The Zishan dsd's corner

[Themilkman46290]

Yup, TPS7Aхх family, both positive and negative one.

These are quite low-noise (compared to DC-DC boost converter), also the way negative rail is arranged in DSD creates uneven voltage on positive and negative rails, and negative rail is only capable of delivering a fraction of current (I believe). So all in all these SHOULD make an improvement, but soldering them in place of the inductors is tricky, so much kudos to So-Hm for getting these LDO going, I would not even dream of attempting it. On the other hand capacitors (both electrolytic and ceramic) should hold enough charge/provide sufficient filtering to clean up +/- rails, so this mod may or may not provide significant gain in terms of quality.
I also note that more recent and advanced chi-fi DAP designs (IHIFI Tuner) use TPS7Aхх family LDO in a similar way, but utilise more recent/better specification QFN form factor ones.
Or you can get one of those:
https://www.ebay.com/itm/TPS7A4700-Low-Noise-Power-Module-RF-Audio-Board-3V-3-3V-5V-12V-15V-1A-Adjustable/153160529239?ssPageName=STRK:MEBIDX:IT&_trksid=p2057872.m2749.l2649

Are you talking about xuelin h8 tuner? Looks prety interesting but if i had $700-800 for a dap, i dont think i would be modding anything, just buy the right dap from the start. Thats why this dsd is my baby, good, cheap and upgradable, and not flashy
So far i feel it sounds better then the fiio x5 (almost bought it used last year) and definitely sounds way better and works alot better then the fiio x1ii (complete junk, sounded digital, froze and crashed if i even thought of it) but they are "sexier" looking for sure, but the x1ii cost me $130 with tax the dsd cost about the same but it cost me about $60 to upgrade so that comes out around $200. The fiio x5ii is about $350, and the x5iii is about $500,so i am still blown away by this player

Cant wait to see if zishan will update to ak4499 chip, hopefully they make it more like the ak4495 version, still the most versatile and best sounding stock, but with the edition of a 2.5 balanced instead of the line out

 

[HeyManslowdown97]

Spoiler: Timing is EVERYTHING

I believe upgrading oscillators will have a bigger impact to your sound since it involves conversions directly to AK449xx chip than the voltage although in the case of the TPS73250 processing is pretty impressive and it ain’t your grandpas LDOs so I would imagine that basic electrical practices have changed in past 30 years or so with Moore’s Law.

It’s also important to note that datasheets are mere standard guidelines set by the manufacturers who make recommendations based on Lab results. I’ll also add that experimentation on this thread doesn’t bother me but I can’t speak for the rest.

“Basic research” often doesn’t start off having practical nor commercial application but eventually become the fountainhead to something useful.

Also, any “rule breaking” in sound practices should also be encouraged since it’s what is aesthetically appealing is also subjective. In principle this should apply to the any Mods fo the Zishan but there’s also such a things as dumb ideas (which I for one are full of them!) [\Spoiler]

I am thinking of using a CMOS op amp as a buffer in the lpf which makes a lot of sense based on unscientific hunch. Didn’t know AK makes their own wonder if they are any good or enhanced their signature velvet sound and overall performance.

https://www.akm.com/akm/en/product/detail/0042/

 

[Ivan TT]

Spoiler

I REALLY struggle with this...

I believe upgrading oscillators will have a bigger impact to your sound since it involves conversions directly to AK449xx chip than the voltage although in the case of the TPS73250 processing is pretty impressive and it ain’t your grandpas LDOs so I would imagine that basic electrical practices have changed in past 30 years or so with Moore’s Law.

What do you mean when you say "conversions directly to AK449xx chip"?

Oscillators provide clock reference, replacing them with more stable versions (like I and some other members here and on russian counterpart forum, which is absolutely safe to use by the way, done) improves jitter, which has positive effect on HF quality (softening highs up) and soundstage. Replacement parts are readily available from arrow and their codes are easy to find using search function in this thread, so the hard work of finding suitable and compatible yet inexpensive components and confirming that they work has been collectively done.

TPS73250 does not do ANY processing whatsoever, it provides AK4497 with 5V power source, used to power analog output section of AK4497 and serve as a VREF. Compared to (genuine, but lower quality, not "fake") stock LDO it has reduced noise, providing yet another incremental quality upgrade.

