SOTA Pyxi Phono Stage Review

[amirm]

@amirm - Regarding the ifi Zen. Can you explain what changed with test conditions to pull it from the rankings? Just trying to understand the evolution of the phono testing protocols.

I just looked at the review and I don't see a reason to exclude it. Yet it was taken out of the list to be plotted in Excel. Maybe was a formatting mistake. I will put it back in. Sorry about that.

 

[amirm]

Adding on, there was a time where I varied the input voltage to get a fixed output voltage. This caused the test signal to be higher than 5 mv, prompting people to say no cartridge goes that high. The benefit of what was doing was that gain factor was neutralized. Disadvantage was the criticism we have now. With 5 mv, almost all measurements are dominated by noise and not distortion so I am not a fan but that is what we have now.

 

[amirm]

Only God knows how many fine high-fidelity speaker systems have ended needing premature woofer refoaming or resealing after one too many episodes of "woofing" caused by subsonic frequencies.

I sometimes see this at audio shows and it is remarkable how much excursion it causes in the woofer. At one show, an audience member complained about it, surprising me that they don't know what this is anymore! I agree that a subsonic filter is a good thing.

 

[Strumbringer]

I was fortunate enough to borrow the prototype Pyxi and got to A/B it for a week with my iFi Zen Phono. I was expecting it to be a very close competitor with the iFi Zen. Boy, was I wrong.

I was so impressed with the Pyxi, so I bought one. Actual user experience (mine): There is ZERO noise. This unit is absolutely dead silent. I like it better than my iFi Zen and by comparison, the Pyxi has more low-mid punch, better defined highs/separation, and sounds wider in presentation to my ears (I listen through a Linear Tube Audio MZ3 and ZMF Auteur headphones).

The Pyxi is a complete winner for me and I'm extremely happy.

 

[wynpalmer]

Adding on, there was a time where I varied the input voltage to get a fixed output voltage. This caused the test signal to be higher than 5 mv, prompting people to say no cartridge goes that high. The benefit of what was doing was that gain factor was neutralized. Disadvantage was the criticism we have now. With 5 mv, almost all measurements are dominated by noise and not distortion so I am not a fan but that is what we have now.

You can't win either way. That's exactly why I specified an "output level" which is 12dB over sensitivity at 1kHz, 5cms/s that essentially represents the highest relative excursion that the LPs/12"45RPM singles that I have digitized produce. Yes, there was a 0.1% level (2 LPs) that made it all the way to 14dB, but that's a 4+ sigma event, so it seemed reasonable to use the 12dB number.
As far as the DC to light. As I wrote, I have a system that is essentially 16Hz to 20kHz, with two subs, two electrostatic panels and two "force forward " woofers.
It can reproduce the BAS CD Saint-Saens 16Hz pedal tone at 16Hz and the Fremaux CBSO LP 16Hz pedal tone (at 24Hz) and its 96k/24bit commercially digitized brethren and there is no problem with excessive woofer excursions. The AHB2s are what, 500+ W equivalent into each panel, and the force forward woofer is I believe 400W equivalent into each channel, as is each sub. I regularly listen at row N symphony levels (85dB SPL average, up to c.102dB SPL peak). No failed woofers, no clipping indicators on the AHB2s. (except when there is way too much bass on a recording- which mostly happens in Pop recordings).
For the Fremaux recording the download and the LP seem to have remarkably similar sub excursions.
Incidentally, I prefer the LP- just. It and the download are very, very, similar, but the LP seems just a tad more spacious....
There are occasions when the records are not acceptable, but they are few and far between, and in my experience, essentially entirely due to warps. Hence the "warp filter" in the Acrux, which eliminates the low frequency LP vertical (stereo) component and retains the lateral (mono) information.
I agree that there are times that an infrasonic filter would be desirable, but I also state that those times are seemingly few and far between, and SoTa clearly did not perceive the necessity.
Empirically (that word again) there have been exactly zero requests from a user base (which is in the hundreds) for such a function.

 

[ocinn]

95% of the questions directed to @wynpalmer I am seeing in this thread, are answered descriptively and with relevant graphs and schematics in his white paper: LINK

For example, page 19-22 describes sub/rumble/bass mono subbing (warp filter), and the reason why sub-filters aren't relevant these days with better tables, and record weights/rings.

 

[pma]

If memory serves me correctly, many 1970s Supraphon classical music recordings were co-productions with Nippon Columbia (Denon). As inventor of PCM (pulse code modulation) digital recording, Denon was eager to build up a catalog of digitally-recorded audiophile LPs in order to showcase the emerging PCM digital technology a couple of years earlier than the big classical companies of the era (CBS, Decca, Philips et al). As such, many of the recording sessions involved simultaneous transcriptions: Supraphon used multi-channel analog tape (to be encoded using SQ stereo/quadraphonic matrix for cutting on LPs), while Denon used its PCM digital taping system to be cut onto audiophile-grade consumer LPs.

Supraphon's recordings from that era, whether on Supraphon or Denon LPs, were quite natural sounding and vibrant. The musicianship on the recordings was very fine as well.

