I removed the regulators and gave 10.5 volts to the opamps.
The result is fine.
Thanks guys. Yes, definitley very dedicated and open for more. I just need a good how2 and some photos. Like from lufubu. I already have the 8139 here. While that would hurt (4 cost 80-90 EUR) not to use them, I cant replace the APS trafos. But you got me curious about the LMP8350: Do you mean I can Just drop and replace them?
What is my aim? I dislike the sunlight firmware. Sounds like ESS Sabre to me. Windom is really relaxed but swollows some details. So I keep Windom and just see where it is going 
Attached is a photo how how it is supposed to look when its finished. Today I wouldnt have ordered those big Nichicon KGs. But back then we started with the Edcores and I really dislked them at first - had a big loss of bass and ordered everything that were supposed to help: Big PSU caps and output buffers (like in the MKii).
As it turned out: the interconnect and power cables were at fault.
Yes you can, but you have to be careful of a few parts in the on-board regs need updating in terms of voltage capability. The on-board regs are fine BTW, I updated them and use them to good effect on the DAC I mentioned above. Also the feed to the regs may need to be updated as well, depending on your goal for output volage. If I recall the stock feed to the regs is only about 9V, once it makes it through the passive filtering to the regs. Think I have about 10V out on the DAC I mentioned above.
TK
I also dislike the Sunlight and agree with you. All my DACs just use Snowmass; it sounds more alive and lifelike with it.
The LMP8350 are pretty much drop in. Youāll have to examine the DS to see what that power select pin needs. If I recall, I think it worked fine left floating as an initial test; but I think I followed the DS and pulled it up (maybe down, sorry, I donāt recall exactly by now).
TK
I should also qualify carefully the above with, I do also change the filter topology on the DACs to a scheme that has perfect step response. The stock MFB topology does not have perfect step response, at least in my simulations, and I think that affects the sound. So your case will not be exactly the same as mine.
What I do know from experience is, once the LPF topology is changed the DAC essentially has no audible distortion, which to me is a very large upgrade.
I also add some new, passive LC prefilter that is not present on the stocker, to help tame the BW the active LPF section actually sees. The extra prefilter also brings the overall high freq attenuation back in line with the stock MFB case and what we need here. I generally see about > -110dB attenuation at 5MHz by now with the Studio version.
However I think the ultra low Dx diff amps are an upgrade vs the 813x in general. The 813x have a large BW advantage however, large enough to even buffer the DSD digital bits basically lol, but we donāt really need that here if weāre using them as low pass filters.
TK
This shows the simulated AC output response and 5MHz attenuation of the current Studio design; recall that this now uses ultra analog performance āerror correctingā gain blocks in the LPF output circuits, not industrial diff amps:
This shows the simulated step response. I dig it, with rapid settling time as well. Note though, these circuit models are NOT even close to perfect however; but useful enough I think:
You need to replace C703 or turn it off, it is designed for 8V.
1 Like
Hi guys,
I got soft and bought another pair of ultra-capacitors for 3.3V or 5V (it was a almost not used kit, so price was attractive).
Can you please advice on which chip it would make sense to apply it on? The FPGA perhaps or I read that there is another chip involved in the PCM to DSD conversion?
I really liked what it did to the sound after applying it to the clock
I think if it were me, I would pay close attention to the little circuit area that converts the bits to the desired format to feed the re-clocker (if I recall correctly). Also the re-clocker, i would treat its PSU practically as a critical analog part basically 
The re-clocker is the device that slings our DSD stream into the output circuits.
Also I think we would want the CLK module to have a low impedance PSU so it can also do itās job properlyā¦ And thereās a CLK speed control line (itās a voltage controlled device) that I also tended some too as well. We donāt want that to have analog band noise on it IMO.
Personally Iām not on the ultra cap bandwagon as they donāt have a low enough impedance for what we need, so thereās lots of bypass that needs to occur around them as well. But I think it can be done well, just not as easy as it seems at first.
TK
1 Like
Thank you Turbotk. Do you have a photo of that re-clocker IC or do you remember where it is ca. placed?
