B1 Buffer Preamp

[Salas]

Originally posted by BFNY
As you can see, it eliminates the peaking, and lowers the regulator output noise floor by over 20dB. It would be very interesting to see if you can hear a difference with this simple mod.

The article says that the 540mOhm 4.7uF electrolytic at the output resonates with LM317's output Z. The guy then replaces the original 220uF at output and then bypasses the adj pin with 22uF and gets a 20dB better performance and no peak.
I already have a 10uF Tantalum adj bypass cap and the output one is MKT with much less series resistance than his elco 4.7uF example.
Anyway I have a couple of spare Elna 220uF/25V Cerafines, if I will lift the board to rewire the pot and fix it in the box, I will try them. But I am not at -60db ripple territory, this is sans adj pin bypass. Its already bypassed to -80dB territory. There is zero ripple roar with my ear pressed on the 95dB speakers. He could get it on his headphones. B1 has a CCS, it has better PSRR than some resistor loaded class A that the link author may have used.

[Diomedian]

Originally posted by salas

There is zero ripple roar with my ear pressed on the 95dB speakers. He could get it on his headphones. B1 has a CCS, it has better PSRR than some resistor loaded class A that the link author may have used.

If I remember correctly, the author of that article amplified the noise on the power rails in order to raise it above the noise floor of his measurement equipment.

-d

[flg]

Well;

http://www.diyaudio.com/forums/showt...79#post1566979

John Curl's Blowtorch preamplifier



John Curl's Blowtorch preamplifier

[Salas]

Originally posted by Diomedian


If I remember correctly, the author of that article amplified the noise on the power rails in order to raise it above the noise floor of his measurement equipment.

-d

Nice one, I was about to try to listen for a phantom then?

[BFNY]

Originally posted by salas
..

OK, the original schematic did not include any adjustment pin bypass cap shown.

I assume the author was using headphones, listening not to the normal output, but to the power supply rail, AC coupled.

The stuff about the LM317 output peaking at 5-10kHz with ultra low ESR caps is very interesting. Certainly appears it's possible that big peaks like those shown could explain what people may hear as a negative treble glare in some designs.

[Salas]

You are right it is not shown correct in the schematic. But if you see the big picture at the bottom of page 19, it is parallel to the 4k7 resistor that shunts the 100 k trimmer, just to bring it down in range. Thus the cap shunts the adj resistor leg.

Yes there are things with PSUs and they get intrusive more or less depending on particular circuits. The B1 does not suffer from glare thankfully.

The more elaborate someone will be for parts quality and PSU design, the more will refine the outcome. My example is almost generic.

I will revise the schematic.

[Salas]

Revised schematic.

[Skorpio]

Originally posted by salas
Revised schematic.

Not that it might change much, but two differences besides DC-coupling:

-Input resistor 1k after ground resistor(direct on gate), not before as on B1

-Output resistor lower than 1k

Are those things for a reason?

[Salas]

Yes there is a reason. The gate stopper must be physically attached directly on the gate pin and better be carbon. Kiwame in my case. The 1K out proved too much for what is needed in my layout, 330R directly attached to the output of the JFETs shown no ringing driving the probe cable even on a 100kHz square wave. Kiwame again. Keeping the output impedance lower is the benefit. I think that somewhere early in this very thread, Nelson Pass said that it can go down to 225R.

[Tolu]

@Salas

Which type of bridge rectifier do you use?