Reply with quote #31
You took the bull by the horns, Kelvin, and I think you will have a lot of fun with this DAC and its options.
You can see in the picture (link) how the DAC can be connected directly (raw output) to a line stage which is a great feature for people who only want tubes. The off the shelf version also has this option as far as I know. https://hifiduino.files.wordpress.com/2015/01/r2routputs.png To your question: To connect the DAC directly to the grid or with a small resistor value for Rg results in the best transients. The resistor value of Rg has obviously an enormous influence on the gain in combination with NFB and I got the 6dB gain back simply by using a smaller value for Rg. Here is my hypothetical answer: In my originally posted circuit (no NFB) Rp (7k6)is also pretty much the output impedance for the 300B OTL input, no influence of Ra (1,26Mohms at 1kHz) because it’s so much bigger. But with NFB and 50k for the pot, Ra might become a problem as the comparatively much smaller 50k or less can load down the tube affecting the original open loop gain of 12dB without NFB ( Rf is effectively in parallel to Ra?). But I cannot take a pot that is a lot bigger than 1,2Mohm in order to stop this as I would have no controllable area anymore. Why this can be corrected by choosing a smaller value for Rg is beyond my knowledge. There was no difference between a 100k and a 50k pot regarding gain (which would kind of support my theory), only regarding controllable area. The 100k pot started to kick in only in the last third of its range. That’s why I changed it to 50k. I can say this: NFB volume control works and I will stick with it.
Reply with quote #32
I have now received my Soekris dac1541 R-2R DAC and am able to give my first impressions.
This fully balanced DAC is a beautifully constructed device with three transformer isolated SPDIF inputs, an optical Toslink and a USB input. My digital sources include a MacMini via USB, a CD Box S CD player via coax SPDIF, a satellite radio receiver via Toslink, and the coax SPDIF output from a Graham Slee 'Bitsie' DAC also connected to the MacMini via USB. The power supplies are onboard and the device does run a bit warm, but the case is well ventilated. As previously mentioned I purchased this DAC to replace an Arcam irDAC. I did not expect a quantum leap in SQ in making this transition and, indeed, the improvement is marginal. If you already have a good quality DAC you may well find the same thing, but you do have the assurance of knowing that this is probably the best you can do. In addition, there are several very nice extra features like the digital volume control, the crossfeed capability in headphone mode, and a selection of anti-aliasing filters. As discussed above, it does have some DSD capability. All in all a very nice, solid package. So far I have used the DAC 'out-of-the-box' and have not yet opened it up to access 'direct' output from the R-2R arrays. A good quality USB cable is included in the package. This is more than adequate: I am not an advocate of 'bijou' USB cables. Kelvin
Reply with quote #33
You should expect a quantum leap, Kelvin, because this DAC can deliver if the in/out is set right for it. While MAC is a good choice for using it as audio computer, SQ still depends on software and quality USB cards. Did you try different music players besides iTunes?
If you have a PC I would recommend to try a free version of JPlay 6.2. JPlay has still the best SQ I know of as it has overcome all the weaknesses of the PC as audio computer...even without tweaking your PC which I would only recommend if you don't use it for any other purpose than playing your files. http://jplay.eu/download/ At least as important as the input is the output. Unless you use the raw output from the DAC via tube buffer right into the OTLs (digital volume control)there is no real fun with this DAC as I have mentioned at other occasions. A simple cathode follower with the 6N6P or any other small signal tube at hand can be made within an hour with a handful of parts.Piece of cake for you.A Choke loaded DHT with battery supply/automatic charger for the filaments sounds stellar but is probably a bit much for most people (although it is as simple as can be). Here is a list of SQ improvements around the R2R DAC in my sound system: about 80%: quality of input source in combination with any kind of tube buffer/raw output about 5%: dedicated USB card with external psu about 2%: very good USB cable about 10%: good interconnects; I tried XLO, Nordost and AcousticZen and the cheapest one (Acoustic Zen "WOW") sounded the best. I use them everywhere (analog/digital). It has very good shielding and sounds rich and balanced with the OTLs. Of course I am taking about my DIY R2R DAC which has linear psus only and other little extras. So one big improvement would be to replace the SMPS with linear psus. It's very easy with this DAC as the designer gives all necessary information about where and how to connect.
Reply with quote #34
Hi Wolfgang. Very good to hear your views and experiences with the Soekris R-2R DAC. Very useful to me and, no doubt, others.
