• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Purifi vs. Pascal - What is going on?

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
The Quest:
I'm calling on the aid of any engineering masterminds that might be in here, to explain what I suppose could be brushed off as anecdotal (given that I don't have the equipment or knowhow to effectively experiments with/test this in a material way), but none the less here goes - as this has been bugging me for a long while!

I've been looking for a new amp for a long while, and had first set my sights on a Hypex (Nad M22) based amp, and the natural progression was the Purifi once released. Both appeared to sound clean, clear and precise, without becoming "grainy, cold or clinical" as class D has been accused of in the past (excuse the silly audiophile terminology).
I had the chance to compare a handful of Class D amps a while ago, including a Hypex nCore (Nad M10), Purifi's Eval Amp (in their own red enclosure), along with a black sheep, a Pascal L-Pro2s - this against my own amps, mainly older AB designs.

Now the Purifi's reputation is already legendary for its technical specifications and measurements - low distortion, high damping factor, etc., although arguable somewhat marginal, but not insignificant in improvement over the nCore circuit. The Pascal is an OEM only, pro audio amp, used in high end studio monitors etc., best known for its efficiency and on board power supply - not seen or used in any hifi products to my knowledge.

The 'Issue' in listening:
Upon listening, first comparing the nCore to the Purifi, the lineage was clear: they sounded quite similar, with the Purify just sounding very slightly "more", so to speak; a bit more precision, detail, etc. all the usual, and slightly more of the same tonal character, slightly warmer, for lack of a better word. Overall both were good, but I could tell that the Purifi was the "improved" version of the nCore basically.
Now the Pascal turned out to be a suprise: It sounded brighter/more sterile, but ran absolute circles around the purifi in every area other than "tonality" or "musicality" (what ever you want to call it, the purifi sounded warmer). Everything sounded MUCH more precise on the Pascal, to the point where on tracks like "Trentemoller - Chameleon" or "Vestbo Trio - Mudlide" the bass line sounded muffled and soft on the Purifi, where as on the Pascal it sounded clearly defined, controlled and had much more "physical" punch to it, which also translated to more precise sense of imaging/space and generally just clearer transients.

Now just to check if it was my own lunacy/bias whatever you want to call it, or my room or speakers, I tested it in multiple systems, and had numerous people listen, without telling them what I was switching or what they were listening for, all with the same impressions of the results (though preferences differed).

Why do I care, you might ask?
I can't decide which I should go for, as both have ups and downs: Apparant tonality vs technicality. So I want to know which is technically objectively correct; if the Pascal's technical precision is a psychoacoustic artefact of some distortion, like one might experience with a bright sounding loudspeaker, sounding more detailed, but actually just highlighting transients through an unnatural peak in the treble, or whether the Purifi is in fact "slower" somehow, fx. if the high amounts of feedback are causing some sort of time domain issues - as I've often experienced a similar sense of "soft" transients with other amps, that I later learned used feedback designs.

Data/Measurements/Objective proof:
There are plenty of differences to be found between the amps in their datasheets - they might by many be considered "insignificant" or small enough that they shouldn't matter - but given how evident the audible differences were, even to uninitiated listeners I subjected to these, I'd personally say its safe to say that something in there is significant -the question is what?

Data sheets for both amps are readily available:
Purifi Amp Modules
The Pascal Data Sheet can be found under the Downloads tap on the product page:
Pascal Amp Module

Test Equipment:
Amps: Purifi, Pascal L-Pro2s, NAD M10 (nCore), Thule IA100, Pioneer M90, Argon SA1, Klank, NCore Eval Monoblocks (briefly).
Pre: NAD M10, MM Audio Pre, Questyle CMA800b
Dac: Lavry DA10, NAD M10, Astell & Kern AK240
Speakers: JBL Ti10k, Dynaudio Confidence 30, Monitor Audio Platinum PL100, Infinity Kappa 6.2i and a prototype.
 
Last edited:

dougi

Addicted to Fun and Learning
Joined
May 31, 2020
Messages
845
Likes
763
Location
ACT, Australia
I do note that the Pascal modules have more gain (28dB-ish) than typical Hypex/Purifi implementations (26dB-ish). Are you sure you matched volume levels accurately enough in the comparisons?
 
