AUDIO
OUT
AUDIO OUT
L
R
By Jake Rothman
Transformers in audio – Part 3
Fig.25. An early prototype of the Steve
Dove microphone preamplifier made by
Student George Perakis.
Fig.23. The inside of a 1970 Rogers Ravensbrook amplifier. Note how the cores of the
driver transformers (centre) are mounted at right angles to the mains transformer core
(top left) to prevent hum pick-up.
I
n Part 2, last month, we looked
at some of the limitations of audio
transformers. This month we’ll cover
techniques manufacturers use to improve
performance in real transformers.
Fig.24. PL84 valve guitar practice amp (see
Fig.17, Audio Out, September 2022). The
output transformer used was the B18A011F
from Carnhill Transformers.
54
Loose windings
First though, it seems the level of technical scrutiny is highest in Surrey, and I
will pass on details of a few errors that
slipped through proofing in recent articles – all reported to me by readers
from Surrey!
Fig.6 from Transformers in Audio –
Part 1 (PE, July 2022) shows the Roger’s
Ravensbrook audio amplifier; note that
the phase-splitter transformer allows output transistors of the same polarity to be
used – not driver transistors – as stated
in the caption. Also, the unmarked lower preset is 100Ω and the resistor above
is 2.2kΩ. This type of circuit is a good
introduction to power amp building
because if you make a mistake in the
driver stage you don’t blow-up your
output transistors – the transformer
isolates the two sections. The interior
of the Ravensbrook is shown in Fig.23.
In Transformers Part 2 (PE, September 2022), Fig.14a, sharp-eyed reader
Peter Brearey has pointed out that the
upper output transistor is inadvertently shorted out by a line. This should
have taken an 820Ω 1W resistor to the
junction of the decoupling network
of the 47µF capacitor and unmarked
(47Ω) resistor, not the positive rail. It’s
the same Rogers configuration shown
in Fig.6 from Part 1. This circuit prevents current being drawn out of the
output stage which would unbalance
the quiescent current passing through
the output transistors. Also, the speaker impedance of the Mullard Class A
amplifier in Fig.18a should be ‘3Ω’, not
‘3W’ and the transformer part number
is PT1 (possibly a Partridge design). My
apologies for the errata, and the absence
of photos for some of the circuits. (Note
Fig.26. Pye Vanguard police radio monitoring unit. Built in 1965, this ran well into the 1980s.
Practical Electronics | October | 2022
Primary
Primary
section 1
section 2
Join centre tap 1500 turns 1500 turns
F
S
Secondary to
loudspeaker
150 turns
22 SWG
F
n
From output
valve
3000 turns
34 SWG
S
Core section, 32mm2
Fig.29. Sectioning of windings improves high-frequency
response. This was a simple valve radio output transformer.
at 50Hz or 60Hz.
Power transformer laminations are
Fig.27. Large, gapped output transformer used in the Vanguard
typically around
police radio monitoring unit. It delivered a couple of class-A
half a millimetre
germanium watts into a Celestion 5-inch speaker.
in thickness, while
audio transformers typically have lamthat I do aim to physically build and
inations of 0.35mm. Very expensive
test all my circuits since paper calcuaudio transformers may have lamilation can only get you in the ballpark
nations just 0.1mm thick. Sometimes
with audio design.) Fig.24 shows the
imperial sizes are used, typically 15 mils
PL84 valve amplifier given in Fig.17 and
(1 mil = 0.001 inch or 0.0254mm) which
the transformer microphone pre-amp
is 0.38mm. Audio cores are run at a
circuit from Fig.20 is shown in Fig.25.
much lower flux level than power transFinally, a bit of history, the 1964 Pye
formers to reduce distortion by avoiding
Vanguard police radio amp (Fig.19a) is
saturation. What matters is obtaining
shown in Fig.26 and Fig.27.
maximum permeability at high frequenNow we’ll get back to the special
cies so that inductance is maximised for
construction techniques used in audio
a given number of turns. This then mintransformers to improve their frequenimises winding capacitances, extending
cy response and distortion.
the response. Mumetal laminations
have about 20-times the permeability
of standard M6 FeSi transformer steel.
Laminations
However, it can only support about a
A good rule of thumb for audio transquarter of the maximum flux density,
formers is the thinner the laminations,
which means it can only be used for
the lower the losses at high frequencies.
relatively low powers. A common audio
This is because the flux tends to concenapproach is to use 49% nickel or ‘Ratrate on the surface of the laminations.
diometal’ cores which provide a good
The result is that audio transformers
compromise between cost, distortion
need to have much thinner laminations
and power handling. This is what is
than mains transformers which operate
used in most of the medium-cost audio
transformers, such as the Vigortronix
VTX-A series.
Brian Sowter, who at the fine age of 83
still consults and deals with technical
support at Sowters Transformers, told
me that the Sowter 5069 transformer
used both Mumetal and M6 laminations
interleaved. This gives the advantages of both. As the Mumetal saturates,
the M6 laminations take over, giving a
seamless transition. They used to call
it the ‘Dolby mix’.