Supraphon's first co-operation with Nippon Columbia was in 1975, recording at small church at Lucany nad Nisou with the Nippon Columbia digital recording system

Schubert - Suk Trio - Piano Trio No. 1 / Notturno

View credits, reviews, tracks and shop for the 1987 Vinyl release of "Piano Trio No. 1 / Notturno" on Discogs.

www.discogs.com

Released later in 1987.

Supraphon has had a recording contract with Nippon Columbia since 1985.

 

[Frank Dernie]

95% of the questions directed to @wynpalmer I am seeing in this thread, are answered descriptively and with relevant graphs and schematics in his white paper: LINK

For example, page 19-22 describes sub/rumble/bass mono subbing (warp filter), and the reason why sub-filters aren't relevant these days with better tables, and record weights/rings.

Unfortunately the explanation as to why a subsonic filter is needed completely neglects the fact it is necessary because of the way cartridges work, as usual these days.
It seems almost everybody doesn't know.

It isn't because records aren't flat or because record players have rumble, it is because all the output of a cartridge below around 2x the natural frequency of the "stator" on its suspension is exaggerated in amplitude and wrong in phase by the nature of the device since the "stator" part of the transducer isn't yet a stationary relative to the LP. At around 2x this natural frequency the headshell becomes a reasonable approximation to an actual stator - remaining reasonable dynamically stationary relative to the LP.
Below that frequency the output of the cartridge is spurious rubbish and should be removed, however flat and concentric the LP or quiet the turntable.

I am astonished at the lack of knowledge about this in the record player enthusiast cohort these days. It is absolutely basic and was well known in the 60's and 70's when I was first interested and then involved.

The need for a high pass filter cut off is entirely to do with the physics of how a cartidge works and nothing whatsoever to do with the LP itself or record player rumble. It is just transducer basics.

 

[pma]

Yes, we need to remind electro-mechanical schemes of the transducers and circuit analogies. And to distinguish between mechanical resonance (tone arm, cartridge) and electrical resonance (cartridge).

Find the right cartridge

Ortofon

www.ortofon.com

 

[restorer-john]

Unfortunately the explanation as to why a subsonic filter is needed completely neglects the fact it is necessary because of the way cartridges work, as usual these days.
It seems almost everybody doesn't know.

It isn't because records aren't flat or because record players have rumble, it is because all the output of a cartridge below around 2x the natural frequency of the "stator" on its suspension is exaggerated in amplitude and wrong in phase by the nature of the device since the "stator" part of the transducer isn't yet a stationary relative to the LP. At around 2x this natural frequency the headshell becomes a reasonable approximation to an actual stator - remaining reasonable dynamically stationary relative to the LP.
Below that frequency the output of the cartridge is spurious rubbish and should be removed, however flat and concentric the LP or quiet the turntable.

I am astonished at the lack of knowledge about this in the record player enthusiast cohort these days. It is absolutely basic and was well known in the 60's and 70's when I was first interested and then involved.

The need for a high pass filter cut off is entirely to do with the physics of how a cartidge works and nothing whatsoever to do with the LP itself or record player rumble. It is just transducer basics.

Thank goodness for Frank.

But what about infrasonic/subsonic filters. Before or after RIAA/phono stage gain? Traditional low filters are after the front end phono stage. Much easier when there is an MM/MC stage and the HPF can be in between that and the flat amp of the next stage. But these days? All bets are off- no logic and they make it up as they go.

If you get a low filter, it's likely to be a 6 or 12dB per octave, set low, not an 18 to 24dB per octave and likely set way too high to be useful.

 

[pma]

After the phono stage. You have then free hands to make a choice of the filter slope/order. You would have hard times to design a filter before the phono stage, without compromising S/N and cartridge loading.

 

[wynpalmer]

Unfortunately the explanation as to why a subsonic filter is needed completely neglects the fact it is necessary because of the way cartridges work, as usual these days.
It seems almost everybody doesn't know.

It isn't because records aren't flat or because record players have rumble, it is because all the output of a cartridge below around 2x the natural frequency of the "stator" on its suspension is exaggerated in amplitude and wrong in phase by the nature of the device since the "stator" part of the transducer isn't yet a stationary relative to the LP. At around 2x this natural frequency the headshell becomes a reasonable approximation to an actual stator - remaining reasonable dynamically stationary relative to the LP.
Below that frequency the output of the cartridge is spurious rubbish and should be removed, however flat and concentric the LP or quiet the turntable.

I am astonished at the lack of knowledge about this in the record player enthusiast cohort these days. It is absolutely basic and was well known in the 60's and 70's when I was first interested and then involved.

The need for a high pass filter cut off is entirely to do with the physics of how a cartidge works and nothing whatsoever to do with the LP itself or record player rumble. It is just transducer basics.

