Purifi Class D

[GUTB]

Okay. I've been told over and over again my knowledgebase re: class D needs updating. I'm struggling under a paradigm from the 90s. I really need to listen to these great new breed of class D modules. One of them is a new design by Bruno Putzeys of NCore fame, the "Purifi". I was told that these "self-oscillating" amp designs overcome the traditional drawback of class D's output filter by doing the filtering in the feedback stage, something about how very high levels of feedback were made possible by something-something.  This, honestly, goes over my head. What the hell is a "self-oscillating" amp? What does feedback have to do with the output filter? Am I really just burying my head in the sand and comforting myself with willful ignorance? I found out about an amp manufacturer, VTV, who puts together a pretty cheap Purifi amp -- the Purifi module and associated NCore switching power supply. They have a 30-day trial period so....why not? I went ahead and bought it.

 

It came today and is now in (actually outside) my rack. I realize that it needs to burn-in. So we'll see if I can update my wrong opinions on class D.

 

 

[GUTB]

Well....first impressions are not good. Restricted dynamics and sound stage. None of the high end resolution and bass control I've experienced with class D in the past. What this tells me is that it simply needs to be burned in. I plugged in my CD player and set it to repeat all. I'll let it go for a day and see how it improves. I hope this amp doesn't need an enormous burin-in period. 

[GUTB]

Okay so....imput impedance. It struck me that perhaps there's an impedance mismatch between my preamp and the VTV. Currently I'm using a Yaqin B-2T, an inexpensive all-tube preamp design. I'm resistant to spending a lot on a pre because I hope one day to get a high-end integrated. This pre has an advertised output impedance of 230 Ohms. So what about the VTV? Their website doesn't say. Searching around google, the figure of 47k Ohms comes up with a few different manufacturers. The Purifi spec sheet states 4.4k Ohms for differential and 2.2k Omhs for SE. Uh....that's insanely low for an amp, is that the right spec? The 47k figures are more in-line with expectations, but is that because of the custom input buffers they use? This thing couldn't possibly be running at 2.2k Ohms, right? I have a Freya which I believe is defective due to what seems like a lot of noise, but it has 75 Ohms on the SE output. 

 

I sent VTV an email about this. I just considered 230 Ohms good enough for most any amp, but maybe not. That might explain why I was getting such weak performance.

[GUTB]

On 12/12/2020 at 7:53 AM, DuckToller said:

Hi GUTB,

 

I've noticed your recenht change of approach from condescending & neglecting to personal experience in regard of Class D.
Clap, Clap - RESPECT!!! Thanks for sharing your adventure in the Class D wilderness!

However I am a bit unsure if selecting the cheapest available appearance aside from the evaluation kit does properly suits your usually high quality criteria that you have shown in the past to devaluate Class D ???

I'd be happy if you could elaborate a bit more in detail your approach and explain i.e. whether you selected the unit for its price or its qualities in engineering before, after and around the PuRiFi module???
Which configuration of the module (Mono, Stereo,3x) did you order, and which buffer did you chose with VTV ET-400A ???

Did you prepare your ears and your bias for an uncolored sound experience and what would be your source/signal chain set up for that adventure ? What will be your test tracks ? What will be the reference (system) you compare it to ?

For the record:
In terms of self-oscillating knowledge / research, the AS-forum search would have lead you here .
Scan down to Bruno's response to Sagittarius, where he explains that term.
The Feedback issue is included in that Q&A as well ...

All the best & curious - stay safe & sound
Tom

 

So, my system.

Digital source: custom audio server with i7-7700k all passive cooling no fans anywhere. Paul Pang USB controller + external battery power. SSD power line isolated to another battery. Running Windows 10, Roon, Fidelizer Pro and HQPlayer. AliExpress linear ATX PSU (with separate power rails, not a micro-PSU).

Analog source: Triangle Art Concerto + upgraded platter, TA speed controller, TA Zeus MC, Reed 2A tonearm. Hashimoto SUT (DIY built from eBay). Chinese all-tube clone of some old model of phono stage, I forget which (Marantz?).

DAC: Schiit Yggdrasil Analog 2 (currently).

Preamp: Yaqin B-T2 + NOS tubes and Herbie Rx tube dampers. Schiit Freya as a backup.

