Modifying the Subbu V3 DAC

[jean-paul]

I have another usb 9023 dac. After heavy modifying it sound almost same as subbu. I used all hints from subbu and get it there, so nothing strange to have similar sound.

Nothing strange but very few have it


I recall you having oscillating regs. Did you solve that ?

 

[Eldam]

Eldam - finally you get soft and try my version.
I sold my dac, so no experimenting from my side anymore...
I have another usb 9023 dac. After heavy modifying it sound almost same as subbu. I used all hints from subbu and get it there, so nothing strange to have similar sound.
Maybe one hint to try. Before 9023 AVCC mic regulator C19 for my ears best combination is 10uf x7r and 0,1uf x7r ceramics. Brilliant highs neutrall midrange and powerfull bass. I have tried 1uf ceramics, 1uf smr ala plastic cornell is the second best ( pronounced midrange, a little bit heavy highs, and lack of sub bas) , 22uf polymer tantalum +0,1uf x7r, 47uf SP + 0,1uf x7r
I have made comparison on hi-end system open baffle speakers Soulsonic, AA referenz preamp, MONO II....and my heavy modified linux player, who easily beat win 7- jriver, foobar.... More detailed, relaxed, fluid, analogue sounded... almost no zzz or ssss sibilance in comparison to windows player.
I do comparison subbu v3 with heavily modified Minimax es9018 ... (my favourite dac... i am also interested in R2R) and winner was Minimax. Low end is more powerfull, more dynamics, vocal have more energy, but suprise...resolution is almost identicall. You have immediately feeling that es9023 is really smaller brother from es9018.
But must be fair and realistic.... SUBBU dac v3 is really really good DAC for that price.

Hé man, I'm always soft... as soft a wookie can be ... No problem of ego I have, I let it for knight jedis... I'm always tweaking my Millenium Falcon

thanks for testimonies & sharing... I doesn't surprise me i like the plastic à la Cornell if it improves the midranges... My second version have the 10 uf X7R... I will try the 0,1 uf in addition hopping it makes not the reg oscillate !

In C32 : the 1 uF MKS2 2.5 mm pitch could be a solution for stronger bass in relation with what you wrote. But still prefer the tonal of Black Gate with C32 !

Finally did you try Nichicon PLE vs Sanyo Oscon SP in C22 ?

see you

 

[Eldam]

i remenber this chineese proverb: "Only on the Ocean the Wise like oscillations"

Do the micrels can ossicilate with 0.1 uf just before ? I liked the specials 1 uF plastic Cornell (who works like 10 x bigger tantalum but less noise..)...

Androa, what is it with the Micrels or others regs in your new pcb (I mean your 10 uf + 0.1 uF...hint)

All those diyers with scopes... I 'm just a beginner with some time and a lot of patience and a nostalgy for the Lego games !

 

[jean-paul]

As a beginner you really should stop changing parts without measuring with an oscilloscope. Changing parts by ear can be very deceiving. The Micrel regs easily oscillate with low ESR caps at their outputs. Any change in parts should be measured. You can't tune the DAC to a certain system but you can introduce oscillation

You are familiar with C17 issue. I used 1 µF X7R originally but then the reg oscillated so I went to 4.7 µF tantalum. I got persuaded by members that 1 µF X7R sounds better. In later builds this works OK but I guess the ESR value is on the edge. Still you can't go wrong with 4.7 µF tantalum after the regs.

So if one wants to play safe and has no oscilloscope the BOM is a safe haven. Never use too low ESR caps after the regs even if they are made from the worlds most exquisite materials. An oscilloscope is a very good tool to check if things work like intended.

 

[Eldam]

oh no, not this discussion again !

Your advices are always precious and I believe we share the same philosophy with different means. I'm more an enthusiast than you... but it doesn't mean I'm always wrong here. And some of my modest changes are better than the genuine BOM... some will say this is personal tastes, I will say try it as we have all different hifi systems and trade offs in it !

