SOTA Pyxi Phono Stage Review

[SIY]

I'm not very knowledgeable here, but could you please comment on my rough measurements here:

Chord Huei Phono Preamp Review

In terms of overall system performance, the noise and distortion of the mechanical parts is usually much higher. The preamp may not be the bottle neck of the system. But the preamp of this bad performance will cause audible coloration. so from your statement of much higher noise from...

www.audiosciencereview.com

with the needle flying in the air i was mostly dominated by the ~90 dB SNR performance of my pre-amp, a stark difference from groove noise.

You need to distinguish between noise floor (which is heavily dependent on measurement parameters, especially FT bin width) and noise, which is a single number integrated measurement. You can roughly correlate spectra like these to noise density, but again, that's a different thing.

 

[AaronJ]

Clipping from surface noise with instant recovery should be better than not clipping from surface noise, it should be closer to the original recording.

How can this be measured? I feel like this is what the Parks Puffin attempts to do but not sure against what metric this aspect can be determined.

 

[Billy Budapest]

A name like SOTA is bold indeed...

It’s been their name for decades and when they first began building vacuum hold down suspended turntables, they were indeed SOTA.

 

[wynpalmer]

I removed the original comment until I have some kind of an explanation. I just repeated the MC measurements too- using an input attenuator to reduce the source impedance to <10 ohms and I was able to more or less duplicate the MC results- except they were better at LF also. This seems to imply a methodology difference, rather than a hardware capability issue, so I need to ponder the issue more.

 

[morillon]

fft 2m..
that s so big... :-0
at 32k (or 64k)?
(seem generaly 256k for armim)

 

[Endibol]

I have that switch on my Zen Phono, but have never had a reason to activate the feature. None of my previous phono preamps had the option. Might be more useful with certain turntables or warped records - very few in my case.

I also have a Zen Phono and for one of my records I really need the rumble filter. If not applied the bass cone excursion gets terrifying large (cm’s). So I am very happy with this option.

 

[wynpalmer]

I'd love to see a dedicated thread on this question. I can imagine a scratch or a very large piece of dirt generating a signal large enough to clip but I have doubts that the "normal" pops and clicks would be affected.

I have digitized c. 2000 LPs using a very similar phono stage to the Pyxi (The Acrux). It has a very similar architecture, with similar overload characteristics, albeit 1.5dB lower SE gain and up to 10dB switched gain boost available. It also has balanced outputs that add 6dB to the gain.
I use an ADC with a +16.8dBv FS input. The "nominal" output from the cartridge is 230uvrms at 1kHz, 5cm/sec (-73dBv) and the highest signal that I have ever detected from the phono preamp is 14dB above this, at -59dBv. The total channel gain is 72dB, and the highest output signal is thus +13dBv. This leaves 3.8dB margin at the input to the ADC.
The phono stage, theoretically, clips at about +26dBV at the output, and -46dBv at the input (ignoring the RIAA characteristic)
I have never had the ADC input clip, except when I drop the cartridge on the LP accidentally . The digitization uses a 96kHz clock, or a period of c. 10us.
I've deliberately overloaded the input using a signal generator- just to test the overload recovery time, and the recovery took about 1us, which is more or less consistent with the TI specs and the low HF closed loop gain of the amps which is +10 or so. If the overload is too long, then the DC correction loop starts moving off nominal, but the rate is extremely slow- the loop bandwidth being sub 1Hz.
So, if there is a problem with overload recovery, I've yet to see it and I'd appreciate some advice on how to expose it in anything approaching a real-world situation.

 

[wynpalmer]

I also have a Zen Phono and for one of my records I really need the rumble filter. If not applied the bass cone excursion gets terrifying large (cm’s). So I am very happy with this option.

A conventional rumble filter sort of comes for free with the DC offset loop. A single resistor needs to be switched. SoTA chose to not have this function.
The Acrux has a second order "elliptic" warp filter that only acts on the lateral signal, but that was too extravagant for this design.

