Devialet Expert 200 Amplifier, DAC and Streamer Review

[Frank Dernie]

Such a pleasure to meet you here.
As a Ferrari supporter, u made me suffer a lot....
You have been one of the most significant F1 engineer in the last decades.

I was glad to heve been working in what I consider to be the golden era of F1, it was more like fun than work.

 

[Overseas]

I did. It is the weekend and with all that is going on, I don't expect a quick response.

Do they know they run a business?! Or should owners use the gear only in workdays...

 

[restorer-john]

This is line-in so is digitized. It produces about 48 kHz of bandwidth which is fine. What is not so fine is that there is a phase shift between channels as frequencies go up:

Some kind of filter is active here that is causing a phase shift between channels. Even if there is some processing active (which there shouldn't be as I disabled tone control and "SAM" speaker EQ), it should affect both channels equally as far as phase.

I can confirm the Devialet Expert 200 I have here also exhibits the same phase shift/delay with frequency (see scope shot at 20kHz below). So there is a sample delay error between channels in the A/D on the analogue inputs.

No such phase delay when streaming the same 20kHz test signal via USB:

(traces offset for visibility)

 

[gianventu]

Any chance that this default selection could cause the strange measurements results made by @amirm ?
Even if you take out the sd card , should remain selected, being a default setting.
This the help description for the selection, I could not show in the picture above.
"Dynamic Power Management : Allows the switched-mode power supply of your Expert system to continuously adapt to the musical content, for greater energy efficiency."
Just a question.

 

[restorer-john]

Any chance that this default selection could cause the strange measurements results made by @amirm ?
Even if you take out the sd card , should remain selected, being a default setting.
This the help description for the selection, I could not show in the picture above.
"Dynamic Power Management : Allows the switched-mode power supply of your Expert system to continuously adapt to the musical content, for greater energy efficiency."
Just a question.

I don't believe so, as this phenomena is unrelated to the power output of the device. @amirm discovered a very interesting issue with the A/D and something I could confirm was not just a one-off, using a totally different unit, half a world away. It's almost like the old 1st and 2nd gen CD players, where one D/A was timeshared between channels, resulting in an increasing phase error as the frequency went up. It's basically a fixed sample delay error. It won't be remotely audible.

As the unit I have here is configured by the owner for bass management (it's rolled off hard for active subwoofers) and a few locked out controls, I'm not over-riding the setup. Just fixing it, testing it, and making sure the subsystems are working to the best of my knowledge. I'm not going the whole "app install and tweak" path.

 

[DavidEdwinAston]

Well, because I skipped from page 1 to page 28 and assumed the point had been made many times before. I was going to suggest he takes my Ford Cortina for a road test.

I think the Muppets on Top Gear, would be a better bet for your Cortina.

 

[gianventu]

I don't believe so, as this phenomena is unrelated to the power output of the device. @amirm discovered a very interesting issue with the A/D and something I could confirm was not just a one-off, using a totally different unit, half a world away. It's almost like the old 1st and 2nd gen CD players, where one D/A was timeshared between channels, resulting in an increasing phase error as the frequency went up. It's basically a fixed sample delay error. It won't be remotely audible.

As the unit I have here is configured by the owner for bass management (it's rolled off hard for active subwoofers) and a few locked out controls, I'm not over-riding the setup. Just fixing it, testing it, and making sure the subsystems are working to the best of my knowledge. I'm not going the whole "app install and tweak" path.

I was not refering to the measurements you posted above. They are quite clear and confirm amirm's conclusions regarding that particular issue.
The question I posed referred to the output drop at 8/10khz.

 

[Soniclife]

As the unit I have here is configured by the owner for bass management (it's rolled off hard for active subwoofers) and a few locked out controls, I'm not over-riding the setup. Just fixing it, testing it, and making sure the subsystems are working to the best of my knowledge.

If you take the SD card out and power cycle you get the default config, to go back just put the card back in, it will reboot on insert to pick up the config. All the config is held on the card.

 

[restorer-john]

If you take the SD card out and power cycle you get the default config, to go back just put the card back in, it will reboot on insert to pick up the config. All the config is held on the card.

Thanks. I wasn't aware you could run it in default config without the SD card inserted.

 

[restorer-john]

The voltage amp is class-A, the current is supplied by 4 (iirc) class D devices transformer coupled to the output and running at different phase to each other to further smooth the output current.

This is correct. There are 4 complete Class D amplifiers per channel (8 in total) transformer to the outputs via multiple multitapped 1:1,1:1,1:1,1:1 transformers.

The idle consumption of the unit is high at 30.2 watts (31.4W with the phono stage active) and in standby/off, it is a whopping 14.3W. Hence without proper cooling, it runs hot, especially in Australia.

 

[BDWoody]

The idle consumption of the unit is high at 30.2 watts (31.4W with the phono stage active) and in standby/off, it is a whopping 14.3W. Hence without proper cooling, it runs hot, especially in Australia.

My cats love it.

Hard to keep the little paw prints off that thing.

 

[restorer-john]

The question I posed referred to the output drop at 8/10khz.

In the interests of not "doing an Amir" and blowing up the PSU of the Devialet just repaired, I've limited my testing to single channels driven, not both, to halve the draw on the power supply during any given test.

