Post by MartinT on Jun 27, 2020 11:36:39 GMT
Situation report: one year ago
The basic components of my system were all there about a year ago. I’d decided to sell my CD/SACD replay system, the turntable and the record collection in order to reinvest in streaming. Cue lots of shock and some doubts from friends, although it all made perfect sense to me. I had mentally moved on from CDs and records and there was no turning back. Although I had never heard a streaming system quite rival a file-based player, I knew it could be done so I was determined to research it and make it work. Meanwhile, my source components and the preamp were off to new owners.
I was left with a good DAC as the core of my system – it’s not perfect, but a building block that could be improved upon by dint of its open construction, and which would not be a bottleneck in the system for a long time. Following it, a simple buffer, the power amp and speakers both of which haven’t changed for a long time.
Up front, the start of a Mk.II streamer (strictly speaking, a Mk.III as I had started with a Logitech Touch) based on the Asus Tinker Board S, a much better single board computer than the Pi for streaming.
What I heard at that point was a huge wealth of detail retrieval, a feature of the Sabre ES9038 dual DACs, but rather thrown at you. Big slam and dynamics, good bass, limited soundstage with a rather two dimensional presentation, and very under-developed depth, ambience, note decay and sense of the recording space. Instruments sounded accurate but didn’t have much ‘shape’.
In short, it was a pretty good hi-fi system but fell well short of communication. Now I needed to make it into a good music system.
Introduction
If there is one parameter I have addressed that encompasses all of the below modifications, it’s noise. Just as with analogue music, but with different outcomes, noise is the digital killer. Noise pollutes digital systems far more than analogue because almost every digital component has 1) a microprocessor, 2) a crystal clock and 3) lots of logic switching circuitry. All of this creates noise and injects it back into the signal and power lines, corrupting the very datastream that carries music to the DAC.
This is over and above all the external noise trying to get into your system manifesting as mains-borne, Electromagnetic Interference (EMI) and radio-frequency interference (RFI).
“Digits are digits, aren’t they?”
What are digital reclockers trying to achieve? Well, they are designed to tighten up the timing of the signal in order to reduce jitter (time domain noise) and reduce phase noise (frequency domain noise) from entering the sensitive DAC circuit, thereby causing conversion issues and polluting the resulting analogue signal.
Just because the data is digital does not mean that nothing can happen to the datastream and it will always sound the same. Far from it, the embedded noise in the stream is critical to sound quality, affecting the performance of the DAC. So we are essentially taking the dirty digital stream and cleaning it by clocking it out against a reference clock.
My first reclocker was at the most critical position just before the DAC. A Mutec MC-3+ USB was inserted into the path, taking the relatively noisy USB output from the streamer and reclocking out the data in S/PDIF format via AES to the DAC using an internal crystal clock reference.
The second reclocker is an EtherREGEN, which inserts earlier into the chain between my router and the streamer. This does something similar to the Mutec but works on the ethernet data, reclocking it and thereby stripping it of a lot of noise. It, too, uses an internal crystal clock reference.
The combined effect of the two reclockers is enhanced over using either one alone. The EtherREGEN makes the streamer perform more cleanly while the MC-3 keeps the DAC conversion as faithful to the data as possible. What happens is a dramatic drop in the noise floor, resulting in finer detail, bigger soundstage and metronomic timing with rhythmic music. The music becomes more vivid and exciting. However, there is still more to come from these components.
“Is that the ticking of a clock I hear”?
Talking of master clocks generally brings furrowed eyebrows and a distant look, or an assumption that you’re discussing studio equipment. To further complicate matters, there are different types of clock: word clocks (operating at a multiple of the sampling frequency) and reference clocks (providing a master reference for computation against all data formats). The former are usually used in the studio to make sure that all the equipment is synchronised to the same master clock and varies according to the digital data format. The latter, almost always a 10MHz reference clock, is used with some studio and hi-fi equipment to provide a very precise, timing accurate and low drift external reference for equipment to use, replacing their internal clock.