A few comments and observations further to my previous post. I always make it a rule to thoroughly evaluate any piece of kit 'out of the box' if only to be fair to the designer, and to inform those who may not wish to attack the inwards with a soldering iron. These are very early days, but I suppose the scale of a 'quantum leap' in SQ rather depends on your starting point. I think the SQ of my original setup was quite good. This was not just the perception of my ageing ears, but also the view of several others. I have recently acquired a pair of Klipsch Heresy lll speakers which sound well with my Masterpiece/300B-PP TS kit. The 'marginal' improvement over my Arcam irDAC is confirmed by others and is not noticeably dependent on the digital source. Of course, not all of my sources are equally good, but the SQ does not appear to be dependent on how the connection is made, USB, coax or optical. I always use good quality, well made interconnects. Here in the UK, Van Damme offer a good range. I try to make all interconnects as short as possible. Now to my MacMini. I use several audio players besides iTunes; Quicktime Pro, Tascam Hi-Res Editor, Jriver multimedia player, BBC iPlayer for BBC digital broadcasts, several others, and I am about to investigate HQPlayer. For BBC broadcasts I actually prefer to use the optical output from my Humax Freesat receiver. I realize some of this is only available in the UK. On occasion I input and output audio via a firewire link with a Canopus ADVC110. Unfortunately, Jplay is only available to run on a PC. I suppose I could run it via Bootcamp, but I am not too sure how it will perform with the Bootcamp port drivers. Worth a try I guess. I changed to Mac when they went Intel and adopted a Unix core. More stable and secure in my experience. With the MacMini I do not have much choice but to use the internal USB ports. This, at least, is my understanding, but I could be wrong about this. It seems that the time is approaching for me to open up the box, take the output directly from the R-2R ladder, and try a simple triode buffer as you suggest. It surprises me that a good solid state output amp can be so deleterious to the SQ, but your experiments do sound convincing. In my pre- TS days I acquired some Bottlehead kit. In particular a 'Smash' preamp which sports a couple of 4P1L DH triode-strapped pentode stages with or without CCS anode loads. This sounds quite good with the 300B PP OTL, but the loctal tubes are dreadfully microphonic. Then again, I could very easily adapt a little used 2A3 'Stereomour' amp which has a choke loaded, para-feed output stage. This might be a bit short of gain for NFB volume control experiments. Or, I might just knock up one from scratch! We will see. Got a few other things to do first. Rest assured though, you have whetted my appetite for further action. By the way. I get a lot of 'fun' from all these audio ventures, whether the results are positive, negative or indifferent! Kelvin
Reply with quote #35
Kelvin, it is really most interesting that the SQ doesn’t seem to change in your setup depending on the connection you use as there is a difference how the signal is handled (S/PDIF/optical/AES vs USB for example). I went to great length to figure this out while I was still using sound cards (like RME Hammerfall ) and Behringer stand alone parametric EQs and active x-overs in connection with JPLay. The difference was so obvious that I finally changed to USB and I2S internal connections and put everything I needed – like x-over for sub, parametric EQ etc - in the DAC; also no more ASIO drivers but instead Kernel Streaming and ultra low buffer settings.
Years ago when I started to replace the opamps in the Behringer with tube buffers something became crystal clear to me regarding SQ and the elements that are responsible for it… good, bad or otherwise. No matter how good the data of the opamps look on paper they always cover the music with some layer of something artificial which makes everything sound kind of the same …like good HiFi.
I cannot imagine that JPlay could work with a MAC even if you could install it (needs win32) simply because of how it interacts with the computer . It’s not like any other software based player I know. That’s also the difference which I could find while testing the HQPlayer with all its possibilities. It sounded really good but I was always missing the final kick which I get from JPlay. I had the two players installed on 2 different operating systems on the same PC, Win8 optimized for JPLay and WIN 10 with HQPlayer. For the test I didn’t change the bios settings (which look quite different for JPlay) so that HQ Player wouldn’t suffer from limitations. I had great hopes regarding the software based room EQ option but it’s more or less worthless as you can only upload room EQ files automatically created by another program without being able to adjust these settings from within HQplayer in real time while listening to the results. This will never sound really good and that’s why many people who tried it complain later that room EQ or DSP in general sounds bad, sterile, flat.
Speaking of microphonic 4P1L: K&K audio mounts the sockets on a decoupled extra board.
