AYA II DS, DIY edition

As announced a few months ago, we have another run of AYA II.

This run, marked as DS, includes one update with respect to AYA II 2014: synchronous DEM clocking circuit is more completely integrated, so now it is not optional but integral part of AYA II project. By using BCK signal from TDA1541A input, this circuit operates properly i.e. synchronously, regardless of selected source.

PCB layout is updated to suit this change, and there are also several other minor changes on PCB, in order to facilitate mounting of possibly different parts.



The AYA II DS is sold out.



Related topics:

AYA 4: 2018 AYA release, featuring 192/384 kHz async USB

AYA II 2014 DIY edition talks

USB to simultaneous data (or I2S) board


189 Replies

  1. Avatar

    Wondefull cap than the bipolar BG N, after a while, try it in the opposite sense : one sens is clear and dynamic the other sense is fleshier and darker… it helps to find the best equilibrium with the whole system!

  2. Avatar

    Hi Pedja,

    I have please a quesion about the divider chip without the 470 pF cap. Is there a BcK speed to respect in relation to the 14 x 0.1 uF DEM caps when one inject its own Bck speed like with Ernest front end above ?

    What happen if the crystals are 45.xxx or 49.xxx Mhz range ?

    Or does the divider can’t work without the Aya II own front end with its 50 Mhz XO ?

  3. Avatar

    Cheers Ernst, and thanks for posting.


    Ernst’s board is -DS, with divider input connected to the BCK sent to TDA. In that way, in -DS edition the divider itself will work with any source. Since TDA1541A BCK input is claimed to have a limit at 6.4 MHz, it is the TDA as a whole that sets the upper limit here, and not the DEM clocking circuit.

    In 2014 version, the DEM clocking circuit input is “open”, so it can be connected anywhere, including the source MCK, which is normally a higher frequency. In this case you should check if divider chip input can accept given frequency, and you have to choose the adequate output as well.

    According to Philips papers, with synchronous clocking necessity for DEM filtering decreases. I have tried going that route too, and tried removing DEM caps, but did not like the results.


  4. Avatar

    Thank you for the input.

    So if Mclk used, does it mean the external MClk should be 5.xxx/6.xxx Mhz XOs max with the non DS version ?

    Better to keep the clock from TDA pin 2 (Bclk) with a coaxial 50 ohms wire with the gnd shield connected near pin 14 (Dgnd) ?

    On the DS I assume this is the /2 output at the divider, so Bclk has to be max : 11.xxx/12.xxx Mhz. Better to use Bclk with the non DS as well (like the DS) ?

    Sorry if I mix the concepts !


  5. Avatar


    Any advice on how to wire the isolation transformer for 230v. Not quite sure and it may be a silly question – sorry for what maybe a silly one.

    The primary of the isolation transformer gets the red and black lead from two coils connected = 230v.
    The secondary of the isolation transformer and primary of the step-down transformer simply connect each wire withs mate – black to black and red to red for each115v coil.

    Do I have it right?

  6. Avatar

    Hey Aron

    Maybe something to consider with your spare 9v supply is running you Amanero off it – much cleaner supply than anything from a computer.

    See here for some discussions regardind this (about half way down the page):


  7. Avatar


    If you use MCK as an input to DEM clocking circuit, you can also use 11.2896 MHz or 12.288 MHz, with probably any regular HC series binary counter, and maybe even twice higher frequencies, but you will have to check the max. frequency specifications of the given manufacturer.

    For digital ground routing, it is always suggested to use the point as near as possible to TDA pin 14. For shielding, assuming it is really only a shielding and not the ground path, it is not necessary to use this point, and in fact some power supply GND point can work better, or you can simply scrape the top side ground plane (actually ground fill), and solder the shield there.

    In DS version the whole DEM clocking part is moved physically closer to the TDA, so BCK path from TDA pin 2 is shorter. However my own AYA II sample is also 2014 version, and I connected BCK from TDA pin 2 to binary counter input (by the resistor), and it works fine too.


    It is no problem, of course, and this should be probably included into the assembly manual.