BTW your BOM gave choice and some were far better than others ! And this is not a critical. Few are able to make like Subbu and you a good sounding dac for this low price and I believe we all thank both of you for that. Here we share our enthusiasm about that and ok some are more pro with their means than others : e.g. Gary use a scope not me... but scope is just to check it...

I never hear yoy say you try my c21/c22/c32 combo and report... why ? I think some psychologic is involved here or goals I can't understand. For me it's just sharing : no problem if somebody say it's not good. Nobody said my tweaks were bad... the opposite in fact ! Finally when we share all together like you do initialy with the project, things are always better... some modest but very good improvement can exist... this is the famous Wheel that you talked about.


We don't talk once again about the BOM etc. You're right and I wrote it myself. Ok some errors can occur as the 1 Uf ceramic X7R which was a bad idea or maybe some subjective like C4 with values > 0.1 uF

But the global modest work of each here, Gary & others really improve the already good Subbu for better ! Few will hear it as lazy, or not having a good enough hifi system.

For the oscillation, Androa talked about the 0.1 before the Micrel not after ! Do the micrel oscillate with small values just before also ?

Ears are good enough for control when no scopes is owned, your' re right every tweaker and diyer should have one, but sometimes money is just missing or few personal project to Worth a buy...

SAL cap didn't work at c32 with me.

Hé where are you now with your new supply design, I will like to try it and trust in it ...

cheers

 

[GaryB]

You are familiar with C17 issue. I used 1 µF X7R originally but then the reg oscillated so I went to 4.7 µF tantalum. I got persuaded by members that 1 µF X7R sounds better. In later builds this works OK but I guess the ESR value is on the edge. Still you can't go wrong with 4.7 µF tantalum after the regs.

JP,
I think we've learned that 1uf X7R is definitely marginal for C17. Several people have built it that way and reported poorer sound compared to the BOM recommendation of a 4.7uf Ta cap. I looked back on the original thread and see that all my experiments were with 0.22uf COG caps and I never actually tried 1uf. Somehow that got translated into a 1uf recommendation but I think I need to modify that at the start of this thread. I just checked my DACs with a 100Mhz oscilloscope and there is no indication of oscillation at C17 but there is a little bit of very low level 50Mhz noise. So I may try the recommendation from androa76 to parallel 10uf Ta caps with 0.1uf X7R and see if there is any difference.
In case people are wondering, the voltage at C8 and C13 also looks clean with polymer caps or Wima film caps, at least in the DACs that I've built.

I never hear yoy say you try my c21/c22/c32 combo and report... why ? I think some psychologic is involved here or goals I can't understand. For me it's just sharing : no problem if somebody say it's not good. Nobody said my tweaks were bad... the opposite in fact ! Finally when we share all together like you do initialy with the project, things are always better... some modest but very good improvement can exist... this is the famous Wheel that you talked about.

Eldam,
I think for many people, the Subbu DAC is a finished project. That includes JP. Only obsessive people like you and me keep experimenting.
I did try some of your recommendations recently and here is a quick update.

I updated the power supply with your recommendation to get rid of C5 and change C8 to a Wima film cap and change C9 to a 47uf BG cap. This is an improvement over the BOM but still falls short of the LT1764a supply that is my reference.

I also experimented with more your cap recommendations in the DAC. I like the recommendation of Panasonic 25SEP10M for C8, C13, C21. I also recently tried Nichicon PLEOJ471MDO1 (6.3v) for C22. I use the 2.5mm 1uf Wima cap for C32, which is one of the best for that position. But overall I don't find this combination the best. I compared it to my reference, the DAC that I show at the very start of this thread that uses 10uf Wima MKS caps for C8, C13, C21, and Nichicon 470uf 16v FPCAP for C22. That one sounds fuller and more fleshed out. When I've got more time to experiment, I may try swapping out parts to see if I can get the Subbu DAC ala Eldam closer to my reference.

---Gary
p.s. you really should try out the JG buffer one of these days

 

[Eldam]

Thank you Gary for sharing and listening review.

