Best yet?
Hi,
In one of the previous threads,
http://www.diyaudio.com/forums/showthread.php?postid=44407#post44407
member ftorres did some extensive testing and concluded that, for digital IC decoupling:
Just to sum up, the absolute cost/quality winner of my contest is
anasonic 120uF/25V/FC // 100nF/X7R/0612 // 220nF/X7R/0805 // Murata 4.7uF/16V/Y5V/1206
Do you guys think this is still the case with X2Y chips?
Hi,
In one of the previous threads,
http://www.diyaudio.com/forums/showthread.php?postid=44407#post44407
member ftorres did some extensive testing and concluded that, for digital IC decoupling:
Just to sum up, the absolute cost/quality winner of my contest is
anasonic 120uF/25V/FC // 100nF/X7R/0612 // 220nF/X7R/0805 // Murata 4.7uF/16V/Y5V/1206Do you guys think this is still the case with X2Y chips?
Cant you use the ferrite between the star ground and rectifier bridge (or center tap of the transformer depending on configuration)?
I.e.
Bridge_____Ferrites_______Caps/starground
+ -----------====---------------- +
- ------------====-------
                                    X--------GND
+ -----------====-------
- ------------====---------------- -
Dont know if it would give any benifit but still?
I.e.
Bridge_____Ferrites_______Caps/starground
+ -----------====---------------- +
- ------------====-------
                                    X--------GND
+ -----------====-------
- ------------====---------------- -
Dont know if it would give any benifit but still?
Interesting idea
Interesting idea- and along the same line of thought that prompted me to ask the question.
You are proposing a single ferrite in the ground return to the transformer. Localised decoupling capacitors and smoothing capacitors would provide a nice short current loop for noise, whilst the ground itself would be isolated from common mode r.f pollution from the mains. It seems like a good idea!
but...
In this thread we are advised that this is not necessary.
I wonder are the ferrites in the supply lines good enough?
Is there a downside to our proposal?
Interesting idea- and along the same line of thought that prompted me to ask the question.
You are proposing a single ferrite in the ground return to the transformer. Localised decoupling capacitors and smoothing capacitors would provide a nice short current loop for noise, whilst the ground itself would be isolated from common mode r.f pollution from the mains. It seems like a good idea!
but...
In this thread we are advised that this is not necessary.
I wonder are the ferrites in the supply lines good enough?
Is there a downside to our proposal?
hjelm said:
I have also considered doing this, but with inductors. You run into a cople of issues - such as peak reverse voltage for diodes, but this is likely a good thing and I plan to do pretty much exactly this for my new DAC. You could also do it in stages before hooking up the various grounds.
Having said all this, I never tried it. I think the most important thing it would be would be to stop noise from getting back into other circuits throught the design.
I am also considering building a pretty serious filter to fix that at both HV side and low voltage side of transformer.
Petter
I am not talking about splitting a ground plane
Guido,
I am not talking about splitting a ground plane- I do intend to use a single ground plane in my forthcoming PCB.
Instead I am talking about a single ferrite in the return cable to the centre tap of the transformer.
I will also use ferrites in the supply lines to reduce the magnitude of current loops.
Guido,
I am not talking about splitting a ground plane- I do intend to use a single ground plane in my forthcoming PCB.
Instead I am talking about a single ferrite in the return cable to the centre tap of the transformer.
I will also use ferrites in the supply lines to reduce the magnitude of current loops.
Hi,
like Oli said, it is the layout which is the limiting factor. Sometimes X2Y caps fit nicely, sometimes it is better to use 0402 or 0306. In my project I mostly use 2 or 3 paralleled 0805 10uf in a layout that resembles 0612 (no hicap 0612 available). I have only 3 0603 X2Y where layout permited. But do not hold your breath, I have'n tested it yet.
To put thing into perspective, I calculated inductance of loop for two paralleled 0805 next to each other to be 160pH. In this view, 50 or 250pH of cap self inductance is not so important anyway.
