• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Sonore MicroRendu Hardware Teardown and Review

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
14,768
Likes
7,959
Location
Seattle Area
#1
Some of you have seen my previous measurements of Sonore MicroRendu (http://www.audiosciencereview.com/f...asurements-of-sonore-microrendu-streamer.577/). If you have not, this is a device that "remotes" your DAC by letting you use Ethernet networking to connect to a USB DAC. Outside of this functionality, many buy this for the presumed improvement in audio fidelity. My measurements did not show such improvement however. And to this date, no one has produced any that show otherwise.

For a part 2 of this project I thought I open the unit and show what is inside. The unit is inside of a small, cheezy aluminum case. Put aside any expectation of high-end fit and finish here. A couple of screws opens the one end and lets the guts slide out. What is inside is a main "I/O" board on which there is a daughter card that holds the main CPU that runs all the software.

Dealing with the CPU part, it is a typical all-in-one SoC (System on a Chip) on a PC Board with very little other parts than memory. This is typical of Single Board Computers (SBCs) or Evaluation Modules (EVMs) that CPU manufacturers design to make it easier for others to use the part.

MicroRendu-.jpg


The board is an off-the-shelf unit (i.e. NOT designed by Sonore) offered for sale by the company, SolidRun. See this document for the full specification: https://www.solid-run.com/wiki/lib/...rosom:docs:sr-data_sheet-imx6-microsom-i2.pdf

In single unit quantities it retails for $66 which is quite expensive for what it offers. It is just CPU and memory with no networking interface. It even lacks actual USB connectors. The much more widely used Raspberry Pi 3 board retails for almost half as much and has all of these features plus Quad Core CPU.

That said, I am puzzled by the speed and capabilities of this CPU. It is "too good!" If you are going to build a computer that is supposed to be more quiet than a PC which is Sonore's objective, then I would go as slow as possible, and as few devices running at high speed. This same board comes in one core for example. Why use dual core? Why clock at 1 Ghz? Odd choices to be sure.

The "value add" by Sonore is the board that this CPU is mounted on which I call the "I/O" card (short for input/output):

MicroRendu IO-.jpg


It looks complicated but it is not. Starting from top left, we have the power supply input using the typical barrel type jack. That supply immediately goes into LTM4625 which is a switching step-down regulator. This allows the unit to accept variable input at higher voltages which then gets converted to fixed output needed for the board. Being a switching regulator, it has high efficiency (>80%).

Of note, no matter how clean your input power supply is, the switchmode "post regulator" above will make it "dirty" again, assuming everyone agrees that switching regulators are dirtier than Linear Power Supplies. In that sense, I don't see the focus on using different external power supplies here. Indeed in my testing with two different linear power supplies, there was no difference in performance.

That "dirty" output voltage gets laundered up through yet another set of linear regulators before feeding other circuits. You see a number of examples of them with all that I have noted with the word "regulator" on them. Those linear regulators seem pricey, high performance units so kudos for going the extra miles and being true to their objective of having "clean supply."

The main new functionality there is on top right which is the SMSC USB2412, 2-port USB hub. While I have not checked, I am assuming this is the same approach and part used in Regen. The USB2412 though is nothing special. It is a standard USB hub with no aspirations to doing anything better than taking one input USB port and providing two outputs.

One improvement could come from the clock/crystal that feeds it. That is right below the SMSC input (the part in gold color). The markings say 24.01 Mhz. The spec sheet for the SMSC though says the input clock is 24 Mhz. It is unclear to me why a slightly faster part is used.

Alas, I could not find any parts matching the designations on that crystal. Maybe the "0.01" doesn't mean that it is 24.01 Mhz. And with lack of data on the part, it is impossible to know if it is a premium part or some ordinary crystal.

The rest of the parts from SD card to Ethernet Transceiver are the parts that should have been on the CPU board but are not.

BOM Costs
My quick, back of the envelop cost of material for this unit is around $120 for low quantities. Sonore retails the MicroRendu at $640. So there is a bit of profit in there :).

