Patrick Turner
September 15th 06, 03:54 PM
Hi all,
I have now included output transformer capacitance calculation method
in my pages dealing with PP OPT and SE OPT.
The transformer winding page has bobbin section details added and
instructions on how to use
polyurethane two pack varnish to be able to apply varnish as the winding
is done
and obtain good hard varnishing of the windings without using a vacuum
chamber and hot baking process.
The capacitance calculations don't change a great deal but anyone
winding OPT should want to know
what capacitance to expect BEFORE winding an OPT rather than find its a
bit much after winding one.
Capacitance in OPTs which are interleaved can be calculated using the
basis that
all the capacitances between the P winding and S sections which are at
0V potential
are transformed to a value at the anode connection depending on the
position of the C
along the P winding.
For example, a capacitance of 330pF between P and S at 1/2 way between
the B+
and anode will be transformed to be 1/2 x 1/2 x 330pF = 82.5pF because
the transformation is
at a factor of the impedance ratio which is the turn ratio squared.
The effect of the capacitances between P and S near the anode
connections will be the majority
of the C and the capacitances between P and S near the B+ where the
signal voltages are low
will be small when measured at the anode connection.
The shunt capacitance at the anode terminal is therefore NOT the
simple sum of all the P to S capacitances that exist in the OPT.
In the case of a PP amp where both tubes are working in class A which
covers
most of the music we hear, the C in at one anode is charged and
discharged
by each tube.
Where the same tubes were biased for class B the capacitance would
double because the one live tube
would see the capacitances of the undriven 1/2 of the OPT primary.
In an SE OPT of identical bobbin winding details to a PP type, but with
B+ at one end, the
C will be very close to double the C seen at one anode connection in the
case of the class A PP
OPT.
Knowing what the C is in each case isn't the end of the matter. This C
is the mass of total
shunt Cs taken as many simple summed quantities within the OPT without
regard for how it reacts
with the many amounts of leakage inductances between P and S which also
sum to give us
what we measure as leakage L which is usually reduced to an equivalent
single value of
inductance in series with the anode load.
The testing of my latest OPTs made for 50 watt amps using 845 confirm
the
method of calculations appear to be correct.
I would have to say its easier than trying to wade through the RDH4
pages on OPT capacitances
because the RDH4 is incomprehensible on this subject afaiac, and I have
never met anyone else who
understands OPT capacitance calcs as spelled out in RDH4.
A common mistake by would be experts when winding OPTs is to go for
close coupling.
So they interleave with say 7S and 6P sections, and then use only thin
insulation between P and S sections,
and then produce a tranny that gives -3dB at 15kHz even with triodes
because the
capacitance is so darn high. They may get low LL but having LL = 1mH for
a load
of 5k ohms results in a pole at 800kHz, but if the C = 0.005uF, then the
load = the
reactance of the C at 6.36kHz, and there will be massive extra
distortion above this F at high power levels
because the load has become predominantly C.
So we would prefer to see that in the case where RL was 5k, the C
shunting this load was 0.0005uF,
or 500pF max, and this is not so easy to achieve in a large OPT.
So we may find that we have to settle for C = 0.0015uF, or 1,500pF,
which will have RL = C reactance
at 21.2kHz.
A triode with Ra of say 1k would not suffer a huge phase shift driving a
5k load with 1,500pF of shunt C
because the response roll off and 45degree phase lag won't occur until
Ra in parallel with RL equals C reactance. This would be where the C
reactance = 1k//5k = 833 ohms,
or at 125kHz.
At low levels such C shunting does almost no harm to the signal but as F
rises above
20kHz the output will become much distorted before rolling off.
At 50kHz, 1,500pF is only 2.12k, and the amp cannot maintain a high
output
into the low value load without serious distortions.
In SET amps the shunt capacitance needs to be known and kept to a
minimum.
In my pages I recommend minimum insulation thicknesses of insulation of
more than 0.5mm
which will stop C from becoming too high nearly all cases.
Once you have 5P sections and 4S sections, increasing the insulation
from say 0.3mm
to 0.6mm will halve the C but have little effect on the leakage
inductance which becomes dominated by the
number of interleavings and not the closeness of coupling.
So if you design an OPT with plenty of room to spare on the bobbin,
think of increasing the PS insulation
and it may benefit the response more than adding another interleave.
