SEUL 22 watt mono bloc with 13E1.
The two mono blocs shown were built in 1997 to prove nobody
needed
to spend an absurd amount of money to achieve the best sound from a
Single Ended
design.
Some audiophiles have said these amps have a midrange you would die
for.
But they also have an excellent 20 Watt full power response from 20 Hz
to 65 kHz.
They produce 22 watts into 8 ohms, and 27 watts into 4 ohms,
which gives them enough power for most listeners with normal modern
speakers of typical sensitivity of 90dB/watt at one meter.
These amps use the rugged 13E1 beam power tetrode which probably
stopped being made in the late 1960s. This tube never been fashionable
among the hi-fi
cogneccenti and magazine paparazzi of tube-land mainly
because it is a beam tetrode, and not a
triode, and the hi-fi press largely
consists of people who have almost zero technical skills,
understanding,
or experience.
This is good, because the stocks of the tubes will last longer and
prices
will remain lower. I would not want to see hoarders and sharks buy up
stocks of such tubes
because they become both popular and rare!
Hoarders often put their purchases in a box under the bed and then
forget they have them, thus depriving the world of good tubes.
I first bought a pair of these tubes in 1996 without knowing
what I would
do with them, and then explored their capabilities for SE triode and
Ultralinear connection by
taking the screen to various taps on an SE OPT.
In pure tetrode the tube is not very linear and generates a large array
of
harmonics even at a few Watts.
I did not find it very suitable for pure SE tetrode with a fixed
screen voltage.
But when the screen is connected to the anode to make the tube work
like
a triode the linearity is no
worse than many other large genuine triodes.
It has a very low triode Ra of only 300ohms, and it will generate a
very sounding nice 16 watts into 1.5kohms. I found that the screen
could
be connected to a tap on the OPT primary
at 66% of the primary turns and
there was negligible increase in the complexity of THD spectra compared
to
triode, where the THD consisted almost entirely of 2H.
33% UL taps gave spectra too similar to tetrode.
So I settled on the 66% UL tap for the 13E1, and got up to about 27
Watts
instead of only 16 Watts with triode, but because Ra with 66% UL is 600
ohms, some global NFB
was essential to keep the output resistance of the
amp low for a good damping factor.
Unlike most transmitting tubes such as the 845 and 211, the 13E1 does
not require a +1,000V supply. 13E1 were designed to work with anode
B+
between +200V and +800V with a screen supply low as +150V.
The low voltages allow the use of
the indirectly heated cathode as
in 6550 or KT88 etc. In fact 13E1 has two cathodes, not just one, to
allow a massive peak current ability of about 800 mA. Some people have
told me they have used the tube for an OTL amp! Although a pair can
make 200 Watts in class AB
tetrode, I wanted to make an SE amp so
efficiency limits the pure class A output
to about a maximum of 40% of
the safe working maximum of plate input power of 72 Watts.
The tube
data suggests plate power max, or max Pda is 90 Watts, but this
figure is a design figure for combined dc and ac Pda in class AB
push-pull
amps where the ac duty cycle is
never continuous as it is with pure class A.
Hence I got a maximum power output of around 28 watts for SEUL
operation.
If 13E1 is set up with Pda = 90 watts at idle the anodes will glow dull
red
and the sound will be quite dreadful and the tubes will not last long.
Its now 2011, and the 13E1 I bought in 1996 lasted well until
2006 with
thousands of hours. Although there is some slight reverse grid current
they
are serviceable
and produced the seame power and distortion before the
owner replaced them as a precaution. The new
pair of 13E1 look like
working for many years.
The 13E1 has a similar power output to a 211 or 845 transmitting tube,
and the same excellent sonic characteristics, IMHO, if set up just
right.
The selected drive tube in 1997 was ECC32 and input was
12SL7,
both with each triode wired in parallel.
In 2006, the drive tube is a 6V6 wired in triode and 6SL7 input
triodes.
The 6V6 driver tube gives slightly less thd, but allows the use of much
lower
bias resistors for the output tubes to prevent slight DC bias drift
after
the last 8 years years of constant use.
The amp has been constructed so that if no 13E1 are available
in 10 years
four EL34 tubes could be used instead in single ended UL or triode
mode.
The tubes are mounted on a sub chassis which could be unscrewed and
replaced by a new chassis for different tubes. The power supply has
several taps for
different HT. However, stocks of NOS 13E1 are not
hard to source, and the owner of
these amps has a spare pair for the
next 10 years of use.
Chassis is brass plus aluminum and the enclosure right around
the
power supply is painted mild steel.
Size is 470 long x 240 wide x 220 high and weight is about 25 Kgs.
A perforated steel cover is normally supplied, and which screws
down
over the tubes, but is not shown.
