[Wood]

Using 14 dacs seems to of highlighted weaknesses. There was a plastic hifi rounded sound that I did not notice so much before.

So I looked for plastic, out came the wimas in the FRM mod. Much better more real. But something need to replace them, that FRM is a keeper. So in went a Bg fk 47uf , much nicer got a kick like a donkey now, gloves off. Also pulled the wima I was using for the TX and changed to a .1uf Bg nx hiq, it will be great in hmm 1000hrs

.

Result overall more transparency.

 

[PatOMalley]

Thanks for that update.
I was a gonna try 12 myself but with the descriptions given and with the trannie on the output I think I will stay with 16.

I think the added resolution helps a lot when going out the transformer.
Which is why I chose the thinner sounding Mundorfs vs the Jupiters.

The Mundorfs have more sliding detail which I am hoping will be shmoozed into thickness by the transformer. I was thinking the Jupiter would be too much of a similar sound as the transformer and thicken up too much.

It's like making soup, ain't it?

 

[Pricklely Peete]

Quote:

Originally Posted by Wood /img/forum/go_quote.gif Glad you started on that power in recabling Pat, I've got some 18awg transformer copper staring at me that I've been to lazy to do anything with yet. I was going to wire the 7.5v directly to the where it inputs did you do that or wire to the PCB? Pete, I think you can see we playing around on the peripherals of that power supply. I'm to dumb to do any serious stuff, but am planning on changing caps and things. What I would like to do is put larger capcity transformer in to make sure those dac chips get everything they need, but dont know what would fit in the case and do the job. But you know this stuff, what would you do Pete?

I'm thinking a true dual mono supply is the way to go but it would require a new chassis and quite a bit of money to complete properly. No doubt IMO that it would make a big big difference compared to the stock supply, not that the stock supply is inadequate for the current job it's just not an all out design. I'm hoping Bill's experiments are along these lines but he has not chimed in in quite awhile so I hope he's still thinking of doing something similar (he hinted at such several pages back).

I would likely split the PSU duties off further by dedicating another separate (PSU board and transformer, 15VA toroidal or R-Core if it can be sourced) low V supply to the receiver chip/clock boards. Getting these supplies (all three of them ) as clean and linear as possible should yield huge dividends.

The expense and size factor is the only limiting factor but it could be done for under 150USD including a half decent chassis I would imagine. Upgrade the DC lines to heavier jacks (like Mini XLR) and use either CC OFC 6N copper solid core (18 awg) or solid silver of the same gauge. Might as well build a decent mains cable for the PSU while your at it. A solid silver mains cable with tinned copper mesh shield could be built for a small outlay in cash. Maybe another 100USD or less depending on length and OEC/Plug choices. The entry level Wattgates are more than good enough. Of course all of these ideas are just suggestions and may represent a level of overkill that some may find unnecessary but I'm pretty sure limitations of this type are not a factor in this thread. What I mean by that is a basic mod rule I learned a while back WRT and it pertains to all audio gear, work on the PSU first, second and third and then move on to the other bits once you've maxed that PSU design out or have run into your budget limitation. It's a good rule that has worked with every mod/project I've managed to complete without blowing the piece of the gear up (I exaggerate of course about blowing stuff up but it happens to the best of us from time to time when attention to detail slips) Kind of an expensive reminder/lesson to be careful and precise and to take nothing for granted.

The hard part is sourcing the right base PSU pcb, proto'ing a board would be too expensive unless some members here have access to such services for a decent price. That being said I'm weak on design but I'm trying to learn as fast as I can.

Peete.

 

[MikeW]

What about adapting a current design, like Sigma 11 or 22? Takes PCB's, BOM's and Chasis issues out of the equation.

 

[Wood]

Ok back to 16 dacs now. Soundstage is wider, less depth, or is that an illusion from the wider soundstage? There is a lot more texture then the other two configs, midrange has less presence, but is less in your face. So the presentation overall is more in the background.

So any point to this diversion? Well its sure now the iv at 175 is fine with less dacs, so if you think you have stress on highs its a way to take that down a touch.

But for me the emphasised midrange, and touch less texture on 12/14 dacs revealed a polishing of the sq that I identified coming from the Wima FRM. I had not notiched this so much with the 16 as they are more textured and have less midrange presence. The Bg FK replacement does the job the Wimas did but lets more texture through. Seems the mantra 'any film cap is better then any electrolytic' does not always apply.