Patrick Turner.
I have now included output transformer capacitance calculation method
in my pages dealing with PP OPT and SE OPT.
The transformer winding page has bobbin section details added and
instructions on how to use
polyurethane two pack varnish to be able to apply varnish as the winding
is done
and obtain good hard varnishing of the windings without using a vacuum
chamber and hot baking process.
The capacitance calculations don't change a great deal but anyone
winding OPT should want to know
what capacitance to expect BEFORE winding an OPT rather than find its a
bit much after winding one.
Capacitance in OPTs which are interleaved can be calculated using the
basis that
all the capacitances between the P winding and S sections which are at
0V potential
are transformed to a value at the anode connection depending on the
position of the C
along the P winding.
For example, a capacitance of 330pF between P and S at 1/2 way between
the B+
and anode will be transformed to be 1/2 x 1/2 x 330pF = 82.5pF because
the transformation is
at a factor of the impedance ratio which is the turn ratio squared.
The effect of the capacitances between P and S near the anode
connections will be the majority
of the C and the capacitances between P and S near the B+ where the
signal voltages are low
will be small when measured at the anode connection.
The shunt capacitance at the anode terminal is therefore NOT the
simple sum of all the P to S capacitances that exist in the OPT.
In the case of a PP amp where both tubes are working in class A which
covers
most of the music we hear, the C in at one anode is charged and
discharged
by each tube.
Where the same tubes were biased for class B the capacitance would
double because the one live tube
would see the capacitances of the undriven 1/2 of the OPT primary.
In an SE OPT of identical bobbin winding details to a PP type, but with
B+ at one end, the
C will be very close to double the C seen at one anode connection in the
case of the class A PP
OPT.
Knowing what the C is in each case isn't the end of the matter. This C
is the mass of total
shunt Cs taken as many simple summed quantities within the OPT without
regard for how it reacts
with the many amounts of leakage inductances between P and S which also
sum to give us
what we measure as leakage L which is usually reduced to an equivalent
single value of
inductance in series with the anode load.
The testing of my latest OPTs made for 50 watt amps using 845 confirm
the
method of calculations appear to be correct.
I would have to say its easier than trying to wade through the RDH4
pages on OPT capacitances
because the RDH4 is incomprehensible on this subject afaiac, and I have
never met anyone else who
understands OPT capacitance calcs as spelled out in RDH4.
A common mistake by would be experts when winding OPTs is to go for
close coupling.
So they interleave with say 7S and 6P sections, and then use only thin
insulation between P and S sections,
and then produce a tranny that gives -3dB at 15kHz even with triodes
because the
capacitance is so darn high. They may get low LL but having LL = 1mH for
a load
of 5k ohms results in a pole at 800kHz, but if the C = 0.005uF, then the
load = the
reactance of the C at 6.36kHz, and there will be massive extra
distortion above this F at high power levels
because the load has become predominantly C.
So we would prefer to see that in the case where RL was 5k, the C
shunting this load was 0.0005uF,
or 500pF max, and this is not so easy to achieve in a large OPT.
So we may find that we have to settle for C = 0.0015uF, or 1,500pF,
which will have RL = C reactance
at 21.2kHz.
A triode with Ra of say 1k would not suffer a huge phase shift driving a
5k load with 1,500pF of shunt C
because the response roll off and 45degree phase lag won't occur until
Ra in parallel with RL equals C reactance. This would be where the C
reactance = 1k//5k = 833 ohms,
or at 125kHz.
At low levels such C shunting does almost no harm to the signal but as F
rises above
20kHz the output will become much distorted before rolling off.
At 50kHz, 1,500pF is only 2.12k, and the amp cannot maintain a high
output
into the low value load without serious distortions.
In SET amps the shunt capacitance needs to be known and kept to a
minimum.
In my pages I recommend minimum insulation thicknesses of insulation of
more than 0.5mm
which will stop C from becoming too high nearly all cases.
Once you have 5P sections and 4S sections, increasing the insulation
from say 0.3mm
to 0.6mm will halve the C but have little effect on the leakage
inductance which becomes dominated by the
number of interleavings and not the closeness of coupling.
So if you design an OPT with plenty of room to spare on the bobbin,
think of increasing the PS insulation
and it may benefit the response more than adding another interleave.
Patrick Turner.