Finally, I have the DAC I have wanted for years.. Signalyst DSC(2)

[Andrew Allen Ballew]

discrete, no DAC, 32tap (33 level unary code) 1 bit FIR (moving average) filter for DSD conversion.  

I am waiting on a better PSU, and as soon as it arrives, I will put the Converter with my iFi ZEN Stream for its excellent low USB noise and isolation.  

 

Even now, on a generic PSU, direct connection to standard, not special HP all-in-one PC, it is a stunning sound.  I am running at DSD512, but I am torn between that and DSD256.  I have not bought HQPLAYER yet, but the Roon MUSE DSP seems more than adequate.  Upconversion to DSD256 means all my native DSD256 files are left as they are, in pure native format... and that is truly a sublime sound on this converter.  

Right now I am using the iFi ZEN CAN Headamp, which I believe has an input impedance around 20k ohm.  I could be wrong on that though.  My DSC2 is setup to have a 3k ohm output impedance, so I am running a bit low on that ratio at around 6:1.  Thankfully there is still plenty of gain, but I wonder about the frequency response.  

In time, I will either mod the output to lower the output impedance, or get the SMSL Headphone amp I have been eyeing, that has 47k input impedance which should be just about perfect.  

Terrific design Jussi.  Thanks for starting this and allowing the DIY community to build around it.  

 

[LowOrbit]

I have one of those (v2.6 as I recall). Output levels are a little low, so needs a preamp but sounds very, very good.

 

 

[Andrew Allen Ballew]

8 hours ago, LowOrbit said:

I have one of those (v2.6 as I recall). Output levels are a little low, so needs a preamp but sounds very, very good.

 

 

yeah the ifi ZEN CAN headphone amp has more than enough gain for it, even at a 6:1 impedance ratio.  ( my DCS2 was originally modified for a headamp with 47k input impedance, thus the higher 3k ohm output impedance.  Currently I am using a cheap SMPS for power, but have a 9 volt iFi XPower on order.  It has good PSU noise regulation (The DSC) so even the cheap supply is working really well.  

[davide256]

Am curious, how did this get made? DIY all the way, kit or did someone else cobble it together? Have been thinking about getting a DSD only DAC

to complement HQPlayer

[dericchan1]

Following!!!

[skipspence]

What kind of USB do you have in your DSC?

Mine is XingCore U30 -  can stream up to DSD1024 to my DSC 2.5

[Andrew Allen Ballew]

On 6/24/2023 at 1:39 PM, davide256 said:

Am curious, how did this get made? DIY all the way, kit or did someone else cobble it together? Have been thinking about getting a DSD only DAC

to complement HQPlayer

 

i did not make it.  I was lucky enough to get it from someone else who had it built by Pavel himself.  As I understand it.  Perhaps I am wrong, but one thing I know, is I did not build it.  

It did not have many hours on it when I received it.  I know some do not subscribe to break-in, but the more it plays, the better it sounds.  Understanding that psychology can play a factor, I don't believe that is the case.  As time goes by, the frequency response seems to be 'extending'.  At first it sounded super warm, which I expected since it is transformer output.  BUT, as I have let it run for about 48 hours straight, the top end seems to be opening up.   

I will be making a full set of measurements on it soon.  (E1DA Cosmos ADC and APU hardware with Multitone, ARTA, and REW as my primary software choices)

[Andrew Allen Ballew]

5 hours ago, skipspence said:

What kind of USB do you have in your DSC?

Mine is XingCore U30 -  can stream up to DSD1024 to my DSC 2.5

 

Mine is Amanero... only to DSD512.   

[LowOrbit]

I think most of these originate from a Group Buy on DIYAudio.

 

The gerbers are/were available also so you could fabricate or get made pcbs. Component count is quite high - I wouldn't be able to hand solder one of these. Mask and flow the only realistic method. 

 

 

[Andrew Allen Ballew]

9 hours ago, LowOrbit said:

I think most of these originate from a Group Buy on DIYAudio.