Amp: VTV Stereo Purifi with stock input/buffer. Odyssey Stratos (upgraded, currently broken), ODL (early Stratos from a different company that split off from Odyssey years ago). Chinese Naim NAP200 clone. Old Chinese 845 triode amp collecting dust.

Speakers: ProAc D30Rs.

Power: Furutech Rhodium wall outlet feeding amp and isolation transformer. LessLoss level 1 power cable + LessLoss 64X firewall into the amp. AliExpress balanced isolation transformer feeding preamp, DAC, phono stage with Audio Sensability Testament feeding these components from the transformer.

Interconnects: Audio Sensibility Statements and a few other types.

Speaker cables: Audio Sensibility Testaments and matching jumpers. 

[GUTB]

Okay so, update on the VTV.

The horrible midrange seems to have mostly solved itself. That's good. There is still a smearing / lack of resolution when things get busy. Midrange sounds -- instruments, voices -- still don't sound really convincing. Another troublesome aspect is that I don't seem particularly interested in long listening sessions. Lack of engagement. This could be a sign of a general lack of euphony which is the kiss of death of anything in my system.

Imaging and soundstage has improved since the early days, but it's still not really convincing. 

Bass however has improved to the point that I'd usually associate with class D. Dynamics have also improved to a level I'd call satisfactory.

Next step I think I'll contact VTV and ask about what changes the input board and buffer options may have. Perhaps a double power supply model will help. If some combination can help midrange I may try that.

[GUTB]

1 hour ago, Blake said:

A different Class D, but this new amp from Boenicke looks promising (Sven would never sell it otherwise):

 

 https://boenicke-audio.ch/products/electronics/p1/

 

https://6moons.com/audioreview_articles/boenicke9/

 

That does seem like a very unique amp. Several high-end tweaks built-in like direct-soldered AC cable, Baybee filters, LessLoss firewalls, etc.

[GUTB]

Another interesting finding. I decided to swap out my reference power delivery to the amp: LessLoss level 1 power filtering cable + Firewall 64x from a Furutech GTX-D(r) NCF outlet to a standard black 14-guage power chord from a standard outlet. The difference on one test track was...almost nothing. Maybe the it was a touch muted / muffled on the standard cable, but nothing you couldn't chalk up to audio memory effects. This hasn't been my experience with my linear amps. I wonder if the switching power supply is just agnostic to mains quality -- or it's already so messed up the mains quality won't make a difference?

[GUTB]

3 hours ago, fas42 said:

 

It's been my experience that the quality of the mains supply becomes more and more important, the closer one is to optimal sound. And that there is no such thing as a "reference power delivery" - the interplay between what the mains is delivering and how the audio component reacts to that is too complex for there to be a simple answer that "solves everything" - in the decades that I've been doing this, I have not yet found a plug 'n' play solution.

 

Which a long-winded way of saying, in your situation I would experiment, experiment, experiment - in all sorts of ways, to try and get a handle on what matters ...

Interesting information, thank you. I'm bringing in yet another Odyssey amp. this one a used Kismet (top of the line). It'll be interesting to see how that responds and compares with the VTV Purifi.

[GUTB]

Just giving you guys an update. I haven't run the VTV for several days now since I got in my "new" Odyssey Kistmet bought used for $1500, it's just 3 years old. No need for burn-in which is a bonus. Frankly...the VTV doesn't stand a chance. I'll do a final, formal comparison before officially declaring it a loser but there a few things I'd like to try first, namely my newly functional CD transport which has shown a huge improvement over my normal digital system. Want to make sure I'm giving things a fair shake. It's just that when music enjoyment comes back immediately after bring in a linear amp it tells me that the fundamental flaw of class D remains unsolved. But we'll see. There is a lack of mid-range quality, an unconvincing nature to it -- missing necessary harmonics or some kind of ugly distortion artifact I can't really say. It struck me especially when listening to the Kurt Elling album Upward Spiral, in parts where the sax player tries to match Elling's baritone and the result is ugly and jarring.

 

Is it this midrange issue what's causing a lack of engagement and enjoyment? I can't really say. Again there's a few things I need to try before making a final judgment.