The response of the cartridge/arm/headshell to the mechanical excitations present have nothing to do with whether the playback electronics have a high pass or not.
We could argue whether this is entirely true or not- for example does a lower load impedance on the cartridge cause the effective mechanical stiffness of the cantilever to increase, as some have suggested, but that's clearly not in play here.
What is in play is whether the infrasonic modes (which, I believe, if present and randomly excited, should actually present themselves in a chaotic way) are sufficiently "audible" to cause listeners to interpret them as extra warmth or something similar- presumably by either some intermodulation or the creation of a background acoustic ambient LF "field" so to speak.
Again, I can only refer to the experiences of myself and many others. This is not what is heard. If anything, the opposite is true. The bass sounds less "muddy", the acoustic "field" is more differentiated. It's in fact, very digital like in some sense, with some of the analog character that people like, added.
What is generally audible is recorded rumble on some LPs and warps. Nothing else.
As it turns out, I have the ability to add LF HP filtering as well as warp filtering to my system, and to measure responses down to the mHz region, and one exercise I went through was to place the arm/cartridge on a stationary platter, with the motor running and the belt removed to explore the mechanical system noise and resonances and the interaction between the speakers (excited by the DAC output) and the cartridge. (As an aside, this was originally done to demonstrate to the manufacturer of the TT I have that it had certain problems, which took two trips back for them to fix as they didn't have a way to measure the problems!) It might be interesting to resurrect that test, or something similar, although the whole thing is becoming severely off topic and probably should just end, now.

 

[cgallery]

I had a Mobile Fidelity StudioPhono that has a switchable subsonic filter.

It had a fairly obvious impact on records with good LF content.

The trick is filtering what you don't want without touching what you do want.

I don't think that is a simple trick.

 

[Soniclife]

The trick is filtering what you don't want without touching what you do want.

I don't think that is a simple trick.

It is in the digital domain.

 

[pma]

What is in play is whether the infrasonic modes (which, I believe, if present and randomly excited, should actually present themselves in a chaotic way) are sufficiently "audible" to cause listeners to interpret them as extra warmth or something similar- presumably by either some intermodulation or the creation of a background acoustic ambient LF "field" so to speak.

Woofer cone excursion (LF excessive) can make it move out of the linear transducer range, resulting in distortion of audible frequencies. THIS is to be avoided.

 

[EJ3]

It is in the digital domain.

Too simple. It's more fun to try to get it right in the analog domain!

 

[pma]

Clipping and overload margin is a bigger issue in digital domain than in an analog circuit.

 

[morillon]

Clipping and overload margin is a bigger issue in digital domain than in an analog circuit.

we can think that the reference tracks found on some test lp at 1khz 113.1mm/s are very often a precious help to correctly adjust our a/d acquisitions
it's not all of course for clliping overload .. but allows adjustments with a welcome margin
;-)

 

[wynpalmer]

we can think that the reference tracks found on some test lp at 1khz 113.1mm/s are very often a precious help to correctly adjust our a/d acquisitions
it's not all of course for clliping overload .. but allows adjustments with a welcome margin
;-)

I have a different approach. I know the cartridge sensitivity at 1kHz, 5cm/sec and as a result I know the max output of the cartridge (at most 14dB above sensitivity) based on multiple inputs (Shure and other evaluations, my own history of digitization). I also know the gain of the phono stage at 1kHz- which is extremely well controlled- and the RIAA characteristic is essentially perfect, so I know the max output.
The ADC has a much better dynamic range than the LP playback system, so I digitize at 96kHz, 24 bits, to allow for normalization, add a few dBs to the expected max output, select the ADC input FS, and adjust the gain using the ADC or the phono stage gain set capability, then let it rip. This is exactly the same for all vinyl.
I then do a software clean on the file to remove unwanted noise/ticks/scratches- being careful to not audibly degrade the "audiophile" aspects of the source and renormalize to a peak level of 1dB below FS, which occurs at the 96k 32 bit level, I then save the FLAC file.
For example, with the Miyajima Madake and the Acrux.
Madake sensitivity - 230uv rms @1kHz, 5cm/sec. or c. -73dBv. The manufacturers reported and actual sensitivities actually more or less agree.
Max possible output is thus -73+14= -59dBv. The gain of the Acrux is 70dB, so the result is +11dBv.
One of the ADC settings is +19dBu FS rms for a sinusoidal input, and I have verified that this is indeed the case within a few tenths of a dB, which is +16.8dBv. I then add 2dB to the channel gain of the Acrux, to produce +13dBv max at the input, and digitize. This should provide at least 3dB of margin.
Works perfectly every time with no clipping except on the extremely rare (thank goodness) occasions when I drop the arm.

 

[wynpalmer]

Woofer cone excursion (LF excessive) can make it move out of the linear transducer range, resulting in distortion of audible frequencies. THIS is to be avoided.

Agreed, but as long as the signal excursion does not exceed the specified usability range of the units, it's not a problem, and the speakers and amps are configured so that the clipping indicators on the AHB2s would flash before such a thing was occurring. The full signal bandwidth is applied to the AHB2s even if the actual frequency range is split between the three speaker units and a DSP correction is applied to the Bass.