Patchwork Reloaded: Circuit Optimization and Board Layout.

[ostripper]

Oh, no.. 100uf will be fine , especially with your cascoded
diff. ,but when you are running 6 amps off the same PS
ANYTHING you can do to lower PSRR is a good thing.

On the 5 channel Japanese HT amps, I notice they don't
even have midsized decouplers (100-220uF). All they
do is cram the main filter caps in the middle,run a solid copper
bus to all OP devices,just decouple the driver chips (LM-XX,
STK-XX) w/.22uF's and rely on the good PSRR of the IC's.

But why make NFB and the LTP work harder than they have to??
I will try utiliboard as I like the way you can make custom
layouts with it ,since you use it, could you give me a few pointers
on it??


Thanks, OS

[MJL21193]

Originally posted by ostripper
,since you use it, could you give me a few pointers
on it??

Me? Give pointers? I'm in need of those myself!

The drawback is that it interfaces with Multisim, and would probably be most efficient that way. There's isn't anything wrong with Multisim, it's just that if you have already figured out how to do things in Eagle, you have another learning challenge ahead. Eagle (from my little experience with it) is much different from MS/UB. Which one is better? I'm biased.

The 6 amps will be running off 2 power supplies that share a common transformer. Each PS has 23500uF per rail smoothing caps. Also, the end use for this 6 channel amp is a pair of 3-way speakers. The mid and treble combined will come close to the power usage of the bass. I haven't had a problem running with it half done (three channels into one speaker) and I don't anticipate any problems when I eventually finish the other 3.

[Eva]

Hi John

I have not checked the small signal layout in detail, but I think that there is little room for improvement in the power layout, unless you go for a larger or a double sided PCB or a daughter board. In any case, it's better than average, it lacks most of the usual mistakes

I would probably do some changes to the ground track in an attempt to improve PSRR a little.

[MJL21193]

Originally posted by Eva


I would probably do some changes to the ground track in an attempt to improve PSRR a little.

Thank you very much Eva.
I tried to put into action all of the pointers you gave in that other thread. It will be nice to have this knowledge to draw upon for future layouts. Starting off in the right direction is always the best way.
Do you see a problem with the length of the feedback trace and how it routes between the P- and the ground?
Can I ask what changes could be made to the ground that would improve PSRR? I'm interested in any advice, barring a top layer, as these improvements are all free.

[Eva]

Decoupling capacitors will tend to carry chopped halves of speaker current at high frequencies. These currents will tend to produce voltage drops across the ground system. If you connect the capacitors to ground in a symmetrical way and take ground references from the middle point, the voltage drop across the ground track will become the sum of both halves with equal weight so it will not contain extra harmonics. Asymmetrical connection results in more voltage drop from one of the halves and harmonics.

[MJL21193]

Originally posted by Eva
Decoupling capacitors will tend to carry chopped halves of speaker current at high frequencies. These currents will tend to produce voltage drops across the ground system. If you connect the capacitors to ground in a symmetrical way and take ground references from the middle point, the voltage drop across the ground track will become the sum of both halves with equal weight so it will not contain extra harmonics. Asymmetrical connection results in more voltage drop from one of the halves and harmonics.

Thanks again,
This makes sense. You are referring to the larger ones, on the output stage I hope. Harder to get that kind of thing happening on the ones for the front end.
I'll do it and post back a picture. I have made a few other small changes.

[Eva]

I had some fun with the image...

You may consider swapping the positions of C4 and C3, and C15 and C16, or even omitting C2 and C14.

Isn't the RC network from output to ground missing? It should be in the amplifier PCB near the power stuff.

[MJL21193]

Originally posted by Eva


Isn't the RC network from output to ground missing? It should be in the amplifier PCB near the power stuff.

Ah, the Zobel. I left it off the first layout, thinking I would put it at the output terminals of the amp. It is added in here:



Better?

[MJL21193]

Some subtle changes. I think I am there now.



Much thanks to Eva for good advice.

[Nico Ras]

John,

I think the design bug has bitten you big time. Once you have finished this amp you will most probably start the next one and the next.... I think you should start a www.patchwork.com web-site.