Power Supply-->Crosstalk on my balanced a47?
Sep 9, 2009 at 10:48 PM Thread Starter Post #1 of 8

Zaubertuba

100+ Head-Fier
Joined
Feb 3, 2009
Posts
449
Likes
20
Responding to a post I made in the opamp thread about my "quest for greater soundstage," murrays pointed out the following:

Quote:

Originally Posted by murrays
There are more subtle issues than just the quoted "separation" of a dual device. (See if you can track down Walt Jung's articles on audio op-amps). There are many paths that act as inputs and can couple signals. The power supply is a common one. A dual device must share its power supply - that is one thing you cannot split. So you must do the best you can with regulation and bypassing. Any ripple or noise imposed on a power rail by one half can affect the other half. Another subtle coupling path is thermal effects. Also there is layout, as mentioned by Tangent.

If you decide to replace a dual device with two singles you will not reap full benefit unless you can also implement independent power supplies. That can be done with local regulators near each op amp. The best / most extreme separation is to go full "dual mono" - two independent mono amplifiers with nothing shared (well, maybe the transformer).



I realized my amp is already wired so L+ and L- utilize a set of opamps and R+ and R- use the other set, so there really isn't a lot of chance for crosstalk within the opamps themselves. This leaves the power supply. Currently, I'm using a 25v Switching PSU feeding a TLE2426/BUF634 rail-splitter:

8821_1221693630784_1482242182_634891_2337175_n.jpg


8821_1221693750787_1482242182_634893_3587923_n.jpg

(once again, excuse the cameraphone pics...and yes, I'm probably going to rewire this to get rid of the "rat's nest"
tongue_smile.gif
)

Would implementing a second identical TLE2426/BUF634 splitter (I would then use one per channel) help, or is there a better implementation (like the regulation and bypassing that murrays suggests) towards decoupling the power rails for each channel? Or will it even matter?

Incidentally, the amp actually sounds really good to my ears. I just can't help tweaking it that extra 1%.
rolleyes.gif
 
Sep 10, 2009 at 12:10 AM Post #2 of 8

the_equalizer

1000+ Head-Fier
Joined
Jul 5, 2009
Posts
1,138
Likes
68
It would help, since you'd have a couple of electrolytics at the output of each rail splitter providing more bypassing/decoupling.

Still if you really want to go all the way for that last 1% you'll have to do what murrays mentions in his post: two independent mono amplifiers with nothing shared (well, maybe the transformer).

Cheers!
 
Sep 10, 2009 at 12:51 AM Post #3 of 8

mugdecoffee

100+ Head-Fier
Joined
Jul 1, 2008
Posts
206
Likes
10
I don't think it doesn't make a lot of sense to have two buffered virtual grounds. The opamps only use the virtual ground to attenuate low level signals. I think (though I could be wrong) that in this balanced configuration, buffering the ground is almost unnecessary since the ground current is minimal. Dual power supplies would be a much more practical upgrade.
 
Sep 10, 2009 at 5:28 PM Post #4 of 8

Zaubertuba

100+ Head-Fier
Joined
Feb 3, 2009
Posts
449
Likes
20
Quote:

Originally Posted by mugdecoffee /img/forum/go_quote.gif
I don't think it doesn't make a lot of sense to have two buffered virtual grounds. The opamps only use the virtual ground to attenuate low level signals. I think (though I could be wrong) that in this balanced configuration, buffering the ground is almost unnecessary since the ground current is minimal. Dual power supplies would be a much more practical upgrade.


So, considering I'm still trying to keep the parts count in line with the relative scale of the amp itself, how about a pair of these, with a suitable transformer feeding both?

EDIT: Or...Since one of these was suggested to me for another build anyway, and I'd probably use it for future amps, maybe I should go whole-hog and build a dual power supply using two of these kits. Or is that silly-excessive?
 
Sep 12, 2009 at 2:17 AM Post #5 of 8

the_equalizer

1000+ Head-Fier
Joined
Jul 5, 2009
Posts
1,138
Likes
68
Quote:

Originally Posted by Zaubertuba /img/forum/go_quote.gif
EDIT: Or...Since one of these was suggested to me for another build anyway, and I'd probably use it for future amps, maybe I should go whole-hog and build a dual power supply using two of these kits. Or is that silly-excessive?


$120 US for your power supply... seems a bit excesive to me... but, hey, that's just ME.

Edit: My bad, it'd be 'only' $90.00 USD.

Cheers!
 
