LM3886 IGC + Pedja Rogic JFET Buffer

[andreszibert]

This is my first post, I can read English, but I can not write much.

However, this is my design

[andreszibert]

This version i believe it is complete. I hope comments, tips.

[Dxvideo]

Hi Andreszibert,

For the buffer PSU;
-You'll need bigger caps for CB5, CB7, as my experiences 10-47u tantals give better ripple rejection for this position.
-Put parallel low esr (or also tantal) 10-47u to CB6 and CB8.
-Before the regulators use low esr high caps (say 470u).
I advice these because the buffer stage is a bit power sensitive and if you have some noise in your PSU than I guarantee that you will hear!
For the buffer stage;
-Put 500R timmmer instead of RB6.. You will need 6mA bias, but depending on J310s Idss you may have different values with this fixed resistor.
-Put ~100K to 2SK170s gate to gnd.. Its not really needed but if you use a potentiometer front of it then you may have some noise problems while adjusting the volume.
-Use high quality MKP caps for both CB1 and CB2.
-You may ommit CB3 and use 100R for RB4 and 22n for CB4.. Of course it depends most of your ear taste, but as my experiences more capacity in this position reduces the treble details a lot.
For the amplifier stage;
-Are you sure just 1000u/50v will be enough for the power supply? You may consider bigger values I guess.
-You must put 100n (or similar) solid caps near the chips power pins to gnd..
....
So, these are my advices. May be other peoples say some other things. But I have made this amplifier (with LM3875) and have good results with it..

[AndrewT]

Hi,
change the polarity of the right hand bank of bridge diodes.
increase the smoothing at the rectifiers
OR
follow the advice of Peter Daniel.

There is no halfway house.

You are using a DC coupled amplifier.
What protects your speaker if something goes wrong?
How much output offset are you expecting with zero ohms on the non-inverting input and 220kohms on the inverting input and a DC gain of 22times?
You have neither a Zobel nor a Thiele Network on the output of the power amplifier.
You have no filters on the input to the buffer.

[Dxvideo]

Originally posted by AndrewT
You are using a DC coupled amplifier.

No! CB2 makes that amplifier AC coupled.

[lineup]

AndrewT is concerned with output DC-offset from LM3886.
And this is DC-coupled.

What DC-offset you get from this circuit depends on each LM3886.
Some exemplar will give less, some will give more offset at speaker terminals.

If you are lucky, with your LM3886, it will be only so small offset, it will not matter.

However, there are ways to add a few components to correct for DC-offset in such a circuit.
So, you need not worry too much.

[AndrewT]

Hi DxV,
the output of the buffer is AC coupled.
The input of the power amp is DC coupled.
However the buffer + Power amp only needs one DC blocking capacitor in the input line. That satisfies the DC blocking requirement.

But the amplifier is DC coupled on both it's input and on it's NFB. The output offset will be the amplifier gain * input offset current * difference between the input resistances.
The max input offset current is 0.2uA @ Ta=25degC with zero output current.
National are clever enough to hide how bad that offset current could become when the chip heats up.
The gain is 22times
The difference between the input resistances is 220k
The cold temperature output offset = 0.2 * 10^-6 * 22 * 220 * 10^3 = <0.98V
I have no idea how bad it could be once you start heating the chip up.

How much output offset are you expecting with zero ohms on the non-inverting input and 220kohms on the inverting input and a DC gain of 22times?

that's why I asked but you butted in to misguide him/her.

[Dxvideo]

Hi Andrew,
I've tried that circuit with the changes I mentioned above and with LM3875.. As I remember there was no noticeable DC offset on outputs (may be 5mV or so)..
However I am still confused about the DC/AC coupling issue for the LM3886.. So, if I use it in inverted configuration where else should I put a cap to run it as AC coupled? I think in this configuration its already running in AC coupled..
You're right about the buffer, it must be coupled with a cap on input. Then this will be a complete AC coupled amp..
And OK, to reduce the possible DC offset on outputs its better to put ~220K (close to Rf) between NI input to GND, however its not a must!
Am I right?

[AndrewT]

maybe that's part of the reason I never recommend an inverting topology power amp.
I don't understand enough about them.

[Dxvideo]

Do you think the inverted chipamps have better sounding (or better THD figure) than the non-inverted ones?
I've read this myth in some places but never saw in any datasheet or technical paper!