1000uF/35V , audio grade , minimum 2000h at 85C. Trust me we looked long and hard to find the right cap.
So can anyone comment on the sound ? Is it better/worst after the "burn in " ..
Do you like it, hate it , could be better..bass not enough perhaps ?
Yes!
Who received the dac?
Please burn it, listen to it, and write reviews about how the dac and the 2 output stages sound.
Everything else is theory and maybe "noise".
@cdsgames : did you already sent a board to all katana 1.1 reviewers? Like Markw4,etc...
I don't doubt it's the right cap. I can really stand behind your cap choice.
My concern is that, when you find the right cap, you make room for it in the PCB! You allocate the correct space for it, even if it will rest sideways. You don't just glue it on top of all the other components (rendering any maintenance useless) and at the random whim of the solderer / gluer to put at whatever spot and angle he dims interesting... and also protruding from the side edge of the PCB.
What kind of case can accommodate a randomly placed component of that size?
How do you service the underneath glued part? What happens if that cap decides to pour all his intestines over all the other components?
I really can't believe that is your 'final layout'. I'm sorry, but that is just sloppy work, IMHO. I will not buy such a product, no matter how good it sounds. It has to be correctly laid out and serviceable to some degree.
without the recommended burn in.. in contrast w/ katana 1
Bass improved dramatically.. bass is a bit fleshy.. all I can say is in the track Brombo.. I can distinguish the base line and the drums clearly..i can appreciate the showmanship of base and drum players.. Brombo is a complex bass test to audition.. Katana 1.2 rendered it with finesse..
I need to burn it in further..
Yes all boards will have this cap. We are talking to manufacturing to make sure that its properly aligned.
India was on holidays (and will be again in 2 days) .All reviewers that got an old Katana will receive a Katana 1.2 + analog stages + isolator for independent review / I would estimate most (if not all) will leave this week/beginning of next
Glad to hear it.
How about noise levels compared with first version ? any improvement in "stage" ? I think that timing has stayed the same..
You still didn't say anything about the minimum voltage on that dirty side. That's a
crucial factor. Especially after now knowing that there is a LDO sitting at the gate.
Backpowering the PI:
The diode taking out 0.25V is a lot. On e.g a 3B+, an iFi under load, with mediocre USB cable, you'd be quickly approaching unstable territories ( unconfirmed: <4.85V).
I'd therefore estimate that you'd need minimum 5.1V at board entry under load to guarantee stable conditions on a loaded 3B+ (without any devices attached!).
Backpowering the PI through the isolator therefore might become a little tricky. You should put a remark into the documentation!
Measuring just the idle voltage won't do of course! The RPI (3B+) load is highly dynamic and the voltage will follow these dynamics.
Last edited:
with regards to noise.. i may not be able to do it with justice since I really cannt hear any noise in Katana 1.. will try to listen to Jen Warnes' Way Down Deep after 72 hour burn in for stage and base..
with regards to timing.. i will just guage it by PRATT.. after burn in..
Last edited:
@cdsgames
Will you still publish the FFT of the two output stages?
It would be useful (along with the subjective reviews) for the choice between "Sound Quality" and "Pure THD" version
DSD playback issue using volumio via DOP
I observed that playing DSD128, Katana V2 cannot play the music file. DSD 64 is fine. I am using volumio. Is this a harware limitation of v2 using isolator?
I didnt encounter this issue in V1 using volumio.
By the way the smurf sound is really solved in Katana v2. Irrespective of booting, garble sound is gone.
I observed that playing DSD128, Katana V2 cannot play the music file. DSD 64 is fine. I am using volumio. Is this a harware limitation of v2 using isolator?
I didnt encounter this issue in V1 using volumio.
By the way the smurf sound is really solved in Katana v2. Irrespective of booting, garble sound is gone.
Some quick thoughts...
I'm not sure if others reported it, but in my listening with the earlier version Katanas I heard what I interpreted as some ringing or oscillation and suggested Allo look for that. Once they had a fix, I modified one of the units I had here and was able to compare an clearly hear the benefits of the added caps. The sound 'relaxed' a bit and I could listen both longer and at a somewhat higher level. With the pre-modified Katanas, I found I got a slight headache and an uncomfortableness in my ears that I could mitigate some by listening at a lower level than normal for critical listening. With the added caps, this effect was gone.
Allo made several other changes lowering the power network noise floors and implementing them were also noticable improvements.
Comparing an original Katana to the updated one made this very clear.
