Devialet Expert 200 Amplifier, DAC and Streamer Review

[Soniclife]

An update. Got a nice call from Devialet support. She said that she had spoken to engineering and they like me to send the unit back to them in France. Unfortunately they are currently shut down due to coronavirus. But once they are open, they will repair it. It was a terrible connection so couldn't converse more than the bare necessities. Happy that there is a course of action at least to get the unit back to health.

That's good to hear for all involved. I wonder if it will come back as a 200, or a 220 pro.

 

[amirm]

This is for the newer 'pro' line, but is the model below what you tested. Other than the 80W line which is approaching it's power rating they all get to 20K without a big rise in distortion. Have they fixed it in the new generation?

We have to find out. A local member has I think the newest generation and once this pandemic is over, I will ask him to test his.

 

[Koeitje]

We have to find out. A local member has I think the newest generation and once this pandemic is over, I will ask him to test his.

And ship another one to France afterwards .

 

[Thomas savage]

And ship another one to France afterwards .

Or just have two fancy french chopping boards ..

 

[gvl]

It is sad the way even people with the measurements don't inform their readers what they all mean, the reviews all conclude these measure excellently, and in the print version of this review there is no data to allow the reader to form their own opinion. Your commentary on each measurement is very valuable, and with the forum nature if you were glossing over things you will get called on it.

Regarding the HF distortion / power have a look at the following measurements.
https://www.soundstagenetwork.com/i...-integrated-amplifier-dac&catid=97&Itemid=154

Chart 3 - Distortion as a function of power output and frequency

(4-ohm loading)
Red line = 2W
Magenta line = 20W
Blue line = 60W
Cyan line = 150W

This is the older original model, which was higher rated. It shows a rise at HF but it get's to 20k at 20W, but only about 15K at 60W, and only 6K at 150W.

Now look at the following.
https://www.soundstagenetwork.com/i...er&catid=97:amplifier-measurements&Itemid=154
Chart 3 - Distortion as a function of power output and frequency

(8-ohm loading)
Red line = 1W
Magenta line = 10W
Blue line = 30W
Cyan line = 60W
Green line = 80W

This is for the newer 'pro' line, but is the model below what you tested. Other than the 80W line which is approaching it's power rating they all get to 20K without a big rise in distortion. Have they fixed it in the new generation?

Back to my earlier question, given the energy of high-frequency components is relatively low in music material, how big of a deal rising THD above 10W for higher frequencies really is???

 

[sergeauckland]

Back to my earlier question, given the energy of high-frequency components is relatively low in music material, how big of a deal rising THD above 10W for higher frequencies really is???

In practice it isn't. However, THD rising as much as it seems in these amplifiers does indicate some 'compromise' in design, or indeed downright poor design if other amplifiers don't suffer from it. Pretty much every conventional Class A or AB amplifier has some increase in distortion at high frequencies, due to a reduction in open-loop gain, meaning that there's less feedback available to reduce distortion at high frequencies. This matters little given both that music has much less energy at high frequencies and that we can't hear harmonics above 10kHz unless distortion is so gross that the accompanying IM distortion becomes audible.

When you look at the distortion graphs for this Devialet amplifier, even if so much worse at HF than at lower frequencies, it's still very small, and therefore likely to be inaudible. In some ways, that contradicts what I wrote at the beginning of this post in that good engineering is doing what's necessary, not just what looks good on a graph, so it could be argued that allowing the distortion at HF to rise, even alarmingly but still inaudibly, is actually Good Engineering if it would have required a lot of effort and therefore cost to avoid it.

I'm not altogether convinced by my own argument, but so many audio products these days are staggeringly over-engineered to provide inaudibly low distortion and noise figures which result in nice graphs and numbers but don't result in any better sound.

S.

 

[infinitesymphony]

@sergeauckland, but if you take that train of thought too far, you end up at Audio-gd; "people like the way our products sound therefore they don't need to perform well."

 

[sergeauckland]

@sergeauckland, but if you take that train of thought too far, you end up at Audio-gd; "people like the way our products sound therefore they don't need to perform well."

SET amp manufacturers have been doing that since 1927!
S

 

[thumb5]

...This is for the newer 'pro' line, but is the model below what you tested. Other than the 80W line which is approaching it's power rating they all get to 20K without a big rise in distortion. Have they fixed it in the new generation?

I believe one of the main changes introduced in the new Pro models was a significant increase in the switching frequency of the class D current-dumping section (the D in ADH), so it's not too much of a stretch to believe that could have improved the way the amp behaves at the top end of the audio frequency range.

 

[Soniclife]

I believe one of the main changes introduced in the new Pro models was a significant increase in the switching frequency of the class D current-dumping section (the D in ADH), so it's not too much of a stretch to believe that could have improved the way the amp behaves at the top end of the audio frequency range.

Interesting, do you have a source for that? I've forgotten what they said they did, just remember the impression given that they fixed a bunch of things they didn't want to elaborate on.

 

[Dialectic]

The Expert Pro units became available in late 2016, right after I bought my Expert 120. Devialet asked for £3500 to upgrade my 120 to an Expert Pro 220. I declined.

I checked today, and it's now a cool $5,990 to upgrade my Expert 120 to an Expert Pro 220.

Fwiw, I don't think the high frequency problems with this amplifier are audible in a normal listening environment with electrodynamic loudspeakers.

 

[Soniclife]

Would an REW sweep be about the biggest test for high frequency power?