 

[wynpalmer]

What's even more important is overload recovery time. If a click or pop clips the preamp but the preamp recovers instantaneously, no harm no foul. Morgan Jones shows some captured waveforms for record ticks in his book Valve Amplifiers, and 20dB is not unreasonably large.

Also a general note: the S/N of a cartridge just sitting there (i.e., not including turntable or surface noise) is maybe 70dB. So getting significantly better noise performance is not particularly important in this application.

When a Miyajima Madake cartridge (DC res c.16 ohms) is connected rather than an input short the Pyxi A weighted S/N ratio in MC degrades from 82dB to 81dB. This is consistent with the thermal noise of the DC resistance being added to the input thermal noise of the phono stage.

 

[SIY]

When a Miyajima Madake cartridge (DC res c.16 ohms) is connected rather than an input short the Pyxi A weighted S/N ratio in MC degrades from 82dB to 81dB. This is consistent with the thermal noise of the DC resistance being added to the input thermal noise of the phono stage.

How are you deriving that S/N? Referenced to what?

 

[wynpalmer]

How are you deriving that S/N? Referenced to what?

Relative to +5.5 dBv output, which is the 12dB above sensitivity of the Pyxi and is the "rated" output. -72dBv (250uv) +12dB (normal max cartridge output due to LP dynamic range) +65.5dB (gain) = +5.5dBv. This is consistent with simulations, hand calculations, and my own measurements, but does not include hum- which in my measurement/home setups is negligible. I just connected the madake up to the Pyxi, connected the ground wire from the VPI TT, and did the measurement.

 

[voodooless]

No. The external thing is just an AC transformer. The rectifier and filtering is inside and that is what needs improving.

You could still feed it with better filtered DC instead of AC. Noise wise that will probably not do much, but you may get rid of the 60 Hz multiples.

 

[SIY]

Relative to +5.5 dBv output, which is the 12dB above sensitivity of the Pyxi and is the "rated" output. -72dBv (250uv) +12dB (normal max cartridge output due to LP dynamic range) +65.5dB (gain) = +5.5dBv. This is consistent with simulations, hand calculations, and my own measurements, but does not include hum- which in my measurement/home setups is negligible. I just connected the madake up to the Pyxi, connected the ground wire from the VPI TT, and did the measurement.

Thanks for the clarification.

 

[wynpalmer]

You could still feed it with better filtered DC instead of AC. Noise wise that will probably not do much, but you may get rid of the 60 Hz multiples.

The 60Hz multiples are not an problem in practice. I don't know why Amirm's setup has this issue, specifically. Perhaps measuring the noise with the input shorted might resolve the question.
As I wrote, in my measurement system I have to attenuate the signal from the generator, in this case using a "rotated pi" (A large R is series with signal and ground, a small shunt R between, in the order of 330 - 1000 ohms and 10 ohms) to make noise and distortion measurements that are consistent with the intrinsically floating cartridge source.
The phono inputs should never have any signal/supply current flow through them. Proper operation depends on the source being fully isolated, which is true of phono cartridges that are not connected to chassis ground. My RME box does not have a fully isolated generator, so the impedance in series with the output has to be sufficiently large that the current flow back out of the phono input into the test source ground is low enough to not create a "ground loop" of sorts, and the attenuation allows a large source signal to be produced, which also helps. The RME box is not isolated. The outputs and inputs do not have a fully broken ground path between them, so provisions had to be made.
The phono inputs are not connected to chassis. The output grounds are connected to chassis and also to the ground lug. The chassis is best grounded to the follow up preamp via the phono cables and the TT ground (isolated from the cartridge signal wires) should be connected to the chassis via the lug.
Sota wanted a system that was cheap and did not use switchers of any sort, but there are two levels of regulation and a hefty amount of post rectifier filtering so the use of a half wave rectifier was inconsequential, and an inexpensive wall wart could be used. It's similar, but not identical, to the approach used in the Mani-2.
The original DIY design used a switched mode supply and filtering to generate the internal +/-15v rails directly from mains.
The hum/ripple on the supplies is almost unmeasurable and the circuit PS rejection is high. It actually does get hum components that are c. 96 dB down from the rated output of +5.5dBv, which is with a 10mv MM or 1mv MC input.