There are quite a few master clocks on the market and all the hi-fi ones are priced well into four figures or more. The principle is mostly the same: there is an internal crystal reference, well selected for low jitter and low phase noise. This crystal is often located in a small oven to keep the temperature constant and operate it at the precise optimum working point. Sometimes there is a double-oven for even more temperature stability and sometimes the clock is synchronised via GPS signals to very precise rubidium or caesium reference clocks in the satellites. Generally, the better the clock the more likely it is to have jitter measured in femtoseconds, an extraordinarily microscopic time measurement. But it is phase noise that appears to be critical to their performance and here the manufacturers play fast and loose with specifications making it virtually impossible to directly compare them.
Firstly, though, remember that the DAC is at the core of a digital system? My LKS MH-DA004 is a typical example of a DAC claiming to have a femtoclock, in this case the almost ubiquitous (and increasingly notorious) Crystek 575. It must be good, right? Fantastic on specifications but there is something about this oscillator that falls far short of great performance. I had it in my mind to replace it with a better Crystek (the 950X) but was persuaded to do the job properly and have a Coherent clock installed plus some power supply rework. That changed the somewhat edgy attack and in-your-face presentation of the LKS into a much smoother delivery, with better detail and some depth. The DAC was now ready for greatness but my system wasn’t up to that performance level just yet.
Now it was time to turn my attention to a master clock to feed both the EtherREGEN and MC-3+ USB, replacing both their internal clocks. I had my eye on a Mutec or Cybershaft but I decided to experiment and have a little harmless fun first. Rather than jumping straight in, I bought a clock that is well known and used by ham radio operators. Its name comes from a Chinese amateur radio operator with the callsign BG7TBL. It’s a ridiculously inexpensive OCXO (Oven Controlled Crystal Oscillator) using recovered crystals from Chinese 4G telecoms masts. These oscillators are specified for precision and low drift and perform pretty well for audio purposes. The price was approximately 1/20th that of a typical hi-fi master clock. You have to do some work, such as provide a decent 12V power supply, isolate it with footers, convert it to 75 ohm if your equipment needs it and purchase some decent BNC clock cables. In my case, a pair of 0.5m Canare LV-61S 75 ohm cables were inexpensive and specified for the job as word clock and master clock cables.
The effect of hooking up the OCXO was way above anything I had expected and quite ridiculously good for the low cost investment. This was an experiment, remember? How could this ham radio oscillator with a recycled clock module outperform the internal clocks of the EtherREGEN and MC-3 by such an extraordinary amount? Yet it does. Of all the upgrades I’ve made this last year, this is the one, and it was one of the cheapest of them all. It just needed a bit of work and application to get going.
How could this be since the OCXO was measured by Coherent Systems as having a jitter performance of around 30ps (picoseconds) whereas the Crystek 575, at least, has a claimed jitter measurable in femtoseconds. Don’t believe the specs and don’t fall for jitter being the sole arbiter of quality. Phase noise is possibly even more important. Regardless, this inexpensive clock outperforms the internal clocks in both reclockers by some margin.
What I hear is a massively developed soundstage in both width and depth. Everything is layered and precisely positioned in space. Instruments and sounds have ‘shape’ and are not two-dimensional. Delivery is fluid and easy on the ear, yet attack and slam are even more apparent. Note decay is so obvious that it draws your attention, from reverb on drum beats to shimmering cymbals to the natural sound of a human voice. Bass extension, tightness and texture are remarkably improved, such that the system sounds like it can excavate the deepest air movements from recordings. Most of all, it makes glorious, vivid, infectious music.
“Where does the current flow”?
If you think cleaning the digital datastream is a contentious subject, try discussing grounding boxes out in the open. Do it on Facebook and you’re likely to be called names within the hour. So, what are these strange boxes and how can current ‘flow’ with just a single terminal?
Try not to think of it that way. Consider a grounding box to be a miniature piece of mother earth. In the same way as we can sink current to ground, a grounding box will sink noise and convert it to heat. Tiny amounts of heat. The idea being that digital circuits, especially, carry a lot of noise on their ground plane – it’s anything but a noise-free environment with all that clocking circuitry around. The ground plane, of course, should be fixed to 0V but it’s a long way back to the mains earth, even if it is directly connected. A local grounding box will represent a lower impedance sink for high frequencies, sponging away noise from the ground plane.
