• The Vendor's Bazaar forum is for commercial offers and transactions. Only unmoderated members can post here.

    diyAudio provides this forum for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members. Use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

Modifying the Subbu V3 DAC

I wonder if a coil with lower DC resistance would make a difference. The DC resistance of the BOM L1 is only 10mOhm. That should not limit current too much. For the LT1764, adding the coil before the regulator improved the sound and did not degrade the bass, so it seems somewhat unlikely that the coil is responsible for the difference. I do agree it is a surprising result.

---Gary


Since I don't use partnumbers or a BOM in my head I thought you were referring to the coils on the DAC board....sorry.

The common mode coil is not to blame I think. On the contrary, it is very useful.

I was talking here about a private discussion between GARY and I about the powersupply of our squeezeBox.

Eldam,
I checked my PM and we never had a private discussion about squeezebox power supplies. Perhaps you are thinking about someone else? We did chat about modifying the Duet in another thread.
---Gary

Ha ha, Eldam you're doing it again! I try to keep up but you are a moving target.
 
Last edited:
Last edited:
Answer to first question: theoretically C0G should be better but try to find them in this size and value. We need 1 µF or more.

Answer to second question: I don't know PEN but I tried PPS and liked them. Too bad they're brittle and break down quickly. Went over to X7R and never looked back.


My question to you: try to find C0G/PEN in 1 µF and try it out and report back what the results were.

Second question to you: what is the purpose to change literally everything on an already tested and, after that, modded design ?
 
Thanks,

1) : just find 0,68 smd COG (so your answer is important)

2) : It's a modyfing thread not really a building thread here; some good improvements were found as you said yourself ; you maid 20 turns of wheel and 3 versions so I suppose the wheel can continue a liitle if any V4.

2 bis) I think it's also a good DAC for didactic. It's surely not to change everything. We saw the BOM work and I highligthed this last point many time as I highligthed the positive changes.

3 answers to 2 questions, because sometimes in a simple question there are more than one... that's why I like often write more than one question.

I'm able to understand they are boring for you because the gap of knowledge between us.

Anyway sometimes, really read me above, it's a better thing to look for improvements on a narrow path. Here as it's a good DAC the path is going more and more narrow, and it's difficult. but we can maybe always learn something; Here it's not an answer it's what I think of discussions between enthusiasts.
 
Hi Eldam, it is not boring but if we would have thought like you do you would not have a good sounding DAC now and we all would have ended up with vapourware. A narrow path indeed ! Still I did try out a lot of stuff but please continue to improve it if you like.

Secondly I don't understand the reasoning behind the way you ask questions so communication is difficult (for me at least). To me it sounds like noise, which does not take away you probably are a nice chap 🙂

Regarding the (0.68 is not greater than 1.....) C0G: try it out, I won't try it as I am quite satisfied with the V3 in its current form with the later added mods. And "satisfied" means I will put my time in a new design, not necessarily being a DAC. V3 is mature and ready for continuous use 😀

For me ready = ready. But never say never, when useful stuff comes up I maybe will try it out. Just go on and maybe you will find the Holy Grail.

http://24.media.tumblr.com/tumblr_m28eb0gfKd1qjonawo1_500.gif
 
Last edited:
Hé always the first J-P !

I can not really understand why some questions will be more important than others as we saw your Ready was not (you know your Demning wheel). One thing is sure, my questions seems not good enough and I can understand it as I said before.

The reasonning behind once again is the humble question : why this particular point can not be improved ? It's ready but improvements exist (subjcet of this thread).

I just don't understand the critics about my questions... it's like if you answer each time : "do we made a not good enough DAC". No of course, it is. It's not about you and not about me. I'm happy with it, you can be prood of the work with Subbu... But you know we are all a little fanatic here. Hé men after finding the Graal, you want let it shine, but it's always a Graal, you know the story: here I try to follow your MP reasonment - Monthy Python reasonment! 😕:Pinoc:

What else reasonment can we find with a simple question like : "why a pen smd cap would not be better here ?"

Think I'm honnest, most of time with my question I explain the reason of it what you translate with : "too much questions".🙁


Ok I stop the questions, i will write only the good finding in this" the Subbu V3 ready moddfyied "😉

The smilies are very helpfull...If I'm good chap I surmise you are the Knight on the attachement movie.

What can i say more... sorry for shaking the coconuts... yes it makes noise !😀
 
Last edited:
Hi Gary, just one more question before I place my order for my final version. How critical is C21? I think I remember you saying at one point you found little difference in polymer and sal-rpm. Thanks, Gary.

Freeman,
What's the decision? I think SAL-RPM or polymer or Wima polyester all sound good. As I pointed out in post #3 in this thread, this is a "Tier 3" change, so important but not that important.

---Gary
 
Freeman,
What's the decision? I think SAL-RPM or polymer or Wima polyester all sound good. As I pointed out in post #3 in this thread, this is a "Tier 3" change, so important but not that important.

---Gary

Yes, but I wanted to make sure I'm up-to-date on latest thinking before I spend my scarce money. I ordered some polymers. Will report back on perception in a few days. Thanks for all your help, man. Been a great learning experience for a newbie.
 
