DIY DAC with premade PCBs, some questions

BTW - in the comments he provides a drive link to his parts list and designs, well worth checking out, his electronics are all together under €1k! o_OBest of luck
the video description just has his speaker in it, did you post the wrong video?

just watched this video:

i actually did not like the terminator compared to the vermeer... bummer... it sounds not as clean and transients seem smeared, it seems kinda oversmoothed

so maybe NOS is nothing for me?
 
You can't tell a huge amount about how a dac sounds from a youtube video. First of all, the sound is lossy compressed. Second, you are listening to it through yet another dac (as well as an ADC and mic).

Look, there is no magical cheap dac that can be turned into winner if only a few better parts are used. Lots of people go through that type of thinking, but only the very easily pleased folks will be satisfied with the outcome. A really good dac is hard to make and will probably cost you dearly one way or another. You can learn how to make your own, but that will come at a cost. Or you can buy something serious, maybe such as a Mola dac, or at least a good Holo.

Getting back to the diy angle, from what people around here keep saying, the Miro AD1862 dac is winner and fairly simple to make. Lots of people seem to have trouble finding anything better, particularly at low-ish cost. Now if cost is not a concern, then pretty sure there is better.
 
Last edited:
You can't tell a huge amount about how a dac sounds from a youtube video. First of all, the sound is lossy compressed. Second, you are listening to it through yet another dac (as well as an ADC and mic).
i know, absolute performance is hard to tell via a youtube video, but you can easly compare two devices imo, of course it wont come trough as in person but it gives you a rough idea
 
Its not necessarily even enough being there in person. It matters what the system sounds like you are listening on. It matters if there are ground loops in the system, if there are Bose speakers, etc.
imo youtube comparisons work fairly well if the recording is good to give you some idea

i did already watch a few comparisons that offer lossless files, its of course easier to hear with lossless but you still get the same idea on AAC it just doesnt sound as refined

well people report more analog sound with nos but im worried that nos is just smoothing the sound/transients a bit more which may make it easier to listen to... kinda the wrong solution for the problem if its true
 
the video description just has his speaker in it, did you post the wrong video?

just watched this video:

i actually did not like the terminator compared to the vermeer... bummer... it sounds not as clean and transients seem smeared, it seems kinda oversmoothed

so maybe NOS is nothing for me?
The drive link is in the comments section, not the description. Incase you have difficulty its below:
https://drive.google.com/drive/folders/158zzLfJvMhsfgb7URPL4A5bm3HV45zHO

I am actually somewhat of a believer that good sound can be portrayed via YouTube videos, you get a sense of dynamism and the sound signature, the same as listening to live performances on YouTube, not ideal and highly influenced by indirect factors, but there is a taste of what its all about, at least on axis!
 
Ghoostnight,

A lot of folks here with a lot experience have been trying to help you. You don't seem to want to listen. You're eventually going to either do something or do nothing. Good luck.
is it about blindly following the advice of others? i have to see what fits best to me, specially regarding future, i probably wont build multiple small dacs... but "the" dac i can retune etc..

@Ghoostknight given how many times the comment "worried" appears in your posts I have to concur with @PJN
JDI (just do it!) and then post for help with data and results
well i wanna be somewhat sure if i gonna spend few hundred euro on this stuff, if its not my taste this stuff probably ends up in the garbage, who wanna buy such china boards used?

imo currently im still biasing towards jlsounds usb + jlsounds ak4493 + ian canadas (3) opamp output stage, well for multiple reasons (no china boards is one of them and i may make something in the future myself, either replace the dac board or add a hardware resampler, maybe play with output stages etc)

also im really curious how a tube output stage on a dac would sound... i guess i could also use a tube buffer for similar effect

With 6 single OPAs I can make 3 DACs. And I have one (DDDAC) without even one OPA. And the ones I made with OPAs, I plan to remove that and put Tubes for I/V stage.
ah well true... well thats another point for ian canadas output stage if i wanna get started... just 3 opamps and still quiet cheap with 50 euro
 
From JLS, I2S over USB is excellent, and I heard that AK4993 and didn't like it. They got some better sound by changing the output analog filter, but I don't know the details. Even that does not sound at the level of these NOS DACs that I proposed. JLS also offers the AD1865R NOS DAC, which is probably a better option, but you don't have DSD. Just don't take power supplies with LM317/337, do with LT1963/3015 because you definitely get a better sound.
 
Regarding LT1963, I personally don't like them. Nor do I use negative regulators such as 3015. That's because its better if both rails + and - have the same electrical characteristics, such as matched output impedance versus frequency. Two positive regulators, oned used up-side-down, can produce a better result. Some pretty decent sounding regulators that can be used that are:
https://www.ebay.com/itm/394029533177?_trkparms=amclksrc=ITM&aid=777008&algo=PERSONAL.TOPIC&ao=1&asc=20230811123856&meid=94afbee509ce4afbaf08b3c05d801a2a&pid=101770&rk=1&rkt=1&itm=394029533177&pmt=1&noa=1&pg=4375194&algv=RecentlyViewedItemsV2&_trksid=p4375194.c101770.m146925&_trkparms=parentrq:a463c4d118e0ac6bae8fd944ffff8bfe|pageci:8ccb3eff-f1c7-11ee-8da9-224bffd943d3|iid:1|vlpname:vlp_homepage

Just use each regulator on the power supply board with its own dedicated transformer winding (no center-tapped windings). There is no shared ground on the regulator boards, which means the two regulators are isolated from each other (a good thing). Run two wires from each regulator to the load (i.e. don't run a shared ground wire instead or you get some power rail crosstalk from the ground wire impedance). For the negative rail, connect the regulator + lead to ground at the load board, and the - lead to the load rail.
 
