the new EL34 Parallel Single Ended
by Andrea Ciuffoli - 2004
contact me for any problem or question: webmaster@audiodesignguide.com
Design targets:
- get an high power from a single ended without high voltage
- always in class A1 so never in positive grid area to use the most
linear area of output tubes
- normal idle condition to have a long tube life time
- Zout less than 1 ohm to drive any kind of loudspeakers with
perfect damping factor
so very good bass frequency
As output transformer has been choiced the Lundahl LL1627/185mA
with secondaries connected in B configuration to have a primary load
impedence
of 2.3K on 8ohm load. so each EL34 will have a load of only 2300 * 4
= 9200ohm.
The LL1627/185mA
have a primary
inductance
of 9H and the EL34 have a nominal plate
resistance about 910ohm
and four 910 / 4 = 227.5ohm so in theory the low frequency cut-off will
be egual to Ra / ( 2 * pi * 9H) = 4Hz (-3db).
Using the LL1627/185mA each EL34 will be set with a bias about 185 /
4 = 46.25mA that produce at 430V an anode power dissipated of Pa = Va *
Ia = 430V * 46.25mA = 20W that is very less than EL34 max power.
You could request without any problem to Lundahl a custom
LL1627/220mA or a 240mA to increase the Ia to 55mA or 60mA per each
EL34 but there is no advantage.
Schematic
I have all gate of output tubes at the same bias volatge because
I have used a matched quad by
but if you want to use spare output tubes without any match you should
use a separate bias trimmer and a separated interstage capacitor
per each output tube on the gates and a 10 ohm 1/2W resistor on each
cathode to measure the current of a single tube.
The complete schematic to implement the separated grid voltage could be
found here.
Before to insert the EL34 tubes in the sockets switch on the
amplifier with only the two little tubes (12AX7/ECC83 and 5814/ECC82)
and set the trimmers to get about -50V from the grids of the EL34 and
the ground.
After switch off insert the EL34 and switch on the ampl. to set the
correct bias.
This amplifier, like the my other designs, should be connected
directly to a CD Player or to a DAC without use pre-ampl. and the
sensibility should be enougth in any combination.
Power supply
Components
The interstage capacitor should be a 0.33uF 630V Jensen Copper film
paper in oil or a 0.33uF 450V Auricap from Audience
All the audio transformers used in this project are Lundahl products:
Lundahl is one of the
world´s leading manufacturer of audio transformers used in many
professional studio recording and radio stations so it has great
experience and all these products use a custom audio C-core.
 |
The input stage is a 12AX7 /
ECC83 Mullard or Jan Philips
The driver stage is a 5814 / 12AU7 / ECC82 Mullard or Jan Philips
The output tubes used are the Svetlana El34.
|
About passive components types, I don't leave many choices:
Caddock resistances on cathode, ELNA Cerafine capacitors on cathode.
 |
A valid alternative to very
expensive Caddock is to use a parallel of
two normal 1% metallic oside resistor with inverted phase to compensate
some of their inductance.
Here is show the connection of two resitors to create a parallel. |
Actually the ELNA Cerafine capacitors are not easy to find because are
no more produced but these could replace with the new Jensen
Electrolytic that I will test soon.
 |
I am using a DACT stepper attenuator instead of the
normal ALPS or NOBLE solution, because the the sound is clearer. |
About wire; forget the teflon ! Use only stranded tinned copper wire
with pvc insulator, and for loudspeakers use the same.
A very good quality stranded tinned copper wire could be bought
directly from E-Z-HOOK that
carries an extensive line
of fine stranded and extra flexible wire.
Simulation
The performances of this design has been optimized with
electronic simulation using the WinSpice
.3.1 software for Windows.
The WinSpice 3.1 has been
develop by Mike Smith and it is a beatiful port of original spice3f4
develop by Berkeley University of
California.
It can generate waveform plots to individual floating windows and
contains a powerful scripting language
download the circuit
this show a distortion about 4% at 20W on
8ohm very near to measurement value.
Distortion measurement
Follows some test made with Clio system by Audiomatica about the distortion
spectrum with 8 ohm resistive load
A good 2.2% at Pout = (Vrms * Vrms) / Rload = 8 * 8 / 8 = 8 watt
A good 3.1% at Pout = (Vrms * Vrms) / Rload = 10 * 10 / 8 = 12.5
watt
An acceptable 4.3% at Pout = (Vrms * Vrms) / Rload = 13 * 13 / 8
= 21 watt
Frequency response measurement
Follows some test made with Clio system by Audiomatica about the frequency
response with sinusoidal sweep and MLS impulse.
The graphic show a low cut-off of 10 Hz at -3db and 22 KHz at
-0.25db that is perfect.
Listening test
Details
The layout of the EL34 of only one channel and the input
stage per both channels.
