AudioLinux and NUC Troubleshooting and Tuning

[lmitche]

21 hours ago, greenleo said:

Thank you Rajiv for the explanation.  I thought that the screen capture was self explanary enough.

Just curious, has anyone else been successful at powering a NUC7I7DNXX board with an LPS1.2 running as a RoonBridge endpoint?

 

I continue to be most impressed with the musicality of this combination.

[lmitche]

8 hours ago, Bricki said:

On the NUC7CJYH bios does anyone know if the package power limit sliders are worth fiddling with? Or just leave them as is? 

Mine runs powered from the LPS1.2 usually without fiddling. However one can set the NUC at 10 watts to be safe without any SQ penalty that I have observed. An ISO Regen means DAC VBUS doesn't come from the NUC thereby lowering current demand on the LPS1.2.

[lmitche]

2 hours ago, Monge said:

Did you leave onboard Audio unchecked or checked In bios? Try leaving it checked

Unless I missed something, there is no reason to check the audio box with a usb DAC or DDC.

[lmitche]

18 minutes ago, Monge said:

My mistake. Sorry. 

I just tried to leave the Onboard Audio unchecked In BIOS and running AL 0.7 it works like a charm.

It tried it a couple of times In earlier Al versions and I could not get it to work.

Thanks for correcting me and have a Merry Christmas.

Hi Monge. Glad to see it worked, Merry Christmas to you and yours. Larry

[lmitche]

9 hours ago, rickca said:

They will probably work if you have something like an ISO REGEN between the NUC and the bus powered DAC.

Yes, the ISO Regen takes very little vbus power itself and creates an independent vbus power source for a bus powered DAC. That's how I run the NUC7i7 on the LPS1.2 here.

[lmitche]

10 hours ago, rickca said:

I got my NUC7PJYH to boot with an LPS-1.2 set to 12V.  All I did was disable SpeedStep, SATA and onboard audio.

 

The interesting thing is it works with an LPS-1.2 if I use the two USB3 ports in the rear for my USB flash drive and the USB cable to my DAC.  If I use a rear USB3 port for the USB cable to my DAC, but a front USB3 port for the USB flash drive, the NUC will not boot with the LPS-1.2.

 

Actually, the USB cable goes to a Berkeley Alpha USB which uses bus power only for the dirty input side.

Sounds like you have more circuits to disable in the bios. At a minimum disable the network interface you are not using, (wifi or ethernet nic), hdaudio, microphones...etc.

[lmitche]

2 hours ago, hifi25nl said:

In this case is easy. The following line can be applied only if partition has not been expanded and only for headless:

 

dd if=/dev/sde conv=sync,noerror bs=1048576 count=5990 | gzip -c  > audiolinux.img.gz

 

Piero, if I add together the blocksize for the first two headless partitions 511705088 +  5767168000 = 6278873088 or 6278873088/1048576 = 5988 blocks.

 

You use 5990 blocks in the command above.

 

Where did the extra two blocks come from? Is there a one block overhead per partition?

[lmitche]

18 minutes ago, hifi25nl said:

Example:

[root@archlinux audiolinux]# fdisk -l -u=cylinders /dev/sde 
Disk /dev/sde: 28.7 GiB, 30752000000 bytes, 60062500 sectors 
Disk model: Ultra USB 3.0    
Geometry: 64 heads, 32 sectors/track, 29327 cylinders
Units: cylinders of 2048 * 512 = 1048576 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: 524A1828-2002-4187-B469-3E7A2DE53ED0

Device     Start   End  Size Type 
/dev/sde1      2   489  488M EFI System
/dev/sde2    490  5989  5.4G Linux filesystem

bs is the bytes number and count is the end of second partition +1: bs=1048576 count=5990

 

Got it. Thanks. So dd starts the count at cylinder one of course!

[lmitche]

2 minutes ago, sig8 said:

Leo: The anodizing on the heat transfer block on the case would act as an insulating layer. Copper has better thermal conductivity than aluminum, and bare aluminum has better conductivity than anodized aluminum. As Dev has pointed out, they did not anodized that surface in another case.

Has anyone had a problem with the heat transfer capabilities of this case?

[lmitche]

1 minute ago, Forehaven said:

If this is the case, just use a dremel w. a wire attachment and eliminate the anodized layer no?

There are a lot of scholarly papers suggesting that the anodized layer improves heat transfer properties.

 

https://www.qats.com/cms/2010/11/09/how-heat-sink-anondization-improves-thermal-performance-part-1-of-2/

[lmitche]

3 hours ago, sig8 said:

This is talking about heat transfer by radiation, we have conduction here. Two very different things.

 

It is about good, better, best. If every degree temperature difference matters in SQ, then it matters. I just pointed out something I saw and did not think was the best thing to do. We fuss about everything, don't we.