Sep 12, 2009 at 3:26 AM Post #6 of 8

mono

1000+ Head-Fier
Joined
Jan 10, 2005
Posts
1,412
Likes
11
Quote:

Originally Posted by Zaubertuba /img/forum/go_quote.gif
Responding to a post I made in the opamp thread about my "quest for greater soundstage," murrays pointed out the following:


I realized my amp is already wired so L+ and L- utilize a set of opamps and R+ and R- use the other set, so there really isn't a lot of chance for crosstalk within the opamps themselves. This leaves the power supply. Currently, I'm using a 25v Switching PSU feeding a TLE2426/BUF634 rail-splitter:

8821_1221693630784_1482242182_634891_2337175_n.jpg


8821_1221693750787_1482242182_634893_3587923_n.jpg

(once again, excuse the cameraphone pics...and yes, I'm probably going to rewire this to get rid of the "rat's nest"
tongue_smile.gif
)

Would implementing a second identical TLE2426/BUF634 splitter (I would then use one per channel) help, or is there a better implementation (like the regulation and bypassing that murrays suggests) towards decoupling the power rails for each channel? Or will it even matter?

Incidentally, the amp actually sounds really good to my ears. I just can't help tweaking it that extra 1%.
rolleyes.gif



A few suggestions.

- Switch your cans, they'll effect soundstage more than most things you can do to an amp or it's PSU(s).

- I'm not so sure channel separation is the key to improving soundstage. Some might say it is the opposite that soundstage involves having the opposing channels effect each other. Your aims seem to be towards a result of less distortion, a cleaner sounding amp.

- To increase channel separation, pick higher impedance cans and wire separate ground leads instead of a shared ground return. You have to, unless I overlooked something, to get your ground return isolated with two railsplitters or else the headphone plug just connects them together again. I might be wrong, been awhile since I took a look at an A47 and I don't know if yours is completely standard. Having/posting a schematic of what you have is often helpful to others especially when it's not a professionally made PCB which locks in the standard design more.

- I'd go with a linear PSU instead of switching but the all-in-one kit seems a bit overpriced to me, that having a double sided PCB with a ground plane (does that kit even have a ground plane?) would be desirable, but the cost is mostly just saving some time hunting down parts or designing your own PCB (or finding another to use) but Tangent has a Tread and Steps designs which many people are happy with.

- Decoupling close to the chip power /gnd pins with low value ceramic or film caps is a good idea. However, past a certain point you have to decide how much time and effort to put into an A47. While it may sound good, I moved on to a little more elaborate amps and never felt the urge to listen to an A47 again as it is easily outclassed by some not so much more elaborate like Tangent's Pimeta design (as a starting point suggestion).

- Your electrolytic capacitors should be on the same board as the amp, including the TLE rail splitter and BUF634, to lower the impedance you shouldn't have this portion of the supply connected by the wires even though it appears the same electrically on a simple schematic. Adding decoupling film or ceramic caps near the chip pins as mentioned previously will increase the distance you can get away with placing the electrolytic caps, but still their leads should be as short and close to the amp pins as possible (except further away than the decoupling film/ceramics since they are contending for the same board real-estate).

Basically I'm suggesting that a change to one power supply per channel will make less difference than most other changes at this point, an unwarranted expense, but you can easily get more than 1% improvement depending on how much time you spend but I recommend a more elaborate amp design next instead of too much tweaking an A47, it's good for what it is but it's hidden purpose is to get you more hooked on headamps so we can recommend even more expensive ways to empty your wallet.
biggrin.gif
 
Sep 12, 2009 at 7:34 AM Post #7 of 8

Zaubertuba

100+ Head-Fier
Joined
Feb 3, 2009
Posts
449
Likes
20
Wow! Thanks for the incredibly thoughtful reply, mono!
bigsmile_face.gif


Quote:

Originally Posted by mono /img/forum/go_quote.gif
A few suggestions.

- Switch your cans, they'll effect soundstage more than most things you can do to an amp or it's PSU(s).

- I'm not so sure channel separation is the key to improving soundstage. Some might say it is the opposite that soundstage involves having the opposing channels effect each other. Your aims seem to be towards a result of less distortion, a cleaner sounding amp.

- To increase channel separation, pick higher impedance cans...



I've actually been thinking of shoehorning a pair of HD 650 drivers into my 555's, just for kicks. Maybe I'm not so crazy for thinking it might be a good thing...if I can make it work.

Quote:

...and wire separate ground leads instead of a shared ground return. You have to, unless I overlooked something, to get your ground return isolated with two railsplitters or else the headphone plug just connects them together again.


My headphones are recabled for the balanced amp, so two separate TRS plugs isolates the ground at least into the amp. You're right though---the L&R channel boards themselves do share the ground, so I'm not sure that's much better.

