[Tomus4]

   
Leave the LT1763 alone, the low noise specs are only guaranteed if the layout , especially the ground plane is done to the manufacturer's specs, there is not enough space on the board to put in another  chip without compromising this requirement.
 
Layout, not datasheet specs determines final performance, please keep this in mind when reading the datasheets.
 
A hacked in regulator is about the worst thing you can do to the DAC, it compromises the noise performance. In the example below the 3/4 inch piece of the lead on the transistor is enough to act as a radio antenna, this injects noise into the regulator. In an area with good FM/AM radio reception this is 300uV or more (This is how simple germanium diode AM radio receivers can function).
 
More importantly it adds unwanted distance between the regulator and the load, this degrades the load transient recovery performance of the regulator.
Transient recovery is the ability of the regulator to quickly recover to the set voltage in the presence of a sudden increase or decrease in load current, which happens when digital logic changes state.
 
 
 

What do you thing about creating supply for clock and D/A converters based on this http://www.ti.com/lit/ds/symlink/tps7a87.pdf?  I'm thinking about removing all LT1763 and stuff around and creating small supply boards which will be connected just before clock and converters.

 

[fjc36]

Bobb thinks it is bad to have long legs of the regulator for good reason. In most of the electronic circuits, the regulators, filter diodes, specially  power transistors are stand tall on the board. Should they be shortened to avoid antenna? I do not think it is possible for shortening all of them, specially for the power transistorsl. What is a solution? shielding?  Or, in low freq. domain, analogy circuitry, it doesn't count.
 
How was the sonic effect of the LKS MH-DA003 mk II (use CSS 9028pro chip) ? One of the reader has purchased it, Please let us know.
 
One thing new, LKS might have the Dual 9038pro  debut before the end of 2017. I am very curious about it. Certainly, some of the implement  of 9038pro DAC already comes out, i.e. OPPO and others, but I believe they are not dual.  

 

[b0bb]

  Bobb thinks it is bad to have long legs of the regulator for good reason. In most of the electronic circuits, the regulators, filter diodes, specially  power transistors are stand tall on the board. Should they be shortened to avoid antenna? I do not think it is possible for shortening all of them, specially for the power transistorsl. What is a solution? shielding?  Or, in low freq. domain, analogy circuitry, it doesn't count.
 

 
LKS has solved all of these issues, so when you mod the DAC please be careful not to destroy the work they have already done.
 
The LT1763 regulators connected to the DAC with very large copper areas for the ground, this is the ground plane and this is the shield you mention above. (The area is 20-50x the size of the protected area)
 
This stops the majority of the RF/Radio pickup and suppresses the noise from ground return current coming from the 9018 chip itself.
Unfortunately to get to the low noise numbers (20uV) the ground plane is customized for the specific layout and chip, this makes the changing out the regulators very complicated and you mentioned some of them (stand tall etc).
 
An external board mounted 90 degrees to the ground plane reduces its shielding effectiveness to zero.
 
Power transistors are not needed on the digital side as the max current is 300mA per rail, the LT1763 supplies up to 500mA, the eliminates power transistors from consideration.
 
You are correct about the analog regulators, the considerations discussed here do not apply.
 
One large ground continuous plane provides a very low impedance path back to the power supply at high/RF (>1MHz) frequencies.
 
This is very important for the LKS as they connected both the digital and analog 3.3V supplies together and supplied this from one regulator.
 
This works as long as the ground plane sinks/absorbs this current and prevent the noisy ground return current on the digital side from going into the analog side.
 
When you replace the LT1763 with an external board, the ground plane is no longer continuous but 2 ground planes, original on DAC motherboard and new ground plane on regulator  connected by wire.
 
RF noise current can flow back and forth between these 2 grounds much like an audio hum loop resulting from multiple grounds, in the RF case much worse as it has alternate paths like coupling thru the air.
 
This is quite nasty.
 
The noisy digital ground return current will also start getting into the analog side of the DAC affecting the sonic performance.

 

[abartels]

As always, great and useful explanation b0bb!

 

[penguin69]

Well,
 
Good news and bad news from my latest round of upgrading.
 
The good news is that my replacement Vishay metal foil resistors arrived today from Texas Components. (The first batch had got lost.)
 
Also, I was able to get my rework station working again, meaning a cleaner removal of parts today.
 
The new 0.005% tolerance resistors are installed. See pic below.
 


On restoring the circuit board, I found that the MH-DA003 unit would not play music.
 
I noticed that the Vdd wire to the Crystek 950 had come away, so I resoldered that, thinking that this would solve the issue. But it did not.
 