    Isolation transformer secondary windings should be connected in series, just like its primaries, when used with 230 VAC – the “outer” poles are then connected to the main voltage (step down) transformer. As opposed to the primary side, which “center tap” should be left floating, the center tap of the isolation transformer secondary side should be connected to the ground. This way the outer poles will have 115 VAC each, with respect to the ground, so they will be the same voltage but the opposite phase, and thus symmetrical i.e. “balanced”, while the voltage transformer will still receive its required 230 VAC.


  8. Avatar

    Pedja, could you make a simple schematics on how to connect isolation transformer to mains transformer in the 230v case?

  9. Avatar

    Thank you Pedja for all these inputs,

    Yes right my mistake, if clock wirering and coax the upper layer is indeed the gnd return 🙂 and not shielding !


  10. Avatar


    Here it is. Colors may differ, but the principle is this. All windings are in series.


    Attached file:

  11. Avatar

    Thank you very much, Pedja!

  12. Avatar

    Thanks as always Predja

    Hmmm. My isolation transformer hums once the load is connected. Is this normal?

    Could it be the earth grounding point?

  13. Avatar

    No, it is not normal.

    Is it the hum that comes from loudspeakers, or mechanical hum coming from transformer itself?

  14. Avatar

    Hello Aya Dac loving folks,

    There is recent talk of output caps here.. So maybe we could explore the options with Line Output Transformers too (in lieu of output caps.)

    I’ve recently discussed Jensens (JT-11-BMCF) with Pedja through email, and he has been graciously very helpful..

    I’m also looking at the Lundahl LL1684.. These have amorphous cores, and are said to be very intolerant of even a small amount of DC offset (saturates the core easily..) But has anyone tried these transformers with the Aya 2014 or DS?

    Another option is the Bifilar wound Cinemag CMOB-1H (High Nickel) types:


    They seem like direct competitors spec-wise to the Jensens mentioned above. They are slightly less expensive, but based on their heritage (winders for old Altec), looks like good quality too. Anyone tried ’em?

    Thanks for any feedback, or anything you guys may share.

    Best regards,

  15. Avatar

    Hi Frederick

    May I ask why ? My pass experience with Sowter did not convince me of its benefit.
    My opinion is for low voltage swings it adds more drawback then benefits. Yes highs & mid might be clearer but you loose drive, dynamics etc. Besides it becomes picky on the requirement on your preamp. Anyway PEDJA has done a fantastic job on AYA, aside from the tweaks that we made, there really is nothing more that u can ask for


  16. Avatar

    Hi Jaffrie,

    Thanks for chiming in.. Well, I like the sound of quality transformers in Dac output stages. But it all falls upon implementation I guess, and how suitable the output stage circuit is.. I tried it in parafeed once in another dac that was modified. But the nearby power transformer in that modded dac makes ’em pick up stray EMI / some 60Hz hum.. So line output transformers needs distance and most likely some shielding (the power transformer) in a fully DIY implementation.. BTW, I had Sowter MC step up transformers, obviously another app, but they were very undewhelming so I sold them (replaced by infinitely better Hashimotos.)

    Anyway, Pedja himself uses Jensens (as an option) in his current roster of commercial dacs. So I would think transformers sounds good to him too.

    I like caps too, don’t get me wrong.. Maybe like the new Jupiter Coppers suggested above, or Audyn Copper foils.. If anyone has tried ’em already.

  17. Avatar

    Hi Pedja

    How can I attach a music file. Like to share with fellow owners on how good the AYA is .


  18. Avatar

    Hi Predja

    The hum on the isolation transformer seems to have gone. I’m checking the voltages on the board. So far Ive resolved some of the issues which were my fault.

    I’m getting pin 28 +4.22VDC and on pin 15 -15.33VDC.

    Are these good enough? I’ve used Eldam’s mod on Caps 308 & 309 & 303 FWIW. Reason I ask is because all the other voltages so far are spot on.



  19. Avatar

    Hi Predja

    Found the pin 28 issue – BD139 put in incorrectly – its collector voltage is 4.2v so tats sorted. According to TDA data sheet the 15.33VDC is within limits so I’m guessing its fine.


