All Hitachi Lateral Fet amplifier for DIY described by Paul Kemble

[AndrewT]

from post128

slightly mod the differential and VAS
current sources voltage references in a way that at powering on
the input stage and upper side of the VAS become functional
before the VAS current source start to conduct.

This way , the negative feedback will be available micro seconds
or so after the amp is powered on and any eventual pop will be
suppressed or at least reduced to a meaningless amplitude.

is this why some amps start up nearly silently and others cause a varying DC to be exported to the speaker?

Is it about getting the NFB circuit up and working first?
C6 achieve the VAS CCS delay?

[wahab]

Originally Posted by dadod

You can do some more improvements.
1. increment LTP tail current as jfet vorks bettter close to the Idss.
2. use floating cascode for the LTP, look at attachment.
3. Use Hawksford enhaced cascode for VAS.
4. You do not need to have the same impedance at the jfet LTP input and FB, you can use much smaller resistor for the FB and could do without capacitor.
dado

I simulated all theses possibilities , apart from 4 and results are that :

1) Increasing tail current will not improve the perfs , it will just
increase the thermals.

2) A floating cascode is no better be it in AC perfs or PSRR
quite the contrary.

3) An Hawksford cascode will bring no improvement.

4) Using smaller resistors is not a good idea , as if ever
there s a DC output fault there could be enough voltage/current
to break the inverting input fet.

Anyway , thank you for the propositions..

[AndrewT]

Dadod,
looks like you will have to build the prototype in 5 varieties and prove each of your assertions.
Wahab is not willing to accept without proof.

[wahab]

Originally Posted by AndrewT

from post128is this why some amps start up nearly silently and others cause a varying DC to be exported to the speaker?

Is it about getting the NFB circuit up and working first?
C6 achieve the VAS CCS delay?

Of course , the usual delayed speaker protection relay
will generaly keep the amp mum but a safe design will always
take care of the standing currents settling at power on
by reducing as much as possible the time necessary for the amp
to achieve stable operation and thus reducing the spike to negligible
time/energy while eliminating all risks of destructive transients currents/voltages within the circuitry.

The case of this thread amp is particularly telling.

In the original schematic the voltage reference is common
for both the differential and VAS current sources.

It can be easily deducted that at power on the VAS current source
will start to conduct before the differential current source can supply
even half of the LTP current so the differential is not functionnal
while the VAS current source has already started to swing the output
mosfet gates to the full negative rail voltage.

This was allegedly corrected by feeding the zener reference with
the second leg of the VAS , hence , the VAS current source could
not conduct at power on since the LTP being not functionnal the second
VAS leg couldnt feed it....but then , the circuit couldnt start at all because
of this chain dependancy :

LTP not functionnal --> VAS non functionnal----->VAS cant supply
the LTP CCS voltage ref.------>LTP non functionnal......

I then proposed to force the voltage ref at power on with a 100K
helping resistor but because of the common voltage ref the VAS
CCS was to kick in first , an undesirable by product , hence the
necessity of separate references.

In the original schematic C6 did extend the time between
VAS CCS starting to conduct and LTP being functionnal hence
increasing the spike energy but in this revised version
the capacitor C6 will effectively act as a further delay for the
VAS CCS conduction , wich is a welcomed effect , but its true role
is to short the AC variations of the second VAS leg ,
otherwise the VAS reference voltage would be very slightly
modulated by the VAS current.

[AndrewT]

Thanks for that comprehensive reply.

Can LTspice look at this start up process and can it predict the voltage fluctuations on the output during the first few tens of milliseconds of start up?

What would the .op look like?
What would the voltage sources be set to?

[dadod]

Originally Posted by AndrewT

Dadod,
looks like you will have to build the prototype in 5 varieties and prove each of your assertions.
Wahab is not willing to accept without proof.

Oooh it's up to him.

[AndrewT]

Wahab presented a schematic.
You made 4 suggestions for improvements.
Wahab claims your "improvements" are not such.

Who should prove what?

[gaborbela]

Hello
Thank you for all of you for the reply.
wahab I plan to use serious separate PS, I'm not after to use large 10 000uF-s on the PC board.
When that large caps start to charge up you all know what happen.
If I get the right transformer I will go with reg front end PS.
Small thump not a problem while the capacitors charged but I did notice your advise and I'll take care of that, thank you for the advise!
Guys please do not say the F5 so bad, I want to build the T1 version.
I collected all the expensive parts, heatsink, transformer, Toshiba power devices etc.
All do to me the Aleph30 was a huge upset (because the money energy I invested it didn't showed up), after 45 hours break in I took the boards out from the enclosure and sold them.
I used BG NP caps on the boards, RPR resistors, exotic caps in the PS large Plitron toroid......... close to $$$$ parts value when I count everything including the heatsink, WBT silver solder, connectors etc I do not talk here about the work I invested.
Probably I didn't gave enough time for break in???!!
That is the reason I will (re)build (give another chance to the Aleph) a similar amp Aleph2.
I read many people still prefer the Aleph series over some of the F Watt.

Greetings Gabor

[dadod]

Originally Posted by AndrewT

Wahab presented a schematic.
You made 4 suggestions for improvements.
Wahab claims your "improvements" are not such.

Who should prove what?

I just suggested, and wahab said he tried it before, so I am good at it, it is his amp.
dado

[wahab]

Originally Posted by dadod

I just suggested, and wahab said he tried it before, so I am good at it, it is his amp.
dado

Suggestions were good but they will work for dedicated designs
that will take advantages of such circuits, one must not implement
a sub circuit because it has the reputation to perform well ,
rather a careful examinantions of how this better performances
are provided by a circuit will tell if it s necessary or not to
implement it.


As an exemple the Hawksford cascode can be thought as being
a possible improvement since it does generaly do better because
the common base voltage reference is enclosed in the local feedback
loop but that works only for single ended cascode not for a
cascoded differential as the differential emitters common point
is a virtual AC ground , there will be no AC variations across the common
emitter resistor contrary to the single ended cascode that will see
AC variations through the resistance that load the emitter , while
a differential VAS emitter resistor will see constant current , hence
the Hawksford cascode will not bring any improvement for this kind
of VAS.