For now will be sticking with the 16, which sounds a touch more natual now then before this little journey started.

 

[MikeW]

I took the reclock out again.. I really dont' much care for it anymore. I think it's staying out for good. I may experiment with wiring the output of the Digital Input board, directly to the dac input board. Instead of following those traces and dip switches.

Edit: Wired it up direct, seems to sound better to me! Try it out. Pic's below This mod is about getting the best possible signal from the Input board to the Dac board.

 

[PatOMalley]

Mike,

You left the connection from the DAC board down to the PCB in place.
Don't you have to remove them or you have a loop?

 

[MikeW]

I removed them pat, it's hard to see from the picutre. But the DATA/LRCLK/BCLK connections have been removed. Only the +5v connection remains. The data/lrclk/bclk, traces actually follow all the way back to the input board, however the +5v does not. The input board does have a +5v on it, next to the other connections, but it's not linked to the one on the dac board, that one actually follows all the way back to the main +5v input.

 

[PatOMalley]

and what you did was to leave the ground pins in the slots while cliping the data pins short but enough to solder to, correct?

And this can be done with either the input or the up-sampling board, I suppose.

Now I gotta get rid of the two resistors I have and use single TX in their place.

And how about jumping from board to board point to point?

 

[MikeW]

I completely removed the 3 pieces of "Wire" that attach the main board to the dac board. Just heat up and they pull right loose. These connections where for Data, LRCLK, and BCLK. I left the "Wire" for +5 volt alone. I then soldered new wires directly from the "Data, LRCLK, and BCLK" located on the digital input board, (The last 3 solder connections), I soldered these wires directly to the input on the dac board, completely bypassing the traces, reclock board "socket" and jumpers.

You could do the same thing, to a lesser degree, with the reclock board by connecting it's output's directly to dac board connetions. This would bypass the jumpers.

Upon listening more, I do have to seriously wonder if there was a really bad connection, or some other malfunction with my reclock board. As it sounds so much better without it, and the direct connection mod further improves it.

Edit: there's no need to "clip" anything they are just wires connected to solder pads. Also, the LRCLK connection is marked as "WS" on the main board, but it's the same connection if you follow the trace. Also, I did not remove any of the Pins on the digital input board. But it does not matter as they lead to no where at this point as that connection has been severed up-stream.

 

[Wood]

Quote:

Originally Posted by MikeW /img/forum/go_quote.gif Edit: there's no need to "clip" anything they are just wires connected to solder pads. Also, the LRCLK connection is marked as "WS" on the main board, but it's the same connection if you follow the trace. Also, I did not remove any of the Pins on the digital input board. But it does not matter as they lead to no where at this point as that connection has been severed up-stream.

Very ingenius Mike, and a clean job to boot. What way does it sound better to you?

 

[qusp]

Quote:

Originally Posted by Pricklely Peete /img/forum/go_quote.gif I'm thinking a true dual mono supply is the way to go but it would require a new chassis and quite a bit of money to complete properly. No doubt IMO that it would make a big big difference compared to the stock supply, not that the stock supply is inadequate for the current job it's just not an all out design. I'm hoping Bill's experiments are along these lines but he has not chimed in in quite awhile so I hope he's still thinking of doing something similar (he hinted at such several pages back).

I have to wonder what benefit there would be to a dual mono supply on this board, its totally single ended with a passive I/V and all the dac chips are in parallel. it would seem to me to be a bit of a waste of time and money. using separate regulation and transformers for analogue and digital sections would work well, but the point of having a dual mono supply is to have separate grounds for physically separate channels, ie. galvanic isolation of left and right channels, this works well with dacs that have chips set up for each differential channel and separate active analogue I/V sections, but you have none of this here. now if you modified the dac so it used 8 chips for R and 8 chips for L, that might be different, but this would be pretty involved and would also require some digital tweaking. would be cheaper to buy a second dac and bridge them.


