After discussion with Jean-Paul, the creator of the Subbu DAC V3 (along with Subbu), it was decided that it would be better to have a separate thread to discuss tweaks/modifications from the "standard" BOM build. That is the purpose of this thread. I previously commented on my experience with alternate parts for an early prototype of the V3 DAC. I'll repeat those comments in the next few posts to kick things off.
---Gary
Additional comments added March 10, 2014
After considerable discussion in this thread, there is a consensus developing on a good set of mods and parts that could be the basis for a new BOM. If you don't feel like reading the entire thread, here is a short summary.
C17 - 1uf X7R MLCC Ceramic capacitor (many possible vendors)
C22 - 470uf 6.3v Panasonic SEPC 6SEPC470
C32 - 1uf Wima MKS2 with 2.5mm lead spacing - MKS0B041000F00KSSD
C35 - you should use either C32 or C35, but not both. If you want to experiment with larger values of Wima MKS2 that only come with 5mm lead spacing then use the C35 location instead of C32. You should not use a value >4.7uf or the sound gets worse.
C4 - 1uf Wima MKS2 with 5mm lead spacing - MKS2-1/50/10 or MKS2-1/63/5T or MKS2C041001F00KSSD
Further comments September 1, 2014
The use of 1uf X7R MLCC ceramic for C17 has proven to be controversial. Some people report better sound with the original BOM recommendation of 4.7uf tantalum. Others have recommended bypassing the 1uf X7R with 100nf X7R but androa17 has done good work looking at this and recommends the following to avoid oscillation.
Best: 10uf tantalum + 100nF x7r
Next: 4.7uf tantalum (the original BOM)
3rd best: 1uf X7R
JP has used 1uf X7R for C17 in all his recent builds and reports no problem. In all of my builds I have used 0.22uf COG caps and like the sound and haven't seen any oscillation. So the 1uf recommendation stands.
---Gary
Additional comments added March 10, 2014
After considerable discussion in this thread, there is a consensus developing on a good set of mods and parts that could be the basis for a new BOM. If you don't feel like reading the entire thread, here is a short summary.
C17 - 1uf X7R MLCC Ceramic capacitor (many possible vendors)
C22 - 470uf 6.3v Panasonic SEPC 6SEPC470
C32 - 1uf Wima MKS2 with 2.5mm lead spacing - MKS0B041000F00KSSD
C35 - you should use either C32 or C35, but not both. If you want to experiment with larger values of Wima MKS2 that only come with 5mm lead spacing then use the C35 location instead of C32. You should not use a value >4.7uf or the sound gets worse.
C4 - 1uf Wima MKS2 with 5mm lead spacing - MKS2-1/50/10 or MKS2-1/63/5T or MKS2C041001F00KSSD
Further comments September 1, 2014
The use of 1uf X7R MLCC ceramic for C17 has proven to be controversial. Some people report better sound with the original BOM recommendation of 4.7uf tantalum. Others have recommended bypassing the 1uf X7R with 100nf X7R but androa17 has done good work looking at this and recommends the following to avoid oscillation.
Best: 10uf tantalum + 100nF x7r
Next: 4.7uf tantalum (the original BOM)
3rd best: 1uf X7R
JP has used 1uf X7R for C17 in all his recent builds and reports no problem. In all of my builds I have used 0.22uf COG caps and like the sound and haven't seen any oscillation. So the 1uf recommendation stands.
Last edited:
This is awesome! Maybe we could also put next to each post the suitability of the respective mod for previous versions of the DAC as well. Since you have both V2.6 and V3 and have "backported" some fixes to V2.6, imho it would be worth it. Maybe others could chime in and talk about mods they did for even earlier versions of the DAC.
Great initiative!
Great initiative!
my experience
I should point out that all of these modifications are based on my ears only. When coming up with these tweaks, I had a version of the V3 DAC using JP's recommended BOM with the SAL-RPM capacitors as a reference and I was able to compare the sound quality compared to that reference as I changed parts. I should also note that the recommended BOM with the SAL-RPM capacitors sounds very good. If you build that you will not be disappointed.
I can't guarantee that your results will match mine. Your experience may be different and each of you does this at your own risk. With that out of the way, here are my findings on how to improve sound quality over the BOM.
Tier 1 - most important change from the BOM
Upgrade C22 - the electrolytic cap for Avcc of the ES9023. I really like polymer caps and I find that the sound got better as I increased the size of this capacitor. I ended up using a 470uf 16v Nichicon capacitor. I think it was Nichicon part #PLF1C471MDO1, Digikey 493-3019-ND. You can get away with lower voltage parts (6.3v or 10v). Panasonic and UCC also make good sounding polymer caps. People have noted that polymer caps like to be bypassed by slightly larger ceramic caps than the 1uf cap used in the V3 BOM. In all of my builds, I used the 1uf cap and it sounded good. In the future I may experiment with a 10uf ceramic cap bypass.
