USB interface and AYA 2021 editions [poll]

According to the feedback we got recently, the interest in these projects may have re-emerged. And in case it can reach 30-40 units per project, we are going to proceed with the given project and release it, as limited series.

Both these projects would be current developments of the previous ones, so those not familiar with them are advised to check them out. (The links are below.)

The USB interface board would now support multichannel operation (no worries about the usual stereo operation, it will work straightforward), and will output Philips simultaneous data or I2S protocol. Other than the set of U.FL, it would include also an HDMI output connector, complying with recent specifications set with the S5 DAC release.

The AYA may be available in the form of the DIY project, whether as a PCB with a pre-mounted USB input stage, or as a completely mounted PCB, and it can be available also as a completed unit in chassis. Will we keep the AYA project DIY friendly, or change it to better suit machine assembling, will depend on the interest. Also, it may and may not become “5” this time, it is on you. The S/PDIF receiver could be replaced with the currently active part, and other possible changes, with regard to the AYA 4, may include more inputs or some tweaks in the supply. You can post here your requests, ideas, and comments.

The price for the USB board would be €150–200, depending on the clocks used. The AYA prices would be similar to those of AYA 4, so the PCB with USB stage would cost about €200, completely mounted AYA board about €600, and completed AYA unit about €900. Shipping by Airmail (International Priority equivalent) for these items would be €15–30.

Please note this is a poll only, and for now you can not actually place an order. If we indeed proceed with some of these projects, and you have signed up, you will be notified by e-mail, in the next two weeks or so, and then you can place an order.


07.07.2021. UPDATE:

The interest in the AYA DAC looks good, so we are going to proceed and release it. The
AYA 5 follow-up is here.

We will probably release the USB interface board too, but it is re-scheduled for September. In the meantime, you can still signup for it, so you will receive an update on this one.



2021 limited editions




Previous projects:

USB to simultaneous data or I2S converter, series 2


65 Replies

  1. 18-06-2021

    Hi Eldam (diyiggy),

    The sma connector will not work. The new clocks have a sine output and they need to be converted to square wave by a separate conversion board, connected with an sma cable, that conversion board has the dil14 footprint of standard clock modules.
    I have the 5xxx and 6xxx clocks with 4 doublers and a set of 22xxx and 24xxx clocks that I can put to good use.
    In the picture you can see the small separate squarer pcb with the dil14 footprint.


    Attached file:

  2. 18-06-2021

    It will work with the square wave converter board Andrea Mori did. It was made for that purpose to swap clocks like the Crysteck, NDK, etc.
    At least that’s what he told me if you follow the thread when I asked 10 days ago on him thread about the AYA 5 as I had some ideas to push it beyond the previous editions.

    The only limitation I see if you want to use the AYA 5 embeded USB to simultaneous board is what Pedja answered above : you need 22 & 24 speed because the roms of the USB Xmos is programmed with those speeds (or use the uf-l but with your own front end and its clock)

  3. 18-06-2021

    Btw I asked also the Crystek (a little larger to flow soldering with flux) pads for the ones using Crystek or DIL14 arengement (Pulsar Clock for illustration). It should have not vias but larger pads if Pedja has room for this purpose.

    He told us to wait a little as he is busy with the A20 amplifier before…

    Feel free to ask Andrea if I made an interpretation mistake about the SMA connectors, it will help a lot.

  4. 19-06-2021

    @ Supersurfer:

    Thanks and you are welcome.

    Clock switching will not employ clocks’ enable pins but the multiplexer.

    If Andrea’s oscillators are originally sine-wave, that only additionally gets squared by the module which fits into the CCHD-957 footprint, I will try to make such clocks footprints.

    @ Eldam:

    I did not plan that this time, but I will try to offer the firmware for other clock frequencies again, so the users can update it via the Windows control panel, so it matches their clock set up. But we should limit ourselves to two, maybe three options.

  5. 29-06-2021

    A quick update, before I post more detailed information, and we start accepting the orders.

    I mostly defined the AYA 5 output stage, so it is fully discrete, and something between the one I published in 2004 (you can now find it at legacy page), and the one used in The Model (the first Audial DAC). We are going down to 0.00x% THD with this one, without feedback, so I believe you will like it. It is not very complicated either, however there are 15 transistors per channel (this includes the output buffer too), and I think it is best if we use SOT23 transistors. Also, there are something like 13-14 resistors within the circuit, per channel, which should be SMD 1206 or Mini-MELF.

    I am not sure how many of you can solder SOT23, but for those who can not, we will offer the option with all these SMD parts pre-installed, at a reasonable cost, I hope.

    We can keep the output stage supply, and most of the D/A parts still classic through-hole. On the other hand, the S/PDIF stage will be mostly SMD.

    Comments are welcome.

  6. 29-06-2021


    Some questions Pedja, please :

    the good carbon comp for the regulator discrete stage are not possible anymore or Melf & smd can be carbon comp as well ? I understand the regs stage are still through hole ?

    Is there room for a through hole I/V resistance and cap as the previous Aya ? As perhaps the serie resistor between the I/V and buffer ?

    SO, sorting out definitely not mandatory for the 15 sot per channels ?

    Is there still the possibility to trim the output of the two channel to avoid DC blocking caps as the previous AYA edition?

    What finally did you decide for the crystal layout options as well as crystal frequencies related to the OS materials ?

    Many thanks

  7. 29-06-2021

    I/V and series resistors in front of the buffer, and caps bypasses across the I/V resistors remain THT. These footprints might be a bit smaller than before, but they surely remain adequate to accept AB resistors and styrene caps.

