at the very least you would want to try and load both outputs the same
seriously why do I have to list out numbers in such an obvious case, everything including the internal layout of the dac and external pin layout, roughly $500 or more (for half) and more again if you include power supplies, all in primo symmetrically matched components tagged onto the output of 2 balanced dacs and then not using it? does that have to be spelled out for you?
just use a transformer, there are several other options but a couple of edcor or even onetics output TX arent that expensive and dead easy to implement, nothing compared to the cost of having so much sitting there completely idle
if you look, weve been through this conversation at least 5 times already (probably significantly more) in this thread.
seriously why do I have to list out numbers in such an obvious case, everything including the internal layout of the dac and external pin layout, roughly $500 or more (for half) and more again if you include power supplies, all in primo symmetrically matched components tagged onto the output of 2 balanced dacs and then not using it? does that have to be spelled out for you?
just use a transformer, there are several other options but a couple of edcor or even onetics output TX arent that expensive and dead easy to implement, nothing compared to the cost of having so much sitting there completely idle
if you look, weve been through this conversation at least 5 times already (probably significantly more) in this thread.
Last edited:
Because it has nothing at all to do with the internal layout of the dac per example. Clearly both outputs of the 9018 are loaded the same, so it is only a question of less output voltage and even harmonics not being cancelled. So, what is this? 6db worse s/n and let's say distortion up to -100db? How on earth does this matter in the context of a 300B amp? It probably isn't the best choice for an I/V stage for a non-balanced amp, but maybe one day there will be a proper balanced amp in his system.
It somehow annoys you that someone may be misusing your favourite toy, but it is the most sensible and practical choice in the situation.
why is it the most sensible? it has just as much to do with the internal layout of the dac as any of these other carefully engineered symmetrical things that end in naught. the most sensible is following it through, its got nothing to do with my favorite toy, its got something to do with answering a shitload of questions and selling this man a bunch of expensive matched stuff, that would be completely without meaning if only half the output was used.
I would have also strenuously advised against such a large expense, the blame for which was placed squarely at my feet a few pages ago by BPHK along with some amount of expectation.
btw BP I know you arent the one pushing for this
if the plan is to fairly quickly replace those amps then sure maybe, but otherwise going the rest of the way is a drop in the ocean compared to the rest of the effort and expense.
I would have also strenuously advised against such a large expense, the blame for which was placed squarely at my feet a few pages ago by BPHK along with some amount of expectation.
btw BP I know you arent the one pushing for this
if the plan is to fairly quickly replace those amps then sure maybe, but otherwise going the rest of the way is a drop in the ocean compared to the rest of the effort and expense.
Last edited:
There are some rough measurements of taking the SE output from one phase of the NTD1 back in the main thread...
Start reading from here:
http://www.diyaudio.com/forums/digi...ssic-pass-labs-d1-ess-dac-17.html#post2199670
Calvin built an early iteration of the circuit and claimed he was getting terrible results... turned out it was because he was measuring SE out.
The THD+N penalty is significantly higher than 6dB... think more in the order of 20-30dB.
Honestly, I would suggest using just one phase as a last resort, and trying pretty much anything else to see what sounds best.
If the next stage were an F5 with SE input, then I would definitely go the op-amp route, but with a tube amp, a transformer might be a little more in-line with the sonic signature of the amp itself.
Regards,
Owen
Start reading from here:
http://www.diyaudio.com/forums/digi...ssic-pass-labs-d1-ess-dac-17.html#post2199670
Calvin built an early iteration of the circuit and claimed he was getting terrible results... turned out it was because he was measuring SE out.
The THD+N penalty is significantly higher than 6dB... think more in the order of 20-30dB.
Honestly, I would suggest using just one phase as a last resort, and trying pretty much anything else to see what sounds best.
If the next stage were an F5 with SE input, then I would definitely go the op-amp route, but with a tube amp, a transformer might be a little more in-line with the sonic signature of the amp itself.
Regards,
Owen
just a quick note/reminder about balanced systems, it is mainly even harmonics as far as THD (from the chip) that are cancelled in a summing stage, but with a completely differential circuit and dac like this, just using one half means you lose ALL the CMRR and that isnt just 6db THD.... rather any common mode error, ripple, noise etc. that would previously be deleted wrt the rejection ratio; shows up in its entirety without a summing stage.
