Forum breadcrumbs - You are here:CommunityGeneral discussionsAYA psu mods
You must be logged in to post to the forum.

AYA psu mods

12

Topic for summarize recommended mods of the AYA DAC psu and further ideas.

One named "Eldam mod" recommends to add parallel 330uf to exisiting 1500uf before -15V regulator with (optional) removing 100uf after the regulator.

Question for the author (Eldam) of the mod, if and what kind of nannofarad bypass cap does he recommends and if holding the original ~1500uf capacitance (so exchanging 1500 to 1200uf afer adding 330) is really crucial or maybe resulting bigger 1830uf is even better.

Maybe such mod on all psu lines would be beneficial?

Moving here Eldam's answer:

Eldam: "After hundreds tests, I liked so much this mix of Pan FM (330/35v long length, 5 mm gap legs) & FC (1200/50v short length) than I had to share. But for me leaving the 100 uF is not as good than remove it and putt the close decoupling: caps according your taste and final result: lytics with good inductance (2.5 mm gap between the legs) should be tested as surface mount of all sorts or film caps with radial leads. But and that’s an important point: according to me, my hifi, my tastes: leave the smd decoupling of the TDA -/+ 5V unpopulated : so juste use the smd decoupling of the -15V at the bottom.
The capacitance from my experience is found from 0.1 uF to 0.47 uF. For me this arrengement sounded better than with the 100 uF"

As you already know, I like this idea, and that's why Eldam's suggestion was included in the AYA II and AYA 4 assembly manual.

And in most cases, I could agree with lowering the local bypass value here, and using the highest possible quality part - only if there were not those critical systems, and recordings.

That is why I still keep (mostly) a usual local bypass caps values. The overall result is just more "forgiving".

But, I can also confirm that this problem indeed might be alleviated a bit by tweaking the reservoir cap(s) value(s) and type(s).

Hello,

Re: Mr. Eldams recipie as stated above, is it a correct interpretation:
1.  Supply +/- 5V: Remove the final decoupling 100uF and SMD bypass altogether.
2.  Supply -15V: Remove the final decoupling 100uF and leave the SMD only, value of which should be 0.1 and 0.47uF.

Re: Pedja's response, if it is okay to ask the 'tweaking of reservoir capacitor cap value and type' for alleviation of 'un-forgiving' sound, does this follow closely enough with Mr. Eldams recipie with FM 330uF + FC 1200uF combo?

With much interest and thanks.
HK

Hello, its me again.

The Panasonic FC and FM suggestions, okay, but why no mention (?) of C01 (also C04, C07 for the 3 supplies) which is at the gate of the first transistor of the Darlington pair (of each supply).

To contribute something: I have pulled (only) the load end of the resistors after the rectifier(s) to have them stand vertical, placed some small chokes in that reclaimed space (both + and - legs of the supply), added 1200uF capacitance above that junction (physically), so electrically before the choke(s) creating an additional filter which then feeds back into the PCB where the load side of the series resistors were removed from.

HK

Quote from Guest on 11/05/2019, 05:09

Hello, its me again.

The Panasonic FC and FM suggestions, okay, but why no mention (?) of C01 (also C04, C07 for the 3 supplies) which is at the gate of the first transistor of the Darlington pair (of each supply).

To contribute something: I have pulled (only) the load end of the resistors after the rectifier(s) to have them stand vertical, placed some small chokes in that reclaimed space (both + and - legs of the supply), added 1200uF capacitance above that junction (physically), so electrically before the choke(s) creating an additional filter which then feeds back into the PCB where the load side of the series resistors were removed from.

Idea came from Pedja's use of additional filtering in the supplies of the Model S, the use of chokes (and this was so long ago that I don't remember mH's but DCR is negligible) from AMR design.  Subjective impression was 'this is better', bias etc, all of that, but it was not worse.

HK

 

Dear Pedja, please tidy the above as you find appropriate.  Thanks.  I would add that I really don't consider the result from addition of chokes as 'could be better', it was/is clearly better.  C at the gate is BG standard, and HiQ NX decoupling, all provided by your good self.

In my personal opinion, although adding additional bypasses to the the main filtering cap I find beneficial - I don't find it sonically nor logically completely removing the local 100uf reservoirs, or small 0.1 caps. I think any mods here or there to the existing stabilizer topology makes changes in the noise suppression/ stabilizing efficiency and so changes the sound, but are still not ideal.

Because of that, I switched to battery powered PSU - this is my recommendation, and I find it superior to the ac-powered. I use 7-cell Ni-Mh packets (8.4V nominal, but it is about 9V), connected just to the legs of the main filtering regulator input capacitors (so NOT trough the bridge and resistors - they are floating). What may be most surprising (like it was for me), the most evident improvement is even from battery powering the first 3 digital stages - that is USB, Isolator and Dem. - So that could be the evidence of how clean battery voltage compares to ac-filtered-stabilized voltage, or how much have to be done with the stabilizer to have the same ideal result.

Quote from Guest on 13/05/2019, 14:24

... again I forgot to log in - so signing the above post by myself 🙂

Okay with your acceptance of battery power yet I would not find that as an acceptable solution, having being down that road before with many things, it has a habit of wearing thin with the charging and maintenance - especially considering that many report, and indeed that I find, certainly WRT DHT filaments, which may or may not be different that these battery supplies are not so low noise or linear in their impedance as their mains powered counterparts.  That said, and battery aside - perhaps we can discuss other elements of the PS which are easily modified by the end user, as per subject of the thread and entries so far.  Kind regards, HK.

12