[GUTB]

I thought the LKV was a custom GaNFET design. Is it a Purifi? I put the VTV back in, currently playing CDs.

 

Dynamics are acceptable, although weak compared to either of my Odyssey amps, and definitely the D-Sonics (Pascal) I tried in the past. It's probably very much a current delivery limitation. Possibly the dual power supply configuration would help with that. However that's something that you should be able to solve. What I'm looking for is if the sound of class D is now good enough to be an alternative to linear circuits. The answer I've always arrived to up to now is "No", and people who say otherwise I suspect simply don't know what good sound is or they're being wow'd by some technical aspect but failing to identify why their system is so boring and uninvolving.

[GUTB]

1 hour ago, fas42 said:

 

"Boring and uninvolving" can be as simple as a static buildup issue - interesting timing here, because I had this turn up yesterday, as a problem. Went through everything, couldn't find a cause - and as always should be done, looked very carefully at the last bit of tweaking ... this was a very long bit of cabling added, which was only casually organised to prevent wrong contact between materials - and lo and behold, the assembly had been bumped and disturbed enough for the cabling to slip out of position, and touch what it was not meant to.

 

Interesting. I do need to get back there to make sure cables aren't touching things they shouldn't...I set that up, but changing things  around frequently and it inevitably gets messed up. I also moved my audio PC up to front rack so that I can use a shorter USB cable. Also I now have a CD player sandwiched between 2 amps, I'm sure that's not optimal. Got to sort all that out. I'm really trying to give things a fair shake.

[GUTB]

2 hours ago, RickyV said:

 

would this help with liveliness and un-boringness??

 

 

 

Possibly. I emailed the VTV guy about my findings and various options which might help. He did mention the tube board greatly increases the sense of holographics. The various opamps are supposedly more about tonal quality which MAY in fact help the problematic midrange. He mentioned that someone is doing an extensive comparison of opamps in this amp in a positive feedback article soon, I should go check that. Also it's a little concerning that Purifi can't come with a decent input circuit -- if all this is to be believed. Someone on audiogon told me the standard amp is just a toy and it really needs at least upgraded boards and opamps.

[GUTB]

12 hours ago, barrows said:

To add to my above comment:  The VTV standard Purifi amplifier build uses the Hypex demo input board.  This input stage is flawed in at least two ways:

 

1.  It has no input filtering for RFI.  All amplifiers should have RFI filtering on their inputs, and this is especially important for class D amplifiers, where RF getting into the inputs can cause serious, audible, consequences.

2.  It uses ancient, 78xx and 79xx voltage regulators with very poor noise performance and high frequency noise rejection ratios.  The opamp used in the input stage is a high speed precision audio opamp, which requires a very good low noise power supply to perform its best.

 

In comparison, the Purifi Eval 1 and 2 input boards have the appropriate RFI filter circuits at their inputs, and they have onboard very sophisticated discrete voltage regulators designed specifically to provide ultra clean low noise power to the input stage opamps.  These discrete regulators also feature very high power supply rejection at high frequencies, which is critical for a class D amplifier running on SMPS to keep high frequency noise out of the opamps.

 

It is an entirely unfair to evaluate the performance of the Purifi modules based on using them with the Hypex demo input board, and this is why VTV, and other amp builders, offer more sophisticated input board options, with proper input filtering and discrete voltage regulation.

Wait, why would Purifi release a compromised product to the public to represent their brand? Are they deaf? No one can listen to this thing and think it's fine, unless you've never listened to a high end amp before.

[GUTB]

BTW I went ahead and ordered a set of input boards with Sparkos opamps. I don't want to drill the case, I'll have to figure out an adhesive solution for the standoffs...

[GUTB]

16 minutes ago, barrows said:

Good move, give at least 100 hours playing time before auditioning, and then give it a listen.  BTW is your source component balanced output?

Currently a tube pre single-ended only. I do have an active SE-balanced transformer unit which is currently in use with the DAC (Yggdrasil). I also have a Freya which is fully balanced but not in system. Does the Purifi really like to have balanced in?

[GUTB]

The upgrade input boards and Sparkos opamps came in today. I installed them and the amp is now burning them in. I’ll refrain from posting early observations because those are always defunct within the first day of burning in so I’ll hold off this time. I’ll check in on it regularly to see where it’s at but I plan on giving it at least 100 hours.