 

[ivanj]

A conventional rumble filter sort of comes for free with the DC offset loop. A single resistor needs to be switched. SoTA chose to not have this function.
The Acrux has a second order "elliptic" warp filter that only acts on the lateral signal, but that was too extravagant for this design.

Pxyi is a purpose built, not universal, phono stage

To reduce the effects of warped records, etc., SoTA makes some of the proven vacuum turntable series. Many of their turntables exhibit low unweighted ruble figures. All things being equal, the Pxyi is cost effective especially when used with the best 'tables in the SoTA range.
********
If folks require a rumble filter in a phono stage they should apply elsewhere.

Better, look at setup/placement and/or turntable/arm/cartridge system. There is plenty written on this and I'm not going to rehearse it here.

 

[The Dragon]

Many thanks to Amir for testing the Pyxi. Having real baseline APx test measurements, executed by an honest and experienced professional (Amir), does a great service to both SOTA and the designers. These measurements will be invaluable to those considering a new phono stage. This review should remove any doubt that the Pyxi is a competent phono stage and that the designers know what they are doing. My hat is off to designers for designing a high achieving phono stage ( with outstanding RIAA compliance) that is also very affordable within reasonble constraints. I think it is worth noting that this particular Pyxi is probably from an early production run, since I was on the original waiting list to buy one.

 

[wynpalmer]

Many thanks to Amir for testing the Pyxi. Having real baseline APx test measurements, executed by an honest and experienced professional (Amir), does a great service to both SOTA and the designers. These measurements will be invaluable to those considering a new phono stage. This review should remove any doubt that the Pyxi is a competent phono stage and that the designers know what they are doing. My hat is off to designers for designing a high achieving phono stage ( with outstanding RIAA compliance) that is also very affordable within reasonble constraints. I think it is worth noting that this particular Pyxi is probably from an early production run, since I was on the original waiting list to buy one.

Thank you. The discrepancies between Amirm's results and mine are minor with one exception (the hum) and I'm certain that is a methodological issue rather than a fundamental test or design issue. For example, when we perform the production RIAA, noise and distortion tests we insert a "pi" attenuator prior to the input.
All currently shipped Pyxis are essentially identical. Noise, distortion, gain, channel match, and RIAA compliance are all fully tested prior to shipping and so far, there have been zero failures.
The published spec 1kHz distortion is a bit better than I, and Amirm, measure, and I thought the spec sheet was changed, but It seems not.
The architecture is unusual, and it results in the very well controlled RIAA characteristics and the "transitioning" (i.e. soft limiting) nature of the HF "clipping".

 

[dougi]

I'd love to see a dedicated thread on this question. I can imagine a scratch or a very large piece of dirt generating a signal large enough to clip but I have doubts that the "normal" pops and clicks would be affected.

Me too. I usual run any phono pre into an ADC, and give the listening level very little headroom before ADC clipping. I have never noticed significant clipping and I don't wash my records. I prefer to have it louder/better matched with digital sources than allowing more headroom.
To truly have headroom would most likely require a completely analog chain.

 

[restorer-john]

You could still feed it with better filtered DC instead of AC. Noise wise that will probably not do much, but you may get rid of the 60 Hz multiples.

It needs an AC supply. (half wave rectifier to produce split rails).

 

[restorer-john]

The gain for MM is ridiculously high at 45dB.

I recently tested a dozen or so phono RIAA stages I had lying around in various integrated and standalone preamplifiers for gain/overload. Typical average MM gain is around 35dB.

I cannot see how a highish output MM playing a few hot 12" 45rpm records isn't going to easily overload this little SOTA box.

BTW, the poorest performing phono stage was number 14, an AT USB turntable's integrated switchable RIAA stage. I knew it would be bad, but at 32mV overload @1kHz, it was was abominable. The SOTA's approx 50mV overload is IMO a disgrace. The result of power supply constraints perhaps, or passive RIAA losses? ,but combine that with such high MM gain and it's not a good choice.