Leo,
Maybe I can help - short extract of relevant posts, schematics from that thread in word doc attached
Thank you, that makes things much easier
BTW guys, my advice regarding the output stage of Buffalo32s, no matter what regs you use , try running its supply for left and right separately
Running temporary until I get Pauls regs.
For each Left and right channel I currently use 1 x E/I core centre tapped transformer feeding rectifier (UF4002 diodes) - 3300uf 35v Panny FM - Burson reg - Buffalo32s
This was better than using single toroidal transformer with dual secondary supplying 1 x +/- main psu board feeding 2 x +/-v Burson regs
Running temporary until I get Pauls regs.
For each Left and right channel I currently use 1 x E/I core centre tapped transformer feeding rectifier (UF4002 diodes) - 3300uf 35v Panny FM - Burson reg - Buffalo32s
This was better than using single toroidal transformer with dual secondary supplying 1 x +/- main psu board feeding 2 x +/-v Burson regs
brgds, I initially wanted to use an external supply at the positions where you have put the two LT1086 but in a thread on the TP forum Russ commented on the possible benefits of this by saying that:
"None that I know of.
I just added the capability so I could adjust the voltage level itself. By increasing to say 3.5V you get slightly better DNR and more output swing. By lowering it you get lower output etc. You could do that very easily with a voltage divider and say 5V from some place . The ref line has a RC filter prior to the reg. That will still be in play even if you use an external ref".This kind of turned me off (maybe wrongly) as I don't need more "output swing" and as the DNR is already quite excellent. Did you find any improvement of this mod?
Why are you feeding the analog rails on the PCB traces? Am I missing something?
Thanks,
Nic
Nice job there, what BOM did you use for the JSR's ? I presume these are the ALW pcb's?
brgds, I initially wanted to use an external supply at the positions where you have put the two LT1086 but in a thread on the TP forum Russ commented on the possible benefits of this by saying that:
"None that I know of.
This kind of turned me off (maybe wrongly) as I don't need more "output swing" and as the DNR is already quite excellent. Did you find any improvement of this mod?
Why are you feeding the analog rails on the PCB traces? Am I missing something?
Thanks,
Nic
Nic,
to put any supply for 3.3V was a necessity - after removing of opamps - the last one LME49710 - buffer - is the one that supplies AVCC dynamiclly being driven by current signal level - AFAI understand. Unfortunately, TP guys weren't so kind to publish the scheme, moreover, they claimed that it's very hard to omit opamps and they wont support such ideas - the product itself is an excellent one and that's all. Why did I removed opamps? They simply sounded bad. AFAIK there is no other way than take the signal from PCB traces (or first opamp nodes after removing them) to feed trafos.
regards,
Andy
Then I guess I like the sound of op-amps! - or maybe the sound of op-amps improve with better power supply
But then again I already liked the B32S in its stock version (with dual supplies for the analog rails).
Thanks,
Nic
The ESS9018 is pretty expensive for a dac chip
Also those are not ceramic caps used in the output stage, they are SMD PPS
Heres proof I DO have a Buffalo32s
An externally hosted image should be here but it was not working when we last tested it.
I also have original Bufallo which I modified adding separate Paul Hynes
shunt and series regs, I intend on rigging the Buffalo32s with the same type regs sometime soon
I was hoping these would be the final dac I buy (I've already spent far too much on various others) and tried a very wide selection of ready built commercial and diy types
Anyway what is the rough price for your pcbs? I may be tempted to build your dac and give it a non biased comparison against the Buffalo32s at some stage
Be aware though I'm honest, if I feel the dac sounds poor I will say so
BTW IMHO the Buffalo32s is sonically superior to the Buffalo24
See.... that's my point. You have a Buffalo, and consider to tweak it. If the Buffalo really was that good, what is the idea in tweaking it?? Didn't the designer spend years and yeard tweaking it
That is the main difference between a mainstream DAC (Even though 32bit) and a DAC like the one we made. 4½ years ago, we had the first version done, at a level of engineering/tweaking like the Buffalo. We could have just finished at that time, like 99,9999% of all designer would do. But we decided to spend 4½ years to ensure, that anything was optimized to the best level. It't kind of like Microsoft launching new OS.... Always much better than the previous. But still it takes about 4-5 years of updating before you have something that really works.
Me too!!! We asked another guy from Denmark in here, but for some reason, he did not want to compare it to his Buffalo... Wonder if he really has got the Buffalo
Anecdoatally, the Sabre seems to respond well to separate & clean PS for each part of the DAC (some parts are more sensitive than others) & it would seem that the Buffalo board does not accommodate this too well - I think Neo's board has all the supplies separated? - maybe he can say something about this aspect of his design. Gordon Rankin with his Crimson DAC incorporating the Sabre also goes this route with something like 8 regulated PS to the DAC. For instance, I have been told by a very knowledgeable digital audio member that it's also important to keep the supplies DVCC_top separate from DVCC_bottom for clock purity reasons - something that may be impossible to do with the Buffalo configuration.
