microrendu

[rab]

I like it. A lot. Will post more later, have been crazy busy at work and no time.

[Crom]

Before I get into details this is my current position (so that people can make their own decisions about my bias/knowledge etc). I currently use a dual motherboard (control and player) JPLAY-based system with Roon running as the front end. I have experimented with HQ Player. I like the upsampled sound and am considering purchasing it. The PCs run entirely off good quality linear power supplies – I think 12 or 13 in total with most of the components being individually supplied (for clarity, I currently have a problem with the 12v ATX feed to my control PC so I have been using a 19v pico on the ATX here…not as good but better than nothing). I use compact flash cards instead of SDD/HDD and all the clocks in the USB, CF and motherboards have been replaced. Neutron star clocks for the player motherboard and USB output. DIY clock replacements for the rest. The DIY clocks, CF cards and the USB output chip all run off lifepo batteries. All components have enhanced grounding and all sorts of EMI/RF/vibration reducing treatment has been applied over the years. Internal cabling is bespoke or bought in (e.g. JCAT SATA leads). Software is Windows server 2012 R2 in core mode on the player and a slightly optimised GUI version of 2012 R2 on the control PC (this used to be core but a few months ago I discovered that it seemed to sound better when it was in GUI mode). The only thing that I haven’t got around to doing that I would like is to individually power the RAM. I have banged on about this for so long, experimented with it, found it improved things but not actually done it on these motherboards. However, Paul Hynes has just delivered on an order from over 2 years ago and the supplies are now here which means that I can start to sort out something that will fit in the box as opposed to hang out of it from all sides!

I’m pretty happy with the sound but not enamoured with the amount of effort it takes to listen to music. So when I heard about the microrendu and the reviews it was garnering, it sounded like the answer to a big chunk of my needs. It was delivered on Wed, was a doddle to set up and is being powered with a Paul Hynes power supply. So, how did it play off against my system?

To test this I set up the microrendu with hard-wired, optical network isolation so that it had the same network connection as my PC. I used my control PC as the playback system and HQ Player as the playback software. I could then choose whether to output from HQP on the control PC via the microrendu NAA or via the JPLAY ASIO driver. All I did was then swap between the mrendu and the PC USB cable going into the DAC. Clearly this couldn’t be done in real time but it satisfied me as good enough for me to identify differences.

The little silver box was nothing short of revelatory. Fantastic detail retrieval, good and deep soundstage, great imaging with pinpoint instrument placement. In turn my system had a slightly shallower soundstage but a wider one, similar detail retrieval and very similar imaging. The bottom line is that anyone wanting to achieve great sounding music for a relatively small outlay should get one of these things (a microrendu that is…not an out-of-control DIY project!!).

But is the microrendu better?
I was expecting my system to be noisier (given that it’s a couple of server motherboards rather than a bespoke system on a chip) and this is perhaps demonstrated by the shallower soundstage but the blackness from which vocals appear was pretty much identical between the systems – no real winner here. Previous experience trying to use and optimise linux for audio told me to expect markedly worse quality than I heard. I have never heard a linux–based system sounding so good. OK, so pretty much a draw so far. I flicked back and forth between the sources. Perhaps there was greater depth of bass from the PC? Or perhaps the bass is just less taught coming from the PC? What was niggling me? After a couple of hours, I worked out that I enjoyed listening to the windows PC more, but ‘why' was harder to pin down. Clearly this could be because my audio pc is a work of love and attention and no little box is going to be able to beat it(!!)…or maybe not. After more hours of listening a few things have become apparent: in comparison to my system the mrendu sounds a little clinical, digital, cold. The start of Maggie Said by Natalie Merchant has got some really delicate string work . From the PC, each pluck was distinct to the extent that together, they danced around in front of me. In comparison, on the mrendu, the engagement I felt was less, it felt a little too controlled. It would be easy to misinterpret what I’m saying here because I am really struggling to explain what I perceived. This isn’t a valve/SS point, a warm/analytical dichotomy, although it is clearly along similar lines – but it’s more subtle than that. Back in Black sounded exciting and dangerous from the PC, on the mrendu it sounded perfect but less alive than I feel it should.

As an aside, one thing that I did discover early on was that plugging in a lifepo-powered regen (orange) improved both the PC output AND that of the rendu. Interesting given that reportedly the output of the mrendu already has a regen built in.

Having said all that, I don’t think I am being very fair on a number of counts. I have literally plugged the mrendu in an switched it on as opposed to spending an inordinate amount of time tweaking things in the PC to get the sound stage to how I want it and it could easily be that the space it produces around the instruments is the difference I am hearing – but I honestly am unable to say that for sure. The PH power supply I am using: although it is one of the best that you can buy and was specifically designed for low impedance power delivery to digital circuits, it was not designed specifically for the mrendu. Finally, although this piece is about ‘the best for me’ and therefore cost doesn’t really have a place in this discussion it seems only fair to point out that it’s a £600 piece of kit being compared to something bespoke - where £600 might pay for a couple of the neutron star clocks.

