Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 3, 2018 13:35:14 GMT
Here I have a nicely upgraded Pi (using the same psu as Martin and John) and have run that into seven aftermarket dac's (from £200 to £25K) so far even used a rather neat re-clocking box which is rather good in the the digital signal stream from the Pi.
Still for the outlay very good no quibble but I still cannot get the non differences between the same bit rate/sample files/Spotify a few chaps are reporting (not just on this forum but a couple of others as well)
Using even a modest streaming streaming bridge produces far more definable results i have found hence my interest.
The tests have been conducted on a pretty modest system, around £6K all in using floor standers and a quality integrated amplifier.
On the more revealing systems the differences are far greater, however they are still clearly audible and repeatable on the more realistic system
I have two ways of connecting to the internet, a 320Meg pipe or as Martin a 4G router (with linear psu similar to the PI/Seg psu) and a top of heap RJ45 cable and Lan isolator
Sometimes the 4G sounds better, other times the fiber does, still the jury's out on that one!
To me I do not understand why some one hasn't delivered a streaming module which encompasses all these aspects and you can have a quality control app thrown in for a sensible price that can be taken up by the DIY fatuity
With this you can build your own platform saves a lot of fannying around in my book
Stan fancy a joint project?
|
|
|
Post by MartinT on Sept 3, 2018 14:14:14 GMT
There are three clocks in the SEG, depending on whether you use USB or SPDIF, the dac and the other clock is for the PIC. That's fair enough and an alternative opinion, Tony. Remember I'm only evaluating on my system with my equipment. YMMV as they say. I am going to superclock the S/PDIF receiver clock as being hopefully the one that matters for me. The TentLabs is not a femtoclock but it has a jitter performance of <3ps and is not to be sneezed at. It is also sufficiently inexpensive, relative to the SEG, that others may be interested in the outcome. That's all I can say for now as I have to install and test it. It may be a dismal failure so I'll say nothing about expectations and will just listen to it as usual once I get it up and running.
|
|
|
Post by MartinT on Sept 3, 2018 14:17:09 GMT
Stan fancy a joint project? Calling stanleyb now THAT would be an interesting project!
|
|
|
Post by sondale on Sept 3, 2018 14:25:11 GMT
A couple of weeks ago I compared the same track (Steely Dan / Ricky Don’t Lose that Number) FLAC file vs Spotify vs Qobuz and the best was Qobuz, then Spotify then my FLAC file.
After some head scratching I think I understand why - my FLAC was ripped losslessly from the Citizen boxed set - further listening made it clear that the SQ of the boxed set is lower than the individual discs. QOBUZ and Spotify may well have the remastered discs as sources.
So I may have to re-buy my SD albums or use QOBUZ - I have already purchased Aja so I will probably get a set of remastered albums.
Rather than using Volumio I used piCorePlayer on two RPi’s, one for LMS (Logitech Media Server) and the other for Squeezelite, this arrangement allows me to upsample everything to 384k. As I said previously I will do a write up on pCP - I may as well start doing that now.
|
|
|
Post by John on Sept 3, 2018 15:10:21 GMT
When I visited Mike the Isis was definitely better. Files were next and Spotify just a bit behind, that was with the DigiOne I think the Signature is an good step up. Where it is now I stay open minded until I hear it against something else You will always get better quality of course. But very hard to beat the value this route offers. For me anything that offers great sound at costs I can afford is something to praise
|
|
|
Post by Stratmangler on Sept 3, 2018 15:21:36 GMT
... this arrangement allows me to upsample everything to 384k. As I said previously I will do a write up on pCP - I may as well start doing that now. Why? All you're doing is adding lots of zero entries at silly sample rates, which takes a lot of needless processing.
|
|
|
Post by sondale on Sept 3, 2018 16:38:00 GMT
... this arrangement allows me to upsample everything to 384k. As I said previously I will do a write up on pCP - I may as well start doing that now. Why? All you're doing is adding lots of zero entries at silly sample rates, which takes a lot of needless processing.