[GUTB]

So, the Purifi is going back.

 

After giving the new input boards and Sparkos opamps several days of burn-in after which it doesn't appear to improve any further I gave it a listen. I tried it with both the Yaqin B-2T all-tube SE preamp and my Freya using all-balanced connections. Digital is from my reference digital source (custom audio PC) to a Yggdrasil Analog 2. Also tried by vinyl system with it.

• Ugly midrange problem: appears to have been solved by the new input boards and opamps. Midrange quality significantly improved and is now almost as good as the reference (Odyssey Kismet / Stratos). Still a little less resolving / muddled compared to the reference.
• Dynamics seems to have improved, but not by a huge amount. It might also be in my head but my perception is that dynamics have improved somewhat.
• No other changes particular stood out to me.

The VTV Purifi with the upgrades appears to be just one step below my Odessey Kismet and Stratos ($1500 used, later $800 used). With upgrades the Purifi is $1400. Essentially this is a budget-performance amp at a budget price. Mind you the Odysseys which are not high-end are still a great value in American hi-fi and if you get them used you can get a decent amp at a budget price. I would say the VTV is reasonably priced for the amount of performance it provides which amounts to "okay". This technical performance. There is one other issue that is absolutely killer:

 

The total lack of engagement. I tried listening to it, but from start to end, I felt like I was forcing myself. I was never involved or engaged in the music. At first I was willing to consider that it was the ugly midrange problem which was doing it. But after that was solved there appeared to be no other excuse. Dynamics, microdetails, etc, is just one step down from the reference. They're fairly close. I fixed the cables and tested various configurations, nothing helped. I just didn't enjoy the sound, and that reason had no particular reason related to sound technicalities. In the end -- it's the usual, traditional death blow to class D, fatigue. There's something it does which is murder on the ear-brain system. I used to associate that with the lack of bandwidth and destructive output filter. I was told that self-oscillating amps don't have this issue by doing the filtering by using feedback; I have no idea how that works but if it's true then there's something else doing it. But it's still there. Basically it's still unlistenable.

 

Those of you who think you like your class D, I recommend being honest with yourself. Do you REALLY enjoy listening to you music? When was the last time you turned on your class D and got sucked in? Has it ever happened? Is your system spending more and more time left alone and you tell yourself it needs this or that upgrade? Once again I'm forced to accept that class D has serious killer problems and I doubt it's a phenomena I'm alone in experiencing. I don't think it's matter of discernment or having experience with high-end audio.

[GUTB]

46 minutes ago, barrows said:

Sounds like you prefer highly distorted and colored sound, there is nothing wrong with that, as long as you are willing to admit it.

In reality, the sound from the Purifi is less resolving and more muddled than from the Odysseys, with or without tubed preamps and with digital or vinyl. It's not awful (after upgrading the input boards) but if you think this is high-end sound you honestly don't know what high-end sound is. Odyssey amps are good, but I've heard better. I also wouldn't describe the sound of the Purifi as being dry, cold or lifeless unlike most other class Ds have struck me as. I believe Purifi had solved that particular problem. But the death-blow to class D remains, the utter boredom and lack of engagement. I can't narrow it down to a particular technical aspect as technically speaking it's okay across the board. The only thing I can come up with is that something about class D is ugly to the ear-brain system which causes listener fatigue.

 

I gave this more than a fair shake. It was found wanting in the end. From a technical perspective the sound was what most would consider "good". It's true that I have higher expectations than many on the forum, but even I can't point to any technical aspect of the sound and call it "bad". Dynamics, microdetail / resolution was a step down from my reference, but just a step. The horrible midrange was solved by the input board and opamps. I guess it images unusually high, but I don't think that's a deal-breaker and besides that it imaged just as well as my reference. Perhaps even the dynamics could be helped by an upgraded power supply. Maybe higher-end opamps could even bring the resolution up to reference. But the killer fatigue IS a deal-breaker, and opamps won't fix that. Tubes didn't fix it as my all-tube preamp using all NOS tubes didn't change that. You can blame it on my personal preference but I think perhaps you should open your mind the possibility that class D still garbage due to intrinsic flaws in the technology which haven't been solved yet. I may still try the the other GaNFETs, those might legitimately be game changers. The Cherry amps might be the lone exception as well as they use some kind of parallel switching tech that allows them to get much higher bandwidth than traditional possible. I have a little Cherry stereo amp but it's too weak for my speakers so I can't really use that as a reference to make a judgment call on it, but I will say that it also has a full tonal quality similar to the Purifi and unlike other class Ds.