The effects of oscillator and LDO replacement are similar (as described above), but reason for improvement is different: less jitter induced distortion (which is not harmonic distortion by the way and thus is much more noticeable); less noise in 5V power supply decreases noise floor and its masking effects (very noticeable on long reverb tails for example), and (most importantly) stable VREF is crucially important in DA conversion, as fluctuating/noisy VREF voltage introduces noise/distortion. Another way to improve VREF stability is to add capacitance, which has been done by few members, including myself. Please also refer to the extract from AK4490 evaluation board manual (older revisions, it's been redacted from current version):

I believe the best way is just replace both oscillators and LDOs and add VREF capacitance and have a good listen, this will leave very little room for speculations as to what effect replacements/upgrades may have and which component has more significant effect on sound quality, again this has been done by a few members already, who posted their impressions in this thread.

As to the basic electrical practices, how are they related to Moore's law? Moore's law is related to transistor count increase in CPUs, while LDO's have had only few transistors for for last 30 years since they were introduced....

 

[Merlin-PT]

The matrix quote I made was 100% a compliment to the mods Ivan did.
https://www.head-fi.org/threads/the-zishan-dsds-corner.826185/page-139#post-14823365

 

[Ivan TT]

The only way forward for the thread and community I see feasible is this:

How everyone who upgraded to 0.4I finds it?

My impressions so far:

1. Fixed crashes/restarts when in shuffle mode and album art is displayed
2. Better stability overall, only one crash since upgrading
3. Lows became better, more punch/definition. I actually took my DSD from "High sound quality mode", as i feel it smears transients
4. Highs are a touch softer/silkier, this also improved soundstage (I feel like floating in a bubble of sound)
5. Strangely enough there's much more noticeable difference between lossless and lossy (mp3) files sound quality. FLACs are like "wow!", mp3s are like "ups, that's embarrassing"
6. Zishan runs a bit cooler too (probably due to CPU code optimised/errors fixed)


Overall sonic profile became very close to F.Audio FA1 (as I remember it), but obviously without any annoying clicks that FA1 had.
Makes me very happy , as the whole point of buying into AK4497 DSD and modding it was to get that mesmerising FA1 sound while FA1 was useless to me due to clicks.

 

[Themilkman46290]

Thought i would mention my impressions on some mods

i did the mods little by little, tested nichicon es, fg and kw elna cerafine and silmic2, with similar values and higher values, tried with and without the ldos and can say every mod makes a difference, some better,some worst but i did find a nice mix

Switched op275 for ad8620, it seemed to give better seperation, soundstage feels a bit wider but bass seemed a bit less and i could still here a little bit of noise in the background ( some distant electric noise during certain music, bassy simple types of beats and a weird fuzziness on some busy beats) the ldos seemed to fix that
Changed input caps, nichicon kw 10v 100uf was ok,i liked the two fg 10v 47uf (94uf) better, they seemed warmer, the two es10v 47uf seemed more detailed but cold. then tried elna cerafine 10v 100uf, they were a bit better to me, warmer but in the end silmic 2, 6.3v 220uf was the best, warm but very detail, and for the output tried nichicon es 10v 470uf (seemed to be take away from the midbass, gave very weird sub bass) settled on the elna cerafine roa 10v 330uf (everything souded good, added a little touch of warmth)
Also tried a few opamps in the dip8 socket, burson v5i, lme49720ha, two lme49710, muses8920, opa2604, opa2134, lt1364, opa827 lme6172 and ad823
In the end, it was between v5i for the mids, 2 lme49710 was the best all around but took alot of battery life, the muses8920 gave me the bass i wanted but i still switch between it, the ad823 for its imaging and warmth.
All sounds exactly how i hoped, the added capacity gave my low impedance iems a nice little boost in the sub bass, everything sounds smooth, really clean highs and smooth mids
Less noise better seperation, gave the beautiful wife a listen, she was surprised

 

[Themilkman46290]

The only way forward for the thread and community I see feasible is this:

How everyone who upgraded to 0.4I finds it?

