 What does upsampling do to the sound quality and how come? And more to the point here, does it pay to buy the Olex Serious G2.1 to do this job? The Serious G2.1 is one of the four products in the G2.1 series. The others being the Aries G2.1 network player, the Viga G.1 streaming DAC and the Leo GX1 master reference clock. As you might know I used the Aries G2 without the .1 edition streamer in my setup 1. The G2.1 series, including the Serious G2.1 reviewed here, has a further refined cabinet on which later on more. Let's first see where the series fits into the stereo setup. The series is to be connected to a digital input on your stereo, like in this example to the digital to analog converter. That connection can be made over SPDIF, TOSLink, AES-EBU or USB. Digital sources are then connected to the inputs of the series, for instance a network player, a CD player and TV. If you own other Olex G2 equipment they can be connected over Olex proprietary Lightning link that uses an HDMI cable for bi-directionally transport of the digital signal, the central clock signal and the control signals. In short the series is connected in between the digital source and the digital to analog converter. The series has a matte black aluminium housing that measures 340 x 320 x 96 mm and weighs 9.5 kilos. On the front we see a standby button, the full color display and a rotary encoder that functions as a volume control or, after pressing, let you scroll through the menus and change settings. On the rear we see the power switch next to a fuse holder and an IEC mains entry. The internet socket is for updates over the web only. Then we get the inputs, AES-EBU on XLR, SPDIF on RCA, Optical on TOSLink and USB Audio Class 2 on USB B. Then the output socket starting with USB Audio Class 2, a USB A output that is highly compatible and a USB A output with galvanic separation. Then the AES-3 outputs, Optical on TOSLink, SPDIF on RCA and AES-EBU on XLR. Lower on the panel the lightning link input and the lightning link output. Both use HDMI cables to connect to other Aurelix G2 and G2.1 products only. They are not compatible with either HDMI inputs or outputs on video gear or I2S on audio gear. Time to show what changed on the DOT1 version. Let's start with the double enclosure. The outside is machined from a slab of aluminium and is opened for servicing by removing the bottom plate. When you done that, you see the inner copper enclosure. This provides improved shielding against electromagnetic and radio frequency interference. The second improvement is the added aluminium base plate with four bonded food assemblies that use springs to mechanically decouple the enclosure from the surface it stands on. Back to the inside. When the copper bottom plate is removed, we see the copper cage. There are three linear power supplies that receive the AC voltage through a mains filter. It looks like the input section on the lower circuit board uses the first power supply. The second one seems to power the Proteus G2 co-processing platform that does the upsampling and the Tesla G1 engine we find in all Aurelix products, handling the system control and data transport. The Proteus G2 is based on the Sionic Arctic 7 FPGA. This powerful field programmable gateway is programmed by bespoke Aurelix software to do the up and down sampling, equalizing and speaker placement correction. After the series is connected, on which later on more, settings like input, output and sampling frequencies need to be set. All the settings are done using the rotary encoder and screen on the series. There is no HTML menu or app like with the Aries G2. Pressing the encoder enters the menu mode and turning selects the menu items. The output sampling frequency can be set globally or through each and every input sampling frequency individually. You can convert all incoming PCM and DSD to for instance 352 kHz PCM or DSD 512. Or all input signals to 96 kHz if that's the highest sampling rate your active speakers support. It is a good thing this only has to be set once since the menu options show up in small characters on the display. Then there are four types of filters to choose from. One for those that listen with their oscilloscope, two others in my opinion pointless filters and one very good one called SMOOTH for those that listen with their ears. Next to the resampling the series also offers an 8 band parametric equalizer to do simple room correction. You can also compensate for less optimal speaker placement by entering the distances from the speakers to your listening position and the differences in loudness. This will then compensate for that in time and volume. The series comes without the remote control but like with the RSG2 you can learn it infrared codes from any remote control you might have. Basic digital to analog conversion is a three step process. The input signal is converted to a signal the converter understands, usually I2S. In the converter plots