OP
Lord Victor

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
I do note that the Pascal modules have more gain (28dB-ish) than typical Hypex/Purifi implementations (26dB-ish). Are you sure you matched volume levels accurately enough in the comparisons?
Very good point, it was volume matched to the best of my ability using a dB meter app, yes.
 
OP
Lord Victor

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
I believe Bruno Putzeys did a writeup somewhere on a relation between iron-cores and the classic "class D sound" which was typically bright and sterile or grainy, which could possibly have some relation to the Pascals sound, in contrast to the Purifi, where I'd assume Bruno would have tried to get around that issue...
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
The noise and distortion of all these amplifiers is almost certainly below any threshold at which they might become audible. The most likely explanations are (1) that levels were not correctly matched and that your brain interpreted this as one amp sounding different from the others, (2) that the Pascal's high-frequency response is a little different into your loudspeakers' load, and/or (3) psychological (and I mean that with no intention of causing offence; we are all, as humans, unreliable witnesses to our own experiences and memories).

Are you able to repeat the test blind and with voltage-matched levels?

EDIT: having looked at the Pascal datasheet now, I would say it's very unlikely, but not impossible, that nonlinear distortion it produces in the high frequencies might be marginally audible under some circumstances. But would still be pretty shocked if this were the cause of the differences you heard.
 

boXem

Major Contributor
Audio Company
Joined
Jun 19, 2019
Messages
2,014
Likes
4,852
Location
Europe
...
Data/Measurements/Objective proof:
There are plenty of differences to be found between the amps in their datasheets - they might by many be considered "insignificant" or small enough that they shouldn't matter - but given how evident the audible differences were, even to uninitiated listeners I subjected to these, I'd personally say its safe to say that something in there is significant -the question is what?
...
Frequency response and output impedance vs frequency curves
Purifi FR:
1612778067809.png

Pascal FR:
1612778205683.png

Purifi OI:
1612778283060.png

Pascal OI:
1612778389438.png
 
OP
Lord Victor

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
The noise and distortion of all these amplifiers is almost certainly below any threshold at which they might become audible. The most likely explanations are (1) that levels were not correctly matched and that your brain interpreted this as one amp sounding different from the others, (2) that the Pascal's high-frequency response is a little different into your loudspeakers' load, and/or (3) psychological (and I mean that with no intention of causing offence; we are all, as humans, unreliable witnesses to our own experiences and memories).

Are you able to repeat the test blind and with voltage-matched levels?

EDIT: having looked at the Pascal datasheet now, I would say it's very unlikely, but not impossible, that nonlinear distortion it produces in the high frequencies might be marginally audible under some circumstances. But would still be pretty shocked if this were the cause of the differences you heard.
Yes, I repeated the comparison multiple times and with different people (some trained listeners too who are acoustical engineers, some just random people like family members with no knowledge or understanding of hifi), and peoples impressions/descriptions were so consistent and identical, without communicating or knowing what they were listening for/to that, even ignoring my own experience, I’m personally 100% certain there is some difference - just not the cause.
Given the purifis reputation for low distortion and accuracy in measurement, I’d have most suspicion that the pascals brightness is caused by some sort of distortion/deviation, but haven’t found a clear culprit.
The pascal has higher output impedance, and a bigger rise in output impedance at high frequency - but it should still be low enough not to matter much, and assuming it does, to my understanding, the output impedance rising at high frequency should cause a decrease in treble energy, not an increase... Hence why I’m currently somewhat stumped.

I’ll have to look into Bruno’s paper tomorrow in case it holds some clues.

I was quite careful about the volume matching, but even tried turning up the Purifi higher to see if that would have an effect, but even when louder, it still sounded more muffled.
 
Last edited:
OP
Lord Victor

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
And just to clarify after some confusion discussing this somewhere else: this is essentially an academic question; I.e. less about figuring out which to buy (though I’d want that too), more about having observed something and wanting to understand it.
Like why one fruit tastes sweet and another sour.
 

