Open and Tolerant: MBL's Juergen Reis on Listening, Measurements, and (Un)Certainty
Juergen Reis is MBL's chief engineer, which means he's the man behind every product from MBL from DACs, to Players, to Preamps, to Amps, to the ever enticing MBL Radialstrahler speakers. I first read Juergen's following remarks on another website where he posted them as a series of comments in response to a discussion about "USB tweaks" and the notion that what cannot be measured cannot make a sonic difference.
As with much of what Juergen says and writes, I found his comments very interesting, informative, and relevant to some of the discussions that have taken place here on AudioStream and elsewhere on the web. With that in mind, I asked Juergen if I could publish them here, in a more finished form, and he very graciously said...yes. Enjoy!
I am an electrical engineer, working in the audio field for over 30 years and working extensively with measurement systems to improve and evaluate audio products, so this is always my starting point. [footnote 1]
But I am also a musician for over 30 years and I also have a small recording studio, so technique and measurement aside, I enjoy making and also listening to music a lot.
From time to time I come across some tweaks and products that I really do not totally understand from the "engineering" standpoint, but I am open to giving them a try. And so it also happens from time to time, that after some time I find a suitable measurement that wasn't in my mind when I first "tested" a questionable tweak or product.
So it's the case with some USB tweaks that do not change the total response but I can measure differences when measuring with probes inside. And there are also some products, e.g. USB and LAN cables, that I still cannot measure, but I have no problem accepting their influence over the years, and don't call them snake oil. Sure, there are some snake oil products out there, no doubt, but they will fail to impress over time, when the hype is over.
"Be open to new ideas, to different ideas, even if they look 'strange' at first sight."
But my main point here is: Be open to new ideas, to different ideas, even if they look "strange" at first sight.
Having been in this business for a long time, I could tell a lot of stories but I would like to share just my first experience concerning what I have heard, but what I couldn't measure at first. This was 32 years ago so looking back this seems to be too easy, but it was my first 'strange' case.
I came relatively fresh from finishing my engineer study and finished my first preamp development. With that I was at a HiFi show demonstrating this preamp with a loan power amp and some loudspeakers. I was proud and happy with the sound.
During this show, a visitor came by and asked: Have you checked the polarity of the mains? I was a bit surprised by this question, but after the end of the show day I rotated the orientation of the mains plug and was "confused" because there was a clear difference in the sound. I had heard of this phenomenon from some "high end freaks", but coming fresh from my engineering degree I was skeptical and thought this is just some "snake oil talk".
After the show, I connected the preamp again to the measurement gear (those times, a HP meter bridge) and couldn't see any difference in the measurement with both orientations of the mains plug of the preamp. I was stumped. Then I repeated listening to the preamp (those years with a turntable as a source) with some other loaned power amp and I could again hear a difference in sound when rotating the mains.
Then I began to read some papers, mainly some older papers from Bell Laboratories that I got on loan from a large library (you must remember, that 32 years ago, there was no Internet or Google, where you can browse), and read some detailed analysis of power transformers, about capacitive and inductive stray coupling, and this led my thinking to go in the right direction.
To cross check my thoughts, I set up the preamp and the power amp together, via RCA, as I had done during listening and with this "complete" setup I was able to measure a difference in performance when rotating the mains power cord. Now it was clear to me that with a hifi setup out of two devices (or more) that need voltage from the mains, and which are connected with an RCA connection, that there will always be an influence on performance depending of the orientation of the mains.
If I checked the power amp alone with this measurement setup, then again I couldn't measure any difference with the orientation of the mains cord. This is because good measurement devices have a very high common mode rejection at the output and at the input so they suppress the influence of the stray capacitance of the mains transformer of the device under test (DUT).
With this knowledge I began to work on a method to test and measure the common mode performance of a DUT between mains and signal ground and also to measure the common mode suppression of a mains transformer. From that experience on, I designed proprietary mains transformers with very high common mode rejection so that the stray signal that tries to enter into the device coming from the mains will be shielded to earth and the stray signal that tries to leave the device into the mains will be shielded to the power supply ground.
