[b0bb]

Keep the following in mind when considering the topic of noise in your specific system

1. If you use speakers, what is the ambient noise level?
2. Can you hear noise from your system with no inputs under the following situations
   1. Listening position
   2. Up close at the speaker

On the 004 I find that the noise floor in the recording itself predominates even with headphones.

 

[b0bb]

IMD is not a new phenomenon, it was first described by Matti Otala in 1977 as Transient Intermodulation distortion in power amps.

A modern chip opamp is defined as such:

• Stupidly high open loop gain, 60dB or higher, often as high 100dB.
• To get to the desired overall gain apply an equivalent idiotic high amount of feedback.
• Example to get 10dB of gain from an opamp with 60dB open loop gain, apply 50dB feedback (60-50 =10 )

Problems start to happen when

• The opamp output stage cannot keep up with signal changes from the voltage amplification (VAS) section
• This is called slew rate limiting,
  • the more output current the worse it becomes
  • the higher frequency/bandwidth the worse it becomes
• Net result is massive signal overshoots due to the delay between input and feedback
• At this point the whole house of cards comes crashing down

Why are we seeing it now ?

• The 9038 runs very high current (100mA vs 5mA)
• 9038 supports DSD512, max 25MHz sampling, 50MHz I/V bandwidth.
  • 44.1k CD Redbook in contrast requires 1.4MHz sampling , 2.8MHz I/V bandwidth
  • 18x increase in bandwidth.
• This causes the small output stages in chip opamps to slew rate limit

Why is it different with the 004 ?

• Discrete opamp, lower open loop gain
• Corresponding lower amounts of feedback
• Very big output transistors
  • Handles the large current
  • The devices have very wide bandwidth, unity gain at 50MHz <-- could be be better

Here is the link to the Otala IEEE Paper
http://hifisonix.com/wordpress/wp-content/uploads/2017/10/The-Theory-of-TIM-Matti-Otala.pdf

Read the abstract and conclusion if the math is too complicated to follow.

 

[ESL57]

Keep the following in mind when considering the topic of noise in your specific system

1. If you use speakers, what is the ambient noise level?
2. Can you hear noise from your system with no inputs under the following situations
   1. Listening position
   2. Up close at the speaker

On the 004 I find that the noise floor in the recording itself predominates even with headphones.

I have about 10 pairs of speakers in my collection (I like vintage ones), by far the most sensitive I have are Altec Lansing model 19 over 101db at 1m, their large horns detect the slightest noise inside the sound chain, Until I fix the problem in DSD, I couldn't put this DAC with the Altecs.
By the way B0bb, what are the smd op amps that our 004 have?. They are of quality or something better to put there?

 

[b0bb]

I have about 10 pairs of speakers in my collection (I like vintage ones), by far the most sensitive I have are Altec Lansing model 19 over 101db at 1m, their large horns detect the slightest noise inside the sound chain, Until I fix the problem in DSD, I couldn't put this DAC with the Altecs.
By the way Bobb, what are the smd op amps that our 004 have?. They are of quality or something better to put there?

004 does not use SMD opamps in the I/V.
Your noise problems are related to the 1.2V regulator and/or the messy application of thermal glue and not the native noise floor of the 004.

Your DAC is a Gen 1.5x where the cost cutting has begun, notice the use of cheap ceramic instead of a proper Aluminum Polymer.
LKS started using the low cost, low power 74AC244 drivers so that they could get away with the cheap ceramic substitution.

If you want to further tackle the noise issue further, start by reversing these cheap shortcuts LKS took, replace the 74AC244 with the 74ALVC244 from Nexperia.
74ALVC244 has superior input signal conditioning and drive capability, very useful for the long copper traces on the large 004 board.

 

[b0bb]

Topics like the DAC noisefloor and IMD are interesting academic topics and may not be audible.

Below are things that audibly improve the DAC performance

• Replace CCHD-575 XO
• Replacing the PET/Mylar caps in the output filter, the blue boxes.
• Low TCR I/V resistor, a bigger high TCR (50-200ppm) resistor only goes part of the way
• Vcom bypass with highvalue Wet Tantalum caps
• Ferrite noise absorber sheets on the auxiliary supplies.
• Fix poor decoupling of ground and powerplanes on Gen2s
• Improve current delivery, especially the handling of load transients

These should be completed before spending time worrying about the effects of noisefloor or IMD

 

[ESL57]

004 does not use SMD opamps in the I/V

Why is it different with the 004 ?