Best regards,
Jaka Racman
like Oli said, it is the layout which is the limiting factor. Sometimes X2Y caps fit nicely, sometimes it is better to use 0402 or 0306. In my project I mostly use 2 or 3 paralleled 0805 10uf in a layout that resembles 0612 (no hicap 0612 available). I have only 3 0603 X2Y where layout permited. But do not hold your breath, I have'n tested it yet.
To put thing into perspective, I calculated inductance of loop for two paralleled 0805 next to each other to be 160pH. In this view, 50 or 250pH of cap self inductance is not so important anyway.
Best regards,
Jaka Racman
Hi,
A very long thread... haven't read it through yet. However, I'd like to add some in the discussion regarding split groundplanes. I've heard the arguments for splitting and not splitting very many times within and outside the company I work for, and I've tried both types. It sounds good to split planes in theory, in reality though, I find it difficult to acheive good split grounding. Not seldom I find myself in discussion which ground should be used where and it's not difficult create ground loops.
To my experience, a single good groundplane is more robust and relayable, split plane gave me EMI problems.
So, I agree with Tent, go for a single, unbroken groundplane.
/Jesper
A very long thread... haven't read it through yet. However, I'd like to add some in the discussion regarding split groundplanes. I've heard the arguments for splitting and not splitting very many times within and outside the company I work for, and I've tried both types. It sounds good to split planes in theory, in reality though, I find it difficult to acheive good split grounding. Not seldom I find myself in discussion which ground should be used where and it's not difficult create ground loops.
To my experience, a single good groundplane is more robust and relayable, split plane gave me EMI problems.
So, I agree with Tent, go for a single, unbroken groundplane.
/Jesper
Hi Guido and all the other experts,
Please help. I have read Guido¡¦s articles and a few other sources, this subject just become more and more complex.
To simplify matters, lets take a single DAC chip for example. The DAC chip and PCB will mainly have the following (power supply lines not indicated):
Analogue Return line (Va-)
Digital Return line (Vd-)
Analogue Ground Plane (AGNDP)
Digital Ground Plane (DGNDP)
-----------------------------
Now, let¡¦s assume there are SPLIT ground planes (both AGNDP and DGNDP exist). In addition, the DAC chip has it own dedicated power supply lines AND power return lines, it does NOT use the AGNDP or DGNDP to carry return current.
Here are the questions:
1. where in the circuit should the DAC chip¡¦s Va- be connected to the AGNDP?
2. where in the circuit should the DAC chip¡¦s Vd- be connected to the DGNDP?
3. Where should AGNDP and DGNDP be connected together?
----------------------------
Then, let¡¦s look at the case where no split ground planes are used, that means AGNDP and DGNDP are all one single ground plane (SGNDP). Again, the DAC chip has it own dedicated power supply lines AND power return lines, it does NOT use the SGNDP to carry return current.
4. If this the case, for a single ground plane, won¡¦t the digital noise on the DGNDP affect the AGNDP?
5. Now, where in the circuit should the DAC chip¡¦s Va- be connected to the SGNDP?
6. Where in the circuit should the DAC chip¡¦s Vd- be connected to the SGNDP?
-----------------------------
Now more interestingly, in X2Y¡¦s materials, it mentioned the GND of X2Y should be connected to a Floating Ground (FGND) for best performance.
7. What is this FGND? Where does it connect to? Nothing? If it isn¡¦t connect to anything, what¡¦s the purpose of this FGND?
Many thanks.
Please help. I have read Guido¡¦s articles and a few other sources, this subject just become more and more complex.
To simplify matters, lets take a single DAC chip for example. The DAC chip and PCB will mainly have the following (power supply lines not indicated):
Analogue Return line (Va-)
Digital Return line (Vd-)
Analogue Ground Plane (AGNDP)
Digital Ground Plane (DGNDP)
-----------------------------
Now, let¡¦s assume there are SPLIT ground planes (both AGNDP and DGNDP exist). In addition, the DAC chip has it own dedicated power supply lines AND power return lines, it does NOT use the AGNDP or DGNDP to carry return current.
Here are the questions:
1. where in the circuit should the DAC chip¡¦s Va- be connected to the AGNDP?
2. where in the circuit should the DAC chip¡¦s Vd- be connected to the DGNDP?