What I already discovered some time ago is that grounding components together on a common terminal is counter-productive. So I ground, using one box per device, all of my digital components from the router to the DAC. I started by using SGS boxes and they still do good service on my router and Mutec. Elsewhere I use the neat Black Ravioli Eflos grounding boxes with different connectors for each ground plane (not the chassis).
The sum effect of having all of my digital components running with grounding boxes is a big drop in the noise floor (after all, that is what they are designed for), revealing more of the soundstage, ambience and fine detail in the music.
“Any properly designed power supply...”
I must have read the phrase “any properly designed power supply…” more often than anything else when justifying a component’s apparent imperviousness to power issues. According to the proponents, any component with a well-designed power supply cannot be affected by mains-borne noise in any way. I’ll just say this: PSRR.
PSRR stands for Power Supply Rejection Ratio. How well does the power supply reject incoming fluctuations and noise? This should include noise up into the Gigahertz spectrum as it can affect digital circuits quite badly. So, a power supply could be considered ‘stiff’ (i.e. doesn’t fluctuate much under varying load) with a high PSRR at low frequencies down to DC, but be completely rubbish at stopping very high frequency noise from getting through (low PSRR). The component’s power supply may be an open window at high frequencies, putting at risk the influence of noise on its sensitive circuits.
The reality is that PSRR is a ratio and even a very good power supply cannot have an infinite rejection ratio. Some noise will always get through. The question then is: how well do the following circuits handle it?
Earlier this year, I mentioned that I sold equipment for reinvestment. One of the things I had my eye on was another mains power regenerator. My reasoning was to separate the huge varying power demands of the power amp from the largely steady-state power demands of my source electronics. I bought a PS Audio P3 from their current line-up for just this purpose, supplying clean and distortion-free power to the source components while leaving the P10 on power amp duty. The P3 is optimised for very low distortion, the P10 for stiff power delivery up to 1.5kW.
My Coherent QP-1 supercap power supply was originally bought for powering a 5V Pi and 12V DAC. It was then re-purposed for twin 5V to power the Asus and Mutec. Recently I had it upgraded again to Tony’s latest specification, providing extraordinarily low noise performance and high current demand capability. The Asus guzzles over 2A while booting up, settling down to 0.9A when playing music. The QP-1 handles it, and the Mutec, with ease while providing a very low output impedance and noise specification.
I also have a couple of LPSUs (linear power supplies) to power the router, EtherREGEN and master clock. They are branded ‘Weiliang’ and ‘BRZHiFi’ respectively but are identical decently made Chinese PSUs at a price that you couldn’t match by building your own. I had Coherent breath on the one powering the EtherREGEN and master clock and its performance is now very good, with a substantial improvement in noise output – just what these sensitive components need.
Finally, I have paid attention to the DC cables and have ditched the stock ones (GX16-2 to 2.1mm connectors) in favour of two MCRU UP-OCC DC cables (router and clock) and one Coherent DC cable (EtherREGEN). All of these add to the clarity of the music, removing a little ‘fuzz’ from the proceedings. Somewhat surprisingly, each one makes a difference despite powering quite different components.
The last thing I attended to in the power department are fuses. Yet another contentious issue. Yes, they do make a considerable difference and no, I’m not going to attempt to answer to those who claim not to hear any and will certainly not rise to those who tell me they cannot make a difference but have never tried them!
Most of my fuses are specialist already and the ones I changed in the last year are the two powering the regenerators, both from SR Red to SR Orange. Then I switched the standard fuses in the two LPSUs from stock to AMR Gold and then most recently to SR Red. Even the switch from AMR Gold to SR Red is clearly audible so a very worthwhile improvement in clarity and vividness.
So, what does all this power treatment bring to the table? Clarity, nuance, dynamic clout, bass heft and scale. There is also the issue of the sound remaining the same at any time of the day or the weekend, riding over variations in power demand and subsequent waveform distortion.
Supports? You mean spikes?”