Summary
Sonore MicroRendu uses an off-the-shelf hobby CPU board as its main computing engine. A simple I/O board with a handful of regulators, some I/O hardware and a USB Hub complete the picture. There is nothing in the design that is unusual or in my opinion, necessary to get better performance out of a high performance DAC. It is the job of a high-performance USB DAC to not care what is upstream of it.

If you are using Sonore MicroRendu for its functionality, i.e. putting your DAC far from the music server computer, that it does. A much more reasonable approach though would be to use Raspberry Pi implementations which will be much cheaper (the Pi 3 retails for just $35).

As usual, I welcome comments/questions/corrections from members and manufacturers.
 
Joined
Nov 6, 2016
Messages
28
Likes
3
#2
I have the microRendu (powered by an Uptone LPS-1 PSU) here. It's a nice-sounding unit, but sounds different to my Win10 audio PC (a Phasure Mach II). I wanted to figure out which was the most 'accurate', but don't have a digital analyzer, so I concocted the following 'test':
a. Feed a file through the microRendu and capture its digital output.
b. Do the same with the Mach II.
c. Subjectively compare the 2 captured files against the original file and determine which one sounds the closest to the original.

Here they are:

1. Original CD rip:

https://drive.google.com/open?id=0B0...UdrMGtRRkZuZkk

2. HQPlayer (no DSP) on music server -> microRendu/LPS-1 -> AR-T Legato USB-to-spdif -> Tascam DA-3000 spdif input:

https://drive.google.com/open?id=0B0...UFxcGs3TDFlclU

3. XXHighEnd (no DSP) on Phasure Mach II audio PC -> AR-T Legato USB-to-spdif -> Tascam DA-3000 spdif input

https://drive.google.com/open?id=0B0...1NkMXhWZTVUd1k

By 'no DSP' I mean that the software player was not purposefully altering the file by upsampling, noise-shaping, format conversion or volume changes. XXHighEnd has a plethora of settings, but I kept as many of these at zero as I possibly could.

If anyone has the time, skills, equipment and/or inclination to analyze file #2 and file #3 on a digital analyzer, feel free to do so. Otherwise just taking a listen to them should suffice, provided your system is of reasonably high quality. Of course your system will add it's own character to all 3 files as they're played back, but this character will be added equally, and you should still be able to determine which of #2 or #3 sounds closest to #1 - I can easily do this on my Ponoplayer!

Hardly scientific, but I found it quite easy to hear the differences between the 3 files, and to make a determination of which captured file most closely resembles the original CD rip.

Mani.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,462
Likes
1,401
Location
Riverview, Fl
#3
Hardly scientific, but I found it quite easy to hear the differences between the 3 files
Why should there be any difference?

Is something broken?
 
Joined
Nov 6, 2016
Messages
28
Likes
3
#4
No, nothing broken...

Mani.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
14,768
Likes
7,959
Location
Seattle Area
#5
Hi Mani. Welcome to the forum and importantly, thank you so much for working to add more data to this analysis. We have company over so I can't do any listening test just now. But did pull the two captured files into Adobe Audition and did a spectrum analysis:

upload_2016-11-6_9-32-35.png


This is for the left channel only (to make the display easier to see). One is in red the other in green. As can be seen, they fall on top of each other so spectrally there is no difference with and without Sonore Microrendu. Would you mind explaining what you hear as far as differences so that I can dig further into that?

Thanks again.
 
Joined
Nov 6, 2016
Messages
28
Likes
3
#6
Hi Amir, thanks for the welcome!

Just to be clear, these are the spectra for the left channel of files #2 and #3, correct?
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,462
Likes
1,401
Location
Riverview, Fl
#7
No difference between track 1 and 2, once the start samples are aligned.

Track three seems to be losing a sample or two along the way, try again on recording #3.
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
14,768
Likes
7,959
Location
Seattle Area
#8
Hi Amir, thanks for the welcome!

Just to be clear, these are the spectra for the left channel of files #2 and #3, correct?
Correct. I was just going to post to clarify that. :)
upload_2016-11-6_9-40-1.png


Thank you for clear naming convention by the way! Too often sample files are provided with names that don't indicate what they are, making it easy to make mistakes.
 