Quote:

I would likely split the PSU duties off further by dedicating another separate (PSU board and transformer, 15VA toroidal or R-Core if it can be sourced) low V supply to the receiver chip/clock boards. Getting these supplies (all three of them ) as clean and linear as possible should yield huge dividends.

as above, just the 2 supplies should be enough, more discrete regulators would work well, but no real need for more power supplies

Quote:

The expense and size factor is the only limiting factor but it could be done for under 150USD including a half decent chassis I would imagine. Upgrade the DC lines to heavier jacks (like Mini XLR) and use either CC OFC 6N copper solid core (18 awg) or solid silver of the same gauge. Might as well build a decent mains cable for the PSU while your at it. A solid silver mains cable with tinned copper mesh shield could be built for a small outlay in cash. Maybe another 100USD or less depending on length and OEC/Plug choices. The entry level Wattgates are more than good enough. Of course all of these ideas are just suggestions and may represent a level of overkill that some may find unnecessary but I'm pretty sure limitations of this type are not a factor in this thread. What I mean by that is a basic mod rule I learned a while back WRT and it pertains to all audio gear, work on the PSU first, second and third and then move on to the other bits once you've maxed that PSU design out or have run into your budget limitation. It's a good rule that has worked with every mod/project I've managed to complete without blowing the piece of the gear up (I exaggerate of course about blowing stuff up but it happens to the best of us from time to time when attention to detail slips) Kind of an expensive reminder/lesson to be careful and precise and to take nothing for granted.

actually with the right thinking and purchases the size does not need to be that bad, buy a nice piece of custom iron (or 2 if you must) from richard sumner (sumR), with multiple secondaries at the voltages needed to cover the voltage drop across each regulator, buy a slab of 10mm copper the same size as the profile of the DC30 and have the digital transformer and PSU on one side and the analogue section on the other, bolt the diodes or ICs to either side (isolated) to keep the temperature steady and even for more matched performance. this would keep heat down and save on heatsinking/space, as well as shielding one side from the other very effectively (ground the sink) use a discrete rectifier to feed the DC voltages to the regs. JSR01 from sjostromaudio.com would save you on space as its a jung super regulator made with discrete SMD parts, so is very small and thin.

contrary to popular belief around here high quality SMD is usually superior and mostly cheaper. crappy SMD is ....crappy of course, but so is crappy P2P wiring, worse even, especially for digital, the ground plane is of huge importance to noise levels and therefore dynamic range in a dac, you completely give away this benefit by airwiring stuff; in fact in many cases I would think that the small 'bypasses' often seen in here to avoid nothing but traces would introduce more noise than any benefit gained by the wire (please remember i'm an avid cable believer too) many DIYers are afraid of SMD though, its not as accessible as through-hole. this dac would benefit greatly from a discrete diamond buffer, mosfet follower or similar at the output IMO, seems many of you are running this into a preamp, with a nice buffer (with current gain) to drive the cables, you could avoid the pre. you can even combine a follower with a transformer in the feedback loop. i'm doing this with my dac was going to go completely passive with transformers, but found a great little circuit by mister jung that gives you the best of both worlds.

Quote:

The hard part is sourcing the right base PSU pcb, proto'ing a board would be too expensive unless some members here have access to such services for a decent price. That being said I'm weak on design but I'm trying to learn as fast as I can.

its not hard at all, i've been collecting regulators and power supply PCBs for the last 6 months or so and i'm amazed (and dismayed, my poor poor wallet) at the shear number of them. sjostromaudio.com is a good source, thats where I got many of my different variants of super regulator, got heaps of different shunts as well. the DC30 has plenty of room for all you could need. house the iron, rectifier and power supplies in the DC30 and then use JSR01 or similar small discrete regs in place of the 3 pin variety close to the target. PSU wiring can be a bit longer, but the regs should be as close as is possible.

i'm playing around with eagle for PCB design as well, plenty of learning ahead there, but I cant wait for the possibilities. in fact getting small runs of PCBs made is quite cheap. you could get 5 FR4 boards large enough for a bipolar or dual regulator for 50-60 bucks and there is no shortage of designs available on DIYAUDIO, can often find blank PCBs in the swap meet section from people who bought too many or went another direction. or get involved in a group buy for a good design.

plenty of fun to be had. i fact i'm currently loving the look of the variable christmas tree lights of my latest power supply

LED references FTW. you will find that LEDs are great low noise references, so power supplies are often quite pretty

especially for lower voltages needed by DACs

yay back to the music

 

[PatOMalley]

I never liked those plugin boards.
Always suspected and had trouble with the connections.
This can be gotten around easily with the right method when issues arise but soldering direct is the best.