These amps were used with pairs of Turner Audio speakers at a
March 2001 meeting of the Audiophile Society of NSW, ( ASON ),
and they provided excellent
sound for a large room with 30 listeners.
Schematic of mono bloc power
amp with the power supply
included :-
The schematic is fairly basic but a few items need careful
attention
should anyone wish to build a sample of this amplifier.
Where possible start earthing and good wiring practice should be used
with point to point on tag strips.
The stability of any tube amplifier including this one will depend on
the quality of the output transformer.
I usually wind all my own OPTs because custom winders cannot be relied
upon to
wind the OPT how I like them to be wound, if ever anyone to
can be found to wind OPT. But the OPT should have more than
15H of
primary inductance, less than 5mH of leakage inductance, less than
7% winding losses, and
less than 1,000 pF shunt capacitance at the anode
connection. That basic specification is not so easy to
find off a shelf anywhere.
Even with a spec like that the circuit needs to have some tweaking of
the open
loop gain and phase shift character so that when global NFB is applied
it
is impossible for the
amp to ever oscillate regardless of whether there is a
load connected or not.
R & C values of the following components must be selected and
checked
for correct maximum stability margins :-
R6&C5, low F phase shift reduction; values shown should suit most
OPTs.
R33&C27, high F phase shift and gain reduction; they need to be
carefully
trimmed.
R20&C14, high F phase advance network for voltage NFB; must be
trimmed.
R19&C12, high F zobel damping network to stop parasitic
oscillations at RF.
R21&C15 high F zobel damping network to provide a load at above
100kHz.
All the values mentioned are fairly critical and cannot be
assumed to be
as I have suggested because your OPT will have different stray leakage
L
and capacitances
compared to what I wound myself.
The schematic shows two global NFB loops, one is negative
voltage FB,
the other is negative current FB.
There are 4 terminals on the rear of each mono bloc amplifier. Two
are
connected to the "COM", or common active output terminal.
Each of the other two are
connected to either the VFB or CFB terminals
shown. With CFB, the Rout of the amp is about 1ohm, and current FB
helps
stabilize the amp at HF, and may give better sound into awkward
ESL loads, or be better to drive
fairly constant impedance midrange
speakers in a bi-wired situation. The
voltage NFB is still operative.
When COM to VFB is used for the speakers there is only voltage FB and
since virtually no speaker current flows in R22, 0.1 ohms, no current
FB is sent back
to the input at V1 cathode. But at above audio F there
IS still some current FB since some HF
currents will flow in R21&C15
and these tend to make square waves into capacitor loads have less
overshoot.
Feedback application can be a complex issue; I make no apologies
for
confusing ppl if I appear to oversimplify. People are welcome to
do their own analysis.
Most people select the COM to VFB speaker connection as being
the one
that sounds best.
NFB isn't evil unless it is abused by ignorant designers or
wannabe
DIY amp experts.
Here is another image of one amp with the psu cover removed,
showing C-core output transformer, large choke, caps, and large
power
supply transformer at the top right rear.
Distortion Graphs.....
in 1960. The 1960 curves usually had a linear scale for both THD and
output voltage which tended make the level of THD look lower than
it really was especially at low power levels of a few Vrms output.
This practice reduced the reluctance of buyers to buy an amplifier which
had high THD. Most buyers thought that all distortion is bad.
I have used a LOGARITHMIC scale for the THD quantities so that the
amount of THD at low levels is VERY easily able to be read off.
All distortion is all bad. Distortion harmonics are always generated in all
equipment used for audio and its better to be honest about drawing graphs
which will show how much total harmonic distortion, THD, is present at
the low levels at which most amplifiers are used.
I have indicated the 1.0 Watt level on each graph above and you can easily
read what the THD is at this low level. A 1Watt average level for 2 speakers
of a stereo system profuces an SPL of about 91dB.
Most ppl think an average SPL level of 85dB is quite loud enough,
with peaks rising perhaps to 100dB. To achieve such levels with a pair
of modern speakers at 3 metres away in an average room needs an
average power level in each amplifier of approximately 0.33 watts,
which is 1.62 Vrms into 8 ohms.
The logarithmic scale for THD cannot ever reach down to zero, because
the scale would have to be drawn too high to reach below 0.01% to
0.001% and 0.001%, and such low levels of THD below -80dB of
the signal are not neeed to be known, and may be smothered by the
noise.
Curve A graph shows THD rapidly rising from what is 0.01% THD at
0.05Vrms output, ie, 0.03 milliwatts, or what is close enough to 0.0
Watts.
Curve A shows that THD = about 0.3% at 0.25 watts, ie, 1.41Vrms
with zero global NFB, 8 ohms.