Remembering what happened when I upgraded the clocks in the PC’s, I think that the mrendu could match/exceed the PC if we could upgrade the clocks – or at least power them separately (lifepo) but this is not going to be as simple as with a big old motherboard.

So, what am I going to do. Difficult decision. No idea. I could take the easy way out and say that on bang-for-buck and ease of use and almost on SQ basis I keep the mrendu and happily live ever after. However, I’m striving for the best for me, and on that basis I need to do some further testing to see if I am doing the mrendu a disservice with the current power supply I am using. I am going to try powering the microrendu with lifepo’s and see what happens.

Please don’t take this post as a mrendu slanging - it is not in any sense meant to be. Quite the opposite in fact. If I was anyone other than someone who likes to tweak things to pieces I would be very happy. If I had a caps4 or some other slightly-optimised-for-audio PC then I am sure that a decently powered mRendu would easily be a match for it – this is clearly apparent in the reception its received. My point in posting this was to help to gather my own thoughts but also to help those similar to myself and to see if my experience rings true with them.

Whatever I find, it is clear that the mass-market democratisation of great computer audio is now happening and that is no bad thing. For too long it has been really tough to get any file-based (i.e. not disc–based) source to sound good and there have been many pretenders to the throne.

[nige2000]

U using this with ians fifo and neutron clocks?
Bound to cover up some sins

[motberg]

Thanks for taking the time for the post Crom.
I have read that the MR requires some significant break in time to come to form....

If you do not mind, do you have a preference with HQPlayer as NAA direct, or through JPlay?
I have not yet tried HQPlayer, but really like the kernel streaming option in JPlay (using WinServer and A/O)

[Crom]

@Nigel. Yes, DAC has Ian's FIFO in it with 2 x Neutron Star clocks running at 45/49M. Audio player PC motherboard has 25M Neutron Star and USB card (your recipe) has 20M Neutron star - might be 24M...can't remember...whichever is correct ;-)

@Motberg - interesting about burn in time, thanks. For future visitors, the impressions above were literally out of the box! I'll leave it on 24/7 for a few days and listen critically again.

My audio player PC is in core mode 2012 and running the NAA file just crashes. I know it needs various runtime DLL's etc and having spent 2 hours trying to install them either through various KB updates or by manually copying the files, nothing works. So, I gave up that path. I then tried to install the NAA on my control PC and had the same problem - absolute no no. The fault is mine because I have messed with the windows install so much that it doesn't know whether it's coming or going. I managed to get all the required runtime files installed but there was some version incompatibility between them and I just couldn't be bothered to spend any more time. It is one of the problems with using Compact Flash as the Hard drive. everything takes forever to install. I am planning on re-installing both audio and control PCs at which point I will do all the installing on an SDD and then copy image across to the CF when I'm happy everything is working.

So, to answer your question, the only NAA direct I have tried is within the mRendu. I did try an install of HQP on the server that is running Roon core so that I could compare running HQP on my control PC against a relatively high spec server. The control PC won on SQ (although HQP starts slower and there were a few of the filters that I remember not working) and I expect the performance gap to increase when I get around to fixing the linear ATX line for the control PC and return the control PC to full linear psu.

I have always found JPLAY (along with the 2nd audioplayer PC) to help to wring the last ounce of performance of pretty much anything I've tried: bughead, HPQ, Roon etc. Audiophile Optimizer is IMO indispensable in such an install! I will give bughead etc a go but for any serious use I need a nice interface. I spend too much time looking at remote desktop to do it more during downtime ;-)

I've just bodged a couple of lifepos to get a 6.6-7.2v supply. Let's see what happens later. FYI, the paul hynes was set at 7.5v - might try altering that a little to see if it makes any difference.

Another thing I'll try is to disconnect the 5v jumper that I put on the regen. My previous USB > I2S card (waveio) needed 5v to be present but I don't think the DIYINHK needs this so I'll remove it.

[nige2000]

@Nigel. Yes, DAC has Ian's FIFO in it with 2 x Neutron Star clocks running at 45/49M. Audio player PC motherboard has 25M Neutron Star and USB card (your recipe) has 20M Neutron star - might be 24M...can't remember...whichever is correct ;-)

usb is 12m family but usually 24m.
thing is almost no-one has a dac like yours, that tries so hard to alleviate upstream flaws
have you a normalish dac to retest?
as the fifo will fix much of the i2s signal
i remember having a rpi with ians fifo on lifepo4s before that was very close to sq of an audio pc quality before but not just there.....
would certainly would likely better some of the half assed caps pcs

I've just bodged a couple of lifepos to get a 6.6-7.2v supply. Let's see what happens later. FYI, the paul hynes was set at 7.5v - might try altering that a little to see if it makes any difference.

might be interesting alright the MR will have on board regulation which takes a lot of the benefit of using super power supplies
if you could bypass these regulators and do clock replacements and power all direct from lifepo4 it would provide a good benchmark

Another thing I'll try is to disconnect the 5v jumper that I put on the regen. My previous USB > I2S card (waveio) needed 5v to be present but I don't think the DIYINHK needs this so I'll remove it.

