NAA with Lampi and HQP on Mac Air for DSD

[bonzo]

Hi,

 

I have just started setting up a new system with the Lampi Big 7 and would like to proceed with all upsampled to DSD, so using HQPlayer. I have tried reading the HQPlayer thread before this posting.

 

Instead of using the SMS - 100 NAA, I am using a Rasberry Pi like thing called Hummingboard, that is doing it's job. My system is in too early a state to assess quality (Shifted from Martin Logan Summits to Verity Audio Leonores and from audio research reference amps to Jadis pre, and will be changing amps a lot over the next 4 months to make any conclusive judgements of computer audio SQ). So prefer to be handheld now

 

Some questions:

1. Is there a difference between different NAAs? Should I expect the SMS - 100 to sound different from the Hummingboard, and will a fitlet-b be different?

 

2. Can a powerful NAA, i.e. a CAPS like computer being used as NAA sound much better than these smaller NUC like NAAs? Can I get Win12 and AO on a nuc like NAA?

 

3. My controller is a Mac Air, no mods, use it noisily. Macbook Air OS X 10.9.5 1.4 Ghz, Intel Core i5

Memory 4GB 1600 Mhz DDR3. Is there anything I can do here?

 

4. Settings: Unchecked the DirectSDM setting, volume max set to -3dbfs. DAC bits to 24, DSD5, max rate for SDM is going to be 5644800. Anything to be done there?

 

5. Anything to be done to settings at Midi level in the Mac?

 

TIA

[EuroDriver]

1. Is there a difference between different NAAs? Should I expect the SMS - 100 to sound different from the Hummingboard, and will a fitlet-b be different?

 

Most of the activity comparing the SQ of various NAA flavors on different OS's is happening on the Italian Nexthardware forum. There is some amazing stuff and testing going on. With some degree of confidence one can say the Win and MAC OS NAA's sound better than the Linux NAA running on the Cubox-i

 

as to the interesting OS choices for the NAA, they are

- WS2012 in Core mode (using AO to make life a bit easier)

- Win 8 in SAFE mode + script to start the needed dll's and services for NAA to run

- Win 7 or 8 using Primo RAM disk

- WS2012 with Primo RAM disk

- MAC OS in "safe" mode

An important aspect not to forget is which is the best sounding driver / OS for your DAC's USB receiver. It seems that ASIO is ahead in this area for some DAC's / USB receiver

 

i have only experience with NAA on WS2012 in Minimal Server mode using AO, and it sounds very, very good. The increased perceived sound level of the bass notes is simply astounding, I have never heard such a difference in perceived bass volume in all my 40 years of chasing better sound. I think that there is a good chance that one of the above OS setups will sound even better

 

2. Can a powerful NAA, i.e. a CAPS like computer being used as NAA sound much better than these smaller NUC like NAAs? Can I get Win12 and AO on a nuc like NAA?

 

The Cubox experience shows that small is not necessarily the most SQ beautiful. What is important is that the network interface needs to be as isolated as possible from the USB. Which USB transceiver works well is very DAC / USB receiver dependant. PPA V2, or SOTM, or a USB regenerator are all potentialy good solutions. Let's also not forget that the power supply of the NAA is very important, and one needs a hardware platform that allows a feed of a clean 5v to the USB. Another RF generating source to pay attention to is the SSD, and here a battery supply can be helpful. Getting the SSD out of the picture, and using SD or compact flash has its attractions too. There are many ways to skin the NAA hardware cat ;-)

 

3. My controller is a Mac Air, no mods, use it noisily. Macbook Air OS X 10.9.5 1.4 Ghz, Intel Core i5

Memory 4GB 1600 Mhz DDR3. Is there anything I can do here?

 

your Mac Book Air does not have sufficent CPU processing capability to do 16/44 to DSD 256 (Lampi will be getting DSD256 capability soon) using Polysinc filter, which still sounds the best in 3.6.1.1 (IME)

 

 

4. Settings: Unchecked the DirectSDM setting, volume max set to -3dbfs. DAC bits to 24, DSD5, max rate for SDM is going to be 5644800. Anything to be done there?

