amps don't exist as there are no perfect parts, so adding a pre amp is going to add noise , distortion and signal degrading no matter what. By simply eliminating the pre amp we also eliminate all related degrading.
Additional questions about PowerDACs:
If you have additional questions about our PowerDAC R&D you can mail your question to: email@example.com and we will try to answer in this blog.
PowerDACs and headphones.
PowerDACs make it possible to directly drive headphones without any analogue amplifier or buffer in the signal path. In other words the headphones are directly driven by a balanced power supply, a number of digital on / off switches and a resistor array. This eliminates all distortion and degrading that is typically introduced by conventional headphones amplifiers. With this distortion and degrading completely removed it is obvious that we end up with a much cleaner signal that lets one listen much deeper into the recording and offers a more realistic representation of the recorded music.
The PowerDAC - R (RCA) is intended for directly driving power amplifiers, eliminating the degrading that is typically introduced by an analogue pre amplifier and the extra set of interlinks. That's why it has RCA - only outputs. However, its low output impedance and 1.4V rms max. output amplitude make it also suitable to directly drive most headphones with impedance between 32 Ohms and up (tested / verified). Depending on headphones sensitivity and impedance there is a possibility that it doesn't play loud enough with the PowerDAC - R.
In that case we can drive it with the PowerDAC - S. The PowerDAC - S is intended for directly driving sensitive (90dB and up) speakers, that's why it has speaker output terminals. However, it can also be used to drive "difficult" headphones with impedance < 32 Ohms or low sensitivity. So depending on headphones properties we can either use the PowerDAC - R or the PowerDAC - S to drive it.
SPL depends on signal amplitude (volts rms), load impedance (2 ... 600 Ohm) and headphones sensitivity. All these parameters must be taken into account to determine if SPL will be high enough with the PowerDAC - R or if it is better to use the PowerDAC - S.
The required RCA / speaker terminal to headphones adapter cable make it possible to use headphones with common ground or separated ground (2 separate connections for each channel) for better channel separation. Over time most jack sockets will oxidise and or spring tension will reduce, creating poor contact (crackling sound). An adapter cable can be replaced easily at minimum cost, replacing a jack socket in an audio product is a whole different matter that could become quite costly.
The PowerDAC integrated volume control.
Volume controls always introduce some degrading and often the output impedance varies with volume setting (potentiometer, stepped attenuator) so the sound quality is likely to change with volume setting. TVC (Transformer Volume Control) introduces band limiting and distortion (it's response is not perfectly linear). Same applies for LDR-based volume controls (comparable with potentiometer).
Software-based volume controls introduce rounding errors and change the audio data in such way that audible degrading occurs. All of the DAC noise is presented to the power amplifier regardless of volume setting (poor SN ratio).
The PowerDACs no longer have any analogue signal path (all digital system) so we had to come up with a different solution here.
It is possible to get 6dB coarse attenuation steps by shifting the audio data one bit with regard to the D/A converter. This way the audio data remains unchanged (bit-perfect) but we simply use less sensitive bits in the D/A converter. With every bit shift we get another 6dB attenuation. No DSP, no rounding, no degrading.
But with every bit shift we loose a LSB (24 bits) so we limited bit shifts to 4, this leaves 20 bits resolution at highest attenuation and that's 2 bits more than any human auditory system can resolve. So we could call it lossless as it is not likely to cause any audible change in sound quality.
The smaller attenuation steps cannot be realised in the digital domain (rounding errors, DSP). But since we have a discrete D/A converter we can simply change the supply voltage a little bit to get smaller 3dB attenuation steps for example. This is also lossless as small supply voltage changes (3dB) do not change the D/A converter output signal quality at all.
By combining both, bit shifting and D/A converter supply voltage programming we have a lossless volume control. Because no attenuator circuit is used, the D/A converter output impedance stays constant, regardless of volume control setting and there is no change in sound quality. The constant low output impedance can perfectly drive power amps (no pre amp required). The PowerDAC - R volume control has a range of 0dB ... -30dB.
The PowerDAC - S runs on a higher supply voltage (up to 18V), so we can add some extra attenuation steps here (0dB ... -39dB).