All electronics are sensitive to vibrations, digital components just as much as analogue. It pays to give attention to how you support the components and a system-wide approach pays dividends. It’s not so much the big vibrations giving feedback or howl-round you need to worry about (unless you run a turntable), but the more subtle microphonics that penetrate inside components to the circuit boards, resistors, capacitors and other electronics that can pollute the tiny signals on their journey to your speakers.
Many years ago, I was introduced to Black Ravioli products by David of MCRU, who markets a range of feet for hi-fi components. They don’t just absorb vibrations like sorbothane, they have a damping characteristic as well as converting vibrations into heat (as with all absorbent materials). When placed hard against the component underside, not under the feet, they perform the dual task of isolating and damping. They are quite remarkable and I use the Pads and Big Pads under everything, including the grounding boxes.
I also tend to use granite mats and coasters on top of every component, which helps with the damping and operation of the BR feet. I buy them as sets and apply black sticky-back felt to the underside. I used a variety of different weights on top but recently came across some nice heavy door stoppers on Amazon. These are German Schramm stoppers weighing 1.2kg each and perfect for the job. I use 8 of them on each mat or coaster.
Just about a year ago, as one of my big sales, I negotiated with David at MCRU to have a custom pair of Townshend Podiums made for the Ushers. These are BIG pieces of heavy steel with adjustable suspension modules at each corner. My speakers are huge and weigh about 160kg each. The story of how I got them onto the Podiums in the right position, on my own, is one that is too colourful to repeat here. Let’s just say that I was kicking myself several times at my obstinacy to get the job done and nearly ended up stranded underneath one speaker or in hospital.
The Townshend’s operation is counter-intuitive. Why ‘float’ speakers when previous best practice was to rigidly spike them to the floor? The reality is that they work spectacularly well. Room loading and unpleasant bass resonances go way down and bass extension and texture are much more evident. Since that moment, decoupling has become my mantra. There are no spikes in my system nor will there ever be again. Decoupling everything from the router to the speakers has been my route to clarity and quite phenomenal bass power and extension.
Supports are a key to getting the most from your components, reducing microphonics and disconnecting them from the room. Big gains are in clarity (again) by reducing the smearing of notes, substantial reduction in room modes and loading by decoupling the speakers.
“So where are we, Captain?”
That brings me to now, one year down the line. What has changed are a myriad of minor changes and one additional major component: the P3 regenerator. None of the core components have changed and yet everything has changed!
In short, it’s a much more musically communicative system now. In a review by Chris Martens in The Absolute Sound back in 2008, I read that my Usher Be-20 speakers were “musical comprehension machines”. This review, read while I was on holiday in Houston, triggered a series of events which led to me buying the speakers. Later, another review in 2014 by Alan Sircom said “this speaker gives away few concessions when put against far more expensive loudspeaker designs.” By then, I already owned them and could only dream of what was to come years later.
All this is a way of saying that I still believe that, as the DAC is the core of my music system, so the speakers are responsible for taking this precious music signal and delivering the best realisation that they are capable of. If they keep up with system improvements, and the Ushers show no signs yet of having hit a performance ceiling, then they just give me more.
That ‘more’ is more musical communication. From fully shaped instruments and voices, hanging in a three-dimensional space, not bloated but focussed, to the intensity of transient detail of anything from tambourine to bass drums, I have so much more insight than before. Note decay, the shimmer of cymbals to the reverb of snare drum to that feeling of being in a big auditorium on the best classical recordings, are all improved. Orchestral trumpets that sound layered at the back of the hall, vocalists coming forward from the mix to being right there in front of me. Rhythmic intensity where the beat is an important part of the music. Air and space where more ambient material is being played, enveloping me in sound.
The system has been transformed and my objective of greater musical insight has been realised more than I had thought possible with the same main system components. Was it worth selling my gear and collection in order to reinvest in streaming? I’m absolutely 100% happy with the decision. I will take great care not to lose this level of performance and will always ensure a ‘rollback’ path before making any future changes.
