[astrostar59]

  DC jack transplant performed.
I need to redo the crimping, as it seems the connection isn't stable (my DAC's lock LED is blinking, when connected to the RN3).
Either that, or the LPS-1 doesn't have enough juice, but I doubt it.
 

 

 

I would use the wiring loom that goes from the SMPS. I cut mine and soldered the first 2 leads to 5V DC positive, and the third lead to earth. I tried to do the crimping as well, but gave up.

 

[Middy]

That definitely tweak of the week award. My favourite thread on headfi always suprised. Lovely work.

 

[Golfnutz]

  Just made an order from Digi-Key. Hopefully the pre-crimped 300 mm cables will fit . I will use the IEC as Clemmaster did. Hopfully the Q204-ND will fit my Mutec, it will be my next project.
 
 

Sorry, those are Molex, and 12awg (pretty big). You want the JST XH you already have listed.
 
If anyone wants a header made, I'll do them for $5 + shipping. That basically covers the cost of wires (2 or 3), pins, JST connector.

 

[Tand2016]

 
Sorry, those are Molex, and 12awg (pretty big). You want the JST XH you already have listed.
 
If anyone wants a header made, I'll do them for $5 + shipping. That basically covers the cost of wires (2 or 3), pins, JST connector.

Thanks! Got it corrected now 

 

[Clemmaster]

 
I would use the wiring loom that goes from the SMPS. I cut mine and soldered the first 2 leads to 5V DC positive, and the third lead to earth. I tried to do the crimping as well, but gave up.

 
My goal was to not do any modification to the RN3, aside from unsoldering the SMPS from the original filtered IEC (they use so much solder on the terminals, with the metallic shell of the connector it takes forever to melt 

)
 
I will not give up, I have hundreds of crimp pins to get it right 

 

[johnjen]

 
Hi JJ:
 
Sorry, but you are incorrect about the UltraCap LPS-1 with regards to the charging and ground-plane noise.  There are 4 entirely separate ground domains on the LPS-1 board, and the only things that cross them are 10 optical isolators.  The bank that is charging contributes ZERO noise to the ground-plane or output of the bank that is actively supplying the ultra-low-noise output (which is via cascaded regulators).  If it did, then the whole point and complexity of our design would have been for nothing.
 
And of course we are not using ANY batteries!  

 
Here you can see the ground planes, and in the board photo you can see the 10 white opto-isolators crossing those moats.
 

Cool!
 
So the 4 planes share nothing in common (ground, power etc.)?
 
And why 4 planes, as in, are there 4 banks of UCaps?
And is it a SMPS or other type of charging circuit/power supply?
 
And I didn't mean to imply that your device had batteries, but instead operated AS a battery.
Sorry for the confusion.
 
JJ

 

[johnjen]

  DC jack transplant performed.
I need to redo the crimping, as it seems the connection isn't stable (my DAC's lock LED is blinking, when connected to the RN3).
Either that, or the LPS-1 doesn't have enough juice, but I doubt it.
 

 

 

So is the negative side of that 5Vdc grounded locally inside the RN3?
 
And if not I wonder if that could be contributing to the 'lack of lock'.
Or not.
 
Just wondering is all.
 
JJ

 

[johnjen]

  JJ
My observation thus far is the LPS on my Rednet may have fully burnt in at 1 week on 24/7. It seems now to be stable. The caps on the LPS and the transformer are quite small items so it may make sense that this will speed up as compared to a big new power amplifier for example with huge capacitance and big mains transformers.
 
Another point, there is a big difference in source material, which can confuse the issue somewhat. For my tests I tend to stick on 6 tracks I know very well, and use those as a datum point. I am possibly odd but seem to have a way of 'remembering' a sonic signature of those 6 tracks. For example, I listened to them on a BHSE with the Esoteric K-01 well over a year ago, but I can remember how they sounded. Clues are length of decay after a loud section in the music, soundstage width and timbre, especially female vocals and depth and frequency response. These things have progressed beyond that on my current DAC and Carbon amplifier, then again after some cap upgrades on the DAC, tube rolling in the DAC and this week, the LPS mod in the Rednet. It has all been a solid sound quality curve in the right direction i.e. better in all areas.
 