  20. Avatar


    This Cinemag looks pretty much like JT-11-BMCF copy. It looks fine, but I never tried it though, so I am not sure how successful copy.

    People from Jensen actually do not recommend anything above 0.1 mA running though its windings, so with 40 Ohm winding resistance the maximum offset would be 4 mV. However, as we already talked, in my experience these transformers work fine with ten times higher values. I guess they are careful with these specifications, which is good, but in the real world of such a DAC, with thermal drift of up to a 10, 20 or 30 mV, Jensen transformers do just fine.

    Also, without output coupling caps this DAC can intermittently send up to 3 V DC to its output, be it associated to the lack of proper clock signals at TDA input, or to the synchronous data mode of operation. Anyhow, Jensen can stand such a torture too.

    And, one more thing to consider here, and the one I probably also should stress in the e-mails, is the output buffer. Please note that OPA861 is not internally limited i.e. protected from short / overload, so care is needed in this regard too.

    With 3 V DC at its output, and 120 Ohm output resistor in series with 40 Ohm winding being a DC path to the ground, the output OPA861 might be intermittently due to provide up to 19 mA or so, which is about to test its capabilities.

    And, you might in fact be tempted to remove this series resistor, and thus let the output transformer operate the best way (with lowest possible source impedance), but in this case you might conclude that the output buffer should be able to supply notably more current, 75 mA or so. This is actually not the case, firstly because the AYA II output stage supply is current limited to about 55 mA, and secondly, even if we suppose that all this could be drawn by one single (“shorted”) chip, it must be shared between two sides of OPA861 current mirror (its “emitter” and “collector”), so in this case the (“emitter”) output would be this way limited to some 27-28 mA.

    BTW, TI tests revealed that the most critical OPA861 part in this regard is not any internal transistor but metal trace, which blows first. Now, is 25 mA or 30 mA enough to fry it or not? I would be the most happy if I could tell you from experience, but I never actually used AYA II with output transformers, so be careful. (I did use them with The Model / Model S DACs, but their output buffer is different.)

    BTW, speaking about possible coupling transformers advantages and benefits, do not forget that, instead of the DAC output transformers, you can also use the (pre)amp input transformers. They will work also for possibly other sources, and input transformers are often easier to design (as they do not require low resistance windings).


  21. Avatar

    Jaffrie, now it should be possible to add media files.

    And, one information… There was a problem with replies notification software here. It is (hopefully) fixed a few hours ago, however all the previous subscriptions are lost. So, to keep an eye on this topic, the next time you reply, please check the notification box below submit reply button, and subscribe again. Thanks.

  22. Avatar

    Hello Pedja,

    Thank you for the very detailed reply.. You’ve given me a lot to chew on. I appreciate the caution. I understand now that while it is possible to use Jensens, there are still big risks (plus it has never been actually tested in the Aya II..) All the while I thought that the Model S’ output buffers are “similar enough” to the Aya II that the risks of using line output transformers are almost non-existent..

    Best regards,

  23. Avatar

    Basically, AYA II was not meant to work with output transformers, but there are people using it this way. Still, transformers do like low impedance source, and things may get more complicated once you go this route, and remove the output resistors.

    The main advantage of Model S output buffer is that it is more robust than AYA II output buffer.

    @ Brent: Yes, -15.3 V is fine.

  24. Avatar

    Even if it works core size of the opt is important to good sound reproduction especially in the bass

  25. Avatar

    Hi Pedja,

    Thanks for the further clarification.. It’s interesting that there are those brave enough to use line transformers with the Aya II. Yes, from what I deduced in your explanation above, Robustness is the main issue.. And not an issue at all with the Model S II / USB.

  26. Avatar

    @ Jaffrie: Sure, that is why these Jensens are quite substantial for line level devices, about half kilogram each.

    The things with signal transformers however get somewhat different, once you move them from source output to the (pre)amplifier input. In that case you do not have a requirement for low winding resistance, and thus you can achieve required inductance by the winding, and not that much by the core.