Quote:

Originally Posted by MikeW /img/forum/go_quote.gif I completely removed the 3 pieces of "Wire" that attach the main board to the dac board. Just heat up and they pull right loose. These connections where for Data, LRCLK, and BCLK. I left the "Wire" for +5 volt alone. I then soldered new wires directly from the "Data, LRCLK, and BCLK" located on the digital input board, (The last 3 solder connections), I soldered these wires directly to the input on the dac board, completely bypassing the traces, reclock board "socket" and jumpers. You could do the same thing, to a lesser degree, with the reclock board by connecting it's output's directly to dac board connetions. This would bypass the jumpers. Upon listening more, I do have to seriously wonder if there was a really bad connection, or some other malfunction with my reclock board. As it sounds so much better without it, and the direct connection mod further improves it. Edit: there's no need to "clip" anything they are just wires connected to solder pads. Also, the LRCLK connection is marked as "WS" on the main board, but it's the same connection if you follow the trace. Also, I did not remove any of the Pins on the digital input board. But it does not matter as they lead to no where at this point as that connection has been severed up-stream.

 

[Wood]

My VR turned up at last. Boy its was a hard squeeze to fit them in, good thing I went for .01, they are big caps. So 16 new caps, on the dacs goodbye Wima. The Wima were not to bad, they were an improvement over the stock smd but a touch to warm and cuddly,cynical as I am, that made me suspicious.

The VR's though have opened up this dac so much, its not the same dac anymore. They haven't removed a veil, they have opened the barn doors wide open.

Very first impression when firing up (apart from the usual wow, it still works!) was 'where has the bass gone?' Well it hadn't gone anywhere it just turned into instruments! Lots of natural contact sound on all types of percussion

Second impression, the hall acoustics, now they were always there, in the background, giving that space around instruments, voices, nice, 3d, realistic etc. But now the hall acoustics are contributing to the whole performance. There is a whole bunch of stuff going on I wasn't aware of before. Its a huge musical feast now, really had no idea that there was so much information stored at 16bit 44.1hz, this is the closest I've heard Cd sound to a good turntable, really room filling stuff.

Did you know symbols are actualy bells, I didn't until a few minutes ago? Had nice shimmer on symbols before, very pleased with myself that I'd grabbed such a nice dac, so undigital. No wrong, actually they ring like a bell as well.

Third Impresson. After a few minutes of burn in I am getting very noticeable height to the soundstage.

Its harder to distinguish the usual, lows, mids, and highs, or even to be motivated to try, just easier to enjoy a gig..... the impression of real humans performing, really grips now.

Those smd have to go, but have to be replaced with good caps to make it worthwhile. Given that the other mods are in place, and contributing, I would say this is easily the most dramatic mod I've heard so far (actually, on reflection, its on a par with the naked texas resistor mod).

Update, think I know whats going on now. Its harmonics. The Yaquin did a nice job of adding natural harmonics, this dac is digging deep enough to find the real harmonics, and all the Vishay-Rodenstein are doing is allowing them to show.

 

[Pricklely Peete]

Quote:

Originally Posted by qusp /img/forum/go_quote.gif I have to wonder what benefit there would be to a dual mono supply on this board, its totally single ended with a passive I/V and all the dac chips are in parallel. it would seem to me to be a bit of a waste of time and money. using separate regulation and transformers for analogue and digital sections would work well, but the point of having a dual mono supply is to have separate grounds for physically separate channels, ie. galvanic isolation of left and right channels, this works well with dacs that have chips set up for each differential channel and separate active analogue I/V sections, but you have none of this here. now if you modified the dac so it used 8 chips for R and 8 chips for L, that might be different, but this would be pretty involved and would also require some digital tweaking. would be cheaper to buy a second dac and bridge them.

Excellent points...a second DAC board might be the practical solution for a dual supply config....possibly giving one an option for balanced operation.

Quote:

Originally Posted by qusp /img/forum/go_quote.gif as above, just the 2 supplies should be enough, more discrete regulators would work well, but no real need for more power supplies

Agreed.


Quote:

Originally Posted by qusp /img/forum/go_quote.gif actually with the right thinking and purchases the size does not need to be that bad, buy a nice piece of custom iron (or 2 if you must) from richard sumner (sumR), with multiple secondaries at the voltages needed to cover the voltage drop across each regulator, buy a slab of 10mm copper the same size as the profile of the DC30 and have the digital transformer and PSU on one side and the analogue section on the other, bolt the diodes or ICs to either side (isolated) to keep the temperature steady and even for more matched performance. this would keep heat down and save on heatsinking/space, as well as shielding one side from the other very effectively (ground the sink) use a discrete rectifier to feed the DC voltages to the regs. JSR01 from sjostromaudio.com would save you on space as its a jung super regulator made with discrete SMD parts, so is very small and thin.