If you can install additional PCIE cards in your MACmini you could install a dedicated USB port for audio. I just don’t know if that’s available for the Mac. All audiophile USB cards I know only work with Windows or Linux OS.
I hope you keep us posted with the outcome of your next steps.
Reply with quote #36
I am afraid I have been a bit ambiguous in my comments on the changes in SQ in going from the Arcam to the Soekris DACs, Wolfgang. I did say that all my sources were not equally good, but did not change the connection type ( USB,coax or optical) for each source when assessing SQ changes on switching between the two DACs. I did realize that each type of connection can sound somewhat different. For all sources the improvement/changes in SQ in switching DACs was marginal.
I note your comments on the effect of opamps on SQ. I clearly have experiments to do there when I 'open the box'. When I use Bootcamp, the Mac is essentially transformed into a PC in that it is then running Windows, in my case currently Windows 10. The only difference is that some extra drivers are necessary to interface with the Mac I/O ports, etc. Jplay works very well and it is sounding very promising. I was a bit taken aback at the prices the Jplay folk charge for their interconnects! There is no way alternative USB cards can be installed inside the MacMini. No slots like in the old MacPros. Any improved device would have to be externally connected. Yes,the 4P1L is one of the most microphonic tubes I have come across. You can hardly breathe near a 'Smash' preamp! I hope this has clarified things a bit. Kelvin
Reply with quote #37
Hi Kelvin, hi Wolfgang,
may i chime in with my experiences in Soekrisland so far... I think I had some luck when due to Soekris' long delay in delivering the 1541 and some financial constraint on my part I changed my order in the last moment to a DIY board instead of the complete dac, for a good 1000 euros less. I added a I2SUSB solution http://jlsounds.com/i2soverusb.html which implements galvanic isolation between computer and I2S connection to the Soekris dac. Three linear PSUs I had lying around, so everything was hooked up in 6 or so hours on a Saturday afternoon. I was astounded when there was music right after plugging in the USB cable and hitting play on my computer! So all very doable this putting things together. Now as to quantum leaps in sound reproduction with the new setup - I got them! I took the R-2R signal from the raw outputs and went straight into my Masterpiece. Before the Soekris my DAC was a heavily modified Oppo 105D which after modification achieved a quality level equalling a Chord Hugo, so not bad at all. I drove the Oppo from HQplayer which gave another small improvement. Now with the Soekris I can only think that the old setup was barely listenable. In comparison the Oppo sounds just plainly wrong, and anything but pure. The feeling with the Soekris is that now finally there is just the recorded sound, and nothing else. (Surely this honeymoon feeling will fade at some point, but for now it's very prominent). So thanks Wolfgang for the heads-up, and Kelvin - go for the raw connection! As to the input side: I found that feeding the Soekris from HQplayer is a no-go. Best is to just send the digital data as-is and leave all upsampling etc. to the Soekris. I will check Wolfgang's suggestion of kernel streaming vs. ASIO. Cheers, Achim
Reply with quote #38
Hi Achim. Excellent stuff! Your changing to the DIY board seems to have paid off in spades, and not just in financial terms. I will be trying the raw outputs in the near future. Can you remember, when you were in Munich, where Soeren ( not Sauron I hope!) showed you the raw output take-off points are situated on the 1541 board? Have you tried connecting to your Masterpiece after the op-amps, which is basically what I am doing at the moment? Also, how much improvement would you ascribe to the I2S isolation/anti-jitter arrangement you have implemented at the input? I take it your main digital source is a PC. Do you use any other non-USB digital sources at all?
Apologies for so many questions! Regards, Kelvin
Reply with quote #39
Hi Kelvin, you're welcome!
I had a look around on the web for the innards of the Soekris ready-made DACs, and I think that it was probably the 1321 where Soeren pointed out the testing points where the signal could be tapped. Please have a look at the image where I marked possible points in the 1541. On my DIY board, there were inly 9 ohms between raw ground and + of one channel. You may find it out like that. Actually, in my case I thought I had short-circuited the output, only later realised that this is probably the nature of the R-2R beast... I did not connect the opamp outputs and will probably give this try a miss, as there is so much else to be tried :-) As to the isolated I2S input - I have no comparison, but looking at the board and Soekris' description in the manual, my guess is that they have implemented the same or similar kind of isolation. As far as I know the digital input gets chopped up and reconstituted in the Soekris severely and thoroughly, so this is probably good as it is. Other input I have not tried yet as I am currently collecting things needed to be ordered from Digikey, the SPDIF isolation transformer being one of those. Sorry to be of no further help ATM, I trust that finding and connecting the raw output is the easiest thing you can do now, and will give a lot! Happy hunting! Achim
Reply with quote #40
Thanks very much, Achim, for indicating possible raw output points on the 1541 board. I had come to the same conclusion as yourself regarding the isolation and re-clocking of the USB input. The blurb describes the line outputs as being driven with zero feedback discrete drivers. I just wondered....