analog voltages that match the values in the digital signal. For more in-depth info watch my video what defines the quality of a DAC. The analog result needs to be filtered at half the sampling frequency to avoid aliasing. This filter has to be extremely steep. For CD quality the filter of at least in theory 96 dB per whole tone, a single white key on the piano, is needed. Quality filter slopes are expressed per octave so these filters have to be about 7 times steeper than 96 dB per octave. Watch my video Q&A is upsampling better. Good sounding analog filters with that specification are very difficult to make at an acceptable price. An affordable way to solve this is to convert the signal to a higher sampling frequency. This way the analog reconstruction filter can be made less steep and thus easier to make. It relocates the steep 20 kHz filtering to the upsampling algorithm where an acceptable filter is cheaper to make. Acceptable to the mainstream market at least. Audio files appreciate a higher sound quality. Digital sources like CD players and digital to analog converters that use ladder converters offer not only upsampling but also bit depth of 24 bit. The analog output stage is almost never capable of a signal to noise ratio equal to 20 bit, which by the way suffices. Especially discrete ladder converters like the ones by AudioGD, Denofrips and HoloAudio I reviewed over the past year or so perform remarkably good. A technique that was introduced later is to convert the digital signal to a pulse density modulation signal, meaning a 1 to 3 bit signal at a far higher sampling frequency than used for CD. This technique is also used for SACD that uses 64 x 44.1 kHz sampling, 2822.4 kHz. That frequency is so high that a very mild filter would suffice. This gets solved using noise shaping, an algorithm that shifts the resulting noise out of the audio band to higher frequencies. Now filtering at 70 to 90 kHz would be fine although most players and DACs use 40 to 50 kHz filtering. This nowadays is the most popular approach since it is cost effective. Unfortunately DAC chips like all integrated circuits need large volumes to become viable, meaning that they are aimed at the mass market and not to the audiophile applications. Therefore the computational power of these chips is rather limited and thus are the algorithms used for the conversion from 16 or 24 bit to 1 bit. Some DAC chips offer an option to use an external processor for the upsampling. Since developing very good sounding upsampling algorithms is extremely time consuming, it costs a lot of money while economy of scale does not exist in the audiophile market so the many years of development need to be earned back with a limited amount of products. That places a high financial burden on each and every product. It is my experience that both ladder DACs and low bit DACs can perform extremely well. There might have been points in time where products from one category outclass the other only to be overtaken by products in the other category the next year. Functionally the series does the same as the processing found in any upsampling ladder DAC or low bit DAC. The only difference is that it according to Aurelix is done a lot better. How did they achieve this? It starts with the mechanical construction that prevents vibrations in the cabinet. This is important since clock crystals are rather sensitive to vibration or as it is called by technicians, crystals are microphonic. Then the copper in a housing that stops electromagnetic and radio frequency interference. System control is done by the Tesla G1 engine that is also found in all current Aurelix products. The Proteus G2 engine's only job is to do the signal processing, upsampling of multi-bit to low bit conversion or vice versa, equalizing and loudspeaker placement correction. All three can be turned on or off individually. To give you an idea of the computational power the Proteus G2 offers. The same side link FPGA is used for processing 4K high definition video and even when bought in bulk costs over 200 euros each. That must be more than sufficient even for 384 kHz 32 bit audio or DSD 512. In the end it all comes down to the code that is flashed into the FPGA and the definite way of judging that is to listen to it. Equipment of this level can only be tested in my setup one. The Air Acoustic AX520 amplifier that powers the AudioSphysic Scorpio loudspeakers over AudioQuest Robinhood Zero loudspeaker cables. The DA converter is the Mitek Brooklyn powered by the Lexicon Acoustic Syntax Icon Power supply and connected to the amp over Grim Audio SQM balanced interlinks. The streamer is the Aurelix Airis G2 connected to the DAC over AudioQuest Diamond USB A-to-B cable. The Serious G2.1 was inserted in between the Airis G2 streamer and the Mitek DAC. Between the resampler and the DAC I used the AudioQuest Diamond USB cable. As said you can switch off upsampling and when used the equalizer and