Quote:

I might be wrong, been awhile since I took a look at an A47 and I don't know if yours is completely standard. Having/posting a schematic of what you have is often helpful to others especially when it's not a professionally made PCB which locks in the standard design more.


Sorry I haven't drafted a schematic--I adopted it from another person's layout, but I really should at least work out a block diagram of how I did it. I'm sure my wiring layout doesn't help much either.
rolleyes.gif
I was thinking I might etch my own board(s) from scratch to clean up this build, but then, at the end of it all you make a convincing case for me to do otherwise...
wink.gif


Quote:

I'd go with a linear PSU instead of switching but the all-in-one kit seems a bit overpriced to me, that having a double sided PCB with a ground plane (does that kit even have a ground plane?) would be desirable, but the cost is mostly just saving some time hunting down parts or designing your own PCB (or finding another to use) but Tangent has a Tread and Steps designs which many people are happy with.


Quote:

Originally Posted by the_equalizer
$120 US for your power supply... seems a bit excesive to me... but, hey, that's just ME.

Edit: My bad, it'd be 'only' $90.00 USD.



You're both absolutely right--the Twisted Pear supply in particular is definitely overkill, but if I used it then I'd have an awesome dual supply for my next amp project. Of course, reading through your whole post, I now see you're already leading me down this evil path...
devil_face.gif


Quote:

Originally Posted by mono
Decoupling close to the chip power /gnd pins with low value ceramic or film caps is a good idea. However, past a certain point you have to decide how much time and effort to put into an A47. While it may sound good, I moved on to a little more elaborate amps and never felt the urge to listen to an A47 again as it is easily outclassed by some not so much more elaborate like Tangent's Pimeta design (as a starting point suggestion).

- Your electrolytic capacitors should be on the same board as the amp, including the TLE rail splitter and BUF634, to lower the impedance you shouldn't have this portion of the supply connected by the wires even though it appears the same electrically on a simple schematic. Adding decoupling film or ceramic caps near the chip pins as mentioned previously will increase the distance you can get away with placing the electrolytic caps, but still their leads should be as short and close to the amp pins as possible (except further away than the decoupling film/ceramics since they are contending for the same board real-estate).

Basically I'm suggesting that a change to one power supply per channel will make less difference than most other changes at this point, an unwarranted expense, but you can easily get more than 1% improvement depending on how much time you spend but I recommend a more elaborate amp design next instead of too much tweaking an A47, it's good for what it is but it's hidden purpose is to get you more hooked on headamps so we can recommend even more expensive ways to empty your wallet.
biggrin.gif



Ack! Now you've done it! I'm being drawn down the path of no return!
eek.gif


Seriously, your line of (very well thought-out) reasoning points me right back to a balanced JISBOS-with-gain that was suggested to me on another thread, or a balanced Pimeta like you're suggesting. Do you have any other reasons for suggesting the Pimeta in particular?
 
Sep 12, 2009 at 8:40 AM Post #8 of 8

mono

1000+ Head-Fier
Joined
Jan 10, 2005
Posts
1,412
Likes
11
Quote:

Originally Posted by Zaubertuba /img/forum/go_quote.gif
Seriously, your line of (very well thought-out) reasoning points me right back to a balanced JISBOS-with-gain that was suggested to me on another thread, or a balanced Pimeta like you're suggesting. Do you have any other reasons for suggesting the Pimeta in particular?


It's very easy to recommend something that Tangent has done such a great job of supporting with documentation and that so many have built, while still being a reasonably inexpensive headamp to build? I see it is the next logical step, a level above an A47, but I will not try to imply you won't eventually want something even better, that depends on how picky you become, some people will end up devoting their lives' worth of free time to this, and fairly so if they want variety more than anything else.

Others would move onto a tube amp, I can't predict what suits your tastes the best. Ideally, we'd all own the best of class in each type, whether tube or solid state. Pimeta is definitely not best of type in solid state but it is an affordable to build, well-supported, definite step up from A47. And it sounds good enough to preoccupy till you decide what to do next, but frankly if I am being lazy it is good enough to use for regular listening instead of being critical, but you have to find your own personal sound signature preference and once you do, you will be stuck on that type of amp... or a few of each type, some people get bored with the same thing every time.

Back to the Pimeta, I suggested it because it is not only easy thanks to tangent, but a professional looking result with the PCB, choices in which buffers you use, and basically a good solid-state topology that will make you want to spend more money to build a PPA2 next.
evil_smiley.gif
Then a tube amp.
evil_smiley.gif
Some people look back and wish they skipped a few steps, and it is true if your skill level allows you to do this, your patience, and your budget.
 

Users who are viewing this thread

Top