Joined
Nov 6, 2016
Messages
28
Likes
3
#9
No difference between track 1 and 2, onece the start samples are aligned.
They certainly sound different! Take a listen...

Mani.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
14,768
Likes
7,959
Location
Seattle Area
#10
As a side note, I didn't know you owned a Pacific Microsonics model 2. As you may know, when I was at Microsoft, I led the acquisition of the company (maker of HDCD) for their speaker correction technology. I recall at the time thinking I should get one of the Model 2 systems but didn't. I regret that now!
 
Joined
Nov 6, 2016
Messages
28
Likes
3
#11
Track three seems to be losing a sample or two along the way, try again on recording #3.
Yeah, I can't for the life of me get rid of the odd click when using the USB-to-spdif with the Mach II. There's obviously a buffer issue somewhere... But in any event, it doesn't change the way the file sounds (other than the odd click).

Mani.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,462
Likes
1,401
Location
Riverview, Fl
#12
Top:

After aligning the starting sample, and inverting track 1, the result is a NULL when track one and track 2 are combined.

Bottom:

Combining track one and three produces a NULL for a period of time, then goes out of sync - missing data. The NULL can be moved along for any section not missing data. Here, I've time alingned during the second period of continuous data - so you see difference, then NULL, then difference again.

Other than the missing data, I expect track 3 to be the same as track 1.

What's broken?


upload_2016-11-6_12-44-13.png
 
Joined
Nov 6, 2016
Messages
28
Likes
3
#13
As a side note, I didn't know you owned a Pacific Microsonics model 2. As you may know, when I was at Microsoft, I led the acquisition of the company (maker of HDCD) for their speaker correction technology. I recall at the time thinking I should get one of the Model 2 systems but didn't. I regret that now!
Yeah, I was aware of your involvement. Too bad more wasn't made of it by Microsoft (or perhaps it was and we're just not aware?). I sold my Model Two last year to a mastering studio in Paris. Phenomenal machine...

Mani.
 
Joined
Nov 6, 2016
Messages
28
Likes
3
#14

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,462
Likes
1,401
Location
Riverview, Fl
#15
Joined
Nov 6, 2016
Messages
28
Likes
3
#16
I have no idea what's causing them to sound different (other than the obvious odd click on #3). I was hoping that there would be some sort of noise difference showing up that is perhaps contributing to jitter on conversion in the DAC.

But I'm sitting here looking at your analysis wondering WTF could I have been hearing consistently and repeatedly across different playback systems. I might have to capture the analogue output from a DAC with an ADC...

Mani.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,462
Likes
1,401
Location
Riverview, Fl
#17
Looking more closely, I see that track three (at least at this point) has added a zero byte of data.

Top, track 2 (same as track 1)
Middle, track 3, with the stuffed zero byte
Bottom - null track where this discontinuity occurs

There's a stuffed zero byte, then no problem for 22 seconds, then something happens again. It's not enough to change the sound, other than to make a a possibly audible click, occasionally.

upload_2016-11-6_13-11-14.png
 
Joined
Nov 6, 2016
Messages
28
Likes
3
#18
There's a stuffed zero byte, then no problem for 22 seconds, then something happens again. It's not enough to change the sound, other than to make a a possibly audible click, occasionally.
Yes, that's exactly what I'm hearing. This will be due to the buffer issue that I mentioned.

Mani.
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,462
Likes
1,401
Location
Riverview, Fl
#19
It's due to "I don't know".

Could be a buffer problem.

Could be clock rate problem: clock on #3 runs infinitesimally faster than whatever is feeding it and it runs out of data to put into the output stream, so it outputs a null sample.

First error is about 45 seconds into the track. Next observed error 20 seconds after that.

I just break stuff and look at the pieces. Amirm might have a better idea of the internals.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,462
Likes
1,401
Location
Riverview, Fl
#20
Track 3 is about seven samples longer than track one and two.

Beginning is time aligned, very close to the end they are slightly off. A few glitches in four and a half minutes, otherwise, the tracks seem to be identical.

Zoomed in to see individual samples on tracks 2 and 3.

upload_2016-11-6_13-29-26.png
 
Top Bottom