I figured when you add up the cost of good regulated supplies for these 8 (or more supplies) it gets costly (just price a PH reg) & begins to make it a complex build. I cut through this cost & complexity by going for High current delivery 3.3V batteries for individual supplies & 1.2V for core supplies - cuts down al lot of the cost & complexity & the batteries are 4mohm impedance. This was not my idea but a member here. This member also uses a transformer output stage rather than any active device (as does Gordon Rankin)
I also decided to use a prototype board as a tweakers platform, which even though it may not be optimal, gives the ability to experiment. Of course this is all vapour at the moment - I haven't built it yet - it's mounted on the protoboard & I have the 3.3V batteries - need to source some 1.2V batteries, connectors, etc
So I agree with Hurtig et al but maybe not in their tone or all their ideas. These are my thoughts on the question posed - hope it isn't too contentious - (I am now going to put on my flak jacket & await the strafing)
Edit: I just see Hurtig has posted while I was composing & I agree with him - if he had said this as his first post then maybe more would agree. I know a couple of Buffalo owenrs and to get it to sound good they have been through an ivy O/P stage, a counterpoint I/V stage, a couple of PS boards, etc - the costs have added up & this ends up no cheap DIY option.
I figured when you add up the cost of good regulated supplies for these 8 (or more supplies) it gets costly (just price a PH reg) & begins to make it a complex build. I cut through this cost & complexity by going for High current delivery 3.3V batteries for individual supplies & 1.2V for core supplies - cuts down al lot of the cost & complexity & the batteries are 4mohm impedance. This was not my idea but a member here. This member also uses a transformer output stage rather than any active device (as does Gordon Rankin)
I also decided to use a prototype board as a tweakers platform, which even though it may not be optimal, gives the ability to experiment. Of course this is all vapour at the moment - I haven't built it yet - it's mounted on the protoboard & I have the 3.3V batteries - need to source some 1.2V batteries, connectors, etc
So I agree with Hurtig et al but maybe not in their tone or all their ideas. These are my thoughts on the question posed - hope it isn't too contentious - (I am now going to put on my flak jacket & await the strafing)
Edit: I just see Hurtig has posted while I was composing & I agree with him - if he had said this as his first post then maybe more would agree. I know a couple of Buffalo owenrs and to get it to sound good they have been through an ivy O/P stage, a counterpoint I/V stage, a couple of PS boards, etc - the costs have added up & this ends up no cheap DIY option.
Last edited:
See.... that's my point. You have a Buffalo, and consider to tweak it. If the Buffalo really was that good, what is the idea in tweaking it?? Didn't the designer spend years and yeard tweaking it
That is the main difference between a mainstream DAC (Even though 32bit) and a DAC like the one we made. 4½ years ago, we had the first version done, at a level of engineering/tweaking like the Buffalo. We could have just finished at that time, like 99,9999% of all designer would do. But we decided to spend 4½ years to ensure, that anything was optimized to the best level. It't kind of like Microsoft launching new OS.... Always much better than the previous. But still it takes about 4-5 years of updating before you have something that really works.
Only thing I'm doing with Buffalo32s is using optimized supplies and upgrading regulation, the rest will be left as it is
Again we can waste countless pages on pointless debates, I think people rather read results of actual sound comparisons
Sorry folks but I cannot help commenting on this.....
If your DAC and its development compares to that of Microsoft operating systems it must be really, really good, stable and up-to-date! Maybe you should call it Vista, or even better Longhorn
Cheers,
Nic
P.S. "Hurtig" is the danish word for "fast". I won't hide that my own somewhat reflects my PC preferences..........
...
Again we can waste countless pages on pointless debates, I think people rather read results of actual sound comparisons
Actually only if you have some new miracle protocol for listening tests that is Scientifically acceptable, reliable, reproducible, and resolving?
plus some info on how many factors you're testing/design of experiments? - I expect the map from circuit topology/parts tweaks to subjective sound impression is a rather high dimensional manifold and not simply connected
I would like to see careful measurements showing reproducible differences in audio output with these power supply tweaks – Audio DiffMaker + studio grade ADC maybe? Or can you point to differences at high frequencies with a Spectrum Analyzer?
Last edited:
How about not bothering listening at all and rely just on measurements
Unlike some I do make the effort to actually try things and listen to them (mostly with friends) before posting opinions, if thats not good enough I'm sorry
Of course going by a bunch of peoples opinions isn't going to be 100% but its going to be more helpful than just claims one sounds better than the other with just a few pictures of the dac to judge with
Sorry folks but I cannot help commenting on this.....
If your DAC and its development compares to that of Microsoft operating systems it must be really, really good, stable and up-to-date! Maybe you should call it Vista, or even better Longhorn
Cheers,
Nic
P.S. "Hurtig" is the danish word for "fast". I won't hide that my own somewhat reflects my PC preferences..........
You really did not get the point in my post. The point is, that 99,9999% of all audio gear is comparable to the way some people develop an OS. In the way, that as soon you have something that is not burning down, you launch it and claim it to be the best of the best. Then you spend the next 5 years on updating
That's what audio-guys do to... Why do you think so many people spend years tweaking their gear, if it was that perfect from the beginning ???
We were at the state of "not burning" 4½ years ago. But instead of launching the design like all others, that needed intensive tweaking to deliver the performance we wanted, we spend the time tweaking it to maximum performance.
This means that you do not need to spend that time/money yourself. Also since you have a pre-tweaked design, you don't have loose connections, extra PCB's mounted on top of the main board. All "tweaks" is on-board! That's the real difference, and I fully understand why you do not understand, because you may very well never have seen anything like that. I know of only very few commercial products, where the designer has spend that much time maximizing the performance.
- Status
- Not open for further replies.