Curve C shows THD falls to 0.05% when 16db of NFB is applied
for the same 1.41 Watts for 8 ohms.
As long as THD remains below 0.1% at normal levels the THD will
not affect the sound quality. The NFB reduces all forms of distortion
by a factor of 1/6.
Now the THD in itself is quite harmless, because all musical tones have
a rich harmonic content above the lowest or fundemental tone present,
and these harmonics may even sum to have a higher amplitude than the
fundemental F.
Therefore altering relative levels of the musical tone harmonics by a
fraction a percent does not change the tone heard. But wherever harmonic
distortion is produced, there is a far worse sounding type of distortion
produced, ie, Intermodulation distortion, or IMD. This means that where
there are two signals produced with say 4 Vrms of 80Hz and 1Vrms of
1kHz, and there is harmonic distortion, then the 1kHz level will be
amplitude modulated. This is the equivalent of having tones produced
at 920Hz and 1,080Hz. These tones are at a low level of about 3 x THD
levels of the larger output voltage. Exactly what IMD levels are is able
to be calculated, but there is little space to devote to that now.
But with many musical tones present, there is a huge number of
"intermodulation products" and if they could be played in speakers
without the wanted undistorted signal present, the sound is like a rustling
noise rising and falling in time with the music. It is a veil on the music,
and multiple IMD tones are often non-harmonious with music being
played. The significance of such IMD may be argued. I find that the
general clarity of music using an SE amp is often better than a a PP
amp with less THD, and less IMD. We should always try to minimise
THD and thus minimise IMD.
Curve B shows the amp loaded with a 5 ohm speaker instead of the ideal
8 ohm speaker. In this case the voltage gain of the output stage is about
3db less than with 8 ohms, so the applied amount of NFB is also 3dB
less because no change has been made to the value of the R20 through
which the feed back signal flows to the V1 cathode.
Thus the THD with the lower load value is higher because of two reasons,
one being that less NFB is effectively applied, and the other reason being
that the load line that the tube sees for 5 ohms at the output is a steeper
load, and one in which more THD products are generated.
Its even a worse situation when 4 ohm loads are used with dips to
3 ohms. But I have demonstrated these amps to audiophiles using such
low impedance speakers that I have built myself, and obtained rave
receptions; they liked the sound from the system very much.
It seems to me that SE tube amps may have considerably more measured
THD than other types of amplifiiers yet they are judged to give far
better subjectively judged fidelity!
The THD with 16 ohm speakers would be a lot lower than what is
measured with 8 ohms because the higher load generates less THD in
the output tube, and there is is greater tube gain hence more effectively
applied NFB. This is generally true of all amplifiers, and when in doubt,
always try to choose speakers with a higher nominal impedance than
the the speaker load labelled on amplifier output terminals, and the
the higher the speaker sensitivity, the less amplifier distortion is produced.
Therefore speakers of 8 ohms may be used where terminals are labelled
"4 ohm" but one should not use a 4 ohm speaker on terminals labelled
for 8 ohms.
Curve A in the above graph sheet is quite typical of many SET amplifiers,
ie, Single Ended Triode amplifiers using just a single triode or paralleled
multiple triodes which boast that they do not have any global NFB.
So the Curve C showing maximum THD = 0.5% at 20 watts is a perfectly
acceptable result.
For comparison purposes, I have plotted curves D and E showing typical
THD curves for Push Pull triode or UL class AB1, D, and class A, D, and
with the similar levels of NFB used for the SEUL amp with two load values.
Notice that PP amps have much lower THD. However it is of little
significance because the THD is mostly 3H, and the resulting mix of IMD
products are alleged to sound worse than those produced by SEUL/SET amps
with NFB and the same power ability.
Nevertheless, the class A PP triode or UL amp will sound excellent.
People who favour SE amps usually say the SE amp will better re-produce
the natural recorded midrange warmth without any loss of detail and
precision in the dynamics, but I have heard too many good sounding
push pull amps to agree which is generally better.
I am at a loss to explain all the opinions I hear.
SEUL Vs SE CFB.
The above graph has a log scales for both THD% and Output Voltage, V).
There are two other scales below the graph for Ouput Power and dB SPL.
For example, where VO = 2.24Vrms, Power from each amp = 1.0 Watt,
and with 2 speakers each producing 90dB SPL at a Watt at 3 Metres in an
average room, ( more SPL than in an anechoic chamber ) then 2 speakers
should measure 93dB SPL.
The graphs show THD from the SE35 amps with local Cathode Feedback,
CFB, is about 1/5 of that of the SEUL with 13E1.
The SE35 and SEUL have been set up with approximately the same total
amount of NFB.