I believe this is one of the thing of the last 60 years, certainly sooner in the USA : don't tweak your toys... buy brand new ones !... I have a problem as I growed with Lego and maid new things always with the same game ! I have to find a specialist to speak about that, even today I'm informatician and do always the same things with just 0 & 1 in a more minimalist approach ...

This is soon a finished project. Have a good news of the nice people of Mouser France at Brive la Gaillarde town : my missing BOM will not arrive in november but those days.... I'm going able to do the JG buffer with the pcb EUVL & Mark sent me in february or march iirc !

I will try your mods also. Just wanted testimoning above than the 1 uF MKS2 2.5 pitch is very better than the 5 mm pitch and defintly a good solution (question of "inteluctual" honnesty even if I prefer personnaly the Black Gate here... personal taste, music is also cultural so subjective...and the hifi System - my clear & very transparent aluminium tweeters ). My best is with OSCON SP at C22 & BG std (as no BG NX) at C32... I believe we enter in the limit of our choice here because our respective systems... But yes, C21 with SEP caps is defintly a non come back hint...

Well I enjoy a lot this DAC and hurry to add the buffer... and beginn soon the AYA2 2014 and the Distinction-1541 .... If only Ian could shipp us quicker the I2StoPCM and the others devices ....

It could be nice for the others t see a Subbu batch IV for the others with the tips here.

See you...

 

[jean-paul]

JP, I think we've learned that 1uf X7R is definitely marginal for C17. Several people have built it that way and reported poorer sound compared to the BOM recommendation of a 4.7uf Ta cap. I looked back on the original thread and see that all my experiments were with 0.22uf COG caps and I never actually tried 1uf.

You do realize that I changed C17 to 1 µF X7R on all my DACs ? It shows that my own testing and the resulting BOM is a good guideline.

... the 1 uF MKS2 2.5 pitch is better than the 5 mm pitch and definitely a good solution...

Ha, got you !

It could be nice for the others to see a Subbu V4 for the others with the tips here.

Many requests but we defined the rules and those mean one has to do something. Apparently this is not what people like as things must be easy. We have done quite some work till now and feel people that want this PCB should participate by giving some energy. Only a group effort can make it happen.

Although I thought the V3 would be the end I will do a V4 with SPDIF transformer and the bells and whistles regarding layout only if we just hear a number and one or more members being distributors of the boards.

 

[Eldam]

If I can help, you ask me but in France you have already Korben69 (Phil.)

I'm going to make 3 DACS... all TDA1541 with two AYA2 for my prefered parts sport .... one reference with Audial BOM, the other to tweak according my system ( my snail simple pleasure, I will keep it for me this time). The third, a Distinction-1541 board printed by Ryan fellow with the shared hints of DIYAUDIO and famous pro " sharing diyers" (John, Pedja, Thorsten...) will be with tubes... pray for me ! There will be this time a box for the safe of family !

I'm asking to myself if the spidf is not dead (pray for it) ? Ok I have SqueezeBox Duet and like it... but no USB, nore separated input line à la I2S with or not L/R separated (simultanous mode = no multiplexing mode) than most of old multibitDAC allow (hé I found a 80 JVC cd player in a garbadge with a 16 bits Sanyo dac chip... it works... like one day the Luxman D105U I found also in a similar way ! Hahaha my first and only cd player with tubes was for zero bucks !
0s CD
Well a choice could be for the V4 to have a unique BOM as some spent their year for the better here and only give an alternativ BOM for the input : I2S with SOTA connector and Layout or USB/SPIDF !

It's not a problem anymore to swap the squeezebox but keep the shareware environment on a cheap device à la Cubiq, RasberryPi...

Is is not a shame that some DIY PowerDAC give only the spidf choice !

Do you enjoy a DIY of shelves or a brand of shelves with yours ?

 

[Eldam]

Hi,

This project is finished, but for the one who have not a JG buffer pcb and want one as a tiny pcb is sounding better than the equivalent point to point on a verroboard, I saw Joro fellow sell a JG buffer on JLsounds shop.