Fig.28. CineMag transformer used in
Some small transformers use ‘T’ and
a condenser microphone. It only uses
‘U’-shaped laminations and occasion‘U-shaped’ laminations. The alternate
ally on some microphone transformers
interleaving gives the misleading
just ‘Es’ are used. These lamination
impression of a magnetic gap. Notice the
shapes give higher permeability, but
twin bobbins/coils.
Practical Electronics | October | 2022
at the expense of maximum saturation
level. This can be seen in the CiniMag
microphone transformer in Fig.28.
Some microphone transformers have
U-shaped laminations with windings
either side. This arrangement reduces
hum pickup by cancelling out the voltage induced by the hum in the two arms
of the core; similar to a hum-bucking
guitar pickup.
It is rare for audio transformers to have
bolt holes punched in the laminations.
These discontinuities in the magnetic
circuit can cause flux concentrations
and raise distortion. To avoid this, the
cores are generally clamped or potted.
(Note, however, that low distortion is not
necessary for guitar amp transformers,
it’s part of the guitar’s ‘sound’ or tone.)
Sectionalised windings
If the windings are just two separate
coils, such as in a split-bobbin mains
transformer, the leakage inductance is
high because the magnetic coupling is
low. This is one reason mains transformers give very poor high-frequency
response when used as audio transformers. An effective technique with
valve output transformers is to split
the high-impedance primary winding
into two layers with the low impedance secondary winding in-between,
as shown in Fig.29. This improves the
frequency response at 10kHz by around
9dB relative to 1kHz. Fig.30 shows a
Sowter 5069 line output transformer
with sectioning. This splitting up of
primary and secondary windings can
be extended to up to 14 sections, as in
Sowter’s 50W KT88 ultra-linear output
transformer for the GEC amp design.
This greatly increases the cost, but it
does enable response up to 40kHz to
be achieved. It is essential to have a
wide flat frequency response (with its
accompanying low phase shift) if a
large amount of negative feedback is
used, otherwise oscillation could occur.
55
Fig.30. If you look down into the bobbin the three sections of
this 1:1 ratio Sowter 5069 output transformer can be seen,
along with the black screen connection. There is a yellow wire
joining the two halves of the outer and inner sections.
Generally, the bigger the turns ratio, physical size and impedance, the
worse the high frequency response. It
is much more difficult to extend the response of a 20W 8000Ω:4Ω valve output
transformer than a 50mW 600Ω:600Ω
line-output balancing transformer.
Interlayer insulation
Good insulation is required between the
primary and secondary windings in valve
output and power transformers – nobody
wants +400V HT on their loudspeaker
terminals. This is achieved by wrapping
polyester tape around each winding. Wax
impregnated paper was once used, but it
suffered from leakage current and moisture retention, just like old waxed-paper
capacitors. Another approach was a special varnish-impregnated glass fibre tape
called ‘Empire Cloth’, which was quite
good. The Empire name is still around
– Empire Tapes make the standard yellow Mylar heat-resisting transformer
sticky tape used today (an alternative is
3M type 56). For prototype low-voltage
audio transformers, I have used gas-fitters PTFE tape, which is good for pulling
everything tight.
Bifilar
Bifilar winding is where two lengths of
wire (ie, the transformer’s primary and
secondary) are wound on the core together at the same time. The windings
thus occupy almost identical magnetic space and have the closest possible
magnetic coupling. This technique can
be scaled up to trifilar and quadfilar, as
in the Ravensbrook and Ravensbourne
amplifiers. It is better to buy special multifilar wire ready for such winding – it
can be a tricky doing it with two rolls.
Driver transformers with centre-tapped or
split secondaries are wound this way to
avoid switch-off spikes from the output
transistors operating in class B. The disadvantage of the bifilar approach is that
the inter-winding capacitance is high and
the maximum isolation voltage is low.
Fig.32. This is the amount of fine wire in a Greenweld X7920
interstage transformer. 1770 turns on the primary and 330 turns
on the secondary, giving a 5:1 ratio (the measured ratio was
somewhat less due to losses). Note the nylon moulded bobbin
and laminations.
56
Fig.31. An old Belclere potted VTX-A-style transformer which I
have ‘de-potted’ (so you don’t have to!). Note the soldered joint
on the copper screen and the lighter-coloured 49% Nickel (Ni)
laminations.
Impregnation
In the old days (from the valve era up to
the early 1960s), the varnish insulation
on enamelled winding wire was inferior
to the more uniform polyurethane coating of today. It was essential to insulate
between each layer of the winding with
paper, then impregnate the whole lot in
wax to render the unit impervious to
damp. However, the wax used to melt in
service (when running hot) and drip out
of the unit. Later, the wax was replaced
with polyurethane which was much better… unless you needed to disassemble
the transformer for repair.