Given this Dac chip and everything after it was designed with this summing in mind (they do not even supply any spec for without the summing) the effects could be severe, the big picture cannot be known without knowing the power supply, grounding, noise on USB etc.
it will not be 6db.
remember also the fairly low PSRR of the circuit, but with everything in place and pretty much constant draw its not a significant issue
Given this Dac chip and everything after it was designed with this summing in mind (they do not even supply any spec for without the summing) the effects could be severe, the big picture cannot be known without knowing the power supply, grounding, noise on USB etc.
it will not be 6db.
remember also the fairly low PSRR of the circuit, but with everything in place and pretty much constant draw its not a significant issue
Last edited:
ahh yes I remember that convo Owen, I knew there was something more concrete and we have covered this ground several times but I forgot the rough measures you made.
from memory (bed time for me) we found out he was using a different dac too wasnt he? lol like the pcm1794 or something? or was that another early detractor?
my thoughts exactly
from memory (bed time for me) we found out he was using a different dac too wasnt he? lol like the pcm1794 or something? or was that another early detractor?
opc said:
my thoughts exactly
Last edited:
I was indeed planning on flipping the phase of DACB and trying it Vout SE into a tube amp. With tube specs in the 0.1% THD, I wouldn't think it matters (the increased THD) would it?.
But I agree, transformer is probably the simplest solution. Way, way back there were some recommendations for low cost (and good for audio) transformers. Anyone remembers?
But I agree, transformer is probably the simplest solution. Way, way back there were some recommendations for low cost (and good for audio) transformers. Anyone remembers?
no, glt I saw your musings on that but resisted at the time =) but I would think it would be subject to exactly the same limitations, but you would get some small amount back from the parallel dacs being added.
SNR and DNR will also be effected I would think and some of the THD and noise will be coincident with the series THD in the tube amp (dont ask me how much. All this was enough for me not to give it any more thought.
if you do a search for posts by me and Russ (may have been Brian, but memory says Russ), I think in one of the early legato threads (or at least in the TP forums here at DIYA), I made some posts along a similar line to what you are suggesting, but with the Ackodac.
None of your solutions for FW control were available and the Buff had limited functionality in this area at that time, the ackodac was the only DIY dac capable of it then. I was pretty green and was corrected pretty completely with info in-line with what we are saying here.
as for transformers I can recommend the ones from BudP O'netics, as a quality but not crazy priced part (maybe $50 each from memory) Pano also made high recommendations and quite a few posts on Bud's units for dac output. if still interested i'll see if I can track down a datasheet, or you could PM BudP here on the forum. they are high nickel core with copper wire, custom wound. very good spec and seem to be also highly regarded in the F6 thread. normally you would pay a lot more for such units but Bud isnt quite as well known as the usual high dollar options. Edcor are also good for something cheaper as I said earlier.
SNR and DNR will also be effected I would think and some of the THD and noise will be coincident with the series THD in the tube amp (dont ask me how much. All this was enough for me not to give it any more thought.
if you do a search for posts by me and Russ (may have been Brian, but memory says Russ), I think in one of the early legato threads (or at least in the TP forums here at DIYA), I made some posts along a similar line to what you are suggesting, but with the Ackodac.
None of your solutions for FW control were available and the Buff had limited functionality in this area at that time, the ackodac was the only DIY dac capable of it then. I was pretty green and was corrected pretty completely with info in-line with what we are saying here.
as for transformers I can recommend the ones from BudP O'netics, as a quality but not crazy priced part (maybe $50 each from memory) Pano also made high recommendations and quite a few posts on Bud's units for dac output. if still interested i'll see if I can track down a datasheet, or you could PM BudP here on the forum. they are high nickel core with copper wire, custom wound. very good spec and seem to be also highly regarded in the F6 thread. normally you would pay a lot more for such units but Bud isnt quite as well known as the usual high dollar options. Edcor are also good for something cheaper as I said earlier.