The texture aspect I spoke about the other day is something I noticed in that progression, but in the Rednet it seems to be very obvious a change. It must be the reduction of hash and general power supply noise polluting the data, bit like taking a photograph with condensation on the lens. Now it is all in focus, tight and clean.
 
Last point, before I got the PS Audio P10 power regenerator, I used to prefer the sound of my system late at night and into early morning much more than the day time (peak mains usage in the street). After installing the P10 I get 100% consistent sound quality 24 hours a day, actually better than the best times of the day. So I wonder if any variances you pick up (on the same tracks) may be due to this?
 
Hope this helps

In contrast, I listen to a mix of ≈ 3600K tracks from ≈ 360 albums, of which probably 30-50 are 'reference' quality, and like you I too have 'memorized' many of them, to the point of singing along, with full intonation and inflection etc.
Frank Zappa is a favorite for this along with Dire Straits.
And many a classical piece has risen to these reference ranks as well, along with several other genre's.
 
This selection of music runs from 3 star to 5 star music (as I have rated them in Media Center) and this preferred selection generally is on a 3-4 week repeating cycle.
 
I have found that when the 3 star music takes a decided step up in SQ, then I know I'm on the right track, and the 4 & 5 star tracks usually follow suit as well.
 
Power ReGen units is one area I have not explored to any degree yet.
I have optimized the short dedicated feed to my system using a variety of techniques and have measured this dedicated branch circuits electrical characteristics using ASCC testing parameters as I posted a ways back.
 
These measurements are aimed at what ability the mains voltage has in being able to dump BIG current in a very short amount of time.
And I do wonder how well any ReGen unit will measure using these tests.
 
Perhaps one day I'll get a chance to get my hands upon one of these ReGen units and put all of these different ways of measuring the electrical response characteristics, along with their SQ results, to the test.
 
But for the most part the power here in the NW and specifically the Seattle area is very clean, stable, and free of noise.
And as it turns out they (City Light) just replaced the transformer 'on the pole' in my backyard with a brand new one just a couple of years back.
Which means my system which is ≈ 80' away from the step down transformer on the pole outside, with the last ≈ 35' as the branch circuit to my system, is about as close to the source of power from the main grid as is reasonably possible.
 
But yes I do notice the SQ improve after about 8:30 and then again after 12:30 another step up is noticed as well.
So there do seem to be times when there is 'better' power, but I wonder how much of that is due to power factor issues vs noise vs other power distribution system related issues.
 
And perhaps your use of 'texture' is akin to my use of the term 'focus' where the micro details become obvious whereas before they were 'buried' in the background of the acoustic soundstage.
 
And as I hear these LPS mod improvements 'settle in' the key aspect that I am hearing is a marked increase in the ability of each 'voice' to provide more of the impact that was buried in the background due to the acoustic energy being smeared thru time.
These mods which result in an increase in leading edge impact also seems to have the effect of sharpening the 'focus', or bringing out the 'texture' of each 'voice' all the more as you noted.
 
Which I figure is the good stuff all the way around.

 
JJ

 

[Golfnutz]

  DC jack transplant performed.
I need to redo the crimping, as it seems the connection isn't stable (my DAC's lock LED is blinking, when connected to the RN3).
Either that, or the LPS-1 doesn't have enough juice, but I doubt it.
 

 

 

Did you try taking the voltage off the PCB to make sure the pins have contact. Black probe is touching the ground wire.
 

 

 

[Clemmaster]

I'll check everything with my scope tonight.
I had to leave 5mn after finishing the mod, so I just put the case back together and set it up in the system, hoping it would just work .

 

[Superdad]

  Cool!
 
So the 4 planes share nothing in common (ground, power etc.)?
 
And why 4 planes, as in, are there 4 banks of UCaps?
And is it a SMPS or other type of charging circuit/power supply?
 
And I didn't mean to imply that your device had batteries, but instead operated AS a battery.
Sorry for the confusion.
 
JJ

The 4 domains are:
1) Input charging circuits
2) and 3) The two separate banks of ultracaps
4) the output domain with its voltage regulators etc.
 
