Amplifier Topologies

[hienrich]

I'm quite lost....

hello,

its bit long to follow through this long thread and misunderstand some

terms.

Is there someone who could tell some stories on Cascoding the VAS.

what its advantages and disadvantages. And in general

is it really recommendable to cascode connect the VAS ?

regards,

hienrich

 

[tmblack]

You can find the info in Douglas Self's book or even his website.
Basicly the cascode buffers the VAS from the input impedance variations of a class-B stage.

He suggests a voltage follower is more satisfactory before the VAS. This increases the amount of feedback in the loop. I've seen this used much more often.

Tom

 

[AKSA]

I've tried both cascode and emitter follower before the VAS and can say with some certainty that these are elegant technical solutions to a problem which does not exist......

They have no impact on the sonic qualities of the amp that I could discern. Stick with simplicity.......

Cheers,

Hugh

 

[hienrich]

hhhhhmmmmmm....

Actually I've been also following some of D. Self's approach and even
R. Slones and other people.

I think the input LTP is more critical than the VAS.
I've tried experimenting on the VAS with the such approaches but nothing seems to be different in the sonics.
It might be better if we will make it simpler.

but how about large supply rails ? Such as the Leach super amp,

it has supply rails of about 90 - 0 - 90 dc.


regards,

hienrich

 

[tmblack]

I am wondering for the diff pair, does resistance load sound better than active load ( with current mirrors or cascode)?

The resistors loaded LTP have the least open loop gain and might be preferable if listening and not just specmanship is involved.

Can anyone with experience please report?

Tom

 

[AKSA]

Tom,

I agree with your observation. I've found that the resistively loaded LTP, with careful attention to current balance, is the best sounding.

Of course, it should be said that this reduces feedback factor by around 6dB - from 70dB to about 64dB in my AKSA.

Cheers,

Hugh

 

[Bricolo]

Something I was wondering about long tail pairs: when you use resistove loading, isn't it optimal if both transistors see the same load?

Sometimes, there are different resistors, sometimes one has one and the other hasn't any...

 

[Mr Evil]

Something I was wondering about long tail pairs: when you use resistove loading, isn't it optimal if both transistors see the same load?

Sometimes, there are different resistors, sometimes one has one and the other hasn't any...

It might be ideal to have identical loads for both transistors, but matching transistors doesn't do that, since the load also includes the input impedance of the following stage. This means that in practice having a second resistor is pointless (although as I have seen pointed out, it can be done just for PCB layout reasons).

 

[Photonicube]

Here is an amp that I have designed, please tell me what you think and were there could be improvements.

 

[hienrich]

here's another stupid question......

hi guy's,

there's another thing here.

there are amps that apply darlington connected bipolars or a darlington transistor on the VBE stage.
what does this kind of connection do with output soundstage,
or is it maybe for accurate thermal tracking for the output stages?

need your ideas....

regards,

hienrich

 

[darkfenriz]

hi Photonicube
interesting circuit
I think that U3/U4 and U5/U6 should be parallel not darlington (guess it's a drawing mistake)
What I would change:
- a 100 ohm resistor between collectors of U1 and U2 to avoid parasitic oscilations of this once called 'darsziklington'
- a 'vintage' Vbe multipier instead of D1 and R14
- 0.22-0.33 emitter resistors
- more output devices parallel

 

[tmblack]

Is there any reason to use JFET or MOSFET LTP in power amps?
Seems the BJTs are easy to match and much cheaper.

Tom

 

[Mr Evil]

Is there any reason to use JFET or MOSFET LTP in power amps?
Seems the BJTs are easy to match and much cheaper.

Tom

High input impedance and low input bias currents are both nice features. Depletion mode devices also make it possible to create complementary LTPs without current sources, which cuts down on complexity.

 

[tmblack]

I've seen some of John Curl designs but matching 4 jfets is a daunting task!

Tom

 

[hienrich]

hello,

since bootstrapping is also an issue here,

is there any way to optimize this part of the amplifier?

Rod Elliot uses two "equal", 3.3k value resistors and a 100uf in the middle in his 300watt subwoofer amplifier.

thus the resistors always need to be of equal values or depends
on output sonic quality?

thanks.

regards,

hienrich

 

[Nelson Pass]

Rod Elliot uses two "equal", 3.3k value resistors and a 100uf in the middle in his 300watt subwoofer amplifier.

thus the resistors always need to be of equal values...?

Not at all. The sum total gives the VAS current in such a
stage (not particularly critical) and choosing equal values tends
to be convenient. As the top resistor value goes lower you
eventually see loss of RC bandwidth and output stage loading,
and as the bottom resistor goes lower you see some loss of gain
and isolation from the output.

There are cases where you make them deliberately unequal,
an example being a VAS stage where it needs to drive the output
with nearly twice the rail offered to the VAS.

 

[Nelson Pass]

I've seen some of John Curl designs but matching 4 jfets is a daunting task!

You don't particularly have to match the N's to P's, and you can
use a single or four ballast resistors to equalize and adjust
the currents through them.

After that, matching the N's or P's is not really difficult, and does
not usually require much of a population to draw from. Also
you can still get 2SK389's and their complement.

 

[hienrich]

NP:

I'm planning to build a new amp with a bootsrapped VAS.

having 80-0-80 DC on its supply rails, what RC values do you

recommend for this approach....?

and is it good if I will use a cascode connection on the VAS stage...?

I will be using BJT's on the outputs, and a quasi complementary

topology.

thanks

regards,

hienrich

 

[Nelson Pass]

I would caution about using BJT's at such high voltages,
since they head into second breakdown at about 100 volts
or so, and the voltages they would see with be approaching
160 volts.

As to what RC values, you can use any C value that you like as
long as it's impedance is low compared to the resistance at the
lowest frequency you want to amplify. R values are arbitrary,
but you can figure that equal values whose sum gives you the
bias current you want for your VAS with about 80 volts across
it. For example 3.9K + 3.9K gives about 10 mA.

 

[hienrich]

NP:

Since mosfets are very expensive and very hard to find in our area
would it be better if I will be cascoding both the VAS and the
Output Stage as well?

but here with me, some couple of MJL21196's and MJL21194's for the output stage
and I'm planning to use MJE15032 for the VAS I'm not sure
if this will fit for the design but there specs shows higher SOA's.

I've posted once 2 months ago an amp using these devices, but I was using a constant current source on the VAS using MJE15033.
And this amp drives my subs currently.
but still I'm aiming for the optimum. Making me think of bootstrapping.

what do you think about my schemes?

thanks.

hienrich