004 does not use SMD opamps in the I/V.
Your noise problems are related to the 1.2V regulator and/or the messy application of thermal glue and not the native noise floor of the 004

Why is it different with the 004 ?
Discrete opamp, lower open loop gain

I mean what model and reference are those 5-pin discrete opamp smd that have 6 holes around to maybe replace them with similar through-holes, I think I remember that long ago I investigated the code on the body (X L) and what I found most similar was a double transistor that did not I remember the reference.

 

[b0bb]

Why is it different with the 004 ?

Why is it different with the 004 ?
Discrete opamp, lower open loop gain

I mean what model and reference are those 5-pin discrete opamp smd that have 6 holes around to maybe replace them with similar through-holes, I think I remember that long ago I investigated the code on the body and what I found most similar was a double transistor that did not I remember the reference.

Not sure what you are referring to, opamp on 004 is discrete.
The thing with 6 holes around is the Jfet diffrential input transistor pair.

 

[ESL57]

Not sure what you are referring to, opamp on 004 is discrete.
The thing with 6 holes around is the Jfet diffrential input transistor pair.

Yes, that was what I meant

 

[ESL57]

004 no utiliza amplificadores operacionales SMD en I / V.
Sus problemas de ruido están relacionados con el regulador de 1.2V y / o la aplicación desordenada de pegamento térmico y no con el piso de ruido nativo del 004.

Su DAC es un Gen 1.5x donde ha comenzado la reducción de costos, observe el uso de cerámica barata en lugar de un polímero de aluminio adecuado.
LKS puso en marcha los controladores 74AC244 de bajo coste y baja potencia para que pudieran salirse con la suya con la barata sustitución de cerámica.

Si desea abordar aún más el problema del ruido, comience por revertir estos atajos baratos que tomó LKS, reemplace el 74AC244 con el 74ALCV244 de Nexperia.
74ALVC244 tiene un acondicionamiento de señal de entrada superior y una capacidad de accionamiento, muy útil para las largas trazas de cobre en la placa 004 grande.

004 does not use SMD opamps in the I/V.
Your noise problems are related to the 1.2V regulator and/or the messy application of thermal glue and not the native noise floor of the 004.

Your DAC is a Gen 1.5x where the cost cutting has begun, notice the use of cheap ceramic instead of a proper Aluminum Polymer.
LKS started the low cost, low power 74AC244 drivers so that they could get away with the cheap ceramic substitution.

If you want to further tackle the noise issue further, start by reversing these cheap shortcuts LKS took, replace the 74AC244 with the 74ALCV244 from Nexperia.
74ALVC244 has superior input signal conditioning and drive capability, very useful for the long copper traces on the large 004 board.

I also followed his advice with the 74..244, 74ALCV244 Nexperia instaled and polymer caps.

 

[b0bb]

I also followed his advice with the 74..244, 74ALCV244 Nexperia instaled and polymer caps.

Did you replace the ceramics in the other locations like

• I2S receiver
• SPDIF receiver
• USB isolator
• Microcontroller

 

[Xoverman]

004 does not use SMD opamps in the I/V.
Your noise problems are related to the 1.2V regulator and/or the messy application of thermal glue and not the native noise floor of the 004.

Your DAC is a Gen 1.5x where the cost cutting has begun, notice the use of cheap ceramic instead of a proper Aluminum Polymer.
LKS started using the low cost, low power 74AC244 drivers so that they could get away with the cheap ceramic substitution.

If you want to further tackle the noise issue further, start by reversing these cheap shortcuts LKS took, replace the 74AC244 with the 74ALVC244 from Nexperia.
74ALVC244 has superior input signal conditioning and drive capability, very useful for the long copper traces on the large 004 board.

Hi B0bb,

I tried this approach three weeks ago. It did not work out for me. I exchanged both AC244 against 74ALVC244 and added 820µF 16V Polymer Caps.
The DAC lost all of its micro details. I build everything back, and the micro details came back.