Yes I also thought about replacing the voltage various regulators on the input board and the analog board. I really like those Ultra Low Noise Voltage Regulator
Regarding impedence: I also heard that regarding battery voltage but it can sound pretty good. One kit costs around 200EUR/USD. Thats almost cheap for our terms
Be aware that most clocks and reclockers canāt help and some definitely can make things worse (or at least cause unexpected behavior: how it sounds is entirely subjective and I wouldnāt argue with someoneās opinion there.) As Turbotk mentioned the DSās clock isnāt a simple oscillator pack.
-
The DSās VCXO (voltage controlled crystal oscillator) has the lowest phase noise of any (voltage controllable) oscillator pack out there. (Phase noise is the part of jitter that matters for audio, long term frequency stability doesnāt matter for audio.) And the DS 's VCXO already has its own power supply (which is a discrete bootstrapped super regulator.) Some of the oscillators I see on some clock replacement modules out there use a much cheaper, lower spec, non-voltage controllable, higher phase noise oscillator. Perhaps ask for a phase noise vs. frequency plot of a candidate replacement clock and compare it to the DSās oscillatorā¦
-
The DS has a feature that partially hides the effects of using a non-voltage controllable oscillator: it duplicates or throws away samples to make up for the oscillator not tracking the average rate of the incoming bits
A casual listen may not hint that something is wrong, but there will be much more jitter and noise in the output than when using a proper VCXO. Also, if the replacement oscillator doesnāt have a pull range of +/- 100ppm it will cause samples to be lost or replicated if the sourceās clock is near 100ppm off of the nominal rate (100ppm is the Redbook spec, and Iāve seen multiple respected CD and SACD players near or past that offset.)
8 Likes
Sorry, I donāt have a good picture of it but Iām talking about this Motorola device that drives the DSD stream into the analog output stage. Its in the middle, near the VSS test point loop.
TK
Thank you Ted. And if I recall, the small chip/circuit between the Crystek and the reclocker is used for logic level conversion?
A couple of audio updates.
Something new DSD (to me), have what could be an interesting DSD DAC arriving on Friday. A Cen.grand Deluxe. Weāll see how it does hereā¦
Iād also love to get my hands on a DS mk2ā¦ Someday? I did hear one last year at the little Lone Star Audio fest and really liked what I heard, even though maybe not through the best playback system.
Some other exciting news, I took on a big amp improvement project and the audible results are very positive so far. An extensive design upgrade of Kinki EXB7 mono blocks. These are a JFET input and MOSFET output and have a circuit topology I like. Among several key things, pushing it more towards an āerror correctingā type of approach. Also halved the rail voltage so could double the output stage bias level the heat sinking can tolerate. Etc. Still some more work to do in round 1. Weāre aiming for ~ the top here sonically; otherwise Iām wasting my time. But oh my, the precious clarity and low noise floor Iām hearing already
If it does ultimately prove to be as hoped we may offer these as a N America special version or something like that. Might even become a new production version of the amp. Weāll see. More work to do.
TK
1 Like
Yes, the small chip does logic level conversion, but it also duplicates the clock to minimize interaction between the reclocker and the transmission of the clock to the digital board. In the Mk II Iāve replaced it with a resistor network for the logic level conversion and use a separate buffer to drive anything else on the board - somehow, I trust passive components to cause less jitter than any active component no matter how good
6 Likes
Ted, it is great to see you back here actively educating (and in some instances saving us from ourselves ) on these parts and design, technical matters.
I am curious as to whether you are back to focusing on the TSSD design/development.
Cheers.
1 Like
Iām not working on new software for the Mk II at this moment I am working on simplifications of the TSS design based on the Mk II work, but I donāt want to say more until I know how it works out.
8 Likes
Thanks for the update.
Looking forward to hearing more when you have something you care to shareā¦
SEE
What a concept! āBubble upā instead of ātrickle downā. 
And what is his TSSD ?
Tedās project Two Chassis Super DAC