speaker placement but not the entire processing where the input signal is converted into a high frequency multi-bit signal to remove jitter, after which it is resampled to the required output sampling frequency in PCM or DSD. I compared the signal with the resampler in non-oversampling mode in and out of circuit but heard no difference. So I programmed the upsampling on off function to an obsolete remote control for easy comparing. Aurelix advises to use DSD output for DACs that use a one bit or low bit converter. So all DACs that are not R2R ladder converters. I have tried both modes on the Mitek Brooklyn and did agree, DSD is the best option for the Mitek. When switched on the stereo image opened up a bit more with more spaciousness. Voices became somewhat cleaner and transitions became slightly faster, which by the way probably is the wrong description. I rather think this is due to less time smearing. Inserting the Mini DSP used as direct processor in between the streamer and the resampler limited the feed to the series to 96 kHz and I was curious what the upsampling would bring here. Well it did correct for the same acoustical problems with further sound improvements remain the same. Overall there was an improvement in sound quality but shelling out almost 7 grand for this improvement I don't know. I would rather spend that money on a higher class DAC, like the Denafrips Terminator Plus that I didn't return to Vinshine Audio yet. So I had the opportunity to try the series on this DAC too and I am very glad I did. So instead of the Mitek DAC with the Sintek's power supply, the Terminator Plus was inserted and I started without the Mini DSP in the chain. If you haven't seen my review of the Terminator Plus, now is a good time to do so. The link is in the upper right corner and in the comments below this video on YouTube. In short at about €6500 it is amongst the best DACs I had the pleasure of reviewing. So what could the series add to that? To my big surprise the series upsampling opened up the stereo image considerable as if a fail was taken away. The spatial representation was larger and even more important, far more credible. The same for the stereo imaging. Instruments and voices took their own place in space better with a clear and natural separation between them. Transients like the snare or glockenspiel came better placed in time or put differently they are rendered with less time smearing, thus being more accurate. This too became even more natural and it happened even with albums like 60 Stop 100. Brown Eyed Girl by Van Morrison, Good Vibrations by The Beach Boys and other tracks from that time, all transcripted from Analog Tape of course, contain so much more information apparently. But also 80 stuff like Simple Mind's Ballad of the Street and Queens Made in Heaven see my video My Favourite Rock Music appeared to be played in large halls with a listener within the reverb radius, the space where the direct sound is louder than the reverb. So no details get lost and just make sure this is an improvement on an already very good sounding DAC. Of course I used the PCM setting for output since the Terminator Plus is an R2R ladder converter. I tried the DSD setting on the series but that resulted in a somewhat thin sound. Said to PCM it unveiled so much more information that apparently is a present in most if not all albums. Apparently the series G2.1 needs a high quality DAC to render the upsampled music to perfection. I wonder if there are high end DACs that do this level of assembling their cells. And if I ever find out, I'll let you know. But as far as the series is concerned, it surprised me twice. The results from the MiTech Brooklyn weren't convincing but the results on the denofrips were far beyond my expectations. The MSRP in Europe is €6699 including VAT and that's a lot of money. For those that have spent equal amounts of more on a DAC amp and speakers, I think it will be almost small change when they hear the improvement the series offers. I hope I managed to explain that this is no voodoo but pure signal processing at the highest level to achieve the best upsampling so to avoid the artefacts reconstruction filters might cause. Which brings me to the end of this video. I love to see you back next Friday at 5pm central european time in a new video. If you don't want to miss that, subscribe to this channel or follow me on the social media and you will be informed when new videos are out. If you liked this video, give it a thumbs up. Many thanks to those viewers that support this channel financially. It keeps me independent and thus trustworthy. Many thanks to those viewers that support this channel financially. It keeps me independent and thus trustworthy. If you like to support my work too, the links are in the comments below this video on YouTube. I am Hans Beekhuyzen, thank you for watching and see you in the next show or on theHBproject.com. Whatever you do, enjoy the music.