My point is that I haven't heard anyone with a heat problem using the Akasa case. Because if this, I am not sure why we are talking about it. That's all.

 

The Plato X7D is a great case, well made, fits nicely in an audio rack, has an attractive design, and keeps things really cool.

 

I am testing hqplayer dsd512 upsampling with a NUC7i7dnbe in this case today to learn about any limitations. If there are any, I'll report it here.

[lmitche]

Hi all,

 

FYI, I have started another parallel thread on AL based servers.

 

[lmitche]

11 minutes ago, Miska said:

 

macOS comes with ssh built-in... Is there a reason to use something else?

 

Termius runs on my Android tablet, which probably has a built in ssh as well, and it's kinda slick with easy to resize text, good keyboard, easy to setup hostname and security details.

 

Recommended for Android!

[lmitche]

@Jud,

 

You can easily make two USB boot sticks, one setup for MPD/UPNPMPDCLI the other for the NAA. Then when you want one or the other, just re-boot with the appropriate stick.

 

Remember to pull the stick out! This matters.

 

 

[lmitche]

2 hours ago, Bricki said:

Is anyone else having stability issues with their server after setting up a network bridge?...it may just be my issue... originally I had no issues with the network bridge for the first 5 days... but I have had to reboot my server 3 times in the last few days....it has frozen after hours of playing music off a micro SD card that is attached to a USB port on the same riser off the mother board as my USB to Ethernet adapter... perhaps this is related, I will have to spend some time setting up bridging on a different port and see if that makes any difference. 

 

Anyone else have a similar problem?

I saw a very unstable usb nic earlier this week. It wasn't bridged. Rebooting the router fixed it. I've seen this happen with a Cisco switch as well.

[lmitche]

18 minutes ago, rickca said:

Has anyone updated AL to the latest kernel (headless version 1.1 with menu version 105)?  Any issues or noticeable change in sound quality?

Yes, I did the update yesterday. I heard a SQ change straight away.

 

Seems there is a bug in the pstate-frequency package.  I reported this to Piero, he elevated the problem and the bug was quickly fixed by the developer at GitHub.

 

Piero suggested executing the following from a $ command prompt to install the fix:

 

You must update database
$yaourt -Syy
and reinstall to latest version
$yaourt -S pstate-frequency

 

The fix worked.

[lmitche]

9 hours ago, austinpop said:

 

What was the bug?

 

10 hours ago, Dev said:

I think I also noticed the pstate issue with the Max vs Performance profile. I can't seem to set the max/min cpu freq and the  pstate governor is always set to performance whatever I do.

 

[lmitche]

1 hour ago, hifi25nl said:

If you have a version before 1.1.1, you should reinstall manually pstate-frequency (because of a bug in a previous version)

yaourt -S pstate-frequency

and reboot

 

Select MPD DAC is working fine here and also LMS. For specific problems of subscribers please contact me.

 

I remember that the menu is not designed for using numbers really, to select an option you should use arrow keys and Enter.

 

Piero, many thanks for the updates. The pstate-frequency module is working perfectly here. This is very much appreciated!

 

Larry

 

[lmitche]

23 minutes ago, Dutch said:

Entertaining the idea that 100mbps may sound better than gigabit; has anyone ever tried to change speed/duplex settings in AudioLinux? I’m not sure about the best way to do it, and do it in a persistent way (‘surviving’ reboots).

 

Installing ethtool makes it easy to try it on the fly but these settings are not persistent. If anyone wants to try see the commands below. You’ll lose the connection for a short while but it comes back. These also work when RAM root is enabled so a reboot reverts back to all original.

 

[switch to root]

su

 

[update packages]

pacman -Syy

 

[install ethtool]

pacman -S ethtool

 

[check current settings]

ethtool eno1

 

[check name of your ethernet interface]

ip addr sh

 

[change eno1 interface to 100mbps/full duplex]

ethtool -s eno1 speed 100 duplex full autoneg off

 

@hifi25nl, could you perhaps chime in when you have the time? Thank you in advance.

 

If you read the novel thread you will see many posts about Ethernet cards, Ethernet speed, FMCs, cable types, duplex, flow control, off loading ...  ad naseum. Generally slower and simpler is better, even 10 Mbps can sound great. The trouble with these Ethernet chains is that many power supplies are required, and the SQ is directly influenced by the quality of these supplies.

 

It is a good idea to somehow isolate the cable modem and related street noise as early as possible in the chain.

 

After spending years looking at alternatives, over here 5ghz wifi has been found to provide the best SQ and simplest to install and support.

 

Keep 2.4 GHz devices away from any audio system. It sucks the life out of SQ.

 

When the EtherRegen arrives, I will take another look at wired Ethernet.

[lmitche]

2 hours ago, bodiebill said:

I now think 'Package id 0' is the highest core temp.

Maybe temp1 is the mobo temp?

My understanding is that temp1 is the PCH.