@ Garry: Hi I spent a little time again on caps on the subbu without the buffer as my others dac projects are stopped because parts are missings. It was just in a spirit of curiosity, not to improve, we finished it afterall ! I'm just looking a good PS for the buffer but unluckily sourcing in USA cost me too much (30 euros minimum taxes for the first dollar)!

According to me & on my system : for C22 : LE caps has the best medium-treble but as it lacks of bass, it lacks of flesh ! That's why I found here the Sanyo SP OScon is the best trade off (I use 800uf/4 V). The United Chemicon has maybe the best bass (better detail, higher low end, more energy and subjective higher volume: bass are highlighted), even better than the Sanyo SP, but the price to pay on my system is a harsch treble and medium with a too much damped mid-bass (like a lack of harmonics : e.g. Bass strings ). The sound with the apsa United Chemicon in the medium treble is parodic on my system, that's why it is not on my 3 first choices. At the opposite the FP caps (RR8 serie tested) is nice to hear, a little "bad tuby spirit" with its flesh" but here as the LE : bass is missing. When I benchmark the both LE and FP, I prefer the LE if a lacke o bass is not a problem... But yes bass register of the APSA united chemicon is addictive, maybe in another speaker it could give good result also !

For the Wolfson, I don't find better alternative than the SEP. SEPC is more transparent but give an agressive sound and I surmise compensation elsewhere with others caps around the dac chips will supress this transparency. The MKS2 : I just try 3.3 uf /5 mm pitch is better : rounder but too damped and lack of finess in the highs in relation to the SEP. I find the FC/FM aluminium a better alternative than the MKS2 3.3 on my system.
The 30 uf FP cap ( you like on C22 with 470 uf) was tested for its special rounder and fleshy sound on the Wolfson : it add flesh, but tonal balance is not good with the rest of my set up.
Well we touch here to the difference on our rrespective systems and are terrorists of sound ! I will say to you when I will find a good PS for the JG buffer... want to try a little noisy non feed-back emitter-follower followed by 10 k Uf to have just a linear supply feed by a constant voltage and maybe just 1 uf MKS2 2.5 pitch for the local decoupling on the JG pcb but certainly try two nimh cells with a simple 9V... my understanding it sounds better with 15V ! cheers

 

[jean-paul]

Well a choice could be for the V4 to have a unique BOM as some spent their year for the better here ...

This sounds somewhat contradictory to me. We had a unique BOM with some alternative paths. It was not us that wished for tens of pages of colorful cap talk

 

[Eldam]

This sounds somewhat contradictory to me. We had a unique BOM with some alternative paths. It was not us that wished for tens of pages of colorful cap talk

This sounds somewhat contradictory to as a unique BOM with some alternative parts is not a unique BOM .... It gives you BOM factor n where n is the number of alternative parts-choice presuming each give a different sounds.

BTW in your non-unique unique BOM some choices are very better than others without a doubt and are talked here and all the alternative BOM goes farer for the better than the original BOM if system is transparent ! This is in that spirit than you can quote my wrotes : to do a good unique BOM. But it don't withdraw any of your work with Subbu: 95% of the result. But the last 5% can allow a device to go very further. You will find the same if you investigated more time like I did. Your talent is to have maid the 95%. My patience 5% with the work of everybody in this thread : no contraditory this time... just a hobby. In any case my little alternative BOM is bad, it is better what you say or not ! I 'm sure you are happy of the honor we maid to you and subbu around our sharing..... Haha but I surmise even 5% could be too much, no ?

Well you can use my BOM or not for V4, don't care and the best conf are with two caps you have but not sourcable anymore. I just write in a spirit of sharing. The SAL seems not sourcable any moreif you want to do a v4, I have no sucess in C32 with the big axial SAL 3.3 uf like I already said !