Sometimes transformers are potted in
epoxy resin which minimises microphony, but it’s rigidity can cause lead-out
wire breakages and its irreversible
solidity makes repair impossible. Impregnants are useful to reduce emitted
noise, but they also increase capacitance. I have occasionally heard output
transformers ‘singing’ due to loose
windings and/or laminations.
Fig.33. There are many Chinese ‘clones’ of classic British audio
products, such as this Neve-copy microphone preamplifier. It
suffers from bad hum due to a PCB PSU earth loop and the
input transformers having just two steel (not Mumetal) cans
which give poor shielding from extraneous magnetic fields.
Practical Electronics | October | 2022
Fig.34. Screening can for Vigortronix
VTX-A transformers. Fitting the
transfomers can be a challenge, but it’s
easier with a little silicone grease.
Fig.35. The base plate of a transformer
shielding can must be soldered to the
transformer’s ground pin before it is
mounted on the PCB.
Interwinding screens
Copper foil screens are often placed between windings on
input balancing transformers to prevent interference being
capacitively coupled; for example, high-frequency switchmode power supplies buzzing digital audio sources. Care
has to be taken to ensure a shorted turn isn’t created, the
ends of the screen must be insulated from each other by
wrapping insulation tape around one end. The screen must
be earthed or the interwinding capacitance between the
windings is made worse. These screens are often referred to
as ‘Faraday shields’. Although a screen is not essential on
output transformers it is still worthwhile connecting it to
earth or chassis if the transformer provides one, especially
if it is being driven by something using an un-earthed power
supply. Fig.31 shows a Beclere/VTX line-output transformer cut open to show the screen under the primary. This is
also connected to the core. (As an aside, since I was in a
dismantling mood, Fig.32 shows a Greenweld interstage
transformer revealing a mass of fine wire.)
I sometimes connect the shield via a 39Ω resistor to signal ground (if that is the only ground available) to reduce
instability in the amplifier driving the transformer caused
by the capacitance from the primary to the shield. Normally, the shield is connected to chassis/mains earth and pin
1 on the input or output XLR connector.
A better approach to isolating the shield capacitance from
the input driving amplifier is to use an LR network. Jensen
sell these, (load isolator, part number JT-OLI-3) – it is simply 40 turns of fine 30 AWG wire wound around the body
of a 39Ω 1W carbon composition resistor (5.5mm x 15mm
non-magnetic). This gives an inductance of around 3.7µH.
This sometimes isn’t enough, so I may use a separate inductor of 10 to 82µH.
Fig.36. The upper part of a transformer
shielding can must also be earthed. This
can be soldered along the base plate or
use a bit of wire from the PCB to the can.
say ‘gently’, because one must never bend or bash annealed
Mumetal because the permeability is reduced. Restoring lost
permeability is possible, but difficult for the home constructor – you have to re-anneal it by heating it to 1150ºC for four
hours, and then cool it down at 250ºC per hour.
Note that these cans must be earthed. This is done by soldering the can base to the transformer’s screen/core earthing
pin (see Fig.35). The upper can must also be soldered to the
base plate, as shown in Fig.36 before mounting. Vigortronix
supply versions with the cans ready fitted and I usually get
these if I need screening, often the VTX-101-003, which is
numbered VTX-102-003 when supplied with the can.
Next month
That’s all for Part 3. Next month, we will conclude with
comprehensive information on sourcing audio transformers,
plus a very handy pair of PCBs for mounting Vigortronix
audio transformers.
Your best bet since MAPLIN
Chock-a-Block with Stock
Visit: www.cricklewoodelectronics.com
O r p h o n e o u r f r i e n d l y kn
o w l e d g e a b l e st a f f o n 0 2 0 8 4 5 2 0 1 6 1
Components • Audio • Video • Connectors • Cables
Arduino • Test Equipment etc, etc
Screening cans
Input transformers can easily act like pick-up coils causing hum. Totally enclosing the transformer in a metal can
is the solution. Mumetal is the best material (ordinary steel
doesn’t work very well). There has been a spate of Chinese
Neve microphone preamplifier clones, such as the one shown
in Fig.33. These suffer badly from hum when mounted in
a rack near other equipment because they don’t use the
correct (pricier) materials needed for effective transformer screening cans – just ordinary steel. Some transformers
add a copper ‘belly-band’ which also reduces hum pick up.
Very expensive input transformers will have many layers
of screening material.
Optional screening cans (VTX-102-000) are available for
the Vigortronix/OEP range, as shown in Fig.34. They are
sized 30mm long x 25.5mm wide x 23.75mm high and it’s
quite a squeeze to insert the transformer – and much harder
to remove. A bit of silicon grease is useful, and they should
be pushed on gently with a vice along each edge in turn. I
Practical Electronics | October | 2022
V is it ou r S h op , C a ll or B u y on lin e a t :
w w w . c r ic klew ood elec t r on ic s . c om
0 2 0 8 4 5 2 0 1 6 1
V is it ou r s h op a t :
4 0 - 4 2 C r ic klew ood B r oa d w a y
L on d on N W 2 3 E T
57