Last edited:
how does the LT1083 compare against something like this lt3042 based one https://www.aliexpress.us/item/1005006222078438.html ?

tho i like the beefy 5A of the LT1083... and specsheet says 40uV noise... seems alright?

i was already looking for used parts i could use.... transformators and the case might be worth looking around for... tho i couldnt find a vintage device or case that fits the project so far but i keep looking...

where do you guys get aluminium cases? i just know modu in eu and the cases are kinda limited
 
Last edited:
Tho with these 5A regulators... the thing that would protect the transformator is the regulator right? so i would also need to have a 5A secondary winding on the transformator or use a fuse (or is the primary winding fuse protecting secondarys too?), im asking because i might use supercapacitors ... these would overload the regulators (or the fuse...) for few seconds or i might need to use a secondary regulator for the supercaps
 
Last edited:
I use fused IEC power inlets. The LT1083 regulators work okay as is for opamp rails, although shunts are a little better. For low voltage stuff, I use the LT1083 regulators to pre-regulate down to around 5v. Then secondary regulators near the loads. LDO secondary regulators may work best with a fairly low voltage drop and a fairly high current. In order to assure enough current through the regulator, sometimes a resistor to ground at the output can bias the regulator pass transistor current to where the sound is best. In other cases a resistor to ground doesn't help. Has to be determined experimentally. The resistor to ground can be fairly low resistance in some cases. I have gone as low as 10R for a 3.3v rail with a 500mA regulator. Regulator input voltage was about 4.7v in that case. The voltage drop across the regulator may have to be kept low when the current is high in order to limit regulator power dissipation.

Regarding the Aliexpress LT3042 regulators, they aren't great. The use a MOSFET output buffer and cheap ceramic caps (maybe X5R or worse). However, they can be used as primary regulators so long as the maximum input voltage is not exceeded. Also, it may help to replace the SET terminal cap with a tantalum cap. Class 2 or 3 ceramics can generate piezoelectric noise in that location, so maybe better to change those out.
 
Last edited:
LDO secondary regulators may work best with a fairly low voltage drop and a fairly high current. In order to assure enough current through the regulator, sometimes a resistor to ground at the output can bias the regulator pass transistor current to where the sound is best. In other cases a resistor to ground doesn't help. Has to be determined experimentally. The resistor to ground can be fairly low resistance in some cases. I have gone as low as 10R for a 3.3v rail with a 500mA regulator. Regulator input voltage was about 4.7v in that case. The voltage drop across the regulator may have to be kept low when the current is high in order to limit regulator power dissipation.
oh sweet trick, i have to test this :)

I use fused IEC power inlets.
but does just a main fuse work out? my thinking says the secondary circuit can still overload transformer secondary? specially if we take a fuse that protects multiple secondarys... it needs to be larger than the secondary max current, no?

Regarding the Aliexpress LT3042 regulators, they aren't great. The use a MOSFET output buffer and cheap ceramic caps (maybe X5R or worse). However, they can be used as primary regulators so long as the maximum input voltage is not exceeded. Also, it may help to replace the SET terminal cap with a tantalum cap. Class 2 or 3 ceramics can generate piezoelectric noise in that location, so maybe better to change those out.
Ah well i wont question you recommending china boards i guess, i will just go with it they are cheap :)

just looked more for transformers, i found 2x9V 100VA from a german manufacture, this ends up at around 11V dc, is alright for regulation down to 5V? or is there maybe generally a difference in sound quality if you regulate from a higher or lower voltage?
 
For most things you can use cheap Aliexpress R-cores (with no center-tapped windings). For low voltage stuff, 9v VAC RMS windings or lower are okay. You can go higher but it may just make your regulators run hotter. Some older regulators worked best with a few volts dropped across them, depends.

For higher end stuff you may want to use premium torroids with electrostatic and magnetic shielding. Maybe a core size or two above what you think you need.

With all these things you can do to tweak performance, many of them are less audible until you fix the biggest problems first. Once you get a pretty clean system, then the smaller level problems start to become more noticeable. Its like, now they are your biggest remaining problems.

If you read through my chronology of powering and driving Marcel's RTZ FIRDAC, you will see the order I recommended for that. Start with power and clean USB. The first example I give is what I called the worst possible. It isn't really. But I explain what's wrong with it, so a reader will hopefully get the right ideas. I suggest to go study that series of posts and pics, since I see no point it writing it all up again here. https://www.diyaudio.com/community/...t-register-firdac.379406/page-94#post-7457164
 
For higher end stuff you may want to use premium torroids with electrostatic and magnetic shielding. Maybe a core size or two above what you think you need.
yea i plan something like this... either with these: https://de.aliexpress.com/item/32631754519.html or the transformers get their own "appartment"

i got a good offer on these: https://www.it-tronics.de/ringkerntrafo-transformator-100va-230v-2x9v-18v-sedlbauer-rso-825025/
2 for 50 euro...

i think 4 seperated power supplys for digital side should be enough..
then i probably need either 2x 12V or 15V for analogue side something similar beefy..