PS Audio Stellar Phono phono preamplifier Measurements

Sidebar 1: Measurements

I measured the PS Audio Stellar Phono phono preamplifier using my Audio Precision SYS2722 system (see the January 2008 "As We See It"). For logistical reasons, I tested a different sample (serial number SPH-B-9J0100) from that auditioned by Michael Fremer. To get the lowest measured noise, I floated the signal generator's unbalanced output from ground and ran a separate connection from the analyzer's ground to the grounding post on the preamplifier's rear panel.

PS Audio specifies the voltage gain for the moving-magnet input as Low (44dB), Medium (50dB), or High (56dB), depending on the setting selected with the remote control. Looking at the Stellar's balanced output, I measured gains of 43.9dB, 49.35dB, and 54.8dB, very close to those specified. The gain at the unbalanced output was exactly 6dB lower at each setting. The nominal MC input gain is 60dB, 66dB, or 72dB. With the MC input's input impedance set to 47k ohms, I measured gains of 59.9dB, 65.4dB, and 70.8dB. The gains at the unbalanced output were again 6dB lower. Both the moving-magnet and moving-coil inputs preserved absolute polarity.

My estimates of the PS Audio preamp's output impedance were lower than the specified values, at 100 ohms vs <200 ohms unbalanced, and 300 ohms vs 400 ohms balanced. The differences will be inconsequential given the much higher input impedances offered by the line preamplifiers with which the Stellar will be partnered. The Stellar's moving-magnet input impedance was 47k ohms at 20Hz and 1kHz, dropping slightly to 42.4k ohms at 20kHz. Set to 47k ohms, the moving- coil input's input impedance was 45k ohms at low and middle frequencies and 40.5k ohms at the top of the audioband. Set to 200 ohms and 100 ohms, the measured MC input impedance was identical to the specified values and consistent from 20Hz to 20kHz. With it set to 60 ohms, I measured 67 ohms, this again consistent across the audioband.

The PS Audio Stellar Phono Preamplifier offers superbly accurate RIAA equalization (fig.1), with an inconsequential 0.1dB difference between the channels in the midrange. This graph was taken with the MM input; repeating the measurement with the MC input reduced the upper –3dB frequency from 80kHz to 45kHz (both frequencies the average of the two channels), but the response was still flat to 20kHz (fig.2). Channel separation was good, at 63dB in both directions below 500Hz, increasing to 73dB at 3kHz, though it was a little asymmetrical above that frequency.

120PSAStellarfig1

Fig.1 PS Audio Stellar, MM input, Low gain, response with RIAA correction into 100k ohms (left channel blue, right red) (0.5dB/vertical div.).

120PSAStellarfig2

Fig.2 PS Audio Stellar, MC input, Low gain, response with RIAA correction into 100k ohms (left channel blue, right red) (0.5dB/vertical div.).

Spectral analysis of the PS Audio's low-frequency noise floor with the MM input set to High gain (fig.3) indicated that random noise components were low in level. However, very low-level, power supply–related spuriae at 60Hz, 120Hz, and 240Hz can be seen in this graph. The PS Audio's unweighted, wideband S/N ratio, measured with the input short-circuited to ground and with the preamp again set to MM High gain, was an excellent 80dB with an input signal of 1kHz at 5mV. Restricting the measurement bandwidth to 22Hz–22kHz increased the ratio to 87.6dB, while switching an A-weighting filter into circuit increased it further, to 89.3dB. Switching the MM gain to Medium, then to Low, respectively reduced the unweighted, wideband S/N ratio by 6dB and 12dB.

120PSAStellarfig3

Fig.3 PS Audio Stellar, MM input, High gain, spectrum, DC–1kHz, of unbalanced output ref. 5mV input (linear frequency scale).

Spectral analysis with these two lower gains revealed that the decreases in the S/N ratios were due to the supply-related spuriae increasing in level. These spuriae must therefore be introduced after the MM gain stage—as the output level increases as the gain increases, a constant level of noise becomes correspondingly lower when referenced to the output. However, even in the Low gain condition, the MM input is still very quiet in absolute terms. The MC input is also very quiet, and by contrast with the MM input, its S/N ratio didn't change by much when the gain was changed. The unweighted, wideband ratio, referenced to an input of 1kHz at 500µV, measured 58.4dB with Low gain, 59.4dB with High gain, these figures increasing to an excellent 73.1dB and 75.8dB, respectively, when A-weighted.