For best sound, I also recommend adding an output buffer such as the JG filter/buffer.
Tier 2 - next level of importance. C35 - the cap for the NEG terminal of the ES9023. I played around with different values of C35 and found that I got best sound with a value of 4.7uf. I like the Wima 4.7uf 50v polyester MKS2 caps. Mouser part # 505-MKS2-4.7/50/10.
I also found some benefit by increasing the size of C4 - the SPDIF input coupling cap. I used the BOM 0.1uf cap and put a 3.3uf 50uf Wima polyester in parallel. Mouser part # 505-MKS23.3/50/10. I mounted this from the bottom of the board since I was worried about fitting it on the top. In hindsight, it would have fit on the top of the board if I mounted the 0.1uf SMD cap first and then mounted the Wima on top of the 0.1uf cap.
Tier 3 - small benefit. Use better quality caps for the 10uf electrolytics for the WM8804 and ES9023 - C8, C13, and C21. Either use polymer electrolytics such as Nichicon PLF1E100MCL2 (Mouser # 647-PLF1E100MCL2) or Wima polyester. I used the Wima 10uf 50v caps (Mouser # 505-MKS210/50/20) but these take up more real estate and must be mounted from the bottom. Of these parts, the most important one is C21 - the 10uf capacitor for the ES9023.
Tier 4 - changes I made that I don't think were worth the trouble. I tried a SAW oscillator for the 50Mhz clock and I liked the sound of it but I'm not sure it's really that much better. Since it's a different pad out than the current clock, it was very difficult to solder this one in. I used Kyocera KC5032A50.0000CM0E00 which is available from Digikey.
I've attached pictures of the top and bottom of my build so you can see how things went together. It's not that neat with parts on both sides but it's quite mechanically stable and sounds very good.
---Gary
p.s. note that this is pretty much a repeat of an earlier post that I made here http://www.diyaudio.com/forums/group-buys/237612-subbu-dac-v3-es9023-wm8804-spdif-power-supply-pcb-group-buy-60.html#post3698767
I should point out that all of these modifications are based on my ears only. When coming up with these tweaks, I had a version of the V3 DAC using JP's recommended BOM with the SAL-RPM capacitors as a reference and I was able to compare the sound quality compared to that reference as I changed parts. I should also note that the recommended BOM with the SAL-RPM capacitors sounds very good. If you build that you will not be disappointed.
I can't guarantee that your results will match mine. Your experience may be different and each of you does this at your own risk. With that out of the way, here are my findings on how to improve sound quality over the BOM.
Tier 1 - most important change from the BOM
Upgrade C22 - the electrolytic cap for Avcc of the ES9023. I really like polymer caps and I find that the sound got better as I increased the size of this capacitor. I ended up using a 470uf 16v Nichicon capacitor. I think it was Nichicon part #PLF1C471MDO1, Digikey 493-3019-ND. You can get away with lower voltage parts (6.3v or 10v). Panasonic and UCC also make good sounding polymer caps. People have noted that polymer caps like to be bypassed by slightly larger ceramic caps than the 1uf cap used in the V3 BOM. In all of my builds, I used the 1uf cap and it sounded good. In the future I may experiment with a 10uf ceramic cap bypass.
For best sound, I also recommend adding an output buffer such as the JG filter/buffer.
Tier 2 - next level of importance. C35 - the cap for the NEG terminal of the ES9023. I played around with different values of C35 and found that I got best sound with a value of 4.7uf. I like the Wima 4.7uf 50v polyester MKS2 caps. Mouser part # 505-MKS2-4.7/50/10.
I also found some benefit by increasing the size of C4 - the SPDIF input coupling cap. I used the BOM 0.1uf cap and put a 3.3uf 50uf Wima polyester in parallel. Mouser part # 505-MKS23.3/50/10. I mounted this from the bottom of the board since I was worried about fitting it on the top. In hindsight, it would have fit on the top of the board if I mounted the 0.1uf SMD cap first and then mounted the Wima on top of the 0.1uf cap.
Tier 3 - small benefit. Use better quality caps for the 10uf electrolytics for the WM8804 and ES9023 - C8, C13, and C21. Either use polymer electrolytics such as Nichicon PLF1E100MCL2 (Mouser # 647-PLF1E100MCL2) or Wima polyester. I used the Wima 10uf 50v caps (Mouser # 505-MKS210/50/20) but these take up more real estate and must be mounted from the bottom. Of these parts, the most important one is C21 - the 10uf capacitor for the ES9023.