I would like to have a preamp that sounds as it should, no matter what orientation the mains cord has to the mains outlet. If you draw a complete signal flow diagram, with all stray currents, you will realize that beside the desired connection between the preamp and the power amp, there is also a "back door" where the signal creeps via the stray capacitance from both power transformers as a parallel path. And if you draw this diagram you will also see that with the typical impedance of RCA inputs at the power amp, and so with typical low signal current between pre and power amp, that these back door currents can have a much lower impedance at mid and high frequencies than the desired main signal path.
This was my first experience, that there are more / different points on a product's performance when connected as a complete system that influence the sound than would be obvious when measuring just one DUT, which is done with most audio magazines for obvious reasons, to see and judge only the performance of the device under test.
A bit later, I set up a hi-fi system where every power transformer was driven by its own battery driven mains (having a battery driven sine wave generator, driving a battery driven power amp, driving a power transformer in inverse mode, so at the output is the mains voltage and frequency, but all battery driven). With that experience I began to optimize not only the common mode rejection (due to stray capacitance) but also the differential mode rejection (due to stray inductance). But this is another story.
Similar things are happening with digital inputs and outputs that are galvanically isolated (meaning no Ohm connection, and so no DC current is flowing), but due to stray capacitance of the digital audio transformers, there is an unwanted leakage path of the digital audio signal. Even this stray capacitance can be lower than 10 pF, but with the Biphase Signal in the MHz area, it can influence the total quality of a DAC.
"...you would be surprised that common mode disturbances and stray coupling does have a larger impact on the total performance than the individual devices."
For example, nowadays nearly every Digital Source, DAC, or Preamp measured alone has fairly good measured performance. But when you connect them together as a complete system, with all sources connected, you would be surprised that common mode disturbances and stray coupling does have a larger impact on the total performance than the individual devices. So the complete setup has to be measured and not only one DUT.
Also, using sine waves does not show the complete picture because even good codecs can deliver good data with sine waves. You must use a test signal that shows you differential and intermodulation distortion, noise floor modulation, aliasing, and jitter, and has a similar DR ratio as typical music. This would give you a more realistic picture of what is going on.
And lastly, using only a pure ohm load for the power amp when measuring the complete setup does not mirror what you are hearing with loudspeakers. Comparing power amp output on real loudspeakers (with complex loads) will show you many more differences than comparing the power amps on a plain ohm load. I could test every phase angle between voltage and current, so simulating everything between a pure capacitive load to a pure inductive load.
"...you need measurements that are much closer to the actual listening setup in order to better compare and find what you are able to identify while listening."
What I want to say is you need measurements that are much closer to the actual listening setup in order to better compare and find what you are able to identify while listening. And with these more complete setups during measurement, you will also see many more things that influence the sound, as compared to when you just set up one device under test for measurement.
So in brief: Measure all devices in this way, i.e. how they are connected together during listing, and use test signals that are closer to the kind and dynamic range of the music you are listening to. Also, use a more realistic complex load for the power amp when the speaker is connected and these three points together will give you a more realistic picture of what you are hearing. Even if you just probe the DAC's output for your measurements, still send the signal to the preamp, the power amp, and to the speakers.
This is just one example, out of many, that if you look deeper into a problem, you can answer questions that are not obvious at first sight, sometimes even afters years.
It is good to stay open to different ideas (also to different cultures, …) and learn from each other. There is much more going on that we do not understand at first sight that is worth looking into.
To summarize the main points:
- Set up the system for measurement in the same way you listen to it, with every component connected (even this component is off, ...)
- Take more appropriate test signals that better cover the content of the music you listen to, and not just "stupid" sine waves.
- Have a "real complex load" to the power amp, even if you only probe measurement on the DAC. This makes a difference.
Footnote 1: John Atkinson has adopted Juergen's test to measure the performance of a DAC's reconstruction filter for his Stereophile measurements.