The optoisolating relays always keep the output domain connected to whichever cap bank is active, and they keep the other bank that is charging connected to the charging circuits.  Draw yourself  a quad of 4 blocks with two in the middle and then it will make sense to you.
 
Of course the actual functioning of the unit and all its state machines is vastly more complicated than this, but in essence the above is how it works.  True isolation and 100% blockage of leakage currents.
 
So yes, you can think of it like you would a battery.  In fact, because these are "floating" supplies, folks can put them in series for hight voltages, or in series and center-tap for "ground" to create a bi-polar, +/- supply.
(Sorry, paralleling for greater current is not possible unless I hand-select units whose output voltage--at identical current--match each other to within 10mV.  Otherwise one unit always ends up carrying most of the load.)
 
By the way, an LPS-1 user confirmed with the CEO of Mutec--and then performed the mod on his MC-3+USB to remove the SMPS and run from the UltraCap supply.  He tells us it sounds great.

 

[Clemmaster]

Summary of my findings:

1- My crimping skills are good. The PSU was steady at 4.94V on the board.

2- The SPDif bitstream was superimposed to a slowly varying AC noise (didn't measure the frequency. Amplitude was in the order of 200mV pk-pk).

3- Johnjen was right: the SPDIF connector is not isolated from the chassis, so the chassis and signal (PSU) ground must be at same potential.
I fixed the AC issue by tying an extra ground pin in the JST connector to the nearby screw holding the PCB to the chassis.
Now the SPDif output is straight like a ruler.

4- the SPDIF output is DC coupled. There's no isolation transformer for it, unlike the AES outputs.

Is the original SMPS grounding scheme to tie the boards GND (0V) to the earth? Without anything in between?

 

[Clemmaster]

 

[johnjen]

Summary of my findings:
snip

Is the original SMPS grounding scheme to tie the boards GND (0V) to the earth? Without anything in between?

The original SMPS provides THE ground connection (the chassis itself) for the mainboard.
And the IEC connector connects its ground straight to that ground post in the right rear corner of the chassis along with the ground wire to the top cover.
 
IOW the mainboard, the ground from the ac mains, and the chassis are all tied together using the chassis as the common connection.
 
But also note it is the entire circuit, not just the SPDIF (or any other) output that is dependent upon a 'stable' and lowered noise floor ground connection.
 
And my measurements showed that with the mainboard 'floating' (not grounded) the noise was considerably higher on the ground itself, which is yet another contributor to adding even more noise to the entire system.
 
JJ

 

[johnjen]

 
The 4 domains are:
1) Input charging circuits
2) and 3) The two separate banks of ultracaps
4) the output domain with its voltage regulators etc.
 
The optoisolating relays always keep the output domain connected to whichever cap bank is active, and they keep the other bank that is charging connected to the charging circuits.  Draw yourself  a quad of 4 blocks with two in the middle and then it will make sense to you.
 
Of course the actual functioning of the unit and all its state machines is vastly more complicated than this, but in essence the above is how it works.  True isolation and 100% blockage of leakage currents.
 
So yes, you can think of it like you would a battery.  In fact, because these are "floating" supplies, folks can put them in series for hight voltages, or in series and center-tap for "ground" to create a bi-polar, +/- supply.
(Sorry, paralleling for greater current is not possible unless I hand-select units whose output voltage--at identical current--match each other to within 10mV.  Otherwise one unit always ends up carrying most of the load.)
 
By the way, an LPS-1 user confirmed with the CEO of Mutec--and then performed the mod on his MC-3+USB to remove the SMPS and run from the UltraCap supply.  He tells us it sounds great.

Thanks for the reply.
It's always good to understand the thinking behind any design, at least from my POV.
 
And in my experiments replacing the SMPS with an LPS has Always resulted in 'Better' SQ, so I can see where removing the 3+'s SMPS would be of benefit.
Which is what I too have noticed when I yanked it's SMPS out of my 3+.
 
And just for my own curiosity what type of PSU does your circuit employ?
 
And do the isolated 4 sections share no power, nor ground, at all?
 
JJ