    Further more, due to their higher winding resistance, I would consider such input transformers mostly safe for use with directly coupled AYA II output buffers. In turn, their reactive impedance components may be higher, so even if AYA II output stability is mostly managed by R417/R418, it is not bad to check that part too.

  27. Avatar

    Think its way better to use transformer as interstage instead either line out of preamp or between input n driver stage of power amp. Thks for your thoughts Pedja

  28. Avatar

    Hi Guys,

    Just like to share. Recording made using Iphone.
    Aya is very well designed & tweaks are easily
    heard. You can tweak the dac to match your system


  29. Avatar

    Another recording. Strangely the recording has
    lots of echo but in reality my room is not really
    that reflective. Anyway hope that you guys
    can hear into the recording itself


  30. Avatar

    AYA DAC heard through MonicaDAC 🙂

  31. Avatar

    Hi is the AYA II DS board still available from here or other sources?

  32. Avatar

    Does the 1545 Monica dac sound this way Ernest 😆

  33. Avatar


    Recording 11 was very sound; clear and bright. Recording 13 was a lightyear from the original sound; pretty muffled.
    Btw, Monica/Mojo is a pretty good DAC despite the simplicity. But I expect AYA II being superior after run in.

    Cheers Ernst

  34. Avatar


    Human mind automatically concentrates on the the source i.e. on the direct portion of the sound, while microphone does not distinguish direct from delayed / echoed part, of course within its polar pattern. So I guess that is where the “difference” came from.


  35. Avatar


    Yes, several boards still remained, and two EI transformers as well.


  36. Avatar

    Hello Pedja,

    do you “burn in” your AYA Dac’s before shipping and what Sound do you use for it?
    Thank you for request.

  37. Avatar

    They usually worked for an hour or two, just to make sure everything is fine with them, but no longer than that.

    I have no special recommendation regarding burning in, apart from listening to the music, and waiting for a week or two. Burning in does not make miracles, but it is certainly audible.

  38. Avatar

    My compliments to Predja on the design of this dac – my Marantz CDA94 sounds veiled when compared to this implementation of the TD5141A. The sound does make me smile and she’s not even properly broken in yet. At some point I will try some other output mods but for now I am happy.

    The DAC runs a bit hot though – around 50 degC for the Transformers and the dac chip – is this normal?
    Means I’ll have to add some serious ventilation to the box when I place her inside it.

  39. Avatar

    Thanks, Brent.

    A 50°C on the surface itself is still fine, even though it is on the high side.

    Of course, the temperature can increase, once you put the DAC into the box, so ventilation holes that make good air flow path might be welcome.


  40. Avatar

    Revised assembly manual has been sent today. Please check your mail.

  41. Avatar

    Finally my Aya DS II have find a nice cabinet☺ i like full sized cabinets.

    Attached file:

  42. Avatar

    Nice Patrik,
    Are you done with your tweaking ?

  43. Avatar

    Thank’s Jaffrie!

    Yes with this one i am. I have one more pcb that i’m going to tweak little different for fun.

  44. Avatar

    Hi Patrik,

    Nicely done! A photo if the innards would be great to see too…


  45. Avatar

    Thank’s Frederick

    Nothing special inside! The sec trafo cables was little short so i made them longer and i had some white sleeving over that i cover them with.

    Attached file:

  46. Avatar

    Hi Patrik,

    If your load has DC blocking cap at its input, you can also try without the DC blocking cap. Different sound presentation.

    Worth to try it and keep the prefered conf.

  47. Avatar

    Thanks Patrik! I guess distance is still your friend even with a toroid.. I might just emulate your faceplate lay out, clean, with small toggle switches at each end, but most likely in black.. Thanks for sharing, and happy listening..


  48. Avatar

    For those interested in using simultaneous data protocol:

    Asynchronous USB to simultaneous data converter board, interest check

  49. Avatar

    Hi Pedja
    Forgive my ignorance, is this going to be a further performance enhancement over the present usb connection even on 44.1 k recordings


  50. Avatar

    Yes, it is the asynchronous USB i.e. master device, with its own master clocks that control the DAC directly (so no PLL), as well as data flow from PC.


Leave a Reply