 

The gerbers are/were available also so you could fabricate or get made pcbs. Component count is quite high - I wouldn't be able to hand solder one of these. Mask and flow the only realistic method. 

 

 

 

yeah, you can't just procure one with ease.  which is why I jumped all over the chance to grab one when i had it.  

[Andrew Allen Ballew]

My Amanero based version on goes to DSD512, however.  The design can conceivably decode DSD as fast as can be imagined, limited only by the input hardware and clock.  

[Andrew Allen Ballew]

I posted this over in my SMSL D300 DAC thread, but it really belongs here.  

 

First, I am not using the cleanest power source yet... iFi X Power 9 volt yet to arrive so I am on a really cheap SMPS of which I do not know the provenance or how noisy it is.  Also, the E1DA Cosmos ADC has a very low input impedance, so it isn't a great match for the DSC DAC output transformer's high output impedance, which can make getting enough 'juice' out of the DAC to the ADC difficult.  

Both of those things admitted, Here is what I did indeed get today attempting a J-test measurement in Multitone.  Very, very impressive especially given the less than optimal situation.  Cleaner, much much cleaner than some other Native DSD DACs we are very aware of that cost close to 10 Grand ;)   The noise floor (note, NOT MUTED) is better than -120db (A)!!!!!  

 

well... on to the jitter graph....






 

[Andrew Allen Ballew]

Being a transformer based solution, this converter is NOT going to wow anyone when it comes to SINAD.  But here is one of those cases where SINAD just does NOT tell the whole story.  First of all, not all harmonic distortion is bad.  I see it from the perspective of a professional musician... these are overtones.  They follow the musical overtone series.  (As opposed to inharmonic or non-harmonic distortion, such as clipping).  For instance, the second harmonic is the octave.  In the right balance, I bet many of you could not hear this distortion at all.  Even the third harmonic, (one of those 'bad' odd order harmonics) is in the musical overtone series a perfect fifth.  From both the fundamental and the second harmonic.  (Yes I know it is no longer a fifth compared to the fundamental, but those of you who know what I mean, you well, know what I mean!)  It is called perfect for a reason.  When tuning there will be no 'beats'.  The mathematical ratio is perfect.  

Elevated harmonic distortion is what makes tubes sound like they do, and the vintage transformers of those tube amps are part of the equation, too.  

GRANTED, there is a delicate balance to be found here.   Even if you like harmonic distortion, there is definitely too much of a 'good thing' at some point.  It can also make for difficult system matching.  You have to be careful with your components and how they interact.  

One thing I CAN say about this DSC2 design, is it doesn't have NEARLY the amount of harmonic distortion as some other transformer DAC designs, including some DSD ones, that cost 7 or 8 times as much.   And the DSC2 has a noise floor and dynamic range that is as good or better than many of the best designs on the market today.  Not gonna name any names, but again, there are some DACs out there purporting to do similar things at a much much higher price that don't come close to the measured performance of the DSC2.  

It is my personal opinion, that once you get over 90db SINAD, you are okay, 100db SINAD is better for sure.  The DSC2 in the area where our hearing is most sensitive, say, 1khz to 6khz, it gets well over the 90db SINAD mark, and once I get a better PSU, I think it can be coaxed closer to that 100db SINAD mark.  

At the same time I will say the balance of sound here is very, very good.  It isn't so warm as to be too 'tubey' sounding, nor is it dry, thin and boring.   Not by any means!   Nor it is 'rolled-off' in its sound.  It has excellent treble extension and detail, without going so far as to be too sibilant.  No, it will not cover up bad recordings.  It has plenty of resolving power to display what is really there.  Bad for the poor recordings, and truly great for the good ones.  

 

 

 

 

 

[skipspence]

12 hours ago, Andrew Allen Ballew said:

Being a transformer based solution

What output transformers are in your DSC2? 

[Andrew Allen Ballew]

8 hours ago, skipspence said:

What output transformers are in your DSC2? 

 

Well, the are not Lundahl.  I wish.  Chinese generic I believe.  Here are my internals. 