Acknowledgements
Tony Sallis (Coherent Systems)
David Brook (MCRU)
John for encouraging me ever forwards
Several friends who gave helpful comments and feedback pre-lockdown
The basic components of my system were all there about a year ago. I’d decided to sell my CD/SACD replay system, the turntable and the record collection in order to reinvest in streaming. Cue lots of shock and some doubts from friends, although it all made perfect sense to me. I had mentally moved on from CDs and records and there was no turning back. Although I had never heard a streaming system quite rival a file-based player, I knew it could be done so I was determined to research it and make it work. Meanwhile, my source components and the preamp were off to new owners.
I was left with a good DAC as the core of my system – it’s not perfect, but a building block that could be improved upon by dint of its open construction, and which would not be a bottleneck in the system for a long time. Following it, a simple buffer, the power amp and speakers both of which haven’t changed for a long time.
Up front, the start of a Mk.II streamer (strictly speaking, a Mk.III as I had started with a Logitech Touch) based on the Asus Tinker Board S, a much better single board computer than the Pi for streaming.
What I heard at that point was a huge wealth of detail retrieval, a feature of the Sabre ES9038 dual DACs, but rather thrown at you. Big slam and dynamics, good bass, limited soundstage with a rather two dimensional presentation, and very under-developed depth, ambience, note decay and sense of the recording space. Instruments sounded accurate but didn’t have much ‘shape’.
In short, it was a pretty good hi-fi system but fell well short of communication. Now I needed to make it into a good music system.
Introduction
If there is one parameter I have addressed that encompasses all of the below modifications, it’s noise. Just as with analogue music, but with different outcomes, noise is the digital killer. Noise pollutes digital systems far more than analogue because almost every digital component has 1) a microprocessor, 2) a crystal clock and 3) lots of logic switching circuitry. All of this creates noise and injects it back into the signal and power lines, corrupting the very datastream that carries music to the DAC.
This is over and above all the external noise trying to get into your system manifesting as mains-borne, Electromagnetic Interference (EMI) and radio-frequency interference (RFI).
“Digits are digits, aren’t they?”
What are digital reclockers trying to achieve? Well, they are designed to tighten up the timing of the signal in order to reduce jitter (time domain noise) and reduce phase noise (frequency domain noise) from entering the sensitive DAC circuit, thereby causing conversion issues and polluting the resulting analogue signal.
Just because the data is digital does not mean that nothing can happen to the datastream and it will always sound the same. Far from it, the embedded noise in the stream is critical to sound quality, affecting the performance of the DAC. So we are essentially taking the dirty digital stream and cleaning it by clocking it out against a reference clock.
My first reclocker was at the most critical position just before the DAC. A Mutec MC-3+ USB was inserted into the path, taking the relatively noisy USB output from the streamer and reclocking out the data in S/PDIF format via AES to the DAC using an internal crystal clock reference.
The second reclocker is an EtherREGEN, which inserts earlier into the chain between my router and the streamer. This does something similar to the Mutec but works on the ethernet data, reclocking it and thereby stripping it of a lot of noise. It, too, uses an internal crystal clock reference.
The combined effect of the two reclockers is enhanced over using either one alone. The EtherREGEN makes the streamer perform more cleanly while the MC-3 keeps the DAC conversion as faithful to the data as possible. What happens is a dramatic drop in the noise floor, resulting in finer detail, bigger soundstage and metronomic timing with rhythmic music. The music becomes more vivid and exciting. However, there is still more to come from these components.
“Is that the ticking of a clock I hear”?
Talking of master clocks generally brings furrowed eyebrows and a distant look, or an assumption that you’re discussing studio equipment. To further complicate matters, there are different types of clock: word clocks (operating at a multiple of the sampling frequency) and reference clocks (providing a master reference for computation against all data formats). The former are usually used in the studio to make sure that all the equipment is synchronised to the same master clock and varies according to the digital data format. The latter, almost always a 10MHz reference clock, is used with some studio and hi-fi equipment to provide a very precise, timing accurate and low drift external reference for equipment to use, replacing their internal clock.