Last edited:
no problems, edcor have a lot of models, so you might need to do some searching for something appropriate, but the spec is reasonably good overall and very good for the money.
the O-netics I know more about, windings are tightly enough matched that they will cope no problems with direct connection to the dac, no caps needed to null AVCC DC current (they wont saturate) and the gandwidth is pretty good really, much better than some. nothing on a good solid state circuit such as this (for THD etc) but there is something to be said for the simplicity of the connection
something similar from Audio Note, Sowter, Lundahl, or heaven forbid the exotic Feastrex/Teramoto/Suzuki Finemet transformers will set you back MANY times more
the O-netics I know more about, windings are tightly enough matched that they will cope no problems with direct connection to the dac, no caps needed to null AVCC DC current (they wont saturate) and the gandwidth is pretty good really, much better than some. nothing on a good solid state circuit such as this (for THD etc) but there is something to be said for the simplicity of the connection
something similar from Audio Note, Sowter, Lundahl, or heaven forbid the exotic Feastrex/Teramoto/Suzuki Finemet transformers will set you back MANY times more
Last edited:
Are you targeting 99% perfection?
I like your concept... I would imagine with the right choices you could get a whole lot of bang for your $297.
hmmm... $99 would get you a pair of Fountek FR-89 drivers.
But I digress... It's OT.
If you're really looking for low price, Hammond also makes some 600 ohm telecom transformers that might fit the bill. I don't know the P/N off hand, but I'll check tonight. I ordered a few to see how they would work direct coupled to the output of the NTD1, but never got around to it.
Cheers,
Owen
I like your concept... I would imagine with the right choices you could get a whole lot of bang for your $297.
hmmm... $99 would get you a pair of Fountek FR-89 drivers.
But I digress... It's OT.
If you're really looking for low price, Hammond also makes some 600 ohm telecom transformers that might fit the bill. I don't know the P/N off hand, but I'll check tonight. I ordered a few to see how they would work direct coupled to the output of the NTD1, but never got around to it.
Cheers,
Owen
Powered the monstrosity up and, no big surprise, direct connection of a summing transformer won't work. It is probably easy to trim most of the dc imbalance at the operating point by applying a parallel resistor to one of the source resistors, but that isn't the problem. As the circuit heats up the imbalance covers a range of 30-40mV, probably because the fets are only matched at the o.p.
I am certainly not an expert on cores but somewhere in the back of my mind there is some info that at least some of the more exotic cores do not take well to magnetization which can permanently alter them. So, any of the worthwhile transformers i can think of won't be too happy.
That leaves parafeed of course, for those who like it
So, as promised i did compare the SE output to a bal to SE using a Lundahl 1676 connected as 1+1:2. In both cases no preamp was used and the following non-balanced power amp had 80k input impedance. Coupling caps were 3.3uF of Clarity MR. Due to unbalanced and unstable dc at the outputs, the transformer primary was fed through a cap as well.
In short, the SE cap only coupled output sounds much better than the transformer summed output, especially at the frequency extremes. Obviously i was biased towards this outcome and in any case it only proves that this particular Lundahl transformer is not so great in this application.
The I/V sounds very impressive - bass, dynamics and surprisingly - warmth. I am genuinely surprised that something powered from crappy 317/337 regs can sound so good. Btw, the regulators occasionally do not fire up on power-on - not sure if it's because of the large caps following them, or the high incoming DC voltage. Will replace them with 1085/1033 at fist opportunity.
In short, the SE cap only coupled output sounds much better than the transformer summed output, especially at the frequency extremes. Obviously i was biased towards this outcome and in any case it only proves that this particular Lundahl transformer is not so great in this application.
The I/V sounds very impressive - bass, dynamics and surprisingly - warmth. I am genuinely surprised that something powered from crappy 317/337 regs can sound so good. Btw, the regulators occasionally do not fire up on power-on - not sure if it's because of the large caps following them, or the high incoming DC voltage. Will replace them with 1085/1033 at fist opportunity.
Last edited:
That was a huge problem for me too, the only thing that wasn't expected... We can discuss what was done in my version, might save you some time.
Hi Guys,
I've never had this issue before with the regs. Can you elaborate on the conditions under which it is happening?
- What is the input DC voltage?
- What is the output DC voltage?
- What is the input capacitance?
- What is the output capacitance?
- What is the bypass capacitance?
- What 317/337 brand / type are you using?
Hopefully we can get to the bottom of it.
Regards,
Owen
I've never had this issue before with the regs. Can you elaborate on the conditions under which it is happening?
- What is the input DC voltage?
- What is the output DC voltage?
- What is the input capacitance?
- What is the output capacitance?
- What is the bypass capacitance?
- What 317/337 brand / type are you using?
Hopefully we can get to the bottom of it.
Regards,
Owen