 

[skipspence]

10 hours ago, Andrew Allen Ballew said:

Chinese generic I believe

Not exactly these ones - the developer of DSC2 recommends Joyin Audio, also chinese, but... at least that's what I have in my DSC2, also tried to experiment with several Lundahls - no go, first wow, then disappointment like lot of distortion, Joyin are good.

[Miska]

That transformer type was selected based on it's properties.

[Andrew Allen Ballew]

On linearity.  

 

Yes, I am still hesitant to post too many measurements, by experience I have become acutely aware of how important clean power is.  I am currently running the DSC2 on a variable output transformer that is of unknown noise quality.   I will be more comfortable when the XPower 9 volt arrives.  Apparently it was out of stock on Amazon and is delayed to a Saturday delivery.  

However, I still have found some interesting things.   

The resolution of the DSC2 is exceptional.  Both the noise floor, and the linearity.  (Both of these are required for true high resolution.  Often we focus on the excellent in-band noise floor of DSD, but ignore the low level resolution that is exposed in a linearity test) 

No worries here. As I said, the linearity of the DSC2 is truly first class.  

 

All these measurements are at DSD256.  

At -100db, it is virtually perfect.  deviations are LESS than 0.1db.  Much less.  

At -110db, we are STILL at <0.1db deviation.   This beats out $30,000 plus DACs.  Like the one if you rearrange the letters a bit in DSC haha.  

At -115db, we finally hit some loss of resolution, with a deviation of an entire -0.5db.  Wow.  

-119db we finally have a deviation of -2.5db.



So with my measurements we have practically perfect 18 bits of resolution, still very, very linear to 19 bits of resolution.   Linearity doesn't quite make it to a 'perfect' 20 bits, but for a 1 bit converter this is exceptional.  Heck, for any converter it is impressive. 

Couple that with the actual noise floor that is lower than -120db, I cannot help but be impressed by the measured performance here.  

SINAD (THD+N) in my latest, better calibrated measurements stays in the mid 90db range, approaching 100db at times. 

 

Even deep into the low frequencies, it only drops once you get under 30hz.  Our ears are not as sensitive to the distortion that low anyway.  Once again, I am thinking of a $8000 DSD transformer DAC that does not come close to this level of performance.  

And on jitter... my final measurements at DSD256, with a 48khz base rate, (48k x 256 or 12,288,000hz),  I got a low result of a mere 18picosecond jitter.  


I will remeasure everything once I get the XPower 9V power supply.  I am not sure the numbers will change much; I do expect the FFT noise floor to be cleaner however, in my graphs.  

[Miska]

You could replace the transformer with analog stage similar to my original one. And DSC is still very simple one, I know how to make it much much better. But then it is not anymore something you could easily DIY.

 

[skipspence]

22 hours ago, Miska said:

know how to make it much much better

What exactly do you mean with this? 

Power supply, output components or something constructively different? 

[Miska]

15 hours ago, skipspence said:

What exactly do you mean with this? 

Power supply, output components or something constructively different? 

 

Modifications to the actual conversion stage.

 

[skipspence]

3 hours ago, Miska said:

Modifications to the actual conversion stage

Let me guess: FPGA? 

[Miska]

On 7/1/2023 at 6:15 PM, skipspence said:

Let me guess: FPGA? 

 

I rather not use FPGAs...

 

[skipspence]

19 hours ago, Miska said:

rather not use FPGAs

Going myself to test a new ladder DAC which converts each format up to DSD1024 onboard through built-in 2 FPGA (XILINX/ALTERA), that's why I asked. Is anyone familiar with Chen.grand DSDAC 1.0 btw?

 

[Currawong]

On 7/3/2023 at 11:54 PM, skipspence said:

Going myself to test a new ladder DAC which converts each format up to DSD1024 onboard through built-in 2 FPGA (XILINX/ALTERA), that's why I asked. Is anyone familiar with Chen.grand DSDAC 1.0 btw?

 

 

This? https://en.cen-grand.com/product/146.html