Measurements & Preview of Holo Audio Spring 3 DAC / wPreamp (Prototype)

[Music1969]

As I have not seen measurement results for Spring 3 L3 anywhere, I'm just curious about the jitter levels. Here's results for my L2 which are good enough for me.

View attachment 151603

Any measurement differences from Spring2 L2 ?

Or both much the same? i.e. very good

 

[sq225917]

Same same

 

[gvl]

Anyone has any theories how they they can get this level of performance on a custom design with discrete components? This is better than best of the R2R chips of the past that were hitting the limits of technology and those were made in a controlled environment.

 

[Veri]

Anyone has any theories how they they can get this level of performance on a custom design with discrete components? This is better than best of the R2R chips of the past that were hitting the limits of technology and those were made in a controlled environment.

Sign magnitude-like R2R with over-engineered error correction, over-engineered output stage with over-engineered capacitors and power supply (holo is BTW not fully discrete; to compare to soekris for example which is "no" or very little feedback into OPA1656. holo audio uses all tricks to get such superlative DAC performance which include feedback and high-performance op-amps(which I don't know)). The first holo spring was already quite good, I suppose due to the financial success they could put money where their mouth is and improve, and improve yet again in spring 2 and 3(trickle-down from the MAY statement DAC).

That to say R2R does not equal "fully discrete" or other such things, to get this good of performance it would need all the technological marvel accessible. Put a tube anywhere in there and the performance will get drastically worse lol from what I gather the spring 3 only has NOS mode though? No more oversampling chip. And actually, makes sense since it was an AKM chip. RIP that factory...

 

[gvl]

I meant discrete ladder only. Perhaps there’s also hand binning of resistors that is taking place prior to assembly. I suppose this shows that the cost is no object approach can sometimes yield positive results

 

[gvl]

from what I gather the spring 3 only has NOS mode though?

Really? So external filtering is the only way?

 

[Miska]

The Spring 3 L3 is said to be equiped with the 'Titanis USB Turbo' module.

If someone wants to donate me one, I can check it out. But as long as I pay for the DAC myself, I need to weight the cost / benefit since I don't have unlimited budget... So far, I can happily live with this.

 

[Miska]

Anyone has any theories how they they can get this level of performance on a custom design with discrete components?

Precisely because of that reason. No need to cram everything on a single small piece of silicon that needs to have selling price around $10/piece.

 

[Miska]

Any measurement differences from Spring2 L2 ?

Or both much the same? i.e. very good

Both are very good, but there are slight differences so it is clear that they are not identical designs.

My favorite feature of Spring 3 is the low power standby mode with display off and cool device.

 

[gvl]

Precisely because of that reason. No need to cram everything on a single small piece of silicon that needs to have selling price around $10/piece.

Soekris couldn’t for some reason. Perhaps it’s because of his output stage with limited negative feedback as Veri speculated above.

 

[Le Concombre]

Sign magnitude-like R2R with over-engineered error correction, over-engineered output stage with over-engineered capacitors and power supply (holo is BTW not fully discrete; to compare to soekris for example which is "no" or very little feedback into OPA1656. holo audio uses all tricks to get such superlative DAC performance which include feedback and high-performance op-amps(which I don't know)). The first holo spring was already quite good, I suppose due to the financial success they could put money where their mouth is and improve, and improve yet again in spring 2 and 3(trickle-down from the MAY statement DAC).

That to say R2R does not equal "fully discrete" or other such things, to get this good of performance it would need all the technological marvel accessible. Put a tube anywhere in there and the performance will get drastically worse lol from what I gather the spring 3 only has NOS mode though? No more oversampling chip. And actually, makes sense since it was an AKM chip. RIP that factory...

correct ; no OS mode with the S3

 

[Le Concombre]

I don't have any experience witht the Titanis. FWIW here's what it does according to Kistune:

The two main functions of Titanis.

1. Regeneration of USB power. Power from the computer USB ports are very noisy with a level higher than several mV. Titanis inverts and processes power from the USB source. The regenerated power noise level is only 8 uV, rendering the Noise level reduced by about 1/1000th.

1. Rebuilding the signal from the USB port. In the computer, the signal consists of a large amount of interference and jitter due to the complicated EM environment and power supply conditions. Titanis receives the signal from USB and puts it in a local buffer, with the use of a local high quality clock, regenerates the USB signal and sends it to output.

The Spring 3 L3 is said to be equiped with the 'Titanis USB Turbo' module.

FWIW I used a Uptone Green USB Regen with my TEAC 501 and I got rid of it for it took of some of the immediacy that makes the S3 so good IMO, it rounded the sound and since I have optimized filters for so called Room Correction I don't need any sweetener. I also replaced my Supra cable by the Holo branded provided USB cable ; can't say I spent much time comparing but if there are differences they are subtle and I prefer to sell the Supra and use the Holo straight from the NUC (ah yes I use HQP in the NAA setting) with the provided cable

 

[Kenmoreira85]

Yeah loving the spring 3 kte. Especially with hqplayer. Got my d90 and usbridge player for sale atm

 

[boXem]

Sign magnitude-like R2R with over-engineered error correction, over-engineered output stage with over-engineered capacitors and power supply (holo is BTW not fully discrete; to compare to soekris for example which is "no" or very little feedback into OPA1656. holo audio uses all tricks to get such superlative DAC performance which include feedback and high-performance op-amps(which I don't know)). The first holo spring was already quite good, I suppose due to the financial success they could put money where their mouth is and improve, and improve yet again in spring 2 and 3(trickle-down from the MAY statement DAC).

Feedback? How so?

 

[Veri]

Feedback? How so?

It uses a highly linear output stage on DAC side. Why would you want NFB myths/nonsense?
Pre-amp module is probably fully "Class A" discrete as is fashionable these days..

 

[sq225917]

Do -120db+ zero negative feedback, discrete class A circuits really exist?

 

[boXem]

It uses a highly linear output stage on DAC side. Why would you want NFB myths/nonsense?
Pre-amp module is probably fully "Class A" discrete as is fashionable these days..

The question is on the technical feasibility, not on what I want
I do not understand how to do feedback on the ladder stage.

 

[sq225917]

You don't, you do it on the output stage.

 

[boXem]

You don't, you do it on the output stage.

That will not bring down the distortion components below - 125 dB on an r2r DAC.

 

[Miska]

correct ; no OS mode with the S3

Yeah, my measurements are with external OS... (with HQPlayer)