 

my favorite is still DSD7, and then ASDM5 for some SQ challenged tracks

 

5. Anything to be done to settings at Midi level in the Mac?

 

sorry not familiar with Mac midi settings

[bonzo]

 

The Cubox experience shows that small is not necessarily the most SQ beautiful. What is important is that the network interface needs to be as isolated as possible from the USB. Which USB transceiver works well is very DAC / USB receiver dependant. PPA V2, or SOTM, or a USB regenerator are all potentialy good solutions. Let's also not forget that the power supply of the NAA is very important, and one needs a hardware platform that allows a feed of a clean 5v to the USB. Another RF generating source to pay attention to is the SSD, and here a battery supply can be helpful. Getting the SSD out of the picture, and using SD or compact flash has its attractions too. There are many ways to skin the NAA hardware cat ;-)

 

 

 

your Mac Book Air does not have sufficent CPU processing capability to do 16/44 to DSD 256 (Lampi will be getting DSD256 capability soon) using Polysinc filter, which still sounds the best in 3.6.1.1 (IME)

 

 

 

Awesome thanks for that detailed reply. On above two points, if the NAA is battery powered, and if the Hard drive is connected to the Mac via USB, but not directly to the NAA, that should solve that part of the problem right? Will a powerful CAPS like NAA still have advantages then? And possibly add the Regen later?

 

And do you think my Mac is fine for DSD128?

[EuroDriver]

NAA battery powered will certainly do no harm, but a good LPS will work well too.

 

Hard drive connected to your MacBook over USB should be fine as your MacBook USB will not be in the data chain feeding the NAA.

 

There is almost certainly an advantage to your MacBook connecting to your LAN over WiFi and not over CAT 5/6 . It seems like NAA's such as SOTM SMS100 ( a good plug and play candidate for your Lampi I suspect) and my DH61DL are pretty sensitive to RF noise coming over CAT5

 

DSD128 might be OK with Polysinc, you will need to try. For sure Polysinc-2s will have lower CPU load and can be handled by your i5 MacBook, but it does not sound as good IME

 

i think a A/B comparison between the SOTM SMS-100 and a WS2012 NAA would be an interesting test for you. Hopefully in the next few months there will be a comparison

[bonzo]

 

There is almost certainly an advantage to your MacBook connecting to your LAN over WiFi and not over CAT 5/6 . It seems like NAA's such as SOTM SMS100 ( a good plug and play candidate for your Lampi I suspect) and my DH61DL are pretty sensitive to RF noise coming over CAT5

 

 

Thanks - So my Mac connects over Wifi, but the NAA is connected to the router using a CAT6. You are saying this is sensitive to noise? How can that be changed, except by adding a laptop for an NAA?

[EuroDriver]

Thanks - So my Mac connects over Wifi, but the NAA is connected to the router using a CAT6. You are saying this is sensitive to noise? How can that be changed, except by adding a laptop for an NAA?

 

As long as you don't have high powered PC's doing heavy lifting connected to the LAN, and a lot of LAN traffic from let's say a NAS, then the network should be pretty quiet.

 

laptops with their built in graphics, are not RF quiet

 

in 2 weeks time I will be testing if a pair of Ethernet Fibre Media Converters can usefully reduce the RF sensitivity of NAA's, SMS-100 and my DH61DL

[Superdad]

Bonzo:

 

Thanks for directing me over to this thread. Edward's (EuroDriver) advice is all good. He has much more experience with NAA variations than I do, so follow him!

 

I too reported a while back that cheap, non-audio optimized NAAs (like CuBox-i) are sub-par for NAA use (even when powered by a great LPS). You may see something like a SOtM sMS-100 but truly optimized from scratch later this year.

 

For best SQ you will need to ditch your laptop. My wife and daughter have MacBook Pro and Air respectively, and I booted them both with the same slimmed-Mavericks SD card I use for my 2012 mini, dismounted their internal drives, and compared them (both on battery and AC adapter). They both fell WAY behind my mini, in the bass especially, but overall lifeless as well.

 

I'm not 100% with EuroDriver on the wifi thing yet. The BlueJeans/Belden Cat6a cables are really head and shoulders above typical cables (affordable too!), and I find bypassing the Ethernet switch/router entirely makes a really nice difference. A few of my clients, not wanting to do my dual-Mac direct Ethernet set-up (described to you in the other thread) have had nice results with the current Airport Extreme--running a good cable directly to it, and hanging a USB HD off the AE to allow their music computer to access the shared drive. Of course this is very different than the NAA route.

 

One permutation that EuroDrive has not really been able to compare due to there not being Linux drivers for his DAC IIRC, is to run test-only Linux NAA on a good machine. Such would have much fewer processes running than under Windows, be far less hassle, and be cheaper than WS2012.