Thanks for the custom iron source ! I'll definitely be looking up Mr Sumner. The dividing shield made of copper doubling as a heat sink for the V regs is very good idea !

Quote:

Originally Posted by qusp /img/forum/go_quote.gif contrary to popular belief around here high quality SMD is usually superior and mostly cheaper. crappy SMD is ....crappy of course, but so is crappy P2P wiring, worse even, especially for digital, the ground plane is of huge importance to noise levels and therefore dynamic range in a dac, you completely give away this benefit by airwiring stuff; in fact in many cases I would think that the small 'bypasses' often seen in here to avoid nothing but traces would introduce more noise than any benefit gained by the wire (please remember i'm an avid cable believer too) many DIYers are afraid of SMD though, its not as accessible as through-hole. this dac would benefit greatly from a discrete diamond buffer, mosfet follower or similar at the output IMO, seems many of you are running this into a preamp, with a nice buffer (with current gain) to drive the cables, you could avoid the pre. you can even combine a follower with a transformer in the feedback loop. i'm doing this with my dac was going to go completely passive with transformers, but found a great little circuit by mister jung that gives you the best of both worlds.

Absolutely agreed about SMD. SMD however is DIY unfriendly compared to P2P or through hole and it presents a challenge to even the most experienced/skilled iron jockeys. Another aspect of machine soldered SMD is consistency of the size and quality of solder joint. Since pathways in SMD designs done right are so short any additional deviation from consistent soldering is problematic and really hard to duplicate by hand. I have a hard time with SMD and I've been soldering for 30 + years. My eyesight is getting worse which doesn't help and I have fingers the size of cigars so my preference is for through hole. TBH if through hole and P2P is laid out properly SMD will not outperform it by any great degree so the real trick is the point you make below (about ground plane importance) which I also agree with 100%. Mr Jung is a popular guy around here and at DIY Audio. He's certainly a innovator on the level of a Nelson Pass. I wonder just how many Jung circuits are out there being used by the DIy community ? Quite a few I'll bet.


Quote:

Originally Posted by qusp /img/forum/go_quote.gif its not hard at all, i've been collecting regulators and power supply PCBs for the last 6 months or so and i'm amazed (and dismayed, my poor poor wallet) at the shear number of them. sjostromaudio.com is a good source, thats where I got many of my different variants of super regulator, got heaps of different shunts as well. the DC30 has plenty of room for all you could need. house the iron, rectifier and power supplies in the DC30 and then use JSR01 or similar small discrete regs in place of the 3 pin variety close to the target. PSU wiring can be a bit longer, but the regs should be as close as is possible. i'm playing around with eagle for PCB design as well, plenty of learning ahead there, but I cant wait for the possibilities. in fact getting small runs of PCBs made is quite cheap. you could get 5 FR4 boards large enough for a bipolar or dual regulator for 50-60 bucks and there is no shortage of designs available on DIYAUDIO, can often find blank PCBs in the swap meet section from people who bought too many or went another direction. or get involved in a group buy for a good design.

Many people (here at least) aren't into this kind of thing at the level you are and that means having loads of spare boards around. Many in this thread likely haven't explored bread boarding their own mock ups. A lot of the focus is on improving what they do have rather building from scratch, although in order the accomplish the latter one must have a form grip of the former's basic building blocks. Something many enthusiasts deem as well above their pay scales (myself included). Leaning how to build your own is the best way to understand how all of this stuff works as it gives one the ability to judge what is worthy of further improvement and what is not. That is something I'm still learning about but with help such as yours and from other knowledgeable members/ enthusiasts the collective suggestions really help speed that journey along

Quote:

Originally Posted by qusp /img/forum/go_quote.gif plenty of fun to be had. i fact i'm currently loving the look of the variable christmas tree lights of my latest power supply LED references FTW. you will find that LEDs are great low noise references, so power supplies are often quite pretty especially for lower voltages needed by DACs yay back to the music

Yep all of this stuff is certainly loads of fun although the wife goes into instant eye role the moment I start yapping about such things to relatives and neighbors

I've been looking at optical based arrays for use as a HQ passive attenuator...there is a small unit available (DIY "Lightspeed Attenuator" - Passive LDR Volume Control (audio optocouplers)) called the Lightspeed that is very promising that you might like to look at.

Thanks for the insight and resource tips Qusp !


Peete.