LM4562 decoupling...

[mikesnowdon]

Cant find any definitive information on this and there is no mention of it in the datasheet.

I currently use 47nF polyester film caps. Thinking of trying some R/C decoupling soon. Maybe even experiment with LC.

People must have experimented with this. What were your conclusions as to which cap type and size sounded best (to your ears)?

Also funny how in the datasheet all the graphs stop @ 20khz. Whats going on there....

Any thoughts on this?

[SoNic_real_one]

That's nothing wrong. That's how wide the analyser was set to sweep.
I did swap the existing OpAmps in my equipment with LM4562 and the existing decoupling caps where sufficient. I have 47uF on the rails plus 0.1uF close to the pins.
Power rails come on the board via ferrite beads.

[Joachim Gerhard]

Sometimes it is better to do only one 0.1uF only from rail to rail and not connected to earth. Saves money too and makes sure that no HF enters earth. That´s the way i am doing it also with very fast OPamps without trouble.
I use bigger elcaps though but 47uF should be fine.

[mikesnowdon]

Joachim,

Have you compared this method with the conventional way of doing it - cap to ground - on the same opamp in the same circuit? Was there an audible benefit?

An interesting note, in the OPA2107 datasheet they recommend 1uF Tantalums for local decoupling...

[Joachim Gerhard]

Tantalums will fail, garantied after some years. Rather thake a modern low ESR like the Pannasonic FM/FC, Robicon ZL/ZLH etc. They are 105° and last a long time.
there are some extremely good Tantal like wet types but then we talk about more then 100,- $ a piece.
No i did no sound comparison but i never whent back.
When i try to make something really good i take 2 x 330uF Nichikon KZ Muse (neutral sound) or 2 x 330uF Elna Silmik 2 ( slightly warmer)and one Röderstein / Vishay MKP1837 0.1uF. Inn case of severe oscilation ( never had one with this arangement and i use very fast OPamps like the AD797 ot LME49713) lokally shunted NPO/COG 4.7nF could help but you have them to solder direct to the pins or a circumferential groundplane.

[Joachim Gerhard]

Sorry for typos. I am a bad typist and did that rather fast, out of my soul so to say.

[mikesnowdon]

Thanks Joachim. Interesting views

[Joachim Gerhard]

Interesting website you have. I think that is a good initiative.

[mikesnowdon]

Its not my site but the site of an organisation that I strongly support. Thanks

[SoNic_real_one]

Low ESR caps on power lines can make more damage than good... The rails inductance coupled with the capacitor needs some damping - usually the "normal" ESR is enough for that. If you go with too lower ESR you will get ringing and other artifacts on the rails.
As for the ideea that caps between rails and ground will "dump" noise there - that is nonsense perpetuated by urban legends.
By the way an OpApm is build, the currents will flow thru the ground back to the PS - on each alternance to the other rail. And that's the current you need to bypass!
Not between the rails, that will ADD more current to bypass at the PS on the opposite rail! Bigger loops instead of short ones.