Phono stage hiss - Measurements

[Dev_Steve]

Hi Everyone!

Alright, so I followed the steps and did my best, but I'm not sure if Audacity recorded the output. However, I also sent over a recording of the hiss that I took from a dB meter I had on my phone. The reality is if I set the phono stage at the required settings for my cartridge without my SUT, there is a massive amount of audible hiss at settings as low as -20 of gain on the Devialet.

I'm glad Mathieu is looking into this, but its just surprising to me that he's indicating this should be a "normal" amount of hiss. In fact, when I got my first 250, I thought the unit was defective and actually sent it back. It seems that anyone at Devialet headquarters can just listen for themselves by plugging a unit into some speakers. I hope he tries this too.

Anyway, I'm really excited to see the number of people on this forum joining in on trying to resolve this issue. Not only are we all passionate about music, we are passionate about what we all know would be the most compelling equipment on the market. I'm glad we're all helping to make the Expert Pro the best it can be.

[thumb5]

...
This also makes me think of some posts a while ago which suggested that the phono stage in the D-Premier was superior to that ultimately used in the Expert series and onwards.  I have no idea if there is actually a technical difference with the D-Premier phono stage, but one thing that would be fascinating to do would be compare a "hiss" recording of a D-Premier phono stage with one from an Expert.
...

Notwithstanding the other posts you mentioned, Mathieu is on record at the OAC event in July saying that the phono stages are "technically identical" in the D-Premier and Expert series (other than the 100/110).  Make of that what you will

[Finn]

It Will be Wery interessant What Mathieu find out about the hiss ploblem.

[Confused]

An update to my measurements.

I have swapped a couple of emails with Mathieu, I have noticed a minor anomaly with my initial measurements, and I have also just received delivery of a copy of the "ultimate analogue test LP".  So a few things to update, I shall take them one at a time.

The minor anomaly with my original measurements is that I had the phono stage set to 30 ohms.  Recently, @Dev_Steve posted on my old "Cadenza Black sudden death" thread.  Irrelevant here perhaps, but in my first post I mentioned how I had used RAM to settle on a cartridge loading of 140 ohms.

https://devialetchat.com/Thread-Cadenza-...dden-death

Reading back through this tread after Steve posted, it dawned on me that I have never set the phono stage as low as 30 ohms, and the gain is slightly wrong at 0.37mV.  I am thinking that I must have accidently changed this is RAM whilst looking at other settings.  A minor thing, and perhaps of no consequence here, but for clarity, any measurements I make from now on will be with my preferred setting of 140 ohms and a phono gain of 0.4mV.

So to the rather strange anomaly in Mathieu's analysis of my hiss recording, an "up tick" in level after 14 kHz.  I mentioned in an earlier post and an email to Mathieu that this "up tick" was not picked up in my Audacity analysis.  Further checking revelled that this statement was wrong.  The graph that Audacity produces does not show the up tick, but the lower end of the scale is too high to show it anyway.  What I have subsequently discovered is that if you hover the curser over the graph in Audacity you can manually read off the dB level, doing this reveals the up tick when I check in Audacity.  So the results are consistent, it is just the case that Mathieu has better measurement kit than I do, perhaps no surprise and certainly preferable than being the other way around!

With respect to my "up tick", Mathieu offered no direct explanation as to why it is there, but he did suggest a further test.  With the dual mono Pro's the master and companion amps are identical, so Mathieu suggested I could retry the measurements using the companion's  phono stage.  An interesting idea, as the two phono stages should in theory be identical.

Then to the test LP.  The idea of the test LP is to make a "hiss" recording, then a second recording using one track on the LP which is a reference level 1 kHz sine wave.  This should allow a signal to noise ratio to be established.  A quick comment on the LP, it has many tracks that can be used to optimise a turntable set-up, in particular if your turntable has lots of tonearm adjustment options and similar.  When I have some time to try all this, I will start another thread.  For now, it is fair to say that the LP should provide me more benefit than establishing a signal to noise ratio figure, which is good news as it is not a cheap LP!  What I can say is that the LP seems to be very good quality, it weighs over 200g, it is absolutely flat and true, and it gives every appearance of being a top quality pressing.  It is possibly one of the highest quality records that I have, it is just a shame there is no music on it.