3. Where should AGNDP and DGNDP be connected together?
----------------------------
Then, let¡¦s look at the case where no split ground planes are used, that means AGNDP and DGNDP are all one single ground plane (SGNDP). Again, the DAC chip has it own dedicated power supply lines AND power return lines, it does NOT use the SGNDP to carry return current.
4. If this the case, for a single ground plane, won¡¦t the digital noise on the DGNDP affect the AGNDP?
5. Now, where in the circuit should the DAC chip¡¦s Va- be connected to the SGNDP?
6. Where in the circuit should the DAC chip¡¦s Vd- be connected to the SGNDP?
-----------------------------
Now more interestingly, in X2Y¡¦s materials, it mentioned the GND of X2Y should be connected to a Floating Ground (FGND) for best performance.
7. What is this FGND? Where does it connect to? Nothing? If it isn¡¦t connect to anything, what¡¦s the purpose of this FGND?
Many thanks.
A floating ground is usually a pieace of groundplane which is decoupled from the main groundplane with a impedence, can be a inductor, a resistor or a pieace of PCB trace.
Using the one-groundplane philosophy, all ground return should be through to the groundplane. The plane should never be split (except for via-holes which normally is un-avoidable).
/Jesper
Using the one-groundplane philosophy, all ground return should be through to the groundplane. The plane should never be split (except for via-holes which normally is un-avoidable).
/Jesper
patwen said:Hi Guido and all the other experts,
Please help. I have read Guido¡¦s articles and a few other sources, this subject just become more and more complex.
To simplify matters, lets take a single DAC chip for example. The DAC chip and PCB will mainly have the following (power supply lines not indicated):
Analogue Return line (Va-)
Digital Return line (Vd-)
Analogue Ground Plane (AGNDP)
Digital Ground Plane (DGNDP)
-----------------------------
Now, let¡¦s assume there are SPLIT ground planes (both AGNDP and DGNDP exist). In addition, the DAC chip has it own dedicated power supply lines AND power return lines, it does NOT use the AGNDP or DGNDP to carry return current.
Here are the questions:
1. where in the circuit should the DAC chip¡¦s Va- be connected to the AGNDP?
2. where in the circuit should the DAC chip¡¦s Vd- be connected to the DGNDP?
3. Where should AGNDP and DGNDP be connected together?
----------------------------
Then, let¡¦s look at the case where no split ground planes are used, that means AGNDP and DGNDP are all one single ground plane (SGNDP). Again, the DAC chip has it own dedicated power supply lines AND power return lines, it does NOT use the SGNDP to carry return current.
4. If this the case, for a single ground plane, won¡¦t the digital noise on the DGNDP affect the AGNDP?
5. Now, where in the circuit should the DAC chip¡¦s Va- be connected to the SGNDP?
6. Where in the circuit should the DAC chip¡¦s Vd- be connected to the SGNDP?
-----------------------------
Now more interestingly, in X2Y¡¦s materials, it mentioned the GND of X2Y should be connected to a Floating Ground (FGND) for best performance.
7. What is this FGND? Where does it connect to? Nothing? If it isn¡¦t connect to anything, what¡¦s the purpose of this FGND?
Many thanks.
Hi
Currents run in loops. Remember that in any A to D or D to A, a current runs inside the chip. That current has to close through the external connections (Kirchoff)
The bigger the loop the worse, so do NOT splitt the planes.
regards
Patwen
In most cases the AGND and DGND pins of a DAC are supposed to be connected at the device and join to the AGND of the system.
The labels refer to the parts of the DAC where they are connected internally NOT the external ground where they are to be connected.
It's an internal bond wire / die limitation, not any conflict over the normal advice to seperate grounds.
In general DACs and ADC's should be viewed as analogue components, not digital.
Andy.
In most cases the AGND and DGND pins of a DAC are supposed to be connected at the device and join to the AGND of the system.
The labels refer to the parts of the DAC where they are connected internally NOT the external ground where they are to be connected.
It's an internal bond wire / die limitation, not any conflict over the normal advice to seperate grounds.
In general DACs and ADC's should be viewed as analogue components, not digital.
Andy.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.