There are quite a few master clocks on the market and all the hi-fi ones are priced well into four figures or more. The principle is mostly the same: there is an internal crystal reference, well selected for low jitter and low phase noise. This crystal is often located in a small oven to keep the temperature constant and operate it at the precise optimum working point. Sometimes there is a double-oven for even more temperature stability and sometimes the clock is synchronised via GPS signals to very precise rubidium or caesium reference clocks in the satellites. Generally, the better the clock the more likely it is to have jitter measured in femtoseconds, an extraordinarily microscopic time measurement. But it is phase noise that appears to be critical to their performance and here the manufacturers play fast and loose with specifications making it virtually impossible to directly compare them.
Firstly, though, remember that the DAC is at the core of a digital system? My LKS MH-DA004 is a typical example of a DAC claiming to have a femtoclock, in this case the almost ubiquitous (and increasingly notorious) Crystek 575. It must be good, right? Fantastic on specifications but there is something about this oscillator that falls far short of great performance. I had it in my mind to replace it with a better Crystek (the 950X) but was persuaded to do the job properly and have a Coherent clock installed plus some power supply rework. That changed the somewhat edgy attack and in-your-face presentation of the LKS into a much smoother delivery, with better detail and some depth. The DAC was now ready for greatness but my system wasn’t up to that performance level just yet.
Now it was time to turn my attention to a master clock to feed both the EtherREGEN and MC-3+ USB, replacing both their internal clocks. I had my eye on a Mutec or Cybershaft but I decided to experiment and have a little harmless fun first. Rather than jumping straight in, I bought a clock that is well known and used by ham radio operators. Its name comes from a Chinese amateur radio operator with the callsign BG7TBL. It’s a ridiculously inexpensive OCXO (Oven Controlled Crystal Oscillator) using recovered crystals from Chinese 4G telecoms masts. These oscillators are specified for precision and low drift and perform pretty well for audio purposes. The price was approximately 1/20th that of a typical hi-fi master clock. You have to do some work, such as provide a decent 12V power supply, isolate it with footers, convert it to 75 ohm if your equipment needs it and purchase some decent BNC clock cables. In my case, a pair of 0.5m Canare LV-61S 75 ohm cables were inexpensive and specified for the job as word clock and master clock cables.
The effect of hooking up the OCXO was way above anything I had expected and quite ridiculously good for the low cost investment. This was an experiment, remember? How could this ham radio oscillator with a recycled clock module outperform the internal clocks of the EtherREGEN and MC-3 by such an extraordinary amount? Yet it does. Of all the upgrades I’ve made this last year, this is the one, and it was one of the cheapest of them all. It just needed a bit of work and application to get going.
How could this be since the OCXO was measured by Coherent Systems as having a jitter performance of around 30ps (picoseconds) whereas the Crystek 575, at least, has a claimed jitter measurable in femtoseconds. Don’t believe the specs and don’t fall for jitter being the sole arbiter of quality. Phase noise is possibly even more important. Regardless, this inexpensive clock outperforms the internal clocks in both reclockers by some margin.
What I hear is a massively developed soundstage in both width and depth. Everything is layered and precisely positioned in space. Instruments and sounds have ‘shape’ and are not two-dimensional. Delivery is fluid and easy on the ear, yet attack and slam are even more apparent. Note decay is so obvious that it draws your attention, from reverb on drum beats to shimmering cymbals to the natural sound of a human voice. Bass extension, tightness and texture are remarkably improved, such that the system sounds like it can excavate the deepest air movements from recordings. Most of all, it makes glorious, vivid, infectious music.
“Where does the current flow”?
If you think cleaning the digital datastream is a contentious subject, try discussing grounding boxes out in the open. Do it on Facebook and you’re likely to be called names within the hour. So, what are these strange boxes and how can current ‘flow’ with just a single terminal?
Try not to think of it that way. Consider a grounding box to be a miniature piece of mother earth. In the same way as we can sink current to ground, a grounding box will sink noise and convert it to heat. Tiny amounts of heat. The idea being that digital circuits, especially, carry a lot of noise on their ground plane – it’s anything but a noise-free environment with all that clocking circuitry around. The ground plane, of course, should be fixed to 0V but it’s a long way back to the mains earth, even if it is directly connected. A local grounding box will represent a lower impedance sink for high frequencies, sponging away noise from the ground plane.