As @amirm discovered, I can also confirm the amplifier will go into shutdown (but not failure) when doing sweeps at somewhat elevated powers and high frequencies.

When driving 4R loads, the amplifier is ruler flat across the audible bandwidth at all powers prior to shutdown. But shutdown will occur at just 20V RMS (100W) output above 5-7kHz.

When driving 8R loads, the amplifier is also ruler flat across the audible bandwidth until 20V RMS (a mere 50W), where the high end begins to droop slightly. Push the amplifier to 25V@8R (78W) at HF will shut it down in a sweep. I did not investigate spot high frequencies at high powers as that is asking for trouble I don't want. The log sweep is clearly tough enough for the amp as it is.

It is clear the amplifier will not achieve anywhere near its rated power across the full audio bandwidth, whereas at low-mid frequencies, it demonstrates excellent linearity and plenty of available power >250W@4R.

I've never had an amplifier shutdown during this type of sensible testing, letalone an expensive product such as this.

 

[restorer-john]

For those of you interested, here's the output stage board. You can see the 8 class D amplifiers, arranged in a 4-2-1 transformer arrangement.

Driver IC:

Mosfets:

 

[Frank Dernie]

This is correct. There are 4 complete Class D amplifiers per channel (8 in total) transformer to the outputs via multiple multitapped 1:1,1:1,1:1,1:1 transformers.

The idle consumption of the unit is high at 30.2 watts (31.4W with the phono stage active) and in standby/off, it is a whopping 14.3W. Hence without proper cooling, it runs hot, especially in Australia.

I have always switched mine off at the plug socket rather than use standby.

 

[Frank Dernie]

Mosfets:

What components are the solder lugs pulling off at the top of this picture?

 

[restorer-john]

What components are the solder lugs pulling off at the top of this picture?

They are the transformers, Frank.

It looks worse than it actually is. The windings are thick, looped and soldered to the tabs where they exit. There is only one transformer slightly offset- nothing to worry about.

The transformers are Versa Pac units. About $10 each at cost and can be configured in about 500 combinations.

I was seriously impressed by the standard of the circuitry and assembly. No expense spared. As long as you live in a cool country, it should last for many decades.

 

[Frank Dernie]

As long as you live in a cool country, it should last for many decades.

I don't know how close to the active circuitry the temperature sensors are - they report between 45C (amps) and 55 C (power supply) in use on mine.
Did you look at them during test?

 

[antcollinet]

I certainly do, but I made it my mission to decide at what level of SINAD an item would be audibly transparent to me personally.
To do that I used some old “hairy arsed” engineering techniques, to use the terminology of a veteran Formula 1 engineer.

What I did was to play music in my listening room at a normal to loud listening level, then lower the level until I could only just hear the music.

Unsurprisingly the audibility of the now barely audible sound was frequency dependant (see fletcher munson) but basically if sound is around 60dB lower in level to my normal listening level I can barely hear it.

Now the “hairy arsed” bit. I reason if I can barely hear it at all, it is pretty unlikely to not credible that I would be able to hear it if I were also listening to music at my normal listening level at the same time.

From this I deduce that for me a distortion level of -60dB (0.1%) is unlikely to be audible, and certainly not obvious.

When there should be silence a hiss or noise level at this loudness would probably be noticed from time to time and be annoying, so I estimate (guess really so maybe not good enough but hey…) -80dB of noise would be a reasonable minimum requirement.

Having “calibrated” myself I can now judge from SINAD results where the measured performance needs to be for me.

It boils down to needing to concentrate on speakers and features/reliability since it is rare for electronics to not be good enough for me

Speaking as one hairy arsed to another, that is a pretty good approach while still leaving you with a fair bit of headroom. There is an online test you can do which applies distortion to actual music so you can check your ability to detect distortion with music still playing at your normal listening level.

I scored about -40dB the first time I used it. The second time I couldn't reliably detect the distortion below about -20dB. The second time I was definitely listening to speakers - it is possible (but far from certain) the first I was using earbuds.

My conclusion - I don't have to worry about distortion on almost any kit out there. If I can't hear the noise with the amp cranked, I'm golden.

Sub conclusion - If I can hear distortion - something is broken.

 

[restorer-john]

I don't know how close to the active circuitry the temperature sensors are - they report between 45C (amps) and 55 C (power supply) in use on mine.
Did you look at them during test?

I used a thermal imager on the whole unit interior and then zoomed in on trouble with the thermal macro lens. The amp section actually ran pretty cool, about 65 degrees C up to shutdown. The shutdown at HF is not heat related, it's a deliberate design decision they've made for some reason. The coils/txfs run hotter than the MOSFETs but the SMPS definitely runs too hot for my liking.

Amir simply may have accidentally pushed the power supply too far at 4R and >250wpc both channels driven. I took a single channel up at 1kHz to 250wpc@4R and there was no problems, but I was holding my breath a bit. I don't think anyone would shut one of these amps down in normal use except maybe one of your famous all weekend rave parties in summer.

I'll post some more internal pics on the 'what's on your bench' thread instead of in this review thread.

 

[Soniclife]

I was seriously impressed by the standard of the circuitry and assembly. No expense spared. As long as you live in a cool country, it should last for many decades.

That's good to hear. There is a deeper standby, but I've forgotten how to activate it, it drops consumption considerably.