In years between 2003 and 2007 I had two customers to whom I had sold
samples of both amplifiers and they both listen to each others amps during
friendly house visits and they had a difficult time deciding which was the
better sounding amplifier. The owner of the SE35 sold it to fund yet
another amplifier purchase because he seemed addicted to regular amplifier
changes.
One cannot expect anyone to keep going to see the same movie, drink wine
from the same vinyard, or decorate the house with paintings from the same
artist, or remain married to the same wife forever. But I was flattered his
ownership of the SE35 was the longest time he'd ever kept a good piece
of audio gear. The fellow with the SEUL snapped it up when it came up
for for sale. He eventually decided the SE35 was a better amp than his
SEUL25. Some considerable modifications were done to the SEUL to
convert it to the SE32 at monobloc-se32-13ei-cfb.html
The performance of both SE35 and SE32 are now very similar, maybe
because both have very similar overall design with two triodes used for
input and driver tubes and the output tube is a multigrid power tube using
local cathode feedback.
SE35 THD figures are described in much more detail in the page devoted
to the SE35CFB amplifiers.
The THD in the SE35 is least when RL = 5 ohms, with THD being higher
when RL is a value either below or above 5 ohms. THD cancelling
between the driver stage and CFB output stage is most effective at 5 ohms.
The result of the work with SE35 with CFB led me to conclude that
the use of a tertiary OPT winding for local CFB for the 13E1 is used
may well prove to make a better sounding amp and one which would
also measure better than a plain UL connected tube. My customer who
owns both SE35 and SEUL thinks they do sound similarly stunning.
And to those who still doubt NFB does any good in SE amps, let me say
that each and every triode already has a local NFB loop within itself.
It occurs because of the electrostatic effect of the anode voltage upon
the flow of electrons to the anode. The control grid voltage changes the
electrostatic field between itself and the cathode, thus controlling electron
flow. So for a positive going voltage at the grid, there is an increase in Ia.
But as Ia increases, the load voltage across the load bettween B+ and
anode increases, thus making the anode voltage move negatively.
The tube is an inverting amplifier. Now the anode voltage swing is
usually much larger than the grid voltage swing. This anode voltage
change also affects the electron flow from the cathode. So as the anode
voltage falls, it tends to lessen the flow of electrons attracted to itself.
In other words, the anode voltage change opposes what the grid is trying
to do. The two electrostatic fields sum together to give a resultant field
that works on electrons in the Ia flow. This is a form of local shunt
feedback, equivalent to having a shunt resistance network of two
resistors between anode and the signal input, with the join of the
resistors going to the grid. I use such local resistance FB networks in my
preamps.
The internal electrostatic FB in the triode is almost entirely prevented
by the addition of the screen grid as used in pentodes and beam tetrodes.
The screen usually has a fixed voltage relative to the cathode to enable
pentodes and tetrodes to have very high useful gain and much more
output power than a triode with a similar Pda rating. The price for this
extra "free" gain and power is much higher Ra giving a very poor
damping factor and a whole lot more harmonics, mainly odd numbered,
and changing in relative mix and phase with load changes.
The screen affects the Ia flow in the same manner as it would if it was
an anode.
So when the screen is connected to the anode, the multigrid tube then
behaves similarly to the genuine 3 electrode triode, because the tubes
internal NFB has been allowed to operate.
If external loop feedback is used for a pure pentode or tetrode tube,
then the complex harmonic spectral content is largely unchanged,
simply reduced in level by the NFB. 20dB of global NFB is needed
for pure pentode amps to get their Rout low enough and to curtail
distortion. IMHO, it is better to try to prevent formation of the odd
numbered H within the OP stage with the UL connection then less
global NFB need be applied.
The pentodes and tetrodes do allow partial application of the anode
signal to the screen, hence there is the Ultralinear connection. In fact,
UL is not "ultra" anything, but does result in the multigrid being able to
be operated in class A1 instead of A2 and to give virtually the same
power as pure pentode or tetrode, but with THD spectra like triode
with very low odd numbered harmonics.
When local CFB is used with a fixed screen voltage, the connection is
called the Acoustical, as it was used in early Quad-II amplifiers. The use
of only 10% of the total anode to cathode signal applied to the cathode
gives 10% series voltage NFB to the control grid to cathode interface.
With screen voltage fixed, there is some FB applied between cathode
and screen, ie, the Acoustical is really like having a 10% UL connection
but with 10% of applied voltage FB to the gid - cathode interface.
The result with KT66 in Quad-II gives the KT66 tetrodes the same
effective Ra as for triode connection, 1k6, much lower than the pure tetrode
Ra = 33k0, and yet power output is double what the A1 triode operation
can be.
Its now 2011 and listening tests have confirmed that the local CFB in the
OP stage worked very well with 13E1.
Back to Power amplifiers.
Back to Index page.