XpanD

Active Member
Forum Donor
Joined
Apr 7, 2018
Messages
146
Likes
171
Location
Netherlands
Phone apps are generally unreliable for this kind of stuff, and tend to spit out wildly different results even in the same setup. Ideally you would be using a calibrated microphone for this (something like an UMIK-1), or measuring and matching output at the terminals.
 
OP
Lord Victor

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
Phone apps are generally unreliable for this kind of stuff, and tend to spit out wildly different results even in the same setup. Ideally you would be using a calibrated microphone for this (something like an UMIK-1), or measuring and matching output at the terminals.
Yea, used a calibrated setup for the tests done at the company, phone whilst at home (though I roughly knew how many tics of difference in volume were neede based in the calibrated tests).
But the characteristics arak persisted irrespective of which amp were louder as I found because I got curious about ether that was effecting it, and so intentionally turned the Purifi up louder than the Pascal to see if it would still appear muffled by comprising, which it did.
The volume change would affect which I would subjectively “prefer”, but didn’t seem to affect the actual apparent character of the amps, if that makes sense.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Yes, I repeated the comparison multiple times and with different people (some trained listeners too who are acoustical engineers, some just random people like family members with no knowledge or understanding of hifi), and peoples impressions/descriptions were so consistent and identical, without communicating or knowing what they were listening for/to that, even ignoring my own experience, I’m personally 100% certain there is some difference - just not the cause.

Ok, perhaps I misunderstood your OP, but as I understood it, you all knew which amplifiers you were listening to during these comparisons. Given the improbability of anything objective other than perhaps high-frequency response being responsible, I think you need to repeat the test - blind and properly level-matched - before drawing any conclusions. Otherwise you'll be chasing rainbows down a rabbit-home IMHO :)
 
OP
Lord Victor

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
Ok, perhaps I misunderstood your OP, but as I understood it, you all knew which amplifiers you were listening to during these comparisons. Given the improbability of anything objective other than perhaps high-frequency response being responsible, I think you need to repeat the test - blind and properly level-matched - before drawing any conclusions. Otherwise you'll be chasing rainbows down a rabbit-home IMHO :)
Yea, no; the people I had listen had no idea what they were listening to, just that I was swapping something, but for all they knew I could be swapping cables, dacs, speakers, you name it, you get my point - I personally knew of course, but other than that, no, I didn’t tell them until after they’d given their impressions what I was swapping/testing.

but yea, it’s only because I feel fairly confident in the amount of testing i did, that I’m still wracking my brain about it.

I also suspect it’s probably mainly a high frequency thing; the perceived brightness of the pascal would suggest that. That’s also why I’m particularly curious/suspicious ether the pascal is legitimately more precise/controlled in the bass, or simply appears to be as result of artifacts in the high frequencies... :)

The effect was also much more pronounced in the JBL Ti10K than on the Dynaudio Confidence speakers - though they were in different rooms, so could entirely be caused by room acoustical differences, reducing the audibility of the effect in one room. Or it’s because the JBL is much more sensitive.
 

BDWoody

Chief Cat Herder
Moderator
Forum Donor
Joined
Jan 9, 2019
Messages
6,948
Likes
22,625
Location
Mid-Atlantic, USA. (Maryland)
but yea, it’s only because I feel fairly confident in the amount of testing i did, that I’m still wracking my brain about it.

Clever Hans needs to be accounted for...

You are confident, but your protocol really wasn't buttoned down in a couple of areas. There are reasons blind test protocols exist.
 

jae

Major Contributor
Joined
Dec 2, 2019
Messages
1,208
Likes
1,508
Always the one with better specs if you can afford it and you're satisfied with its "features". So in this case there is no reason not to go with the purifi.
 

Shorty

Active Member
Joined
Sep 25, 2019
Messages
171
Likes
269
Always the one with better specs if you can afford it and you're satisfied with its "features". So in this case there is no reason not to go with the purifi.
.... and live forever after with that nagging feeling that you liked the Pascal better... ;)
 

jae

Major Contributor
Joined
Dec 2, 2019
Messages
1,208
Likes
1,508
.... and live forever after with that nagging feeling that you liked the Pascal better... ;)
Get a sharpie and write "Pascal" on the Purifi and then you can get the best of both worlds- reality AND fantasy.
 