My first thought was to revisit my resistor handiwork, perhaps by restoring one or two of the old resistors. However, before trying this I would like to hear a view on the following items that make me suspect that something may not quite be right now with the Amanero board.

a) My Mac Mini is the transport for the DAC, running Audirvana+. When I start A+ up, the Mac OS now deselects the Amanero 384 interface for output and selects Internal Speakers instead. I have never seen this behaviour before. (FWIW, the MH-DA003 now permanently shows 44K on the display with USB selected.)
 
b) My optical input seems to remain in UNLOCK mode. I have not thoroughly tested it, I just tried a source which previously has worked.
 
As an aside, all the usual motherboard lights come on when the unit is powered up 
 
 
So the questions I have are as follows:
 
1) Do the symptoms described indeed suggest a problem with the USB interface. The Mac Mini recognises the interface for what is (Amanero 384) so it can't all be bad.
 
2) If this is a USB interface problem, what are the best diagnostic to run?
 
3) If this is not a USB interface problem, where should I start?
 
 
Thanks in advance for any help provided. (And thanks again to b0bb for the sage advice to only make one mod at a time. I was going to do more today, but I am glad that I did not as it would have made my problems harder to decode.)

 

[b0bb]

  Well,
 
Good news and bad news from my latest round of upgrading.
 
The good news is that my replacement Vishay metal foil resistors arrived today from Texas Components. (The first batch had got lost.)
 
Also, I was able to get my rework station working again, meaning a cleaner removal of parts today.
 
The new 0.005% tolerance resistors are installed. See pic below.
 


On restoring the circuit board, I found that the MH-DA003 unit would not play music.
 
I noticed that the Vdd wire to the Crystek 950 had come away, so I resoldered that, thinking that this would solve the issue. But it did not.
 
My first thought was to revisit my resistor handiwork, perhaps by restoring one or two of the old resistors. However, before trying this I would like to hear a view on the following items that make me suspect that something may not quite be right now with the Amanero board.

a) My Mac Mini is the transport for the DAC, running Audirvana+. When I start A+ up, the Mac OS now deselects the Amanero 384 interface for output and selects Internal Speakers instead. I have never seen this behaviour before. (FWIW, the MH-DA003 now permanently shows 44K on the display with USB selected.)
 
b) My optical input seems to remain in UNLOCK mode. I have not thoroughly tested it, I just tried a source which previously has worked.
 
As an aside, all the usual motherboard lights come on when the unit is powered up 
 
 
So the questions I have are as follows:
 
1) Do the symptoms described indeed suggest a problem with the USB interface. The Mac Mini recognises the interface for what is (Amanero 384) so it can't all be bad.
 
2) If this is a USB interface problem, what are the best diagnostic to run?
 
3) If this is not a USB interface problem, where should I start?
 
 
Thanks in advance for any help provided. (And thanks again to b0bb for the sage advice to only make one mod at a time. I was going to do more today, but I am glad that I did not as it would have made my problems harder to decode.)

 
Check to see the soldering on the XO is still good, did you use a socket for the XO ?
Post close up photos of the XO, adapter etc to show the condition of the XO.
 
The blue LEDs closest to the DAC turn off when there is signal lock.
 
The original I/V resistors on the other side of the board have to be removed before adding the Vishays.
 
Measure the voltage output of the primary regulators for the digital logic

 
Output voltage appears on  the metal tabs in the picture above
 
The Amanero might be faulty, remove it from the DAC and plug it into the Mac and see if the behavior changes.
 
The stock Amanero USB board is powered by the USB bus.
 
Not too expensive to replace.

 

[penguin69]

Thanks, b0bb.
 
Yes, I removed the original I/V resistors using the rework tool.
 
The crystek 950 sits on a rhea board, which in its turn sits on a socket allowing interchange of xo's. The awkward part of this is that my DAC already had a regulator 'mod' which meant I had to disable one of the pins on the DIP socket so that I could get the voltage feed from the regulator mod. But since the xo only uses three pins, I would be surprised if I have messed up anything here. I will double-check nonetheless. This is an older pic prior to a few mods, including the addition of the DIP board. The MH-DA003 I acquired had the 575 xo sat out of picture, on the extra regulator board which can just be seen in the bottom right corner. I could always remove this extra board and just take the original voltage source from close to the xo socket, but it seems to have worked ok prior to yesterday.
 

 
Anyway, I will do some testing later today and tomorrow.
 
If the USB board is the issue, I'm not too fussed as I was going to buy the new version anyway.
 
Thanks as ever b0bb. I'm amazed I've even got this far, and without the mods you have suggested I would have sold the DAC on long ago due to the average stock SQ (IMO). 

 

[dacusb]

Hi b0bb,
 
I've been trying to contact Volent HK via email to enquire about the Amanero USB power supply, but have not received any feedback from them at all. I am wondering whether the said PS has been discontinued.
Can you suggest any other boards as good replacements for Volent USB PS and what would be the mods needed?
Thanking you

 

[penguin69]

Update on my side.
 