    Regulators may and may not involve SMD, but I guess the DIY-ers will appreciate it if not?

    Transistors matching (and resistors matching, while we are here) may be beneficial, and may yield somewhat better numbers, but everything will work well without it too.

    Offset nullifying current sources are again included, however the coupling caps removal is still not an officially suggested solution (even though it might be possible, under some conditions).

    As for the USB stage master clocks footprints, the goal is to make them so they accept both the 5×7 mm parts and CCHD-957. I will get back with the outcome in a few days.

    Standard clock frequencies will be 22.5792 / 24.576 MHz, so Fs up to 352.8 / 384 kHz will be supported. As I said before, we can make the firmware migration file so some other frequencies may be used too, but we should define one or two most typical alternative options.

  8. 29-06-2021

    Thank you for these inputs.

    I surmise through hole for the regulator stage will indeed be more friendly.

  9. 02-07-2021

    Hi again,

    about that Crystek clock foot print. IMHO, what diyers would prefer most is not really the pads for the bare can. Most of us are using a pcb adaptator – as you did with Kyocera Xtals IIRC – that has the size of the DIL16 socket. The 4 vias/leads of these pcbs adaptators many still use for their DIY clock projects are of course at the corner of that DIL16 square.

    That’s because many people started to experiment with the Crystek 957 8 years ago but now use also more modern oscillators that still fit on a DIL16 size

    If not vias are possible for any reasons, perhaps pads at the corner of that square size could be very welcomed in order to flow soldering instead vias and leads. Vias are of course useful because it raises the heigth with the DIL16 socket in order to put sometimes crystals pcb that can be a little bigger and so the main pcb parts like caps or resistors are not touched by a direct stack. Some prefer direct stacking bu th eones using DIL16 support of course use extra decoupling embedded on the little crystal pcb on the oscillator hat. Many things were done in almost 10 years and we are not younger for sure – pehaps more deaf, go figure 🙂 -.

    Just to precise as I was the one who asked about clock pad option and I was not so precise – thanks Supersurfer to have rectified-

    Hope that helps.

    best regards

  10. 04-07-2021

    Hummm, just checked : it’s a DIL14 spacong, not a DIL16.

  11. 05-07-2021

    So, the outcome – with a big thank you to everyone for your input!

    The AYA 5 PCB design is finished, and in the next day or so we will start accepting the orders. We are going to set up another page for this purpose (I will post the link here), and everyone who signed up will receive the notification by e-mail.

    This AYA includes more changes (with respect to the previous version) than I originally planned, and even if we will possibly manage to ship some items in August, September is now more realistic.

    As for the design itself, most shortly, I can confirm the AYA 5 will have USB, S/PDIF (BNC), and optical S/PDIF input. Alternatively, instead of either USB or these S/PDIF inputs, it can have PCM (I2S) direct inputs at U.FL or HDMI connector (per Audial specs). The D/A and output stage supplies are improved, and both the I/V and output buffer stages are discrete, both employing a diamond transistor topology.

    Unfortunately, I had to give up on installing the DIL14 footprints for the USB stage clocks, there was just not enough space for this (without more significant maneuvers in the rest of the board). There will be however a U.FL input at both clocks though, so external clocks can be still used.

    You will of course find more information on the forthcoming AYA 5 page.

    We will most probably proceed and release the USB interface board (Mk3) too, but taking into account the amount of work required for this AYA 5 release, and the interest in the AYA is higher than in the USB interface board, I settled on re-scheduling this one for September. I will try to include there the above said DIL14 footprints for the clocks, and increase the supply current capabilities for OCXO too.


    (Click on the image to enlarge it.)

  12. 05-07-2021

    Many thanks Pedja for all these novelties.
    I am to pull the trigger for the USB board poll.

    Some people surely will made a DIL14 to uf-l adapter. If staying short with 2″ uf-l cable, it could make a good trade-off as well on the AYA5 board for those wanting to keep their special clock VS a CCHD-957 or the crystals provided with the AYA5.

    I hope some people will fill in the poll as well with the USB standalone interface as well….

  13. 06-07-2021
  14. 07-07-2021

    Looks like a good step forward in the AYA evolution Pedja!

    I’ll be keeping an eye out for the USB interface board though, that’s where my interest lies at the moment.

  15. 07-07-2021

    Looking forward to the coming USB interface with its DIL14 4 vias arrangement as well.

    Tip to try with the Crystecks according the main power supply : despite the embeded Crysteck oscillator has local decoupling under the can, I experienced it sounds very better if a near local decoupling is avoiding Ceramic class B for 0.47 uF FCA caps from Cornel Dublier (smd package) per CHDD-957. My Aya 2014 is running like that with the Crysteck.

    The uf-l inputs near the crystal should provide an input for the old Mori’s Discroll for enjoying the USB to simultaneous on board stage. I beleive it’s a great novelty.

    The new USB to come if having at least a Gnd vias and local vias around the 14 dil size can provide this time the newer SC-Cut crystal Discroll oscillator some use.

    Whatever the AYA5 and the New USB board, both should provide the 3.3 V/mA amount they are able to provide and maybe a strap with a zero ohm smd for people wanting to disconnect local 3.3V for a more powerfull third party reg that support support their third party oscillator hungry, imho.

    At least one can always cut the 3.3V pcb trace, of course if needed for DIY.

    Have the AYA 5, Darlington Tip chips yet ?

Leave a Reply