I think the Problem is, that the line driver has to drive against those 4 ceramic caps next to the most right 244.
But during these tests I did have success with getting rid of smeared middle tone. The HDMI receiver chip is supplied from the same regulator as both AC244 line drivers.

I added a 100µF 16 Polymer cap, and exchanged the ferrit bead against a small 5µH 0.2 Ohm choke I took from a broken hard drive.

 

[b0bb]

Hi B0bb,

I tried this approach three weeks ago. It did not work out for me. I exchanged both AC244 against 74ALVC244 and added 820µF 16V Polymer Caps.
The DAC lost all of its micro details. I build everything back, and the micro details came back.

You made more than 1 change at a time

• Do you know which change caused the problem?
• Was it the 74ALVC244 or the 820uF caps?
• Did you cover any of the signal traces going to the 9038's with ferrite sheets?

A couple more questions comments.

• Did you complete the removal of the blue PET/Mylar caps [1] in the picture? They cause a lot of transient smearing.
• A weaker line driver (74AC244) will make the top end more pronounced, balancing out the negative effects of the blue PET caps.
• Makes it hard to get more smoothness from the DAC later on if blue caps are not eliminated.
• One of the blue caps is damaged.
• Area [2] has up 50MHz running thru it, has it been cleaned up? High frequency response of the DAC will be reduced if you don't.
• You should not be relying on the side effect of the weaker driver to compensate.

Did you remove this resistor, your analog output filter is quite different/changed if you did.
Did you do this to change the sound?

 

[MartinWT]

• Replace CCHD-575 XO
• Replacing the PET/Mylar caps in the output filter, the blue boxes.
• Low TCR I/V resistor, a bigger high TCR (50-200ppm) resistor only goes part of the way
• Vcom bypass with highvalue Wet Tantalum caps
• Ferrite noise absorber sheets on the auxiliary supplies.
• Fix poor decoupling of ground and powerplanes on Gen2s
• Improve current delivery, especially the handling of load transients

These two are next on my list to do. I have the parts already.

 

[ESL57]

Did you replace the ceramics in the other locations like

• I2S receiver
• SPDIF receiver
• USB isolator
• Microcontroller

I still need to change them in some of those circuits, currently enjoying the exuberant music that comes out of this dac, I made many changes at the same time and this is not a good habit to know if something is a step forward or backward. The problem is also having a good ear memory, sometimes the changes take time to establish, time passes and your auditory memory must be good to see what happened, I will have to buy another original used GEN1 without touching anything, so the I will be able to compare in direct AB mode, for me a 004 of your second hand below € 700 is a bargain of the hifi, that if, only for electronic handymen, abstain those who do not know anything, when Sabre (ESS) gets a better chip contrasted, I will sell or buy them the same dac that B0bb and Xoverman acquire, the game could continue in the future .
By the way, in the firmwares after mine, LKS put memory of the independent DPLL number for PCM and DSD?, it seems that it already did it in 005, it is a drag to have to put 1 in PCM and 6 in DSD and be changing.

 

[Xoverman]

I still need to change them in some of those circuits, currently enjoying the exuberant music that comes out of this dac, I made many changes at the same time and this is not a good habit to know if something is a step forward or backward. The problem is also having a good ear memory, sometimes the changes take time to establish, time passes and your auditory memory must be good to see what happened, I will have to buy another original used GEN1 without touching anything, so the I will be able to compare in direct AB mode, for me a 004 of your second hand below € 700 is a bargain of the hifi, that if, only for electronic handymen, abstain those who do not know anything, when Sabre (ESS) gets a better chip contrasted, I will sell or buy them the same dac that B0bb and Xoverman acquire, the game could continue in the future .
By the way, in the firmwares after mine, LKS put memory of the independent DPLL number for PCM and DSD?, it seems that it already did it in 005, it is a drag to have to put 1 in PCM and 6 in DSD and be changing.

By the way, in the firmwares after mine, LKS put memory of the independent DPLL number for PCM and DSD?, it seems that it already did it in 005, it is a drag to have to put 1 in PCM and 6 in DSD and be changing.

Wher did you get this information ?

@b0bb Did you hear anything about 2 memorys ( registers ) in the 004 ?