5v handshake is not required

[Crom]

@Nigel. Yes, DAC has Ian's FIFO in it with 2 x Neutron Star clocks running at 45/49M. Audio player PC motherboard has 25M Neutron Star and USB card (your recipe) has 20M Neutron star - might be 24M...can't remember...whichever is correct ;-)

usb is 12m family but usually 24m.
thing is almost no-one has a dac like yours, that tries so hard to alleviate upstream flaws
have you a normalish dac to retest?
as the fifo will fix much of the i2s signal
i remember having a rpi with ians fifo on lifepo4s before that was very close to sq of an audio pc quality before but not just there.....
would certainly would likely better some of the half assed caps pcs

Not sure what you're hoping for here. The DAC is the common element in the testing setup, so swapping the DAC would just change both results? oh hang on... are you saying that the FIFO is clearing up the signal to the same level but is working harder and there are potentially more opps to improve when connected to the MR? ie the resultant signal is the pretty much the same? I think that's what you must mean...yes, agreed I guess that could be a leveller in this test..unfortunately I don't have another DAC at the moment.

I've just bodged a couple of lifepos to get a 6.6-7.2v supply. Let's see what happens later. FYI, the paul hynes was set at 7.5v - might try altering that a little to see if it makes any difference.

might be interesting alright the MR will have on board regulation which takes a lot of the benefit of using super power supplies
if you could bypass these regulators and do clock replacements and power all direct from lifepo4 it would provide a good benchmark

Yes, for sure, powering directly and using multiple cells would be massively preferential and I bet that there is more to gain here. It's going to be tricky to achieve that though and a challenge that I probably won't rise to until everything is run in. I'll need to spend a few hours with a multimeter to find out what's going on in there and even if it's possible ;-)

Another thing I'll try is to disconnect the 5v jumper that I put on the regen. My previous USB > I2S card (waveio) needed 5v to be present but I don't think the DIYINHK needs this so I'll remove it.

5v handshake is not required

Thanks and I can confirm that this worked!

I've now tried the lifepo's and can confirm decent improvement over the Paul Hynes regulator. I was quite surprised by this because in the past PH regs have, for me, come v close to the lifepos but in this case - and this may be an indication of how critical the power supply is for the MR - there was an improvement which, to me, closed the gap on the PC and removed the feeling that I was 'missing some excitement' from what I was listening to. I ran out of time yesterday so serious AB'ing didn't happen. I'll revisit once run in.

[nige2000]

Not sure what you're hoping for here. The DAC is the common element in the testing setup, so swapping the DAC would just change both results? oh hang on... are you saying that the FIFO is clearing up the signal to the same level but is working harder and there are potentially more opps to improve when connected to the MR? ie the resultant signal is the pretty much the same? I think that's what you must mean...yes, agreed I guess that could be a leveller in this test..unfortunately I don't have another DAC at the moment.

yea pretty much i would suspect your dac could substantially but not completely level out the differences between sources

I've now tried the lifepo's and can confirm decent improvement over the Paul Hynes regulator. I was quite surprised by this because in the past PH regs have, for me, come v close to the lifepos but in this case - and this may be an indication of how critical the power supply is for the MR - there was an improvement which, to me, closed the gap on the PC and removed the feeling that I was 'missing some excitement' from what I was listening to. I ran out of time yesterday so serious AB'ing didn't happen. I'll revisit once run in.

wow....
direct power sure would be interesting then so...

[Sloop John B]

Wonderful post in the mRendu Crom, very interesting to hear such opinions.

If you decide to move the mRendu on I would be interested in acquiring it.


SJB

[Crom]

Not sure what you're hoping for here. The DAC is the common element in the testing setup, so swapping the DAC would just change both results? oh hang on... are you saying that the FIFO is clearing up the signal to the same level but is working harder and there are potentially more opps to improve when connected to the MR? ie the resultant signal is the pretty much the same? I think that's what you must mean...yes, agreed I guess that could be a leveller in this test..unfortunately I don't have another DAC at the moment.

yea pretty much i would suspect your dac could substantially but not completely level out the differences between sources

OK, I understand. However, a few timesI have tried standard mac laptop as source in the past and without any trouble been able to tell the difference between this and PC streamer. It would be interesting with a less-optimising dac. I haven't compared laptop to mrendu although I'm not sure it's worth it given the number of existing comparisons against standard kit.

I've now tried the lifepo's and can confirm decent improvement over the Paul Hynes regulator. I was quite surprised by this because in the past PH regs have, for me, come v close to the lifepos but in this case - and this may be an indication of how critical the power supply is for the MR - there was an improvement which, to me, closed the gap on the PC and removed the feeling that I was 'missing some excitement' from what I was listening to. I ran out of time yesterday so serious AB'ing didn't happen. I'll revisit once run in.

wow....
direct power sure would be interesting then so...

Agreed ;-)

@Sloop John B - thanks for the kind words and the offer. I'm not sure it's going anywhere at the moment but I'll keep you posted.