So what I have done now is made another "hiss recording" with my cartridge loading at a more practical 140 ohms, a reference level recording with the same settings, then repeated with the companion's phono stage.  One point to note, switching to the companion phono stage was not as hard as you might think, it was a case of just swapping the SD cards between the two amps, and swapping the turntable's RCA cables also.  OK, I would have needed to swap all cables if I was to permanently switch between master and companion, but it was less hassle than you might think just for this test.

Here are the results:

Hiss master

   

Hiss companion

   

1 kHz master

   

1 kHz companion

   

Interestingly, it looks to me like my master phono stage has marginally less hiss than the companion.  That said, the graphs do not make total sense to me.  Indeed, the 1 kHz tone graphs do not look anything like what I would expect from a 1 kHz tone.  For the record, it does sound like a 1 kHz tome when you listen to it.  I presume Mathieu's analysis will be better, I shall send the files to him and see what he make of it all. 

Also, I am not sure how you derive a signal to noise ratio from the above data.  Obviously the 1 kHz tone is louder, but what is the correct methodology to calculate a valid signal to noise ratio?

Also, both the master and companion plots still show the 14 kHz "up tick" when checked in Audacity, so this holds true for both amps and a 140 ohm loading.

[Soniclife]

Interesting stuff.

Indeed, the 1 kHz tone graphs do not look anything like what I would expect from a 1 kHz tone.  For the record, it does sound like a 1 kHz tome when you listen to it.

I assume that is with the sub-sonic filter off, I think this shows you should have it on.  If it is with it on I think you have a TT rumble problem.  I say problem but I expect lots of us might do with vinyl setups, they are just so rarely measured we have no idea.

[Confused]

Interesting stuff.

Indeed, the 1 kHz tone graphs do not look anything like what I would expect from a 1 kHz tone.  For the record, it does sound like a 1 kHz tome when you listen to it.

I assume that is with the sub-sonic filter off, I think this shows you should have it on.  If it is with it on I think you have a TT rumble problem.  I say problem but I expect lots of us might do with vinyl setups, they are just so rarely measured we have no idea.

Looking at the Audacity plots I know exactly what you mean, however, I can confirm that these measurements were taken with the RIAA 1976 curve, which includes a subsonic filter by default.

Frankly, the Audacity plots are not very good, I will certainly be interested in seeing how things look if Mathieu Pernot runs these through his test gear.

As an aside, if I hover the curser over the actual plot in Audacity, it shows a peak of - 7.3 dB at 1003 Hz, this is not clear from viewing the graph itself.  I would like to find some better software for performing the analysis, I am sure there would be a few issues you could discover and potentially mitigate with measurements like this.  Subjectively, listening to the test tone you would say there is zero output from 1 to 100 Hz, but that is not what the measurements say.  It is interesting stuff.

EDIT:  I tried listening back to the 1 kHz recording via headphones.  Yes, it still sounds like a 1 kHz test tone, but on headphones you can clearly hear some vinyl surface noise, and a very, very, quiet, vey subtle, thump sound.  Ah the joys of vinyl "character".  I am not sure if the subtle thump thing is on the test record or my TT, although I suspect it sounds more like something on the record.  Something for further investigation another time …..

[Soniclife]

Ah, if the real peak level for the 1KHz is way higher then it makes much more sense. Are the peak values elsewhere as per the graph?

I have forgotten how 1976 vs subsonic comes out at the low end, I use 1953 + sub filter.

[Soniclife]

P.S. If you hover over the 2 and 3 K levels you can see how much distortion your setup has.

[Confused]

P.S. If you hover over the 2 and 3 K levels you can see how much distortion your setup has.

Trying to extract the worst possible figures, I can see -78 dB at 1995 Hz and -95 at 2989 Hz.  This compares to -7.3 dB at 1003 Hz, as mentioned earlier.

I have no idea how good or bad this is though ….

[Soniclife]

So 2nd is 70.7 db below fundamental, and 3rd is 87.7 db below, which according to an online calc is 0.03% and 0.004% distortion, I think that's excellent. Things probably look way worse at lower and higher frequencies though.