Tier 4 - changes I made that I don't think were worth the trouble. I tried a SAW oscillator for the 50Mhz clock and I liked the sound of it but I'm not sure it's really that much better. Since it's a different pad out than the current clock, it was very difficult to solder this one in. I used Kyocera KC5032A50.0000CM0E00 which is available from Digikey.
I've attached pictures of the top and bottom of my build so you can see how things went together. It's not that neat with parts on both sides but it's quite mechanically stable and sounds very good.
---Gary
p.s. note that this is pretty much a repeat of an earlier post that I made here http://www.diyaudio.com/forums/group-buys/237612-subbu-dac-v3-es9023-wm8804-spdif-power-supply-pcb-group-buy-60.html#post3698767
Attachments
Last edited:
retrofitting these changes to a Subbu V2.6 DAC
The V2.6 and V3 DACs are very similar so one can also retrofit some of these changes if one is willing to desolder some parts and replace them with new parts. I did that to my V2.6 DACs with good results. Note that the space of the holes or pads on the V3 DAC isn't always optimized for the parts that I am suggesting, so the build is not simple. Here are the changes that I made more or less in order of importance.
1) Avcc capacitor (C10B) - change from 100uf 16v to 470uf polymer cap. The spacing on the board is 2.5mm so one should pick polymer caps with this lead spacing. The Panasonic 470uf 6.3v part # 6SEPC470MW or Nichicon 470uf 6.3v part # RS80J471MDNA seem like good choices. I actually used the Nichicon PLF1C471MDO1 even though it didn't have the right lead spacing. Its leads are 5mm which means the capacitor leads need to be bent slightly to fit on the board.
2) NEG capacitor (C32) - use 4.7uf 50v Wima MKS2. Again there is a pin spacing problem. The board supports 2.5mm pin spacing while the Wima part has 5mm spacing. So one needs to bend the pins to fit.
3) SPDIF input capacitor (C1) - I put a 3.3uf 50v Wima MKS2 in parallel with the 0.1uf SMD ceramic cap on the board. There are no mounting holes for this capacitor - only the SMD pads for the ceramic 0.1uf cap. One needs to solder the 3.3uf Wima leads to these pads, which is a bit tricky. For those worried about their ability to pull this off, one could try using bigger SMD caps, but I haven't tried this. I just checked on Digikey and found a 3.3uf 10v X7R cap that would fit the pad pitch. It's Digikey # 445-7589-1-ND, TDK part # C2012X7R1A335K125AC. Note that I have not tried these ceramic caps. There are many other SMD parts that might work.
4) Vreg capacitor (C9) - change from conventional 10uf electrolytic to 10uf polymer electrolytic. This is another 2.5mm lead spacing part. I actually took some SMD 10uf polymer caps I had in my parts box and soldered wires to them and then soldered them to the board. It's hard to find 10uf polymer caps with 2.5mm lead spacing but there are parts with 3.5mm spacing. These can fit with a bit of bending of the leads. Parts like the 10uf 25v Nichicon PLF1E100MCL2 or the 10uf 25v Panasonic 25SEP10M+TSS seem like good choices.
That's where I stopped in upgrading the older V2.6 DACs.
---Gary
The V2.6 and V3 DACs are very similar so one can also retrofit some of these changes if one is willing to desolder some parts and replace them with new parts. I did that to my V2.6 DACs with good results. Note that the space of the holes or pads on the V3 DAC isn't always optimized for the parts that I am suggesting, so the build is not simple. Here are the changes that I made more or less in order of importance.
1) Avcc capacitor (C10B) - change from 100uf 16v to 470uf polymer cap. The spacing on the board is 2.5mm so one should pick polymer caps with this lead spacing. The Panasonic 470uf 6.3v part # 6SEPC470MW or Nichicon 470uf 6.3v part # RS80J471MDNA seem like good choices. I actually used the Nichicon PLF1C471MDO1 even though it didn't have the right lead spacing. Its leads are 5mm which means the capacitor leads need to be bent slightly to fit on the board.
2) NEG capacitor (C32) - use 4.7uf 50v Wima MKS2. Again there is a pin spacing problem. The board supports 2.5mm pin spacing while the Wima part has 5mm spacing. So one needs to bend the pins to fit.
3) SPDIF input capacitor (C1) - I put a 3.3uf 50v Wima MKS2 in parallel with the 0.1uf SMD ceramic cap on the board. There are no mounting holes for this capacitor - only the SMD pads for the ceramic 0.1uf cap. One needs to solder the 3.3uf Wima leads to these pads, which is a bit tricky. For those worried about their ability to pull this off, one could try using bigger SMD caps, but I haven't tried this. I just checked on Digikey and found a 3.3uf 10v X7R cap that would fit the pad pitch. It's Digikey # 445-7589-1-ND, TDK part # C2012X7R1A335K125AC. Note that I have not tried these ceramic caps. There are many other SMD parts that might work.