Why? I do this when running a DAM1021 R2R dac which as far as I am aware will upsample to 384K and then it applies a filter to the result. Opinion seems to be that if the upsampling can be done by software (for example the excellent SOX program) then the result is much better as the filter is not applied. Now I cannot in any way give you an Objective set of results as I do not have any equipment to do so BUT to my ears using Stax SRM-T1s into Sigma Headphones the difference is clear - it may be placebo effect but if it makes me happy listening to music then Why Not? As far as processing is concerned the RPi seems to handle it with ease - I have to say that I have only been doing this on RPi 3B and 3B+ machines.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 3, 2018 16:39:40 GMT
Hi Martin If you are using the Xo-dac upgrade you may find quite tight in there, the case height and spare space in the SEG is minimal at best even with the supplied stand off's as it is a generic upgrade for a wide range of dac's which have more space. Mountinghole pitch: 51,0 x 81,0 mm which is almost half the size of the SEG board! The clock will be positive without question, is it just getting them to fit satisfactory, while keeping the rfi noise at bay. I have fitted one clock without case work adjustments, however I had to design that one from scratch on a dedicated board for the SEG. A lot of hassle, which is why for real updates then a change of casework pays off, also socket changes and other items can befitted. However that then begs the question why not start at a higher level in the first place? The power supply to clock is AS important if not more so than the clock change imho. The TENT kits are quality so it will benfit form the change without question, though it will be accumulative with the upgrades you have already made to the SEG However I am really intrigued with these comments, I am talking a straight comparison, no up-sampling as that usually involves some form of SRC or FPGA that is doing the number crunching (remember 1 million taps is an audiophile's dream ) and again contains another PLL or re-clocking latch which has an effect. Keeping the original file in a direct multiple of its original size will imho be more beneficial than say 2.5 or 5 times up or over sampling.
|
|
|
Post by Stratmangler on Sept 3, 2018 17:05:14 GMT
Why? I do this when running a DAM1021 R2R dac which as far as I am aware will upsample to 384K and then it applies a filter to the result. Opinion seems to be that if the upsampling can be done by software (for example the excellent SOX program) then the result is much better as the filter is not applied. Now I cannot in any way give you an Objective set of results as I do not have any equipment to do so BUT to my ears using Stax SRM-T1s into Sigma Headphones the difference is clear - it may be placebo effect but if it makes me happy listening to music then Why Not? As far as processing is concerned the RPi seems to handle it with ease - I have to say that I have only been doing this on RPi 3B and 3B+ machines. Well for starters, 44.1 kHz does not multiply up to 384 kHz. Which means that 2 processes have to take place.
First off there's a new data construction required to change the 44.1 kHz data to 48 kHz data. The word change is very appropriate. The data is not the same as the original input file.
Secondly, once the data has been changed it's undergoes a fourfold upsample. No new data added, just 0s, so there cannot be any new details hidden within the resultant files. What this upsample does is trick the simple computer that is a DAC into using one of a number of of different filters to block out of band noise.
If you were to upsample to 352.8 kHz you'd have the occasional sample which would be identical to the one you fed it with.
|
|
|
Post by MartinT on Sept 3, 2018 17:23:09 GMT
I upsample red book files to 24/88.2 which makes sense and doesn't generate as much data as to overstress the DAC and start undoing the good clocking work with rounded waveforms.
|
|
|
Post by Stratmangler on Sept 3, 2018 17:49:25 GMT
Some DACs can't cope with the 2nd upsample to 176.4 kHz, Martin. My current one can (I say current - it's been sat in the saddle for quite a long time now). My current one is on the M2Tech obsolete list.
It ain't broke, and it don't need fixing or changing
|
|
|
Post by MartinT on Sept 3, 2018 18:11:35 GMT
Yip, the Caiman-II couldn't but the SEG can.
I'm beginning to fully understand how much more important clock timing and jitter reduction are than original file/stream format. It's seriously critical to overall quality.
|
|
|
Post by MartinT on Sept 4, 2018 6:24:04 GMT
Mountinghole pitch: 51,0 x 81,0 mm which is almost half the size of the SEG board! The clock will be positive without question, is it just getting them to fit satisfactory, while keeping the rfi noise at bay. Yes, it's a 12MHz TentLabs XO2.2 module. It does look a bit of a squeeze, I may try to mount it upside-down over the main board but I'm not going to know for sure until I try it. If I get home not totally shattered tonight, I'm going to open up the SEG and take a proper look at fitment.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 4, 2018 7:58:55 GMT
You can use a tap off the incoming ultra clean 15Vdc Martin as the X0 requires 10Vdc
|
|
|
Post by MartinT on Sept 4, 2018 9:06:59 GMT
Actually the module takes 9-35V DC so I'm going to give it the direct 15V from the supercap PSU after the SEG power switch.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 4, 2018 9:19:18 GMT
Either that or I put in another dedicated tap from the PSU 15Vdc for you that works good to
|
|
|
Post by MartinT on Sept 4, 2018 10:45:27 GMT
Either that or I put in another dedicated tap from the PSU 15Vdc for you that works good to Thanks, I'll hold that in reserve. It only needs 30mA so not much of a current draw.
|
|
|
Post by John on Sept 7, 2018 16:23:15 GMT
The new DigiOne Signature was not compactable with the Seg. It started to have slight drop outs causing timing errors and making it unlistenable. So far Allo support has been excellent but we could not solve the issue. So I will send it back for a refund on Saturday. I think for now that me done with the pi. Whilst I like the tremendous value to sound performance I have not has the best of luck with various issues.
|
|