As you might expect, both overload margins and distortion levels varied according to which input was in use and which gain had been selected. With the MM input set to Low gain, the overload margin, ref. 1kHz at 5mV, was an extraordinary 30dB at 20Hz and 1kHz, dropping to a still good 13.25dB at 20kHz. Surprisingly, the MC margins were 4dB higher, despite the higher gains. The distortion signature was virtually pure second harmonic (fig.4). Increasing the gain by 6dB reduced the overload margins by the same 6dB and while the distortion was still predominantly second harmonic, its level had increased by 9dB (fig.5) and the third harmonic appeared at –90dB (0.003%). With the MM input set to High gain, the overload margins were reduced by another 6dB, which meant there was almost no margin at 20kHz. The distortion also increased (fig.6), with the second harmonic now lying at –56dB (0.2%) and the third at –80dB (0.01%). However, reducing the load impedance to the current-demanding 600 ohms didn't increase the levels of these harmonics. The PS Audio Stellar phono preamplifier must have a beefy output stage.

120PSAStellarfig4

Fig.4 PS Audio Stellar, MM input, Low gain, spectrum of 1kHz sinewave, DC–10kHz, into 100k ohms for 10mV input (linear frequency scale).

120PSAStellarfig5

Fig.5 PS Audio Stellar, MM input, Medium gain, spectrum of 1kHz sinewave, DC–10kHz, into 100k ohms for 10mV input (linear frequency scale).

120PSAStellarfig6

Fig.6 PS Audio Stellar, MM input, High gain, spectrum of 1kHz sinewave, DC–10kHz, into 100k ohms for 10mV input (linear frequency scale).

Fig.7 shows how the PS Audio Stellar's MM input, set to Low gain, handled an equal mix of 19kHz and 20kHz tones, at a peak input level equivalent to 1kHz at 10.5mV. This is 7dB below the 20kHz input level where the harmonic distortion reaches 1%, but while higher-order intermodulation products are low on level, the second- order difference product at 1kHz has reached 1% (–40dB). This graph was taken from the single-ended outputs; as with the harmonic distortion measurements, the behavior was identical from the balanced outputs, apart from the doubled output level.

120PSAStellarfig7

Fig.7 PS Audio Stellar, MM input, Low gain, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 1V peak output into 100k ohms (linear frequency scale).

Overall, the PS Audio Stellar appears to be a well-engineered, versatile device, and is among the quietest phono preamplifiers I have encountered. The way in which its measured distortion and overload margin vary with the gain setting suggests that its owner use the lowest gain setting that will give acceptably loud listening levels with their preferred cartridges, both moving-magnet or moving-coil types.—John Atkinson

COMPANY INFO
PS Audio
4865 Sterling Drive
Boulder, CO, 80301
ARTICLE CONTENTS

COMMENTS
Ortofan's picture

... a subsonic filter?

Do the various EAT E-Glo hybrid phono preamps exhibit an "overexposed" sound quality with "far too much high-frequency energy" and "a grey sheen"?
Consider the Petite model that costs $1K less than the Stellar:
https://www.absolutesounds.com/pdf/main/press/HFN%20Feb%20EAT%20E-Glo%20Petite%20Reprint[1].pdf

georgehifi's picture

At least it's not a few opamps like many are, though this board looks fairy well stuffed.
https://www.analogplanet.com/images/styles/600_wide/public/0419DM.JPG

Cheers George

JRT's picture

It is not easy to exceed the performance of some of the best monolithic operational amplifiers when properly implemented in a circuit.

The now obsolete (shame on you Texas Instruments) TI / National Semiconductor LME49990 was truly excellent for many uses in audio circuits.

LME49990 datasheet:
https://upverter.com/datasheet/7cd4a4f0b32df3909631d7b5fda12936b94811b9.pdf

Jack L's picture

.....monolithic operational amplifiers when properly implemented in a circuit." quoted JRT

Chip op-amps? Thanks but no thanks.

First off, We have to know an op-amp is built up with MANY bipolar junction transistors with built in feedback loops all over.

So however immaculate measured data published only show the STATIC performance working on sine/square waves test signals in the lab. But when used in audio amps to handle music signals which are for ever fluctuating DYNAMIC complex multi harmonics. It will be another story!

The published static performance data do not show how the op-amp in question will operate with dynamic music complex harmonics at all !!!!!!!!

Logic will tell the huge number of bipolar transistor junctions & the feedback loops inside the op-amp chip the complex music harmonics have to pass through, only cause harmonic & phase distortion to the music complex harmonics, substantially compromising the music sound quality. This is physics.

My critical ears just can't tolerate the clinical sound of any op-amps. Period.

Listening is believing

Jack L
Canada

hollowman's picture

It's good to see such an extensive Measurements section for a phono component. I don't recall Stereophile ever publishing THIS many measurements before. In fact, I only recall the late Audio (USA) magazine publishing this many phono metrics in their reviews.

Jack L's picture

...... a phono component." quoted hollowman.

Measurement however "extensive" can never tell how it will sound to our ears. So let yr ears have the final say !!!!!

Jack L

Anton's picture

This review generated hobby excitement for me.

