Review and Measurements of Purifi 1ET400A Amplifier

[JimB]

Hello all, I'm new to the DIY amp/speaker world, would like to try out building some of these kits, I am not an electrical engineer but I have soldered on a PCB. I'm just trying out new things whilst in retirement and thought maybe I'd give Purifi a go. I'm learning as quickly as I can, so if my questions annoy the more seasoned of you, well, what can I say, try to keep in mind that you were once where I am knowledge-wise. ...

This is the review topic for Purifi's evaluation amp. Please be sure to check out the DIY thread for the amp modules.
https://www.audiosciencereview.com/forum/index.php?threads/diy-purifi-amp-builds.10478/#post-289105

I'm adding some comments over there to those already made by @Julf and others, here.

 

[LondonLad]

Perfect. Thank you, JimB.

 

[VintageFlanker]

https://nadelectronics.com/nad-electronics-and-purifi-audio-announce-cooperation-agreement/ I wonder what product this may be in

That one:

NAD just announced the M33 at CES!

 

[kaka89]

NAD's M33 available at US$4999/€4999/£3999.

Compositors are
- Apollon's mono amp is €1190 each = €2380
- miniDSP SHD Studio for DAC & streaming = US$1199 US$899

I think it is not too bad for the value.

 

[mocenigo]

NAD's M33 available at US$4999/€4999/£3999.

Compositors are
- Apollon's mono amp is €1190 each = €2380
- miniDSP SHD Studio for streaming = US$899

I think it is not too bad for the value.

Of course it depends on your needs. If you already have an excellent DAC (I do, Soekris 1541) and a streaming/play from HD solution that fits your needs (I do, a raspberry pi 4 with moOde) and you do not need a separate headamp or subwoofer amp (neither do I), so th Apollon's are a much much much much better value. All things considered, the NAD price, from an established commercial entity with a traditional sales model, is also a very good value. I think I see more than two modules inside the NAD amp: six or something? What am I missing? They have separate amps for each of the biwiring outs and the subwoofers? That would make it indeed almost "cheap" (of course they pay less per module, because volume).

 

[Julf]

Of course it depends on your needs. If you already have an excellent DAC (I do, Soekris 1541)

This one?

 

[mocenigo]

This one?

View attachment 44909

No, I meant the 1541. It has a different set of R2R networks. Head to SBAF and you will read measurements of the 1541 that are in line with the "faves" of the ASR aficionados.

Also, if is quite well known outside this forum that Amir does not really know how to use his equipment (a few regulars have as well remarked sometimes his measurements were not consistent) and the results are nearly irrelevant. It is not difficult to get something measuring quite well (note that I said "quite well", not "excellent") on some static tests, but the measurements done are all on signals that are static on the spectrum domain, except for simple (the key is SIMPLE) transients (and, again, only simple ones). I would be interested in seeing tests made on more complex signals, and, no, a simple 19+20khz signal is not complex, as also a constant superimposition of several frequencies: these are both ludicrously simple and unrepresentative with respect to actual music.

I suspect that the 1421 measurements you report are botched, since they are quite far from those by Soren Christensen himself, who is a well respected engineer in the field, not an amateur with a scope.

The DAC1541 is known as an excellent DAC, and sounds way better than the SABAJ DA5, which may (or may not) measure better on test tones.

For full disclosure: I come here to stay informed and take the measurements published here with a pinch of salt, as an indication that some parameters are ok, or MAY not be ok (it is easy to take a wrong worse measurement, but a good one is seldom a mistake). But i do not take the measurements as a *proof* of quality.

Roberto

 

[tw99]

Amir doesn't know how to use his equipment.

 

[mocenigo]

Amir doesn't know how to use his equipment.

To be more correct, technically he does, and he also probably knows what the numbers mean. an apology to him may be warranted, and I extend it hereby.

The main problem is that the audience take this as a SUFFICIENT proof of quality. No, if the measurements are done properly they are a NECESSARY condition (but I suspect that in the case of the Soekris 1421 they are defective). Theme measurements do not even consider the simplest variable signals. I would like to see them to be persuaded, but, curiously, they are always absent. Not only from here.

 

[Julf]

No, if the measurements are done properly they are a NECESSARY condition (but I suspect that in the case of the Soekris 1421 they are defective).

What makes you think that? R2R ladder DACs are really simple (I designed and built my first one in 1975 or so), but it is a really primitive architecture that has long ago been superseded by better, more advanced designs.

Theme measurements do not even consider the simplest variable signals. I would like to see them to be persuaded, but, curiously, they are always absent. Not only from here.

What "variable signal" measurements do you suggest, and what additional information would they provide?

 

[mocenigo]

What makes you think that? R2R ladder DACs are really simple (I designed and built my first one in 1975 or so), but it is a really primitive architecture that has long ago been superseded by better, more advanced designs.

The measurement of the 1421 are too bad than believable, that's it.
And my reasoning is not about R2R vs DS, it is more general.

Regarding R2R being primitive, if you say so. DS is not necessarily better, it is cheaper. That's why it is dominating also in integrated designs. And they have some problems, esp with complex transients. That's why they have been made more complex, such as multi-bis delta sigma, and with various types of oversampling and processing. OTOH a good R2R design will, as you probably know, convert the the upper bits with a single resistor, and then move to a "real" R2R chain, use various types of aliasing and averaging, together with self-measurement and correction. But they will (at least mathematically and with ideal resistors) represent transients more faithfully. OTOH with delta sigma you can get more easily "perfect" measurements.

What "variable signal" measurements do you suggest, and what additional information would they provide?