OP
Lord Victor

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
Clever Hans needs to be accounted for...

You are confident, but your protocol really wasn't buttoned down in a couple of areas. There are reasons blind test protocols exist.
Don't quite know what you meab by "Clever Hans"? :p

But yes, to be precise: I'm confident within the scope of what I'm doing here, i.e. making a forum thread and digging into the data, based on my personal results, and the last 30 years of peoples reports/impressions of what Class D amps tend to 'sound' like (the reason they've been looked down on in hifi for so long).
I'm not confident in the sense of; I would write a white-paper on the topic based on my current data - I'm aware that there are protocols for such things, like ideally doing blind ABX tests, preferably using a randomized ABX machine and so forth...

But I just got to borrow this equipment for a weekend, so I simply did what I found possible/useful within that limited scope. I'd love to own a true blind ABX machine at some point, but just not a practical option as is.
Plus I do think quick AB have merit in teasing out even smaller differences for me personally, but that's all a topic on methodology for another time. Not really anything I can do here.

If anyone thinks my sample size/method not enough to be valid at all, even on this relatively casual, though serious, level, then fair is fair - its your time :p
 
OP
Lord Victor

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
.... and live forever after with that nagging feeling that you liked the Pascal better... ;)
And there you hit the nail on the head for me in terms of the purchasing decision for - Realistically I'll probably have to end up getting both, sooner or later, if nothing changes :p

That being said, as I noted later, this isn't just about purchasing one - I might not even be able to get the pascal since its not sold to private individuals sadly. More so an academic curiosity, if you will!
 
OP
Lord Victor

Lord Victor

Active Member
Joined
Apr 19, 2019
Messages
140
Likes
99
Frequency response and output impedance vs frequency curves
Purifi FR:
View attachment 111229
Pascal FR:
View attachment 111230
Purifi OI:
View attachment 111231
Pascal OI:
View attachment 111232
Several posts ago, I showed measurements highlighting that pending the speakers, the Pascal can be bright.
Speakers are not resistors.

Would you care to elaborate please? I'd considered the exact same measurements, but came to the conclusion that, assuming purifi's "1m! on the scale means 1 miliohm, the purifi has much less output impedance, and much less rise in output impedance, relatively speaking - however even though the pascal has a greater amount and a bigger rise, its quite far up in the frequency spectrum, and if anything, rising output resistance would reduce, not increase treble?
But on the other hand if the Purifi's rise in output impedance is actually great enough to be audible, the way it graphs, it starts so early that under a low resistance load, if it rolled off frequencies in that same inverse manner, that would I think correspond pretty well with what I heard...

Because assuming I'm correct in saying rising output resistance, would reduce voltage/linearity in the given area, relative to the rest, i.e. would roll off upper frequencies, and more so the lower resistance the connected load is, and assuming the output resistances of both amps are high enough to affect linearity to an audible degree, due to the non-linear nature of the resistance (which I'm aware would be dubious, as both with conventional "good enough" reasoning, should be low enough as a whole and would be best to actually test.), and presuming the non-linearity of either amps impedance is equally audible (fx. if the purifi scale is ohm, not miliohm), simply looking at where they would be rolling off:
Then the purifi would theoretically start seeing some degree of roll-off or reduction in relative level already from 3-400Hz, increasing at the frequencies rise, which would produce a warmer/more rolled off sound, more so than the Pascal arguably, which only begins to really have rising resistance at 2-3kHz,, an only steeply rising much later.

I don't know if that is what you hinted at, but assuming all of the above is true, and is audible, then I think it would line up with, and likely explain what I was hearing.

It would also possibly explain why the difference was more audible on my JBL Ti10k vs. the Dynaudios, as the JBL is very sensitive, and from 70hz up is around 4Ohm, where as the Dynaudios are not efficient, but fairly stable at 6Ohm, going to 5 at their lowest - in other words, the JBL's lower impedance, would exacerbate the above issue, relative to a higher impedance speaker like the Dynaudio.
Though this may also simply be a result of the Dynaudios low efficiency, which I know has been correlated to speakers being much less audibly affected by characteristics of the chain they're connected to...
 
Top Bottom