 

BTW Edward:

You mention the NAA permeation of booting OS X in Safe Mode. But then you are of course taking about running the new OS X version of the NAA daemon. I have not tried that yet (when I want to NAA I boot my mini with Miska's text-only Linux NAA from USB thumb drive), but since I recall Miska saying that the OS X NAA still goes through CoreAudio, I have my doubts about that being a sonic contender.

 

Best,

--Alex C.

[bonzo]

Hi Superdad, so the Mac mini comparisons you mentioned, were they while using the Mini as a controller for a NAA? Given that you are connected through a cable that might make more difference?

[Superdad]

Hi Superdad, so the Mac mini comparisons you mentioned, were they while using the Mini as a controller for a NAA? Given that you are connected through a cable that might make more difference?

 

No. The comparisons were done with Audirvana+ 1.5.10 (well before I switched over to HQP). The computers were all just booted with the same SD card (then shut down and the next one tried), and my DAC's USB cable attached. Test tracks were also stored on the SD card to keep all things equal.

 

Looking back at my notebook, I see that the Mac mini which beat the laptops was in fact my 2010 unit, not my current 2012. So the distance is even greater now. What is not clear from my notes is whether or not the 2010 was running on external LPS at the time. Might have been using an early prototype of the JS-2 spread out on a board in the corner.

[Jabs1542]

It seems like NAA's such as SOTM SMS100 ( a good plug and play candidate for your Lampi I suspect) and my DH61DL are pretty sensitive to RF noise coming over CAT5

 

How did you figure this out - by listening? I have a RFI meter I will run over my Blue Jeans Ethernet tomorrow to see if my RFI is even measurable. I think I picked up better packet movement (slightly faster yet consistent speed) when I changed the power supply on my local switch to a LPS.

[EuroDriver]

How did you figure this out - by listening? I have a RFI meter I will run over my Blue Jeans Ethernet tomorrow to see if my RFI is even measurable. I think I picked up better packet movement (slightly faster yet consistent speed) when I changed the power supply on my local switch to a LPS.

 

the postulation is based observation of clear sonic differences between Mac Pro running HQP, DH 61DL running HQP and Surface Pro 3 running HQ Player.

 

The most startling observation was when MAC Pro was running HQP on Yosemite, and feeding to a SMS-100 in NAA mode, as you open and closed Safari, you could here the SQ change immediately. There were no dropouts or stuttering, the CPU was able to handle the load, but something was depressing the SQ immediately as Safari was launched.

 

Surface pro 3 running HQP is almost unlistenable when connected via USB to the DAC, but sounds great when it is streaming over WiFi to the LAN where the SMS 100 is residing.

 

with the DH61DL running in NAA mode, the MAC Pro running a Win 8.1 and HQP desktop, sounds much better than when running Yosemite and HQP desktop. Again Surface Pro 3 connected over a WIFi to the LAN sounded crystal clear.

 

from these observations, there is a suspicion that the heavy CPU processing of HQP doing Redbook to DSD256 is generating a lot of RF which is finding its way in to the DAC. It could be via the LAN cable, or it could be via the AC cords, the Surface Pro 3 was running on batteries whilst the MAC Pro was plugged into the mains

 

it will be worthwhile to see if hooking up the Surface Pro. 3 SMPS charger affects the SQ noticeably and also to test the SQ effect of turning off the SP3 wifi and using a USB NIC to handle communication tootoo

 

what is the frequency range of your RF meter ?

 

Have the Blue Jeans Ethernet cables made a SQ difference compared to standard CAT5 in you system ?

[Superdad]

Have the Blue Jeans Ethernet cables made a SQ difference compared to standard CAT5 in you system ?

 

Well Edward, you know they have for me. I've only been raving about them for the past 9 months! I compared the BlueJeans/Belden Cat6a to 4 other carefully chosen EN cables based on various constructions. Everything from Cat 5, 5e, 6, 6a, and the supposed Cat 7--but moreover looking at their actual construction (stranded, solid, shielded, wire gauge, etc.).

 

I saw that someone has opened up a Mac Pro and extracted the SMPS to feed it with a clean LPS, that might improve things for you on that machine. Might have to be a BIG LPS though.

[Lionelh2]

I run NAA on OSX and saw a mention of running in safe mode. I gave it a shot and HQP cannot find NAA in this mode. Any other details I'm missing?