[b0bb]

Sorry, I'm going to have to disagree with you b0bb. Time and again ASR have demonstrated that they barely, if at all, actually listen to the review products. The golden example is the Uptone EtherREGEN reclocker. ASR found it to have zero effect. You have to be thoroughly cloth-eared not to hear the benefits of reclocking, reducing jitter and phase noise, by using an ER ahead of the LKS or any other DAC. Since that review, I have no time for them.

I like the objective data like SINAD, SNR, noise floor and the product comparison tables of these parameters.

Plenty of other places to go for subjective impressions.

 

[b0bb]

It's worth mentioning that in the 005 the FPCAPs bypassed by MKPs are everywhere, even the processor got one.
Values before the regulators are 270uF, 100uF after the regulators. Near the line driver I've noticed also only 100uF.

Additionally there are many inductors as a part of power supplies.

LKS also used inductors in the power supplies of the 004 though not as many.
These are small SMD inductors that can overload too easily.

If the current demand increases too suddenly it can saturate the cores in these inductors.
Noise filtering is lost when that happens.

 

[b0bb]

Topping seems to have done something different than what I recommend ESS to put the internal dacs in parallel, and it is an independent i / v conversion by each internal channel with multiple op-amp chips and do the parallel later, I don't know if this was done by others previously or is a novelty. Your data is spectacular in tests. What do you think B0bb of this method? even not being discrete op-amps.
Topping D90se
https://shenzhenaudio.com/products/...r-usb-iis-cpld-ultra-low-noise-preamp-decoder
Make 16 i / v converters for a dual mono version possibly nobody implements it or if? We will see

This technique allows the designer to use off the shelf parts with limited current handling capability.
Topping did nothing special, all they did was combine the analog outputs of the several low powered chip opamp I/Vs.
Gustard did similar on the X26 and X26Pro.

9038Pro has a built in ADC for automatic gain calibration of multiple I/V channels just for this use case. 004 does not use the ADC.

The big turn off for me is the use of ceramics in the audio path, reminds me of the horrible little brown circular disc caps on old audio equipment.

 

[piaseczek]

LKS also used inductors in the power supplies of the 004 though not as many.
These are small SMD inductors that can overload too easily.

If the current demand increases too suddenly it can saturate the cores in these inductors.
Noise filtering is lost when that happens.

Maybe these small inductors are a reason why Xoverman experienced sound problems with the 820uF FPCAP before the clock's regulator?

 

[b0bb]

Maybe these small inductors are a reason why Xoverman experienced sound problems with the 820uF FPCAP before the clock's regulator?

That might explain why adding a current limiting resistor resolved some of the issues.

LKS used the large 470uF Polymers in Gen1, they would have to adjust the inductor current carrying capacity for this.

 

[piaseczek]

That might explain why adding a current limiting resistor resolved some of the issues.

LKS used the large 470uF Polymers in Gen1, they would have to adjust the inductor current carrying capacity for this.

Even in the Gen 1 there is only 47uF before and after the crystek's regulator. Why such a small values were chosen?

 

[piaseczek]

Add resistors, diodes and caps

B0bb,
What is the purpose of the wima cap below the LT1763/LT1764?
I dont see any filter/set capacitor in the LT1763/LT1764's pdf.

Maybe it's the bypass for the big cap...

 

[piaseczek]

Caps values left to right 10,000uF 35V x2, 22,000uF 25V x1, 47,000uF 25V x1

B0bb,
Is the 680uF original output capacitor for the LT1763 ok and you won't replace it?

 

[Xoverman]

First, these are op amps, so DC current or/and voltage are fine. Don't worry.

Second, ESS DAC
each pin of complementary current output (D1 and D1b in Figure 1 of datasheet) has a current offset, that is a bias current.

The VCM (Vcom) bias voltage that is common to both I/V transimpedance amplifiers
(U1, U3 Figure 1 ESS datasheet) is used to cancel out these current offsets.
This is the job of VCM (Vcom) !!!
That is why the Vcom capacitor has such a profound influence on the DAC's sound.
That is why this Vcom voltage reference must be frequency independent.

A large Vcom capacitor causes these common bias currents to trickle downstream.
I have worked out the details, be happy to post if interested.


Now the numbers, from the datasheet:
At idle, ESS DAC has an "current mode offset" of ~9mA per pin. ~18mA total.
Multiply this idle current with Rf=68.1 Ohm I get 1.24V, which is in the ballpark
of VCM (Vcom) that I measured. With Rf=61.9 Ohm I get 1.13V which is a bit off.
So there might be something details that I am not aware of.


Third, resistor R6 is connected to VCM.
By superposition priniciple,
it is easy to see a VCM term emerges from DAC1,
given any ideal/perfect difference amp.
Why is VCM there ? My guess is: the complementary outputs D1, D1b may not be
perfectly balanced. VCM is connected to R6 just in case ???

This is as far as I want to go in this Vcom study, really enjoy the discussions. Next study is jitter....

Very nice work!!!

 

[piaseczek]

Another alternative to the wet tantalum, relatively small sized 100uF MKT:
https://pl.mouser.com/ProductDetail/EPCOS-TDK/B32524R0107K000?qs=rXlmONB1aiHOwKOLC/J4bA==

 

[b0bb]

Even in the Gen 1 there is only 47uF before and after the crystek's regulator. Why such a small values were chosen?

The stock CCHD-575 XO current consumption is low between 15-25mA, 47uF is adequate.
CCHD-950x consumes about the same amount of current, this makes it a straightforward swap.

Higher performance XOs like the SIT-5357 (65mA) and the OCXOs (150mA -1000mA) require more current, the filter capacitance has to be increased to keep the supply ripple and hence noise at the same level.

 

[b0bb]

B0bb,
What is the purpose of the wima cap below the LT1763/LT1764?
I dont see any filter/set capacitor in the LT1763/LT1764's pdf.

Maybe it's the bypass for the big cap...

Bypass for the input and output caps and voltage setting resistors.

150nF 100V Polypropylene (MKP)

 

[b0bb]

B0bb,
Is the 680uF original output capacitor for the LT1763 ok and you won't replace it?

Optional, it is for the XO.

Keep the stock unless you have something like an OCXO requiring higher supply current.

 

[tdrood]

The stock CCHD-575 XO current consumption is low between 15-25mA, 47uF is adequate.
CCHD-950x consumes about the same amount of current, this makes it a straightforward swap.

Higher performance XOs like the SIT-5357 (65mA) and the OCXOs (150mA -1000mA) require more current, the filter capacitance has to be increased to keep the supply ripple and hence noise at the same level.

I have an SIT in mine, what do you recommend?

 

[b0bb]

I have an SIT in mine, what do you recommend?

I used 100uF on the output side and 820uF on the input side.
You may have to experiment with the value on the input side try values between 100uF and 820uF

I used Nichicon FPCAP RNL, these have very high pulse current capability.
https://www.nichicon.co.jp/english/products/pdfs/e-rnl.pdf

The larger values may pull in too much current on recharge causing the small inductor LKS used to overload.
If that happens you may need to reduce the size of the cap on the input side.