My impressions so far:

1. Fixed crashes/restarts when in shuffle mode and album art is displayed
2. Better stability overall, only one crash since upgrading
3. Lows became better, more punch/definition. I actually took my DSD from "High sound quality mode", as i feel it smears transients
4. Highs are a touch softer/silkier, this also improved soundstage (I feel like floating in a bubble of sound)
5. Strangely enough there's much more noticeable difference between lossless and lossy (mp3) files sound quality. FLACs are like "wow!", mp3s are like "ups, that's embarrassing"
6. Zishan runs a bit cooler too (probably due to CPU code optimised/errors fixed)


Overall sonic profile became very close to F.Audio FA1 (as I remember it), but obviously without any annoying clicks that FA1 had.
Makes me very happy , as the whole point of buying into AK4497 DSD and modding it was to get that mesmerising FA1 sound while FA1 was useless to me due to clicks.

I did notice the filters also make a bigger difference
And i didnt have alot of freeze ups before but now it seems to not have those issues

 

[Merlin-PT]

I also noticed the softer/silkier highs mentioned, to me the highs sounds better in some passages, but I'm not sure about all songs, I need to listen more.

 

[silverfishla]

Does anyone hear a Left side favor in the channel balance? After the update? Could be my ears.

 

[Themilkman46290]

Does anyone hear a Left side favor in the channel balance? After the update? Could be my ears.

Nope, sorry, but it could be ears, or maybe something shifted on the board, something got overheated

 

[HeyManslowdown97]

Does anyone hear a Left side favor in the channel balance? After the update? Could be my ears.

Check the caps with a multimeter. Would be my guess? It must not be firing out all cylinders. Or somewhere the board is making contact eventually. It happened me. And the 2.5 and 3.5 is faulty at times.Also I’ve found it takes a while for the Zishan to warm up; it’s got some kinks.

 

[silverfishla]

Check the caps with a multimeter. Would be my guess? It must not be firing out all cylinders. Or somewhere the board is making contact eventually. It happened me. And the 2.5 and 3.5 is faulty at times.Also I’ve found it takes a while for the Zishan to warm up; it’s got some kinks.

Thanks, I’ll check it out tomorrow. It’s slight but noticeable.

 

[Merlin-PT]

I switch the headphones L R ear sides to rule out music, ears and bad placement in ears.
If possible also try with other headphones.
If you need to open the box, try to press or change the socket opamp, bad opamp connection to the socket can do that.

 

[HeyManslowdown97]

The matrix quote I made was 100% a compliment to the mods Ivan did.
https://www.head-fi.org/threads/the-zishan-dsds-corner.826185/page-139#post-14823365

The EXACT Matrix Quote used to referenced the love of circuitry.

“You take the blue pill, the story ends. You wake up in your bed and believe whatever you want to. You take the red pill, you stay in Wonderland, and I show you how deep the rabbit hole goes. Remember, all I'm offering is the truth.”

How I used it was comparing the difference between Basic Research to commercial Research and Development to bring a product to market.

For instance, It’s like studying neurotransmitters in a PhD program and publishing compared to pharmaceutical company developing a useful drug that builds on that research. It’s the same for circuits and micro processors.

How it relates to Ivan experiments it’s neither an insult nor a compliment I don’t understand what he’s attempting to do nor care. I’m here for the Zishan mods. All I said was it isn’t wise to play with octaves on the DAC. Again, I’m entitled to my opinion and it wasn’t meant to be taken offensively but somehow it was twisted to be as such??

Texas Instruments and a plethora supplier sites like Arrow have experimental forums and already have schemes to make cool stuff on them. You can do it yourself if you are interested or build on them your own designs. They also have sources to educate yourself on ANY electrical topic. So I don’t have to visit dodgy websites when I have cutting edge ones here. But again we are encouraging in off topics. Personally, I don’t care what he does or post but he’s wagged his finger at me and accuse me of going off Zishan topics and it was fair criticism. People may not be interested in my musings. And that’s fair enough. Here for the good vibes not bitter sour Notes in high frequencies.

 

[Themilkman46290]

Yeah what merlin says, its also good to check the legs on the op amp, one of the legs of my ad823 was a little off and had same issues in my z3, also had something fall into the socket hole where the opamp legs go in and it had similar effect on my z1