Hiraga "le monstre"

[Lumba Ogir]

Sigurd,

Here we disagree. JFETs work fine for cascode work. I have many amp with JFETs for cascodes. Even MOSFETs could be used.

They work, but perform poorly, should be the same type as suitable for voltage amplification. You don`t use FETs in VAS. It´s about conduction losses, Gm and slew rate.
(As a matter of fact, it was insightfully pointed out by Hiraga himself. That man is obviously right about everything).

I guess your 5V "Zener" is not a Zener but something else, right?

It`s a voltage reference. High dynamic impedance current sources in conjunction with low dynamic impedance voltage references form effective filters. The voltage at any sensitive base or gate should be clean. Ripple and noise don`t sound nice (much worse than THD).

[lineup]

Originally posted by Lumba Ogir
Sigurd,
zeners are nasty noise generators causing ugly distortion.
They stay on the top of my blacklisted parts toghether with ceramic capacitors and opamps.

IC voltages references should be used.
LEDs are noisy too, with a zener-like behavior.

Hello, Lumba
I am not at all afraid of using RED LED for V-Ref.
They are excellent when wanting a simple + good solution
and not very noisy, in my opionion.
( See Christer, Sweden, investigation link below )

One of the positve things, is when they are put in close contact to the transistor body,
when used as a constant current source, RED LED have a fair temperature compensation.

Preferably I use flat 5mm RED LED, with the flat surface tightly attached
to the flat surface of the TO-92 transistor body.
See pratical example how to in my attached picture.


diyAudio 2004-06-10, Christer:
Some noise measurements for LEDs and zener diodes

[Sigurd Ruschkow]

Lineup - I agree with you. LEDs are superb voltage references in a CCS,
but also when one wants to have a super-low-noise voltage source.

If anyone knows of a more low noise voltage source than a blue LED, do let us know.




Sigurd

Originally posted by lineup


Hello, Lumba
I am not at all afraid of using RED LED for V-Ref.
They are excellent when wanting a simple + good solution
and not very noisy, in my opionion.
( See Christer, Sweden, investigation link below )

One of the positve things, is when they are put in close contact to the transistor body,
when used as a constant current source, RED LED have a fair temperature compensation.

Preferably I use flat 5mm RED LED, with the flat surface tightly attached
to the flat surface of the TO-92 transistor body.
See pratical example how to in my attached picture.


diyAudio 2004-06-10, Christer:
Some noise measurements for LEDs and zener diodes

[Sigurd Ruschkow]

Lumba - I do not agree that JFETs perform poorly in cascodes.

Correctly implemented, they perform very well.

I guess that me and Hiraga do not agree with each other then
Not that he says much about JFET vs BJT cascoding in the le Monstre article....


Sigurd

Originally posted by Lumba Ogir
Sigurd,

They work, but perform poorly, should be the same type as suitable for voltage amplification. You don`t use FETs in VAS. It´s about conduction losses, Gm and slew rate.
(As a matter of fact, it was insightfully pointed out by Hiraga himself. That man is obviously right about everything).

It`s a voltage reference. High dynamic impedance current sources in conjunction with low dynamic impedance voltage references form effective filters. The voltage at any sensitive base or gate should be clean. Ripple and noise don`t sound nice (much worse than THD).

[Lumba Ogir]

Sigurd,
they don`t have the ability, are sluggish, cannot handle large voltage swings linearly, their operating area needs to be limited. It`s a hard task even for the best suited video devices. Tubes do that superbly.
What about a hybrid amplifier?

[qwad]

THANKS SIQURD AND LUMBA
for your replies, l asked only because l have some spare devices lying around: i.e; 2n 3055/ m2955's etc as well as some mj 802/ 4502, not to mention some excellent toshiba output devices as used in the aksa pp a/b amp circuit that l could use to build a circuit. but one more question: the second circuit with the bc330/ 560 /TIP 2n3055/ 2955 devices shows no bias adjust pot , is this necessary? or would the circuit work as is? naive of me to ask l know forgive a senile old mans silly questions thank you in advance cheers TC

[Sigurd Ruschkow]

Which schematic with the BC330 do you refer to?

There is no bias pot in the le Monstre design. There is a pot to adjust DC offfset to zero, though.
Output stage bias is set with the resistors 1k and 1 Ohm. Bias is also dependant on the pot's value - but not its setting.

I do not know about the AKSA design, but I can guess that the "excellent toshiba output devices " you mention are the 2SC5200/1943. If so, I would use these instead of the oldish 2N3055/2955.




Sigurd

Originally posted by qwad
THANKS SIQURD AND LUMBA
for your replies, l asked only because l have some spare devices lying around: i.e; 2n 3055/ m2955's etc as well as some mj 802/ 4502, not to mention some excellent toshiba output devices as used in the aksa pp a/b amp circuit that l could use to build a circuit. but one more question: the second circuit with the bc330/ 560 /TIP 2n3055/ 2955 devices shows no bias adjust pot , is this necessary? or would the circuit work as is? naive of me to ask l know forgive a senile old mans silly questions thank you in advance cheers TC

[qwad]

thanks for your quick reply SIGURD, the schematic l am talking about is the one that was posted earlier in the thread on a light blue background: though l do not have the toshiba devices in front of me l think you are probably right as for part numbers go
thanks again cheers TC
PS; do a search on this forum for the aksa, there was a long thread about it a while ago, which if still existing may be enlightening for you

[HKC]

Originally posted by Sigurd Ruschkow
HKC- here are some ideas:

* make the tracks that carries heavy currents MUCH wider. As wide as you can make them, ie make copper pours.

* Put the input JFETs so that they "kiss" eachother, so you can later on glue them together

* Move the power resistors to the sides and put the drivers close to the output transistors

* Move the long feedback wire more to the middle of the PCB.

* Move the cascode transistors more to the middle an d closer to the JFETs.

* Maybe a ground plane?

Hirga has a 47u cap on each rail. Did you skip that or did I miss it?

You might want to decouple the cascode resistors with a small cap (say 100n) so that the cascode voltage going in to the small BJTs is low pass filtered. It can give slighly better sonics.


Attached is my layout which has more PS decoupling (incl input section low pass filtering) than the original le Monstre and also an input filter and a feedback capacitor. I also use TO220 power resistors with a heatsink instead of the standard cylindical ones,
and have LEDs power-on indication.
Size is 160x100mm.




Sigurd

Hi Siqurd

Thank you for the suggestions. They are all good. I will make revise before I send it to factory and post pcb when it is ready.

Best Regards

[Sigurd Ruschkow]

Lumba,

please post a hybrid amplifier design so I / we can see what you have in mind.




Sigurd

Originally posted by Lumba Ogir
Sigurd,
they don`t have the ability, are sluggish, cannot handle large voltage swings linearly, their operating area needs to be limited. It`s a hard task even for the best suited video devices. Tubes do that superbly.
What about a hybrid amplifier?