 

Package id 0 is from the heat spreader above the cores, so it is likely to be the highest core temp.

 

On some NUCs the temp of the WiFi card will show up as well.

[lmitche]

3 minutes ago, bodiebill said:

Thanks, in that case temperatures look fine using HQPe DSD256 with EC modulator: CPU from 35 (DSD orig.) to 60 (PCM 192 orig.) and PCH around 60. 

Yep, those temps won't damage anything.

[lmitche]

On 9/9/2019 at 11:45 AM, ericuco said:

New Audio Linux user here. Purchased and installed AL LXQt about two weeks ago. Sorry for the length of this but wanted to share my experience in hopes of getting advice and encourage new users that is relatively easy to do.

 

Everything has gone fairly well with a few lessons learned items along the way. I run HQP 4 Desktop and Roon Server on this server. Output is via fiber optic NIC to an opticalModule and then to microRendu (HQP NAA).

 

I used Win32DiskImager to create the USB boot flash drive. Then used my Ubuntu Studio installation for the second part of repartitioning ext4 using gparted. No problems.

 

I originally was booting from a USB flash drive until I had my music server upgraded to 16MB RAM (up from 4MB) last week so now I am successfully running in Ram Root mode. (FYI - Roon database is 3 GB). Also, removed my internal SSD so everything is booting/running off USB drive.

 

For the Roon database, I had been doing automatic backup to my NAS. I copied one of the backups to a separate USB flash drive which took some time with a 3GB database. Once AL was up and running, I used my iPad (Roon remote) to connect to the new Roon server (core). Using the Roon menus (Settings > Backups), I pointed to the backup files on the USB flash drive and everything was updated which again took some time with a 3GB file.

 

One question: I was updating the kernel and was presented with two choices, one being the BFQ. Initially I installed the non-BFQ version but when reading through the documentation, seems that I should be using the BFQ version so I installed that version. Can anyone provide suggestions as to which version I should be using?

 

Some of the lessons learned have been associated with the Realtime Priority settings.

 

Lesson 1:
Given that I use Roon as a front-end to HQP, the documentation suggested that HQP be listed first and that Roon could be removed from the RTAPP list (well that was my intreptation at least). First off, the list )see below) is not correct if you are using HQP 4 Desktop:

 

APPLICATIONS="jackd mpd hqplayer hqplayerd RoonAppliance RoonBridge sox mediacenter24 networkaudiod deadbeef a2jmidid ardour-5.12.0 rosegarden audacity"

 

Need to replace "hqplayer" with "hqplayer4desktop".

 

I had removed RoonAppliance from the list for which Roon was NOT happy and non-responsive so I added it back in front of hqplayer4desktop and everyone seems happy now. I might swap the order of HQP and Roon at some point but for now everyone seems happy including me listening to music.

 

Lesson 2:
For the RTIRQ configuration, the fiber optic NIC is the main output device so I had changed per suggestions in the documentation:

 

RTIRQ_NAME_LIST="usb" to RTIRQ_NAME_LIST="enp3s0"

 

This made my USB ports unhappy to the point they all shutdown and became unresponsive.

 

Changed to RTIRQ_NAME_LIST="enp3s0 usb" and everyone is happy again.

 

The journey continues but that is all for now. Any and all suggestions are welcome noting that I am not looking to go to the nth degree on this and pretty happy where everything sits at least for now.

 

The BFQ kernel sounds better here.

[lmitche]

32 minutes ago, cat6man said:

anyone using the new 16G image successfully on 16G usb stick?

AL only takes 2 partitions of approx. .5 and 6 gb. You should have plenty of space left free on a 16gb USB stick.

 

What am I missing? How are you making these images?

[lmitche]

4 minutes ago, cat6man said:

 

I wish I knew what I was missing and why a separate 16G image was needed.

 

I'm on Ubuntu 18.04 and using Balena Etcher to burn the image.

I see no partitions created and just a large unallocated disk (16G usb)

 

When burned the image to a 256 USB, I got the 2 partitions and a lot of space left free.

No idea why

 

I have no experience with Etcher so can't help you there. These days I tend to manually create partitions using parted on new USB sticks and then dd the two partitions over. This could be done from your 256 gb stick. You will need to learn parted, which isn't rocket science. Starting blocks and sizes can be found using existing sticks that are properly formatted.

[lmitche]

14 minutes ago, cat6man said:

 

thanks.  i actually started to do that last night before going to bed.  i manually created the partitions but wasn't sure how to dd the partitions...........will have to research that if updated etcher doesn't work.

 

edit:  from etcher developer:

The way that Etcher works is that it copies an image byte by byte to the drive, without doing any manipulation to it. The result is that the partition table from the image is copied as-is to the drive

Just dd by addressing the partitions level device like: /dev/sda1 . Use lsblk to find the device names and block sizes.