4) Vreg capacitor (C9) - change from conventional 10uf electrolytic to 10uf polymer electrolytic. This is another 2.5mm lead spacing part. I actually took some SMD 10uf polymer caps I had in my parts box and soldered wires to them and then soldered them to the board. It's hard to find 10uf polymer caps with 2.5mm lead spacing but there are parts with 3.5mm spacing. These can fit with a bit of bending of the leads. Parts like the 10uf 25v Nichicon PLF1E100MCL2 or the 10uf 25v Panasonic 25SEP10M+TSS seem like good choices.
That's where I stopped in upgrading the older V2.6 DACs.
---Gary
Thanks for sharing experience, Gary.
What sounds good is a subjective matter, and our ears sensitivity and personal preferences are different, too. I don't think there is any absolute opinion. I am going to apply some of the above to one of the 2.6 boards for some fun
Please come back once it's done with details and result. Will you be able to compare with a V3 or unmodified V2.6?
Best of luck with the mods!
As mentioned over in the "V3 build thread", for 75 ohm BNC jacks, I like the Amphenol 112432 jacks for mounting in DACs or digital sources. Any good 75 ohm coax cable and 75 ohm BNC will work fine for the cable. I use RG179 coax cable and whatever 75 ohm BNC connector looks good when it's time to build new cable. Crimp cables are better, but you need to buy a crimping tool which adds to the expense. Quite a few companies in the US sell premade 75 ohm BNC cables for very reasonable prices. One that I like is Markertek, which sells RG179 cables of various lengths with Trompeter 75 ohm BNCs already attached.
Regarding the power supply, I used one of the Subbu group buy 5v power supplies.
---Gary
My Mouser order is in, so I shall have some parts before Thanksgiving long weekend; however, V3 is likely to be a X'mas project due to work.
I did replace C10B with a Sanyo OS-CON SVP 330uF and C32 with a Visay 4.7uF polymer. My first impression was these 2 changes together enriched the mid. Is it better? It is too early to say, but it is definitely different.
Thank you for reply Gary
Amphenol 112432 Is to be enclosure mounted ?
Is there a limit for a coax length - can I use 9 m
do you hear / measure difference between ordinary cable with RCA and coax 75 ohm with BNC
Panasonic has a very good overview of polymer caps on their website. Take a look here: http://industrial.panasonic.com/www-data/pdf/AAB8000/AAB8000PE24.pdf
I have quoted one passage which seems relevant:
"The OS-CON is an electrolytic capacitor that has excellent frequency characteristics. It improves ESR greatly, and provides the excellent frequency characteristics because the OS-CON uses a high conductive polymer as electrolyte. Fig.A: The OS-CON’s frequency characteristic shows a nearly ideal curve. When compared at 100kHz, The OS-CON 56μF, and low impedance aluminum electrolytic capacitor 1,000μF nearly have the same feature. Fig.B: The resonance point of the OS-CON is at 100kHz to 10MHz. The ESR is an extremely small value approximately 5mΩ at 100kHz of 560μF."
---Gary
Attachments
Lucky guy, can't find those anymore.
I concur with GaryB that BNS 75 Ohm is the way to go but only if you use them at both sides. I use another variant which performs good and it is easy to mount as well. It is the Tyco/Greenpar 1-1478048-0 which is available through Farnell. I choose this one as it is also insulated and drilling just a hole is enough to mount them. It is also one of the more easily available here. I found that there are BNC 75 Ohm connectors that can only be mated several tens of times while this one is higher specced (500 times). It is rated "professional" FWIW. Bandwidth is 2 Ghz while the Amphenol is 4 GHz but I doubt if this will give benefits in this application.
When going BNC make sure you add a BNC 75 Ohm to the source as well. The worst combination is RCA to BNC cabling ! Better stay with RCA in that case.
And Gary ... you really should clean the flux from the PCB's
Last edited:
Don't know what you exactly mean but you can put a jumper wire for the input cap if your source has a cap in the output. I suppose you mean the CS8416 chip ?
My mouser arrives. I changed the big cap but the PS ones for OScon SEPC/SEPF and expensive top red organic from Panasonic when no exist at Sanyo.
When I look at the small smd caps and W8004, I really don't know if i will sucess, but doing myself is a part of the adventure...Flux FLUX...save me !
When I look at the small smd caps and W8004, I really don't know if i will sucess, but doing myself is a part of the adventure...Flux FLUX...save me !