Signals whose spectrum varies at more than one point in time. Even a very fast glissando would be interesting to observe, since there would be no periodic signal. They would provide information about the behaviour of the DAC with a quickly varying signal. Second example: two very, very close transients. Examples: when in an orchestra you have different percussion instruments playing at the same time – two players never hit a note exactly at the same moment in time, or if you have an ensemble of strings and they all play a pizzicato. There is never perfect synchrony. How does a DAC (or an amplifier) react to dozens of transients which are very very close in time? Even a string playing usually has a vibrato. A singer does vibrato as well. The pitch is never constant. This is never tested. I want to see the results, and then we can say what they provide. Oftentimes "perfect" DACs fail at reproducing strings in a realistic way (many DS, BTW), and even though this is a subjective assessment, this *should* point at something. This should tell us that maybe something is missing.

TBH it would be interesting to see these studies and a correlation with the other measurements and subjective perception. Everything has been simplified to just THD, noise, maybe IMD and behaviour under a few other test signals. Maybe is the mathematician in me that sees the incompleteness, but all papers I have seen claiming the "sufficiency" of some limited tests suffer from logical fallacies (they claim to prove that by stating "these are two frequencies which are common in reproduced music, and all within the hearing range of an adult, so let us add them together and measure"). i would be curious to read more.

 

[tw99]

"even though this is a subjective assessment".

You should read more background on the limitations of your abilities, and the fallacies of such assessments.

 

[Julf]

The measurement of the 1421 are too bad than believable, that's it.

In what way?

Regarding R2R being primitive, if you say so.

As I wrote, I designed my first one 45 years ago. They are very simple - just very challenging to make linear.

DS is not necessarily better, it is cheaper. That's why it is dominating also in integrated designs. And they have some problems, esp with complex transients.

Tell us more. What kind of issues? What is a "complex" transient as opposed to a normal one?

Signals whose spectrum varies at more than one point in time. Even a very fast glissando would be interesting to observe, since there would be no periodic signal. They would provide information about the behaviour of the DAC with a quickly varying signal.

As a mathematician, I assume you are familiar with the work of Joseph Fourier?

How does a DAC (or an amplifier) react to dozens of transients which are very very close in time? Even a string playing usually has a vibrato. A singer does vibrato as well. The pitch is never constant. This is never tested. I want to see the results, and then we can say what they provide.

I also assume you, as a mathematician, are familiar with the Shannon-Nyquist sampling theorem?

 

[March Audio]

How does a DAC (or an amplifier) react to dozens of transients which are very very close in time? Even a string playing usually has a vibrato. A singer does vibrato as well. The pitch is never constant. This is never tested. I want to see the results, and then we can say what they provide.

Oftentimes "perfect" DACs fail at reproducing strings in a realistic way (many DS, BTW), and even though this is a subjective assessment, this *should* point at something. This should tell us that maybe something is missing.

.

Sorry but with respect you have a fundamental lack of understanding which is going to be extremely difficult to cover in the context of this thread. I'm not even sure where to start.

The time waveform is the time waveform. All that dynamic change (within the defined limits of sampling theorem and the bandwidth limit) is in the waveform. The dacs accuracy in reproducing that waveform is quantifiable by various measurements, many of which you see here. R2R is not fundamentally better.

Btw I have blind tested people with a soekris R2R V an Ess based DS DAC and they couldn't tell the difference.

"oftentimes"? You are gonna have to provide some evidence to back that up.

 

[March Audio]

Having a brain fade. Anybody have the link to that digital audio 101 video?

EDIT. Found it

 

[mocenigo]

"even though this is a subjective assessment".

You should read more background on the limitations of your abilities, and the fallacies of such assessments.

What tells you I did not?

 

[March Audio]

What tells you I did not?

So you tested blind with precisely matched volume and instant switching?

 

[mocenigo]

In what way?

-80db SINAD?

As I wrote, I designed my first one 45 years ago. They are very simple - just very challenging to make linear.

Ok, so we have a misunderstanding here.
I never a accept mediocrity. in fact, I do not tolerate it, because in many cases it is just the result of laziness. I meant of course designing a GOOD one. Linear, precise, with low noise, and sounding good.

Tell us more. What kind of issues? What is a "complex" transient as opposed to a normal one?

I already described it.

As a mathematician, I assume you are familiar with the work of Joseph Fourier?

Yes I am, and here's your fallacy: When you are performing certain measurements, you are basically measuring both linear distorsions and non-linear ones together. Fine.
The latter ones are the key.
If there is a non-linear interaction, we do not have distortion(signal A + signal B)=distortion(signal A)+distortion(signal B), in fact there is a discrepancy.
So if distortions added linearly, the measurement of a single-point-in-time spectrum change would tell us the whole story.

But they do not, as we all well know, unless explicitly proved. Has this been ever proved?

Therefore, these measurements do not tell the whole story unless we have proof that the distortion adds linearly. Which we can reasonably assume it does not.

I also assume you, as a mathematician, are familiar with the Shannon-Nyquist sampling theorem?

Yes, but I also understand how it is applied.[/QUOTE][/QUOTE]

 

[mocenigo]

So you tested blind with precisely matched volume and instant switching?

I responded to "You should read more background on the limitations of your abilities, and the fallacies of such assessments". I am not the one with reading problems here.

 

[March Audio]

I responded to "You should read more background on the limitations of your abilities, and the fallacies of such assessments". I am not the one with reading problems here.

"The DAC1541 is known as an excellent DAC, and sounds way better than the SABAJ DA5, which may (or may not) measure better on test tones."

I was just taking the point further.