What I already discovered some time ago is that grounding components together on a common terminal is counter-productive. So I ground, using one box per device, all of my digital components from the router to the DAC. I started by using SGS boxes and they still do good service on my router and Mutec. Elsewhere I use the neat Black Ravioli Eflos grounding boxes with different connectors for each ground plane (not the chassis).
The sum effect of having all of my digital components running with grounding boxes is a big drop in the noise floor (after all, that is what they are designed for), revealing more of the soundstage, ambience and fine detail in the music.
“Any properly designed power supply...”
I must have read the phrase “any properly designed power supply…” more often than anything else when justifying a component’s apparent imperviousness to power issues. According to the proponents, any component with a well-designed power supply cannot be affected by mains-borne noise in any way. I’ll just say this: PSRR.
PSRR stands for Power Supply Rejection Ratio. How well does the power supply reject incoming fluctuations and noise? This should include noise up into the Gigahertz spectrum as it can affect digital circuits quite badly. So, a power supply could be considered ‘stiff’ (i.e. doesn’t fluctuate much under varying load) with a high PSRR at low frequencies down to DC, but be completely rubbish at stopping very high frequency noise from getting through (low PSRR). The component’s power supply may be an open window at high frequencies, putting at risk the influence of noise on its sensitive circuits.
The reality is that PSRR is a ratio and even a very good power supply cannot have an infinite rejection ratio. Some noise will always get through. The question then is: how well do the following circuits handle it?
Earlier this year, I mentioned that I sold equipment for reinvestment. One of the things I had my eye on was another mains power regenerator. My reasoning was to separate the huge varying power demands of the power amp from the largely steady-state power demands of my source electronics. I bought a PS Audio P3 from their current line-up for just this purpose, supplying clean and distortion-free power to the source components while leaving the P10 on power amp duty. The P3 is optimised for very low distortion, the P10 for stiff power delivery up to 1.5kW.
My Coherent QP-1 supercap power supply was originally bought for powering a 5V Pi and 12V DAC. It was then re-purposed for twin 5V to power the Asus and Mutec. Recently I had it upgraded again to Tony’s latest specification, providing extraordinarily low noise performance and high current demand capability. The Asus guzzles over 2A while booting up, settling down to 0.9A when playing music. The QP-1 handles it, and the Mutec, with ease while providing a very low output impedance and noise specification.
I also have a couple of LPSUs (linear power supplies) to power the router, EtherREGEN and master clock. They are branded ‘Weiliang’ and ‘BRZHiFi’ respectively but are identical decently made Chinese PSUs at a price that you couldn’t match by building your own. I had Coherent breath on the one powering the EtherREGEN and master clock and its performance is now very good, with a substantial improvement in noise output – just what these sensitive components need.
Finally, I have paid attention to the DC cables and have ditched the stock ones (GX16-2 to 2.1mm connectors) in favour of two MCRU UP-OCC DC cables (router and clock) and one Coherent DC cable (EtherREGEN). All of these add to the clarity of the music, removing a little ‘fuzz’ from the proceedings. Somewhat surprisingly, each one makes a difference despite powering quite different components.
The last thing I attended to in the power department are fuses. Yet another contentious issue. Yes, they do make a considerable difference and no, I’m not going to attempt to answer to those who claim not to hear any and will certainly not rise to those who tell me they cannot make a difference but have never tried them!
Most of my fuses are specialist already and the ones I changed in the last year are the two powering the regenerators, both from SR Red to SR Orange. Then I switched the standard fuses in the two LPSUs from stock to AMR Gold and then most recently to SR Red. Even the switch from AMR Gold to SR Red is clearly audible so a very worthwhile improvement in clarity and vividness.
So, what does all this power treatment bring to the table? Clarity, nuance, dynamic clout, bass heft and scale. There is also the issue of the sound remaining the same at any time of the day or the weekend, riding over variations in power demand and subsequent waveform distortion.
Supports? You mean spikes?”