On the SQ versus THD versions, while there was a lot of speculation on what Allo did and what it affected, it is really a very simple modification. IF you search on DIYAudio for the 'Rasmussen' mod or effect, you'll find 3 closed threads that discuss it*. While the pictures are no longer available in the original thread, I saved them locally. While I don't know the full schematic of the Allo / Sparkos output stage, I believe it is similar to that of the attached circuit. The mod is nothing more than a couple of resistors and a capacitor in each channel providing a gentle rolloff in the analog signal out of the DAC chip before it goes into the output stages. In my listening tests, I found it improved both musicality and dimensionality with very little negative effect. All of my listening was with fairly high-test power supply setups, with some comparisons between using the Allo +-15V DC-DC stages and a separate +-15V DC stage. My impressions held up for both configurations. Over the next several weeks, I'll do additional listening tests with power setups more representative of what most will use (as Allo almost exclusively did). BUT we both reported similar impressions even using significantly different power setups.
Finally, one has to be VERY careful in separating the boards. The interboard connectors' pins, while effective, can easily bend and break if one misaligns or exerts significant side force on them. To safely separate them with little chance of damaging them, once I removed the posts securing a board, I VERY CAREFULLY inserted a very thin flat screwdriver blade in-between one of the connectors and levered it up slightly, NO MORE THAN .5mm! Then I did the same to the other, but went to about 1mm. Back to the first and levered it to 1.5mm. Back and forth until the two boards basically are un-attached and you can then gently and carefully separate them. IN MY EXPERIENCE (and I had to separate boards a lot during the listening tests), any other method will ultimately result in bent pins and they will break at some point.
AND also, yes, with really good power supplies, the Katana is now very good!
Greg in Mississippi
*P.S. Joe also talks about it on his commercial website in reference to DVD-Player modifications.
Attachments
Last edited:
The isolator limits the SR to 192kHz according to Allo.
DOP-128 requires 352k8Hz.
DSD-native doesn't work on I2S-HATs.
What started to be a minor output filter upgrade on Katana-V1 turned out to be a major
overhaul ending up in Katana-V2.
Great developments. I really like it.
Also nice to see is that the "Golden Ear" fraction contributed valuable feedback to these developments.
########
Obviously all this modulation and buffer-cap stuff seems to be related to the much higher load currents Allo got to handle with Katana. Yep, that's a game changer...
...especially if people hook up poor USB cables and power supplies.
However. What I don't really understand is the role of the "Supercaps" .
I would have assumed that these would be more then sufficient to cope with load variations. Perhaps these are just sitting at the wrong spots to cope with the reported issues. Anyhow.
Again. Great to see that Allo seem to have it all under control by now.
overhaul ending up in Katana-V2.
Great developments. I really like it.
Also nice to see is that the "Golden Ear" fraction contributed valuable feedback to these developments.
########
Obviously all this modulation and buffer-cap stuff seems to be related to the much higher load currents Allo got to handle with Katana. Yep, that's a game changer...
...especially if people hook up poor USB cables and power supplies.
However. What I don't really understand is the role of the "Supercaps" .
I would have assumed that these would be more then sufficient to cope with load variations. Perhaps these are just sitting at the wrong spots to cope with the reported issues. Anyhow.
Again. Great to see that Allo seem to have it all under control by now.
Yes load is dynamic but any RPI will have maximum current draw at boot time. After boot RPi3B+ takes about 1.5A that makes voltage drop be around 0.2V. Also we made sure to add more caps (10uf multiple) to help.
Generally speaking wallwarts (like ours) output 5.1V at 2A. Stay away for the cheapest ones that might output less under load. IFI , is not a great smps *but better than most*
What started to be a minor output filter upgrade on Katana-V1 turned out to be a major
overhaul ending up in Katana-V2.
Great developments. I really like it.
Also nice to see is that the "Golden Ear" fraction contributed valuable feedback to these developments.
########
Obviously all this modulation and buffer-cap stuff seems to be related to the much higher load currents Allo got to handle with Katana. Yep, that's a game changer...
...especially if people hook up poor USB cables and power supplies.
However. What I don't really understand is the role of the "Supercaps" .
I would have assumed that these would be more then sufficient to cope with load variations. Perhaps these are just sitting at the wrong spots to cope with the reported issues. Anyhow.
Again. Great to see that Allo seem to have it all under control by now.
Supercaps act as a reservoir of electrons. We found that main rail (that feed all rails/LDOs) had noise of 4mV under normal load condition. With the supercap added , we went down to 1.2mV (and then of course we filter it and have LDOs)
Adding the supercap was a not something that was necessary , but I felt that lowering the overall noise is a positive.