For example for the few who like to test (I m sure there are no more than fingers on one hand and that's a shame as you dac is very good and has a good margin of amelioration without the say good JG buffer, in the spirit of a cheap but very good DAC) : for the Wolfson if you like my conf or if you like the lake of bass of the Nichicon LE or FP serie : replace the SEP around the Wolfson by two KZ Muse 20 uf/100 V. You wil gain a lively sound with a full mid bass and strong bass without the trebles was delayed, this last staying non agressive and better than the MKS2 with longer notes (harmonics) with the drums. Especially here complex instrument like saxophone has this time this mix of warm & acidity which is difficult to do for every dac : listen to "Time to Time, Ben Webster on the Webster & Associates album !

If the JG buffer is really better you should just put it on the pcb like our other fellow friend which maid a xlose to you but with I2S... I believe if the spidf is maybe a default in relation to some usb stuffs, I find it cool for everybody as many have also cd-reader or stuffs like SB.

If I can help you in a modest maneer don't hesitate as I already said... you maid a very nice bidule with Subbu and I listen it everyday (and I have also dacs to be cooked with others dac chips ! So I surmise it's a compliment)

 

[PJN]

I have both a Subbu and a curryman. I tried the curryman with I2S input via a ministreamer, and eventually picked up a twisted pear board to be able to get PCM input because it fit the equipment I was using it with. I could not tell the difference between the ministreamer I2S input and the PCM input, even on a fairly hi res system. I thought that the Subbu without the JG buffer sounded a bit more transparent than the curryman with the buffer. Other than that they sound very similar. But I suppose all well designed DACs should sound similar if they do their job well.

PJN

 

[Eldam]

. But I suppose all well designed DACs should sound similar if they do their job well.
PJN

If it could be true .

Do you try some hints from this thread and heard any difference ? What is the impedance input of your pre/amp ?

What I find funny and interresting with this dac chip is the lack of external outputstage so it is a fantastic tool to try variation of the PS stages and firstly the compensation between caps to know what the limit if the chip is... but assume some say a good layout and shematic is proof about the choices of caps..
According to me that confirms each can with a little patience and work- if the base is good like here- to tailor the sound in relation to his system. But also some global qualities can go further : transparency can be improved even if it is not according to me a major quality in a source, more" flesh" also but here its a personal taste and it is just a question of tonal balance (in relation to the rest of the system).

My own experience about this DAC is I started from a good DAC and arrived with maybe a twice better DAC measuring the pleasure I had before and after the mods. Like a F1 the engineer maid a good job but each road need a little set up from the pilot to adapt to the race (sytem & ears of the owner !)

Unluckily, two caps of my personal alternative BOM are not sourcable for everybody : two black gates (especialy a 1 uf in C32) and a Sanyo SP OSCON for C22.

But the 10 uF Panasonic SEP for each C21 and for the WOlfson receiver chip + C32 with 1 uF MKS2 2.5 mm pitch (one of the alternative of the genuine BOM)+ C22 Panasonic SEP 470 uF - better for my than the SEPC here - is a good choice.

If with C22 a Nichicon FP cap or a LE serie ( I prefer myself the last) is choosed, two Nichicon Muse KZ 20 uf for the Wolfson will add the energy missed in the low range (here C21 keep the 10 uf SEP).

Maybe for the ones with DSPs, all those mods atre certainly unusefull but maybe a gain in transparency from the source !

Note I don't solder the caps I talk about here but use high quality copper + berylium contact pipe when caps legs are very thin (like MKS2 2.5 mm pitch e.g.) when legs are floating in the vias for a better electric contact ! So I can replace the caps without soldering and de soldering which coud destroy the pcb & of course which is the sentence of the day... (but as I talk maybe for less 5 personns so....)

 

[Eldam]

My God....listened to Abbey Road naked version with the two KZ muse for the Wolfson chip.... really near a very good turnable ! Le "je ne sais quoi en moins" but not so far !

I mean the 10 uf polymer SEP are better here but with the KZ muse it's no more about audiophile détails... it becomes a lifely sound on my system which remind me the best turnable I heard !