All electronics are sensitive to vibrations, digital components just as much as analogue. It pays to give attention to how you support the components and a system-wide approach pays dividends. It’s not so much the big vibrations giving feedback or howl-round you need to worry about (unless you run a turntable), but the more subtle microphonics that penetrate inside components to the circuit boards, resistors, capacitors and other electronics that can pollute the tiny signals on their journey to your speakers.
Many years ago, I was introduced to Black Ravioli products by David of MCRU, who markets a range of feet for hi-fi components. They don’t just absorb vibrations like sorbothane, they have a damping characteristic as well as converting vibrations into heat (as with all absorbent materials). When placed hard against the component underside, not under the feet, they perform the dual task of isolating and damping. They are quite remarkable and I use the Pads and Big Pads under everything, including the grounding boxes.
I also tend to use granite mats and coasters on top of every component, which helps with the damping and operation of the BR feet. I buy them as sets and apply black sticky-back felt to the underside. I used a variety of different weights on top but recently came across some nice heavy door stoppers on Amazon. These are German Schramm stoppers weighing 1.2kg each and perfect for the job. I use 8 of them on each mat or coaster.
Just about a year ago, as one of my big sales, I negotiated with David at MCRU to have a custom pair of Townshend Podiums made for the Ushers. These are BIG pieces of heavy steel with adjustable suspension modules at each corner. My speakers are huge and weigh about 160kg each. The story of how I got them onto the Podiums in the right position, on my own, is one that is too colourful to repeat here. Let’s just say that I was kicking myself several times at my obstinacy to get the job done and nearly ended up stranded underneath one speaker or in hospital.
The Townshend’s operation is counter-intuitive. Why ‘float’ speakers when previous best practice was to rigidly spike them to the floor? The reality is that they work spectacularly well. Room loading and unpleasant bass resonances go way down and bass extension and texture are much more evident. Since that moment, decoupling has become my mantra. There are no spikes in my system nor will there ever be again. Decoupling everything from the router to the speakers has been my route to clarity and quite phenomenal bass power and extension.
Supports are a key to getting the most from your components, reducing microphonics and disconnecting them from the room. Big gains are in clarity (again) by reducing the smearing of notes, substantial reduction in room modes and loading by decoupling the speakers.
“So where are we, Captain?”
That brings me to now, one year down the line. What has changed are a myriad of minor changes and one additional major component: the P3 regenerator. None of the core components have changed and yet everything has changed!
In short, it’s a much more musically communicative system now. In a review by Chris Martens in The Absolute Sound back in 2008, I read that my Usher Be-20 speakers were “musical comprehension machines”. This review, read while I was on holiday in Houston, triggered a series of events which led to me buying the speakers. Later, another review in 2014 by Alan Sircom said “this speaker gives away few concessions when put against far more expensive loudspeaker designs.” By then, I already owned them and could only dream of what was to come years later.
All this is a way of saying that I still believe that, as the DAC is the core of my music system, so the speakers are responsible for taking this precious music signal and delivering the best realisation that they are capable of. If they keep up with system improvements, and the Ushers show no signs yet of having hit a performance ceiling, then they just give me more.
That ‘more’ is more musical communication. From fully shaped instruments and voices, hanging in a three-dimensional space, not bloated but focussed, to the intensity of transient detail of anything from tambourine to bass drums, I have so much more insight than before. Note decay, the shimmer of cymbals to the reverb of snare drum to that feeling of being in a big auditorium on the best classical recordings, are all improved. Orchestral trumpets that sound layered at the back of the hall, vocalists coming forward from the mix to being right there in front of me. Rhythmic intensity where the beat is an important part of the music. Air and space where more ambient material is being played, enveloping me in sound.
The system has been transformed and my objective of greater musical insight has been realised more than I had thought possible with the same main system components. Was it worth selling my gear and collection in order to reinvest in streaming? I’m absolutely 100% happy with the decision. I will take great care not to lose this level of performance and will always ensure a ‘rollback’ path before making any future changes.
Acknowledgements
Tony Sallis (Coherent Systems)
David Brook (MCRU)
John for encouraging me ever forwards
Several friends who gave helpful comments and feedback pre-lockdown