The grounding scheme in my Arcam Alpha+ CD player looks severely compromised. I have already rerouted one ground wire which resulted in a noticeable improvement in sound quality. The DAC board is single sided and has analogue and digital ground traces which are commoned at the PS caps. The digital ground runs from the PS in a loop to the left, ending under the TDA1541 and the analogue ground runs in a loop to the right via the output op amps and I/V and eventually ending under the TDA1541. I have read that this is completely wrong and the analogue and digital grounds should be commoned at the DAC (this is difficult to achieve without galvanically isolated PSUs for digital and analogue).
I am now thinking about adding a full groundplane to the DAC board by adding a single sided plain copper pcb sandwiched underneath the DAC board (copper side down) with the appropriate ground connections connected through. Obviously I will need to cut the relevant ground connections on the DAC pcb to avoid ground loops.
Guido Tent seems to recommend a full ground plane - i.e. don't try to separate digital from analogue or signal from power - and this is the easiest in practical terms.
Do you think this "brute force" method is the best way to go?
I am now thinking about adding a full groundplane to the DAC board by adding a single sided plain copper pcb sandwiched underneath the DAC board (copper side down) with the appropriate ground connections connected through. Obviously I will need to cut the relevant ground connections on the DAC pcb to avoid ground loops.
Guido Tent seems to recommend a full ground plane - i.e. don't try to separate digital from analogue or signal from power - and this is the easiest in practical terms.
Do you think this "brute force" method is the best way to go?
See 'Ideal Decoupling of Digital ICs'
I started quite a debate between Guido Tent and others regarding the ground plane issue in a recent thread on
'Ideal Decoupling of Digital ICs'- loads of different views here.
I believe I will give it a try! (although I have been warned)
Through my discussions I have come to realise that every time we dump current to ground we are really forming a current loop.
If multiple grounds are used and all of these are joined at a single point on the board, then this will be the only path for the return currents to travel. This will certainly not be the shortest route. The larger the current loop, the more noise it radiates and the noise floor will rise according, along with jitter and poor sound quality.
The only precaution seems to be that you must avoid placing digital circuitry between analogue components. 'Delicate' return currents must not disturbed e.g. PLL. I will place my various grounds closest to the main ICs that they serve to further avoid such crosstalk.
I started quite a debate between Guido Tent and others regarding the ground plane issue in a recent thread on
'Ideal Decoupling of Digital ICs'- loads of different views here.
I believe I will give it a try! (although I have been warned)
Through my discussions I have come to realise that every time we dump current to ground we are really forming a current loop.
If multiple grounds are used and all of these are joined at a single point on the board, then this will be the only path for the return currents to travel. This will certainly not be the shortest route. The larger the current loop, the more noise it radiates and the noise floor will rise according, along with jitter and poor sound quality.
The only precaution seems to be that you must avoid placing digital circuitry between analogue components. 'Delicate' return currents must not disturbed e.g. PLL. I will place my various grounds closest to the main ICs that they serve to further avoid such crosstalk.
Thanks Oli
Oli - Many thanks for the link - it's just what I was looking for and it looks like you have already spent a long time trying to get to the bottom of exactly the same problem.
It was Guido Tent's note on grounding (and a long term niggle about the Arcam grounding) that set me off down this particular avenue. However having read all the documents referred to in your thread I think (humble opinion etc etc) that the single ground plane method is not the best solution even though it is temptingly easy to implement.
In my Arcam GND-A and GND-D for the TDA 1541 have at least 8" of thin track separating them (before my 1st track mod it was more like 11"!!!) so this leaves plenty of room for improvement.
I did not originally want to add an extra isolated PSU , but I think this is the only way to properly sort the grounding without a total redesign of the pcb. This would allow me to move the common point from the existing ps star to the TDA1541 where I can connect GND-A and GND-D directly together.
BTW - on your thread you kept looking for an answer on the "best" decoupling components. My experience with power amps is that correct grounding and layout makes far more difference than the selection of passive components. You also need to try things for yourself, I have found some circuits or techniques to work well in some situations but not others, for example the best supply for the OPA627s in my I/V stage is an RC (22R and 47u 63V Pana FC fed from LM317). I'm not sure why this gives better sound than the many different active regs I've tried, in other situations I prefer opamps with shunt regulators. It is possible that shunt regs sound bad with my I/V because all the extra current they shunt away has to flow down the tracks of such a bad layout.
Hope the DAC is going well.
Dave
Oli - Many thanks for the link - it's just what I was looking for and it looks like you have already spent a long time trying to get to the bottom of exactly the same problem.
It was Guido Tent's note on grounding (and a long term niggle about the Arcam grounding) that set me off down this particular avenue. However having read all the documents referred to in your thread I think (humble opinion etc etc) that the single ground plane method is not the best solution even though it is temptingly easy to implement.
In my Arcam GND-A and GND-D for the TDA 1541 have at least 8" of thin track separating them (before my 1st track mod it was more like 11"!!!) so this leaves plenty of room for improvement.
I did not originally want to add an extra isolated PSU , but I think this is the only way to properly sort the grounding without a total redesign of the pcb. This would allow me to move the common point from the existing ps star to the TDA1541 where I can connect GND-A and GND-D directly together.
BTW - on your thread you kept looking for an answer on the "best" decoupling components. My experience with power amps is that correct grounding and layout makes far more difference than the selection of passive components. You also need to try things for yourself, I have found some circuits or techniques to work well in some situations but not others, for example the best supply for the OPA627s in my I/V stage is an RC (22R and 47u 63V Pana FC fed from LM317). I'm not sure why this gives better sound than the many different active regs I've tried, in other situations I prefer opamps with shunt regulators. It is possible that shunt regs sound bad with my I/V because all the extra current they shunt away has to flow down the tracks of such a bad layout.
Hope the DAC is going well.
Dave
I did the separate PSU mod....
As is common with any mod first impressions were good - tighter bass, more detail, all the usual stuff. However as so often happens I breathed a sigh of relief when I went back to the (almost) original setup - the scale, fullness and correctness returned to the sound.
Just to make sure it was the layout and not the add on PSU, I tried the original layout with and without the extra PSU - there was only a small difference is sound quality. So I'll stick with 8" of track between D-GND and A-GND on the TDA.
Maybe the Arcam engineers know a thing or two? Maybe the IC manufacturers are not quite as clever as we thought? Maybe Guido is right and I should try one solid ground plane? Maybe I just prefer 8 bit audio to 16 bit?
Hmm... wonder what 4 bit would sound like? Or how about undersampling?
As is common with any mod first impressions were good - tighter bass, more detail, all the usual stuff. However as so often happens I breathed a sigh of relief when I went back to the (almost) original setup - the scale, fullness and correctness returned to the sound.
Just to make sure it was the layout and not the add on PSU, I tried the original layout with and without the extra PSU - there was only a small difference is sound quality. So I'll stick with 8" of track between D-GND and A-GND on the TDA.
Maybe the Arcam engineers know a thing or two? Maybe the IC manufacturers are not quite as clever as we thought? Maybe Guido is right and I should try one solid ground plane? Maybe I just prefer 8 bit audio to 16 bit?
Hmm... wonder what 4 bit would sound like? Or how about undersampling?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.