Dx Blame MKIII-Hx - Builder's thread

[bonfis]

Originally Posted by canonnica

For the specific design of this HX amp with miller compensation, is it very important to have the lowest possible Cob for VAS transistor?

Let's say I have the choice between:
A) a transistor with a Cob of 5pF and a fT of 100Mhz
B) a transistor with a Cob of 16pF and a fT of 200Mhz
... and all the other specs being similar, what would be the best choice?

We can read all over this board that the lowest the Cob and the highest the fT, the better the results, but in this particular case, what spec shoud be preferred over the other?

Martin.

I would go with the low Cob transistor for the VAS. There is an article describing the modulation of collector base capacitance by VAS collector voltage at post 1570. The larger the Cob the greater the distortion from this effect. I don't think there is a comparable concern with fT but other's might disagree.

[AndrewT]

Does Cob=16pF and Ft=200MHz exist? Can it be sourced economically?

Cob=5pF and Ft=200MHz does exist. There are quite a few devices around those numbers that are economical.
Some go as low as Cob~1.7pF (@30Vce) and some go as high as 500MHz (@ 2mA to 10mA of Ic).

Back to your 16pF & 200MHz device:
What would be the Ic required to get that speed of response?
What would be the Ic required to get into the near linear hFE range?

You need a device that at the chosen operating current gives the desired parameters that enable good VAS duty.
Usually we are around 2mA to 5mA for a three stage output topology.
Quite a bit more Ic, maybe around 10mA to 30mA, for a two stage output topology.
But do keep in mind that the two stage does not suit driving low impedance loads due to the beta droop of both the output device and the beta droop of the driver device when high transients currents are demanded by the reactive speaker load.

The load seen by the VAS must be equally good. i.e. the load must have good impedance consistency for all the signals that the VAS sends to it's sink/source.
Resistors have a very good reputation as the VAS sink/source, This is probably due to consistent impedance over an enormous range of slow and fast signals.
Where an active sink/source is substituted for the resistor/s, then I suspect that good amplifier performance can ONLY be achieved if the active sink/source behaves as well over that very wide range of slow/fast signals.

This is a quite opposite opinion to that in post1638.

Cannon, you have to decide which, if either, makes more sense, or decide that an alternative and more scientifically based argument determines the combination of VAS and it's sink/source.

[canonnica]

Originally Posted by AndrewT

Does Cob=16pF and Ft=200MHz exist? Can it be sourced economically?

Yes, the 2SC5171, it's very actual.
I'll re-read your post a few times to understand it fully...

...the other I'm referring to is the 2SC1819

[Vostro]

I'll stick to my advice, if you still using the clamping diode, and want better, Iv'e looked at designs that use this diode and is fairly easy to source,
http://www.fairchildsemi.com/ds/BA/BAV21.pdf

Also about the dip on positive peak, if its small then id ignore it and be satisfied.
I dont suspect it will produce gross instability.

About using VAS transistors, they are NOT as easy to source as they once were,
the amplifier manufacturers source the VAS transistors from the suppliers.
If you want nice VAS transistors I advice doing the same, maybe a group buy in bulk

I also think that when advising on devices that links or at least names are given.

Regards

[canonnica]

Thanks Vostro. My target is to get rid of the rail sticking without having to use a clamping diode.

I've found original Matsushita 2SC1819, I have what looks like genuine Toshiba 2SC2238B. Their 2SA968B genuine counterparts are on their way too. I also ordered a set of 2SA1837/2SC4793 and a set of 2SA1930/2SC5171.

The 2SA1819 will hold the VAS, and the others will be tried in the CCS and drivers positions.

Will post results as usual.

[Vostro]

Originally Posted by canonnica

Thanks Vostro. My target is to get rid of the rail sticking without having to use a clamping diode.

I've found original Matsushita 2SC1819, I have what looks like genuine Toshiba 2SC2238B. Their 2SA968B genuine counterparts are on their way too. I also ordered a set of 2SA1837/2SC4793 and a set of 2SA1930/2SC5171.

The 2SA1819 will hold the VAS, and the others will be tried in the CCS and drivers positions.

Will post results as usual.

Have you had success with getting rid of sticking, by changing transistors?

[a.wayne]

Interesting and good Info Andrew , do you have a link to this , i would love to read into this a bit more ...

Originally Posted by AndrewT

Does Cob=16pF and Ft=200MHz exist? Can it be sourced economically?

Cob=5pF and Ft=200MHz does exist. There are quite a few devices around those numbers that are economical.
Some go as low as Cob~1.7pF (@30Vce) and some go as high as 500MHz (@ 2mA to 10mA of Ic).

Back to your 16pF & 200MHz device:
What would be the Ic required to get that speed of response?
What would be the Ic required to get into the near linear hFE range?

You need a device that at the chosen operating current gives the desired parameters that enable good VAS duty.
Usually we are around 2mA to 5mA for a three stage output topology.
Quite a bit more Ic, maybe around 10mA to 30mA, for a two stage output topology.
But do keep in mind that the two stage does not suit driving low impedance loads due to the beta droop of both the output device and the beta droop of the driver device when high transients currents are demanded by the reactive speaker load.

The load seen by the VAS must be equally good. i.e. the load must have good impedance consistency for all the signals that the VAS sends to it's sink/source.
Resistors have a very good reputation as the VAS sink/source, This is probably due to consistent impedance over an enormous range of slow and fast signals.
Where an active sink/source is substituted for the resistor/s, then I suspect that good amplifier performance can ONLY be achieved if the active sink/source behaves as well over that very wide range of slow/fast signals.

This is a quite opposite opinion to that in post1638.

Cannon, you have to decide which, if either, makes more sense, or decide that an alternative and more scientifically based argument determines the combination of VAS and it's sink/source.

[AndrewT]

I made it up.
Or put another way:
think about where the VAS current goes.

I suspect that good amplifier performance can ONLY be achieved if the active sink/source behaves as well over that very wide range of slow/fast signals.

I think this is the critical part that LOADS the VAS.
Yes, we have the next stage and that next stage MUST have a high impedance to allow good performance from the VAS. It's this requirement (or rather lack of it) that kills the performance of a two stage output trying to feed a low impedance load.
Question:
Is the VAS sink/source in parallel with the next stage?

I've over-used my allocation of posts. I must stop.

[canonnica]

Still waiting for my driver transistor to arrive, I'm trying differrent things in the simulator. I've found that the bootstrapped VAS CCS provides around 3 more Volts of peak voltage on the positive side. Nothing on the negative side, hence asymmetrical clipping. Removing the bootstrap capacitor leaving only the CCS gives symmetrical clipping.

I can't explain, it's just an observation.

If someone have a theory i'd be interrested to read it...

Martin.

[AndrewT]

The bootstrap does what it's supposed to do.
It supplies a nearly constant current through the VAS as the output signal voltage varies.
It does this by swinging the supply voltage to the resistor junction (or in this case the resistor to CCS junction) to maintain a nearly constant voltage across the resistor/CCS. That voltage divided by the resistance/effective resistance is what the VAS wants to see.

The result of this is that the bootstrap creates a virtual voltage that can swing outside the supply rail voltage. That virtual voltage is what delays the onset of clipping.

The bootstrap has no effect on the VAS supply rail voltage. The VAS side does not benefit from the enhanced non-clipping characteristic.

Asymmetrical clipping of a bootstrapped VAS sink/source is not a sign of bad design. It is a natural result of the topology.
If you run your amplifier at levels such that it never clips then you won't notice the asymmetrical clipping.

As a footnote;
I don't know why I waste my time with these unhelpful posts, since few seem to read what is posted.