One of the aftermarket regulators had come loose. (b0bb's tip about the blue LED's and locking made me notice that one was not on). The soldering had broken away. After some awkward resoldering, I managed to restore power.
 
However, the orignal problem of no lock (and thus no sound) remains. I resoldered the Crystek 950X in case that was the problem. No joy. 
 
It's almost certainly not the USB board, as I couldn't get any success with the digital coax input either.
 
I'm close to ordering another unit.
 
I will maybe first order another Crystek XO and see how I get on.
 
Frustrating, after getting so far with all previous modifications.

 

[Tomus4]

   
The DC supply voltage is below 30V, this within acceptable limits for the Burson, contact them about the failures to get replacements.
 
The blue trimpot on the left channel may be faulty, given the sudden jump in output you are seeing, however 1.5mV is not going to cause any DC problems down stream.
I do not expect that you will be able to reduce the offset any further on the left channel without component replacement.

Alex from Burson wrote:
I am not sure about that. But this particular DAC chip is outputting excessive DC.
 
Cheers,
 
Alex 
 




On 24 Feb 2017, at 4:21 AM, wrote:
 

 



Hi,  
Do you think that problem is with this specific item oraz LKS Audio MH-DA003 generaly?
 
Best regards
Tomasz



   



2017-02-23 1:21 GMT+01:00 Burson Sales <info@bursonaudio.com>:

 



Hi Tomasz,  
Based on your measurement below. our V5 and the audio imagery opamp are both not to blame. 
 
The problem is with the DAC chip.  This DAC chip is outputs excessive DC and it will burn up any field effect transistor designs.  In fact, this DAC places any down stream ampifier and speakers at risk if the DC continue to increase over time.
 
We will replace the defective V5-Dual for you.  That's not a problem. However, please don't use our V5D in this DAC anymore.  The customer can try our V5i-D which has higher tolerance for DC or he can use any opamps that are not based on Field Effect Transistors. (FET)
 
Best regards,
 
Alex
 
 



 



  Burson Audio| www.BursonAudio.com | info@bursonaudio.com 
Supreme Sound Audio| www.SSAudio.com.au | opamp@bursonaudio.com
Follow Burson on Facebook and Twitter!
 




 



On Thu, Feb 23, 2017 at 1:30 AM, Toma wrote:

 



Hi,  
I have found time today and made folowing measures:
 
1) Model 994Enh-Ticha  I/V, default output :
- Left channel
              6,0mV no music, music playback up to 50mV between pins 1 and 3 
              3,1mV no music, music playback up to 500mV (mostly 50..250) between pins 2 and 3 

- Right channel. 
              1,2mV - no music, music playback up to 50mV between pins 1 and 3 
              0,2mV - no music, music playback up to 500mV (mostly 50..250) between  pins 2 and 3 

 
2)  Model 994Enh-Ticha I/V, default output - set up DC was made
- Left channel
              0,4mV no music, music playback up to 50mV between pins 1 and 3 
              3,8mV no music, music playback up to 350mV (mostly 50..250) between pins 2 and 3 

- Right channel. 
              1,5mV - no music, music playback up to 50mV between pins 1 and 3 
              0,1mV - no music, music playback up to 350mV (mostly 50..250) between  pins 2 and 3


 
3) default I/V, V5 output -
- Left channel
              9,2 mV no music between pins 1 and 3 
              3,9mV no music between pins 2 and 3 

 



4) default I/V, V5 output - set up DC was made
- Left channel
              0,4mV no music, music playback up to 50mV between pins 1 and 3 
              3,9mV no music, music playback up to 390mV (mostly 50..250) between pins 2 and 3
 
5) V5 I/V, default output 
- Left channel
              6,5 mV no music,  between pins 1 and 3 
              5,2mV no music,  between pins 2 and 3 





 
6) V5 I/V, default output - set up DC was made
- Left channel
              0,1 mV no music, music playback up to 50mV between pins 1 and 3 
              5,2mV no music, music playback up to 350mV (mostly 50..250) between pins 2 and 3



 
7) Model 994Enh-Ticha  I/V, V5 output :
- Left channel
              17,5mV no music, music playback up to 40mV between pins 1 and 3 
              3,8mV no music, music playback up to 350mV (mostly 50..250) between pins 2 and 3 

 

8)  Model 994Enh-Ticha I/V, V5 output - set up DC was made
- Left channel
              0,4mV no music, music playback up to 50mV between pins 1 and 3 
              3,8mV no music, music playback up to 350mV (mostly 50..250) between pins 2 and 3 

 



When music playback  there is no matter between left and richt chanel and no diiference between opamps.
Te maximum value of DC was once per 20..30s playing
 
Best Regards
Tomasz



   



2017-02-17 1:07 GMT+01:00 Burson Sales <info@bursonaudio.com>:

 



Hi Tomasz,  
Different opamps have different tolerance on DC.  Some IC opamps will simply amplify DC and pass it down to the next component.  This sometimes can result in destorying the power amp or the speaker drive. 
 