I am actually very satisfied with the sonics of my present setup. I suppose my problem is I have difficulty in imagining what a 'quantum leap' would sound like from here! We will see. Many thanks again, Kelvin
Reply with quote #41
My own Soekris R2R project is progressing nicely, just waiting for a custom wound toroid to arrive before final assembly/resting.
I'll be using the OEM version of the R2R, the DAM1121, with I2S input courtesy of a JLSounds USB board (as per Achim's build). In this initial build the raw output will be passed to a Broskie ACF-2 (Aikido cathode follower) using 6SN7 tubes. The ACF-2 uses a bipolar supply (B+/B-), enabling it to be used without input and/or output capacitors. It has no gain (in fact it has a gain of <1) but with the high sensitivity of the 300B SE-OTL amps I don't see that as a problem. Subsequently I will probably experiment with a build using Wolfgang's suggested DHT buffer stage. I am also planning to rework a Masterpiece to use as a headphone only amplifier and, if all goes to plan/expectations, I will include another Seokris R2R DAC board within the Masterpiece chassis. Ray
Reply with quote #42
I just tried the free Jplay demo.
I have no idea why, but there is indeed quite a sound difference compared to addressing the DAC over USB ex ASIO or WASAPI. I am confused...
Reply with quote #43
JPlay looks like any other software player but it is different as it connects to the hardware in a way which we might compare to how the motherboard is connected by specific software drivers to the OS. That's why you can tweak the sound by tweaking your computer (bios, disconnecting background processes, changing windows priority settings and many more settings in the registry). That's also why I think it doesn't work with MACs as it works with PC. And that's why JPlay belongs to a different category of player.It sounds extremely "real", reaches down into the depth of the music, throws the driving elements of the music without warning in the middle of the listening room as if it were live, creates connections between single events that simply weren't audible with other players (HQPlayer for example, my most recent disappointment) and it doesn't surf gently along on the nice sounding HiFi wave which has no depth. Buffer settings are audible as we know from sound cards with ASIO drivers (less buffer-more impact, presence, dynamics). Drivers which have been written for a specific sound card always sound better than generic drivers and allow in most cases lower buffer settings. If a player offers basically direct streaming (depending on your computer, USB) it will be audible. Version 6.2 lets you even select the processor type (Intel).
Reply with quote #44
thanks for the explanation, which I will try to get a handle on with more auditionining later. That the way data chunks are sized up and spoon-fed to the USB port and EM secondary effects of the way the computer runs leave their mark across at least two barriers of galvanic isolation and as many reclockings and flip-floppings - has metaphysical proportions. Nonetheless, it's clearly audible. Something to explore once the evenings get longer and the weather a lot colder... Many greetings, Achim
Reply with quote #45
I have to say that after some testing I am delighted with JPlay running on my Windows 10 booted MacMini outputting to the Soekris 1541 via USB audio mode 2. Windows 10 does not have mode 2 support native and I am using the appropriate driver for this from the Soekris site. I am illiterate when it comes to 'audio speak', but I can say that, to my ears, there is a distinct gain in sonic quality-more 'presence' if pushed. I am not a believer in 'absolute' sonic quality. To me, beyond a certain point, the perception is highly subjective.
I do not consider a lengthy discussion on legacy BIOS or EFI startup is appropriate on this TS forum, but I would be interested to hear exactly what specific PC hardware, apart from the CPU of course, is invoked by JPlay. I have set up JPlay with the Intel core option (2.7GHz Intel core I7) and I have full access to the Windows 10 registry. As far as I can tell JPlay on my setup runs as advertised. In connection with all this, I have requested from Soekris info on possible raw R-2R take off from the 1541. It may require some signal tracing/retro-engineering on my part with my Tek scope, digital tone source, a big magnifier, and a steady hand with a 'needle' probe! Will keep posted on this one. Some are starting to ask me if I am ever going to be satisfied! They have a point.... . Kelvin