I could see this or the Mu Fi Nu Vinyl being a good foundation for a 20 year system.

davip's picture

Capacitance switching for that MM input -- an extraordinary omission for "...a product that will stand or fall on its own merits, regardless of price", particularly when money is wasted on a useless feature like remote gain-switching...

JRT's picture

Pass Labs XP-27 product marketing web page:
https://www.passlabs.com/preamplifier/xp-27

Product specifications:
https://www.passlabs.com/sites/default/files/Product_specs_17_27.pdf

Owner's manual:
https://www.passlabs.com/sites/default/files/xp-27_om_prelim.pdf

It still needs a good AD converter to get your FLAC files onto the server. I would suggest an RME ADI-2 Pro FS. It was on sale for $1499 in a few months prior to 2020, and you might still find it for that, but be sure that it is the newer FS variant. It is also a good DA converter and headphone preamplifier, so has other uses in the system.

RME ADI-2 Pro FS marketing web page on the new poorly done website:
https://www.rme-audio.de/adi-2-pro-fs.html

RME ADI-2 Pro FS marketing web page on the older and better, but now archived website:
https://archiv.rme-audio.de/en/products/adi_2-pro.php

User Manual, includes section with product specifications, and includes a section describing and showing advantages of digital volume control when properly implemented.
https://archiv.rme-audio.de/download/adi2profs_e.pdf

Ortofan's picture

... short shrift in many modern preamps.

The venerable Apt Holman preamp offered adjustments for both load resistance and capacitance - plus a switchable subsonic/infrasonic filter.
https://www.kenrockwell.com/audio/apt/images/holman-preamplifier/D3S_5281-back-1200.jpg

Michael Fremer's picture

Most buyers will end up using MC cartridges. The fixed capacitance is probably fine for most MM cartridges too.

Bill Leebens's picture

...for an extraordinary, heartfelt review.

One thing I notice from most of the comments—thanks, Anton, for being the exception—-can’t we focus on what is right, rather than constantly pissing on good work? Celebrate what it is, rather than diminishing it by saying it should have been something else?

Have any of these folks ever agonized over a design as Darren did for a year, working to ensure that a brilliant prototype became an amazing piece in production??

I love audio, its designers, and the people who devote their energies, heart, and soul to it.

Armchair experts? Not so much.

Ortofan's picture

... offend you. If, in your opinion, this magazine or its website should only contain positive comments, then it becomes no more than a cheerleading section or a form of advertising.

In my several decades long experience of designing and developing products, prototype units have always been submitted for evaluation by focus groups and/or field tests. Feedback - both positive and negative - is reviewed and changes are incorporated as necessary and where feasible.

Did PS Audio conduct any outside testing? It seems as though the design was deemed complete once it met Mr. Myers "one note" criteria. Who determined the feature set for this product? Had PS Audio solicited my input, I would have suggested eliminating the remote control and including adjustable resistive and capacitive loading for fixed-coil/high-output cartridges. While the remote control selection of MC cartridge loading might constitute a 'unique selling proposition', the load only needs to be set once for any given cartridge. The unit already has separate inputs for MC and "MM" cartridges. How much more would it have cost to include a pair of switches and a few extra resistors and capacitors to enable adjustable loading for the "MM" input?

Bill Leebens's picture

There are always a number of outside evaluators of proposed/new products at PS. Beyond that, I shouldn’t comment further as I’m no longer with the company.

I’ve probably already shot my mouth off wayyy too much.:-)

Catcher10's picture

For me and my Lyra Delos at 0.6mV, only need about 56-57dB to match my preamp input sensitivity. My Nova II is set at 56dB and anymore I could be over loading.
I don't understand the need for so much gain for MC carts....Starting at 60dB does me no good.

Jack L's picture

...... with a phono preamp that always presents the music in the correct light." quoted Michael Fremer.

Both JFET & MOSFET devices are nonlinear devices with transfer curves
kinked sharply, leaving the linear ohmic operation zone pretty narrow for linear music signal swings, even more nonlinear than bipolar junction transistors.

Like it or not, ONLY truly linear active device which exhibits linear transfer curves is triode vacuum tube. period.

That explains why critical ears, like yours truly's. can distinctly tell amps employing triodes only active devices sound so much more musical than solid state amps.
But please forget the measurement data which, IMO, bear little relevance, if any, to what our ears perceive.

Mind you, not all tubes are as linear. Pentodes & tetrodes are nonlinear like JFETs & MOSFETs, also with kinked transfer curves, but the ohmic linear tranfer zones there are much milder & wider than their solid states counterparts. So more musical friendly than sold state devices.

Yes, passive RIAA EQ is the better EQ topology than conventional active
RIAA EQ which employs some feedback loops across active stages.

Listening is believing

Jack L
Canada

X