Dynamic bias stabilization was Re: Brook sliding bias operation.



Alex wrote:

I first tried zeners with a revised ST70 schematic shown at my website.

The trouble with zeners is that they do not react to current. They
react to voltage.
So if a tube short circuits itself, Ek is held down but Ik goes high.
In the ST70 protection circuit I developed, series R between bottom of
zeners
and 0V sends an error signal off to an SCR, and amp meltdown is avoided.

The transistor dynamic bypass stabilizers use very non critical setting
for the
transistor bypass. It only reacts to ac current generated in the low
value sensing R.
So Ek can vary quite a bit before triggering the protect circuit

But if a hard sudden drum beat occurs, the high AB current is nicely
bypassed
and Ek barely moves at all, but with a zener, a considerable rise in Ek
has to occur
before it conducts at all.

Yes, that makes sense. The transistor circuit reacts to the
instantaneous current. The threshold can be higher than a quescient
current, while with the zeners it shall be right on the mark. With the
zeners you do not need 1000uF||750R resistors at all -- it will work
as fixed bias, which is, if I understand correctly, critical to tubes
replacement and B+ variation.

Regards,
Alex

The idea of using zeners for cathode biasing tubes to give a form of
fixed
bias lacks the ability for adjustment.
I'd never use zeners to do it. The zeners are slightly non linear and
damn noisy.
You get zener coloured sound.

Fixed bias isn't fixed because its adjustable, but once its adjusted,
its fixed, right?

Makers have found making a negative grid supply cheaper and better than
having
say 35V at 50mA dropped across zeners for each OP tube. Kinda wasteful.

But let's go back to my page in the issue...
http://www.turneraudio.com.au/schem-300w-5-bias-stabilizer.html

The first schematic is for a one pair of OP tubes and has 1,000uF and
750 ohm C&R cathode network.

R9 and R10 don't have a value shown, an ommision on my part which I must
fix soon,
but from the text they are both 10 ohms.

10 ohms is a good value to begin with.

If Ia = 30mA, the Vdc across R9&10 is =0.3Vdc. The bases of the BD239B
have 1k connected
between R9&10, and the base to 0V caps, C7, C8, could be 0.01uF,
giving a pole at about 16kHz because one does not want to have the
circuit work at above
this F.
The base current will be zero if the Vbe = +0.3V. Collector current also
is 0.0mA
But current will begin to flow at about Vbe = +0.6V, and the current
flow versus
10 ohm voltage transfer is not very linear, but its nonlinearity
complements the non linear tube current as it goes into class AB
operation from class A.
As the bjt draws base current during its turn on, the base current
causes a voltage across the 1K base resistors,
R11, R12. 1mA of base current gives a whole volt across the 1k.

Under normal operation, base current is kept low.
The collector current Ic = hfe x Ib
The max Ic is when bjts are turned on enough to pull nearly the full
cathode bias across the
R13 and R14, 47 ohms, so if Ek = 35V, max Ic = 0.744 amps peak.
If the hfe of the BD239B was 20, then max Ib would be 37mA, which
means 37V needs to be developed across the 1k R12, R13, and also across
R9 and R10,
and this is quite impossible for the tube to ever do,

The tube might manage 0.7A peak, giving 7V peak across the 10 ohms of
R9&10.
So adjusting R11&12 will determine just how smartly the bjt will turn
on.

The values I've shown work OK, but if you play with them and analyse on
your own,
you see what happens. I should not have to spell every tiny little
detail out.

With Ia dc = 30mA, the 10 ohm Vdc = 0.3Vdc, and the cathode dc could
rise
by 20mA before dc turns on the bjt, so a rise of Ek from 22.5Vdc to
37.5Vdc is possible without
the bjt turning on, but let's say we set the excess cathode current
detector to trip the protect circuit at 30Vdc, then we'd have some
variability
allowable in idle dc which would regulate the Ia.

The text below the schema says additional R from each base to 0V will
form a divider to reduce the Vdc at the base, and make the circuit
less prone to bypassing ac at a small peak current in excess of the
idle.

There should in fact be no AC shunting by the BJT where the audio
signals cause
less than + or - the idle current, and so R9 or 10 can be varied to
accomodate this principle.
If the bjts turn on at too close to the idle dc, you'll find the
Ek will FALL a bit, because the signal current in the tube is triggering
the
release of energy store in the 1,000 uF caps to 0V, and in fact
idle current and class A PO is actually increased.

The circuit works best when a load is chosen that produces
maximum class AB PO.
So in a "50W" PP amp meant for 4 ohms, one may find 60W is possible at 3
ohms

The transistors and their resistors attached should keep Ek no no than
10%
above the idle value when the amp clips into 3 ohms, ie, at 60W.

Between idle and 60W, the Ek should never fall slightly before slightly
rising
to the max at 60W clipping.

So these things have to be methodically checked when anyone tries this
circuit.

It only has a "handful of easy to get parts"
but there are at least 3 important things someone has to trim and get
right.
And diyers hate things like this, they want it easy,
because they hate learning, and hate spending time to learn.
I don't do easy.
I do hard,
I do mentally challenging.

Patrick Turner.
.

Quantcast