Field Effect Transistor FET Opamps such as the V5 does not pass any DC down to the next component.  Instead it converts that DC into heat.   In situations of excessive DC. It can heat up the components beyond it's tolerance levels and damage it.   
 
No opamps IC or discrete can tell you how much input DC they can tolerate.  This is because DC fluctuates during music playback (that's why I asked you to measure DC during playback).  And heat builds up over time so time is a big factor in DC tolerance.   
 
All I can say is that with no music playback if the DC is 4mV then you can expect a lot of DC during music playback which in term heats up that V5 in that channel. 
 
So, if the customer whats to use the V5, then he needs to consider these options below. 
 


 



* Swap places between the V5 and the Sparko (Imagry) so the V5 goes to the upstream and outputs to the Sparko.  
* Or use the default opamp in the place of the Sparko, or 
* Use all V5 opamps in all 4 positions, or
* Use our V5i opamps which can handle DC better. 
 
Best regards,
 
Alex
 
 

 

[b0bb]

  Alex from Burson wrote:
I am not sure about that. But this particular DAC chip is outputting excessive DC.
 
Cheers,
 
Alex

 
ES9018 DC level with no signal is 1.65V
 
As this voltage is present on both opamp input terminals this is known as the common mode voltage.
 
This parameter is documented for the Ticha 994 and the stock OPA1612
 
Ticha994: Handles up to 12V
http://www.sonicimagerylabs.com/products/product_images_docs/994Enh_Discrete_HD_OpAmp/994Enh_DiscreteOpAmp_Datasheet.pdf

 
OPA1612: Handles up to 13V (15V-2)
http://www.ti.com/lit/ds/symlink/opa1612.pdf

 
I think you should discontinue the use of the V5D as it is not able to handle a common mode voltage of 0.004V
Burson did not supply this info on the V5 datasheet.

 

[b0bb]

  Hi b0bb,
 
I've been trying to contact Volent HK via email to enquire about the Amanero USB power supply, but have not received any feedback from them at all. I am wondering whether the said PS has been discontinued.
Can you suggest any other boards as good replacements for Volent USB PS and what would be the mods needed?
Thanking you

 
Other vendors are selling it on Ebay.
 
This one is from Shenzen Audio
http://www.ebay.com/itm/112189661866

 

[b0bb]

However, the orignal problem of no lock (and thus no sound) remains. I resoldered the Crystek 950X in case that was the problem. No joy. 
 

 

 
The pins on the XO socket are not coming out on the other side, if they are too short you may not have full contact on every layer of the PCB.
You may have a hidden dry solder joint in there.
 
Notice the pins for the opamp socket is clearly visible
 
Alternatively buy or borrow an oscilloscope of at least 500MHz bandwidth and see if there is a clock pulse train coming out of the XO.
(Some of the old analog scopes might be had for cheap if you look around)
 
Basically if there is no XO signal, there will be no lock.

 

[dacusb]

Hi b0bb,
 
Thank you very much for your help.
I will immediately place order for the PS board, the Antek transformer and all the parts to be swapped.

 

[penguin69]

   

 
The pins on the XO socket are not coming out on the other side, if they are too short you may not have full contact on every layer of the PCB.
You may have a hidden dry solder joint in there.
 
Notice the pins for the opamp socket is clearly visible
 
Alternatively buy or borrow an oscilloscope of at least 500MHz bandwidth and see if there is a clock pulse train coming out of the XO.
(Some of the old analog scopes might be had for cheap if you look around)
 
Basically if there is no XO signal, there will be no lock.

 
Thanks b0bb, excellent tip.
 
The DIP socket I am using will not pass through one of the holes due to some kind of blockage. I don't know if this is down to the original modder leaving something there when moving the 575 xo to the discrete regulator board (discussed earlier in this thread) or whether it is debris from some of my earlier efforts. Anyway, I made some alterations to get around this and I am still not getting a lock.
 
Further to this, one of the regulator boards from the original mod performed by a 3rd party has worked its way loose and so I no longer have power to one channel of the output stage. This is idential to the problem I corrected yesterday. It's a very fiddly correction, as two of the pins span a tiny resistor and the third goes to a fairly small contact where presumably the output of something like an LT1763 should be. These discrete regulator boards will likely come loose again as I continue to make alterations.
 
Since correction of these issues is eating up all of my spare time, and with no guarantee of a positive outcome, I have put in an order for a replacement MH-DA003. That might sound a bit extreme, but I will at least have a much cleaner platform to work with, and the benefit of hindsight + experience.