 

[thumb5]

I'm going partly from memory, but it is also hinted at in a patent at the time the Pro was launched: http://www.freepatentsonline.com/y2017/0207750.html. The patent does make reference to how the original Expert design was stressed above 10 kHz, and that the new design was intended to reduce this stress. I think the point of the patent was to allow the class D section to run at higher frequency, but I must say quickly scanning through it again I can't see that in black and white.

Quoting what I wrote on another forum to try and decipher the patent:

It's about an improvement to the ADH technology. In a nutshell, the idea is to reduce the stress on the class A amplifier, particularly at higher audio frequencies, so that it has to deliver less current and therefore is more accurate. This is done by adding a carefully designed loading network to the output of the class D amplifier.

Here's my attempt at translating the patent into something resembling normal English...with apologies for any misunderstandings on my part. (By the way I think there may be several typos that escaped the proof-reader, which threw me a little on first reading.)

Figure 1 shows the class A amplifier (in the box marked 18) and the class D amplifier (19). Their outputs are coupled to each other and the loudspeaker at point 14. This is the ADH core of all Devialet amplifiers. The class A amplifier is a very linear (accurate) voltage amplifier and the class D amplifier provides current to beef-up the output of the class A so it can drive real-world loads like loudspeakers. The class D amplifier inherently produces an on-off, pulse-width modulated (PWM) output which is merged with the class A's inherently analog output via an inductor (32A/B in the figure).

The class D amplifier is switching at a high frequency - effectively 2 MHz in Devialet's design - which is an engineering compromise between two opposing factors. First, at higher switching frequencies the class D amplifier is less efficient, i.e. generates more waste heat, because the transistors that implement the switches take a fixed time to turn on and off, during which they're dissipating lots of power (I x V). To make the class D as efficient as possible, then, you don't want it to switch at too high a frequency. On the other hand, at lower switching frequencies there is a higher ripple current through the inductor that couples the class D and class A amplifiers. The class A amplifier has to absorb that ripple current, which stresses it, causes it to dissipate more heat, and makes it less accurate.

Interestingly as a side note there is a hint (in paras 0016-0018) that operating the ADH system in a bridged, configuration is one way to reduce this ripple current and thereby allow the class A amplifier to be more accurate. Maybe that is a reason why the dual mono set up has a surprisingly large effect on sound quality, more than the increased power capability would suggest?

What the patent covers specifically is the addition of a damped RLC network (the components labelled 44x) that is carefully designed to reduce the ripple current at the class D switching frequency while having no noticeable effect at audio frequencies. This new network is presumably in the Pro amplifiers, while the original D-Premier and Expert series just have a resistor, according to para 0020.

As a result of this new RLC network, the class A amplifier has to "deal with" much lower ripple current from the class D, which makes it both more accurate and less power-hungry. I think this explains why Devialet said the Pro series sounded better because the current delivered by the class A amplifier was reduced relative to the class D.

Hope that makes sense...

In retrospect it looks like I mis-remembered (sorry) and the class D section isn't running at higher frequency, but nevertheless the new design has the effect of improving accuracy above 10 kHz.

 

[Soniclife]

I'm going partly from memory, but it is also hinted at in a patent at the time the Pro was launched: http://www.freepatentsonline.com/y2017/0207750.html. The patent does make reference to how the original Expert design was stressed above 10 kHz, and that the new design was intended to reduce this stress. I think the point of the patent was to allow the class D section to run at higher frequency, but I must say quickly scanning through it again I can't see that in black and white.

Quoting what I wrote on another forum to try and decipher the patent:



Hope that makes sense...

In retrospect it looks like I mis-remembered (sorry) and the class D section isn't running at higher frequency, but nevertheless the new design has the effect of improving accuracy above 10 kHz.

Thanks. My limited understanding of all that is it sounds a lot like they added the output filter most class D uses, but they didn't use before, because the class A did that for them.

 

[jjo]

The 200 came out in 2015 it seems so less than 5 years old.

Actually... 200 was just a software upgrade to 170 which came out in 2013. Of course the original point remains, but just to be accurate, the hardware is slightly older.

Anyways, interested to see how this ends. I owned 170 -> 200 and I liked it a lot. But the upgrade to 220 was totally worth it. It is way better all around. I hope you get a chance to measure that as well.

 

[RigorDude]

(G)ood engineering is doing what's necessary, not just what looks good on a graph, so it could be argued that allowing the distortion at HF to rise, even alarmingly but still inaudibly, is actually Good Engineering if it would have required a lot of effort and therefore cost to avoid it.

I'm not altogether convinced by my own argument, but so many audio products these days are staggeringly over-engineered to provide inaudibly low distortion and noise figures which result in nice graphs and numbers but don't result in any better sound.

Terrific insights that imo go right to the heart of the controversy of this thread, and much else besides - amps, DACs, etc. I believe there is an adage to the effect that an airplane can be engineered to be perfectly safe, but it would be so heavy it would never get off the ground.

 

[infinitesymphony]

Terrific insights that imo go right to the heart of the controversy of this thread, and much else besides - amps, DACs, etc. I believe there is an adage to the effect that an airplane can be engineered to be perfectly safe, but it would be so heavy it would never get off the ground.

Maybe, but in this case the airplane has engineering problems that are not present in far less expensive competitors.

 

[gvl]

Is an airplane that can't get off the ground really an airplane? I think that's an entirely different problem altogether.

 

[RigorDude]

Maybe, but in this case the airplane has engineering problems that are not present in far less expensive competitors.

What I'm seeing is a lack of unanimous agreement as to whether there are "engineering problems" in this instance, or a different set of engineering design goals and priorities.

 

[RigorDude]

Is an airplane that can't get off the ground really an airplane? I think that's an entirely different problem altogether.

Yeah I think that's the point of the adage.