[GUTB]

30 minutes ago, fas42 said:

 

Class D can deliver satisfying sound ... my current, very cheap Edifier active digital speakers, using very ordinary class D chip amps of some variety, are doing quite nicely - they largely get out of the way of the sound of the recording; I'm close to hearing only the signatures of the recordings spaces and environments. BUT, like all setups, they need some level of tweaking - if this is not done, then only luck, or extreme care in selecting good combinations of components will likely lead to a thumbs up.

 

My experiences to date tell me that anything that relies on high frequencies in its operation, which means DACs and class D amplifiers, are very sensitive to noise on the mains, and other cables - pristine waveforms are essential; anything less is highly likely to be the kiss of death - and lead to the "lack of engagement" issue.

 

The "killer problem" is the poor understanding by most with technical knowledge of how important it is to get these things right - you can't get away with the shortcuts and lack of attention to detail that circuits operating essentially always at audio frequencies will allow - vinyl and class A, no global FB amps may have easily measurable distortions; but the ear/brain can easily account for, and discard these.

 

I was using my reference mains which is a Lessloss level 1 filtering cable + Lessloss Firewall 64X plugged into a Furutech outlet. My listening room has only audio gear plugged in and no lights / fans turns on. I suppose I COULD try it through my 500VA balanced isolation transformer before I pack it up...but I had tested it against a stock black cable plugged into another common outlet in the room and I don't think there was any difference. On my reference there were large upgrades to sound with this setup.

[GUTB]

17 hours ago, fas42 said:

 

What I do if I want good understanding of how much mains or other interference may be a factor is to completely shut down the house electrically - also, no smartphones or other wireless devices operating. A further step is to pull the fuses, or circuit breakers in the switchboard, meaning only the circuit feeding the audio is functional - and every device and cord is pulled out of that circuit, apart from the audio components. The point is to completely silence the house electrically, and minimise any cable acting as an antenna. ... Does this make a significant difference? If the SQ does improve, then restore the electricals a step at a time - to get understanding of where the interference degrading the sound is coming from.

Okay so I plugged my reference mains into the input of the 500VA balanced isolation transformer and the VTV into one of the outlets using an Audio Envy power cable.  This is the transformer I have: https://aliexpi.com/ugvH

 

Also, by happenstance, I got in a new preamp, a Mbl 6010 clone: https://aliexpi.com/pxKz

 

The new preamp has balanced input and output and low noise / distortion specs. I thought I'd throw this into the chain as well. The Freya is high-bandwidth and has a JFET buffer stage option, but I never trusted I was getting good results from this unit. It's been burning in for a day now.

 

So, there MAY be something to this power quality argument. I listened to some tracks, some Bluegrass that makes me especially happy to see if engagement has improved. And....I think there may be an improvement. I felt I was able to get into it at least somewhat, something that the amp has failed to do up until now. Maybe the preamp has something to do with it. So I'll hold off on returning it just yet. Dynamics are even worse but that has to be the preamp burning in (I hope and I don't need a larger transformer and those get pricey to ship from China).

 

[GUTB]

On 1/5/2021 at 4:30 AM, RickyV said:

Curious to know what you think about the opamps. What kind of regulator are feeding them?

Just opened it up look and I can't see them, they're fully covered by another component directly in front of them so I can't read the text on the front. I don't even know if they're regulators I just assume they are because they're attached to heatsinks.

 

Also I've decided NOT to return the VTV. The reason is that on Audiogon someone commented that the Weiss opamps destroyed all the others and was a game-changer for him. So okay, I went and ordered some Weiss...these are very large and very expensive parts. $170 for each one, and they are single opamps meaning I need a total of four, so that's $680 for the lot...going into a $1k amp. So between these and the input boards the actual price is now around $2k. They're on their way and I should get them next week.