 

[jean-paul]

This sounds somewhat contradictory to as a unique BOM with some alternative parts is not a unique BOM .... It gives you BOM factor n where n is the number of alternative parts-choice presuming each give a different sounds.

Exaggerated. Please be happy we gave some alternatives as some were hard to source. It were other people that made versions of the BOM. I can only speak for myself but I don't think of different sounds or "tuning" to a certain system by using other parts. I just want to use good industrial parts and have a good performing product that performs good on any amplifier. I guess discussion is useless and I think at least a part of my point has been proven.

My own experience about this DAC is I started from a good DAC and arrived with maybe a twice better DAC measuring the pleasure I had before and after the mods.

"Twice as good" also seems exaggerated. It could be that you enjoy fiddling around with DACs but I am quite confident the DAC won't measure twice as good when parts are changed.

 

[GaryB]

"Twice as good" seems exaggerated.

I agree. No more than 1.9x better .
---Gary

 

[jean-paul]

I thought that the Subbu without the JG buffer sounded a bit more transparent than the curryman with the buffer.

Thanks for reporting, I heard that before but thought it would be a lone case. Personally I never bothered to use the JG buffer (made some attempts to obtain PCBs though) but I did build a Curryman DAC which is still unfinished.

 

[Eldam]

Exaggerated. Please be happy we gave some alternatives as some were hard to source. It were other people that made versions of the BOM. I can only speak for myself but I don't think of different sounds or "tuning" to a certain system by using other parts. I just want to use good industrial parts and have a good performing product that performs good on any amplifier. I guess discussion is useless and I think at least a part of my point has been proven.



"Twice as good" also seems exaggerated. It could be that you enjoy fiddling around with DACs but I am quite confident the DAC won't measure twice as good when parts are changed.

Sorry.... maybe just 1.95 x !
What I wrote is never a critiq and I'm happy with the DAC... so just try to do the same with the mods. You don't trust about it ! Just try... Do you know this story about the man who talk to God and ask him why he never had the numbers at Loto gambling ? One day God respond to him : "But play !"

Please just try, caps interaction is compensation : ok we know all about it, but like food some mix are better... I just write hopping the one I found can help !

But you just can not say it doesn't without tried it ! If you don't trust to its, just don't break the effort of the others. I repeat it : take it like an honnor to your work and never as a critiq like you seem often take it . It is just noy in my spirit. I'm a french, I only can make a good cook if the basics are SOTA : it's impossible to make a "Poule au Pot" with a bad chicken, bad wine, and bad vegetables ! ahha : our role is minimum in relation to yours ! So just try and take it like a turn of the Demning wheel.

But I surmise you understand already this and just joke about it.... some provocations, no ?

Try it if curious and report it even if bad... never did ! Hé you stay my heroes with Subbu to maid this DAC and my contribution is just cooking, but eating is art, no ?

Maybe i'm wrong, maybe it's just my system, try and report : I often listen to your advices.... and never write as to critism !

Myself i'm verry surprised how some caps swaps can change the things for the best or the worst. I really think with my long experience with its game that esr, inductance, etc are not a sufisant data too choose. Because interactions atre more complex. It's not about magic, it's just we don't master all the data and sometimes : experience and tries and errors can help... a little also !

So what to say : changes are a minimum important and I report or so subtle and Im wrong and borring about the changes? In my system, if I believe I am in the first position I report. Gary often try and don't share each time my opinion. So I believe it can be also a setup in relation to the system or some system which highlights some band frequencies in relation to the room... So only a try can say to you. Finally you can just have a critiq if you try yourself.

I would like you say to me if you find it bad or not.

And I don't try the JG buffer yet hé this time I have the caps but not a good PS yet for it !

I'm a turtle..... : http://www.youtube.com/watch?v=Lq8OlkzmX6k

 

[PJN]

I built my version with some of Gary B's recommended tweeks, using it with an Akido tube pre and a Alph J class A amp. Can't remember the pre's impedence right now, but it is a great combination. Very revealing.