The best capacitors for digital decoupling close to the pins (ie the small value caps 0.1uf etc) are X7Rs. The one exception to this is decoupling an oscillator, here the ultimate clock signal fidelity is achieved buy using a very low value GOG (0.001uf)(and thus small case) next to the pins, with X7R decades (0.001, 0.1uf) further out. The use of X7Rs allows the use of the smallest possible case, this is critical at these frequencies' as it reduces the parasitic inductance, also when used together in decades (2 or 3 at 10X value from lowest to highest) the lossy nature of the X7R dielectric minimises the resonant peaks.

Some info:
http://www.avx.com/docs/techinfo/mlcbypas.pdf

Here is how I would decouple a 50MHz osc, with 3 caps next to power pin, a very low value COG, and two larger X7Rs, the smallest case size is 0603 which in reality is rather large for decoupling, 0402s and 0201s are much better, but far to small for DIY work unfortunately.
🙂
 

Attachments

  • ESS_Osc.png
    ESS_Osc.png
    35.5 KB · Views: 371
The only difference between my two V3 DACs is the cap used for C3, C18. Nichicon FP in one and UCC NPCAP in the other. So I'm set up to do a direct comparison of the difference between these 2 caps. Both DACs have all the other changes that I described in my earlier note. So this morning I swapped DACs and couldn't hear much if any difference. The Nichicon FP version sounds very good as did the UCC NPCAP version.

I spent a bit more time listening and I think I need my to update my comments. I now agree with Eldam (and the original BOM) and can say that the UCC NPCAPs as specified in the BOM are better for C3 and C18 than the Nichicon FP caps that I used in one of my builds. So I'll replace the FP caps in the one DAC and then start experimenting with FPcap vs. SEPC for C22.
---Gary
 
What is the current consumption of the V3 DAC(I have mine in a box so I can´t messaure it).
I am going o try it with a BiB shunt psu.
I think the dac consumes around 75ma total.
If you follow the Salas rule of thumb, add at least 100ma on the top.
For all my salas reg, they are ranged from 225ma to 280ma. Since they different from each other, I can't tell whether more current sounds better or not.
 
That looks excellent, and will be beneficial to the clock integrity IMO.
I spent four years doing some comms. equipment where we needed a very stable and jitter free clock, the engineer in charge of the radios spent a few weeks looking at clock and stability issues and found, as most are aware on this site that; clean power and the correct decoupling were paramount, also the smallest possible trace length from clock to chip input.
 
Hi Marce

Thank you for your input 😉
According to your post explanations, is my suggestion correct ?

Regards

Phil

Phil, Marce,

(I use myself a military COG 1 uf, 5 mm between pads which is too big for the value and the size if I follow the discussion.)

I read the doc of Marce, then others about digital decoupling for my own culture (not to critize the DAC itself). i have please a question in relation to the mods with the 1 uf MLCC X7R of this thread then your picture above.

As the crystal has its own powersupply from a tantalum cap filtered by a ferrite bead : would it not be better just to put on C17 a ceramic which self resonance frequency is above the 50 Mhz of the crystal. So 10 nf max (with a 1012 smd size SFR is 53 mhz for X7R and 52 mhz with NPO (COG) ?

About the aging of the value with time (lake of capacitance): only the NPO seems good, here X7R or X8R are bad. But I read elsewhere than the MLCC X7R have no piezo electric effect or less cracking with time ?

What do we have to think ? can we heard the difference or is it just the last outrages to drosophils ?

After reading your doc Marce, it seems clear it could be better put the lower size a man can solder near the crystal. but the temperature is hard to control for the security of the cap : I read no more than 160° for such a little ceramic cap no to destroy its characteristic. too much solder can get off the result of the wanted minimized size of the cap as well.

Is there a compromise for an enthusiast which use solder ? the ratio: low size:high frequency of the X7R 0603 smd size ? Or NPO is still better ?

What do you think of the silicium IPDIA caps for this use, seems to habe best characteristic than NPO and less fragile for soldering (enthusiasts) ? IPDiA Silicon Capacitors | Mouser

thanks
 
I spent a bit more time listening and I think I need my to update my comments. I now agree with Eldam (and the original BOM) and can say that the UCC NPCAPs as specified in the BOM are better for C3 and C18 than the Nichicon FP caps that I used in one of my builds. So I'll replace the FP caps in the one DAC and then start experimenting with FPcap vs. SEPC for C22.
---Gary

For the integrity of the reading for other people, my tests were maid with the genuine Powersupply of the BOM. I don't know if it's important here to compare apples with apples.

The C36 decoupling cap of C22 was always the BOM one. As it has an important role too as you know.

Here my test was maid with the more expensive 390 FP cap with the lower ESR < 10 m ohms). Which is good in theory. I don't know about the ESR of the C36 decoupling cap. JP said the 1 uf value is important here. I was thinking of NPO O.68 uf with the FP cap, because O.68 is the lower value I found with NPO smd caps. My though was to move the resonance interraction : just a theory, I have no scope and few knowledge and mix up sometimes with theory). But as I said before I give up here as it seems non important details.

But I heard some litlle harchness with dynamics attack in mid treble which cause very non comfortable pic of ringing energy for the ears = distorsion (female voice attack, dynamic attack of saxophone in mid treble...). Strangly with its higher ESR, 470 SEPC was better, harchness (which I believe is a resonance ringing in the interaction between caps) dissapeared.
 
Last edited: