Guitar Amps 2017.
Tubed guitar amps are not for Hi-Fi but are an essential part of
the gear used by countless musicians.
I have repaired or modified hundreds of them.
The Internet is overloaded by a serious amount of bullshit about
what makes a good guitar amp and if I had
one dollar for each time yet another wannabe expert tells everyone
else how he must change his amp, I would be rich.
Unfortunately, 90% of the mods proposed online do not do very
much, can make sound worse, and possibly make
the amp oscillate or become unstable or blow up. Almost all
wannabe experts have zero idea about fundamental
behavior of L, C and R, and all of them may disagree with me.
The best factory made guitar amp may have been the 1969 Fender
Deluxe and other similar types.
NONE had any active solid state devices, and none could compensate
for their owner's lack of musical ability to
play well without just creating noise. Since the 1960s, there has
been an abysmal trend of makers to fill their creations
with ever more features and opp-amps costing ever more dollars
while reducing the construction integrity of the circuitry,
and always trying to use ever more complex and miniaturized
printed circuit boards making service work difficult,
unpleasant, and expensive.
There has been a slight pleasant trend to production of "boutique"
or "hand made" amps. Nearly all these show their
maker's ability to copy 1960s designs using tag strip circuitry.
Unfortunately, the typical "custom amp maker" uses the
very cheapest and worst quality of tag strips and resistor made in
China and purchased from Jaycar for a pittance.
The best tag strips and 1W metal film R come from Taiwan. There is
ridiculous worship of "orange drop" caps etc.
Polyester or polypropylene caps with 630V in box filled with epoxy
I can't say which potentiometers are best, but never use anything
less than the 25mm dia types. They should be sealed
with tape to stop solder fumes ingress during assembly and their
contacts tightened with vice grip pliers and never over
soldered lest heat loosen the contact lugs.
Tube sockets should always be bolted to holes in a metal chassis,
and the layout of circuit should be well thought out
without cramming of any parts to prevent easy access to any one
thing without having to move any other thing.
Printed circuit boards are only a benefit to a manufacturer to
lower construction labour. In many guitar and Hi-Fi amps
the printed circuit board must be moved to get to soldered joints
and there are often added wires from board edges
to sockets, and the final result still gives a rat's nest of wires
and lower quality of product which always more difficult to
repair or analyze.
Below, I show the mods I did to a Marshall MkII head amp made
after 1968, before 1985. The Groove Tube book
with many schematics had the original schematic which may also be
available somewhere online if you search for it
so I won't bother spending a day to scan the schematic from a page
in a book, then produce a readable schematic
just for you.There are many existing amps which could use my
schematics and ideas, all of which may be intelligently
Without further bullshit, I shall get on with describing a good
guitar amp to suit most good players, and which is
able to be used with a player's collection of Effects Boxes ( EFX
), each of which contain vast numbers of transistors
in chips which now do an amazing number of things which would
otherwise need a fridge sized box of countless
tubes to do the same thing.
Fig 1. Guitar preamp.
Fig 1 has V1 1/2 12AX7 to amplify input x 50 so that for a typical
5mV input at g1, the Va output = 250mV.
This was found to be an ideal input signal to a couple of EFX
boxes. Many EFX will work from say 10mV
produced by the guitar pick ups, and do not have any gain, so 10mV
comes out of the EFX unit, but with added
effects, could be echos, like an endless tape unit, or could be
reverb, and the EFX is then fed into the sensitive input
if the tubes guitar preamp. This suits some ppl, but the guy here
wanted more than one EFX unit used, and it was
decided by trial and error that inserting his favorite echo box
after one stage of preamp was the best place, because
some other EFX could be be installed between guitar and amp. The
impedance in and out for EXF box and data about
optimum operation Vac levels were all not available. I have made a
SEND circuit with V2 1/2 12AU7 cathode follower
with Rout = 600r. V3 is a cathode follower with high Vin and which
drives VR1 preamp volume control. If no EFX box
is used then R10 82k allows the V1 output to be applied to VR1
driven with low Zout of V3. V2+R10+V3 do cause a
signal loss of about -2dB, a negligible amount, So without any EFX
box, the amp can be used as a normal amp - but with
the big benefit of the added preamp VR1 volume control which
allows the guitarist to crank up volume to create over load
distortion without overloading the output stage at all, if
desired, and it IS desired by a number of players in a small
The original Marshal only had a master volume control just before
the power amp, and that just does not suit everyone.
The owner of this Marshall who had me remove 1/2 the original
horrible PCB to allow the point to point wiring mods said
he much liked the distortions available by overloading the preamp,
and being able to use low volume with 10W output
without the 2 EL34 being pushed into overdrive and class C
V4 amplifies 50 times, and clips at 42Vrms, so that max g1 input
without clipping = 0.84Vrms. V5 is 1/2 12AU7 and
cathode follower which loads V4 anode with high RL, and drives the
tone stack with low Z. Tone stacks should always
be driven by a strong cathode follower so that the variable
input Z of stack does not cause IMD well before clipping.
It always sounds well, and all good guitar amps have CF drivers to
tone stacks while worse sounding amps omitted the
CF and they sound brittle and hard. 1/2 a 12AU7 makes an excellent
CF better than 1/2 a 12AX7.
The tone control knobs are marked 0 to 10 evenly spaced on the
front panel. The positions of controls to get a near
flat F response from 50Hz to 10kHz were not what anyone might
guess, or determine by ear, and are listed on the
schematic. There is a high amount of treble boost available above
1kHz, the most wanted feature. The headroom of
the preamp must be high so treble boost can accommodated -
something that is helped by use of the V5 CF, and also
V6 CF with 1/2 12AU7 after the treble control. The output from
treble pot is high impedance and variable, so for best
sound the loading of tone stack after treble pot output should be
high Z, and in this case its about 2M, because the
V6 grid bias R22 220k is bootstrapped to become at least 2M0.
V6 CF output Z > 1k0, and easily drives the Master Vol pot
which drives the power amp.
Fig 2. 40W Guitar Power Amp.
Fig 2 is different to the original Marshall from 1968, although it
does have the same tubes of V7+V8 12AX7 as
long tail pair differential driver amp and a pair of EL34
output tubes for class AB ( or class A if the speaker load is
4 times the value on S2 speaker Z select switch. )
The original B+ values are about the same as original.
I have made some drastic changes.
(1). TWO bias pots are used for "fixed bias". TR2 is a 10k0 x 3W
wire wound pot mounted on the remaining
portion of PCB board is on the power amp under the chassis which
must be removed to make adjustment of TR2.
TR2 allows the possible nominal bias to each EL34 to be adjusted
between -36Vdc to -46Vdc, and once it is set,
wners should never be encouraged to twiddle TR2 bias pot because
many will get it wrong, and cause tubes to
overheat. The size of OPT in this amp indicates possible use of
6550 or KT88. BUT, the screws holding tube
sockets to chassis MUST be replaced with c/s head types and
springs and clips used to hold tubes in sockets MUST
be anchored further away from sockets to allow 6550 or KT88 tube
bases to fit properly with pin penetration to socket.
Use 6550 or KT88 also requires Eg1 bias to be between about 45Vdc
and -58Vdc and to obtain the Eg1 increase
R4 56k on SHEET 3 Power Supply schematic must be disconnected at
For EL34 or 6CA7 with Ea = Eg2 = 440Vdc, and for Ikdc = 30mAdc,
average Eg1 will be -42Vdc, with +/- 3Vdc
for typical tube samples which do not need to be exactly matched,
or of the same age, but be quite usable.
TR1 is a 10k 3W wire wound pot. This is mounted on rear panel of
chassis along with tp1 and tp2 and a 0V terminal.
This allows an owner or service person to measure Idc in each
output tube and to equalize the Idc in both output tubes.
To achieve equal Idc, Vdc between tp1 and tp2 to 0V can be made
equal by adjustment of TR1 with a screw driver.
Both Vdc should equal 0.30Vdc. This indicates Ikdc through
R14+R15, both 10r0, is 30mAdc.
When TR1 has equalized the Idc, the output circuit has BALANCED
Idc, and will give the best sound because there
is no significant Idc flow across all primary turns in OPT. With
Idc balanced, each output tube will generate the same
heat at idle, give the least hum, and prolong tube life. The
action of TR1 is to increase Eg1 of one tube while reducing Eg1
of the other, and having one tube grid at -41Vdc and the other at
-44Vdc is typical with an average pair of healthy tubes.
Where the TR1 balance pot cannot be turned enough to achieve equal
Vdc at tp1 and tp2, then there is a problem
with one of the output tubes at least. If Vdc is balanced, but say
0.5Vdc, when it was once 0.3Vdc months before, then
at least one tube is conducting too much Idc and is becoming
difficult to balance with enough Eg1 bias Vdc.
In many amps I have serviced, one output tube may be running too
hot with Ikdc = 60mA, and the other too cool with
25mA. There is 45mAdc of unbalanced Idc in OPT primary which
caused the core to be dc magnetized on one direction.
The amp still works, but with a low threshold for distortion on
low bass notes, and there will be a much higher amount
of distortion at all mid and higher F. The hot tube has much
higher peak current during overdrive and the guitarist
does hear the problem until the overheated tube shorts internally
a fuse blows, and amp is silent.
(2) S1 has been added to allow choice between pentode or triode
operation. "Normal" pentode operation gives about
0W at clipping and triode operation gives about 20W. Although
triode operation gives less power, the sound at levels
under clipping may sound cleaner and more precise, with better
Overdrive while in triode mode will give similar rapid increase of
distortion. The original amp was for pentode only.
There were very bad oscillations at low RF when I first used
triode and R11+R13 3k9 "grid stopper" resistors were
added which completely prevented the oscillations. Most tube power
amps use a series R of 2k2 to 10k between from
driver anode to output tube grid. I think Marshall made a mistake
to omit them.
(3) I have connected a Zobel network of R18 4k7 + C5 1n0 across
the OPT primary. This means that above 33kHz,
there is a load of 4k7 across the OPT primary. This added load has
no effect below 20kHz where C5 becomes an open
circuit. But the 4k7 loads the OPT and helps prevent HF
oscillations if tubes over heat, or if the amp is left turned on
with an open speaker connection; ie, a 6mm lead plugged in, but
not plugged into a speaker cab.
The 6mm switched output jack automatically shorts the OPT
secondary to 0V when no speaker is plugged in.
If there is any input signal, the output tubes could overheat and
it would be better to have 8r0 in the switch to 0V circuit,
so that with no speaker, there is 8r0 10W to absorb any power if
there is a continuing input signal. It is not likely if only
guitars are used to input.
(4) The presence control boosts the HF above 1kHz to brighten up
the sound with a maximum boost of about +7dB at
7kHz in pentode, with +3dB in triode. The value of C6 22n acts to
shunt the global NFB which is across VR3, 5k0.
Many amps have C5 = 0.1uF, so that all F above 300Hz can be
boosted. This means the amp just gets louder, and can
more easily be pushed to overdrive. There is not much change of
tone except for increased distortions with less NFB.
Some may prefer this, but the presence control was intended to
slightly boost F between 2kHz and 10kHz, and it is an
easy available way make an additional change to tone, not
loudness, and to add to what the treble control also does.
The HF content from electric guitar strings and from speaker cabs
may be low, and the HF must be boosted to
compensate for the low amount of HF naturally present.
Fig 3. PSU for 40W amp.
Fig 3 is similar to the original Marshall MkII because the
original PT, Si diodes and electro caps are all retained.
(5) In Australia, I often measure 245Vac mains most days,
sometimes 250Vac. This means that if the mains PT
of a tube amp is set for 230Vac for use in UK, the B+ after the
diodes will be 6% higher than in UK, say +480V
instead of +450V. Fortunately, the dual unit electro caps have a
high +500V rating which allow higher than normal
mains Vac. Use of 450V rated elcaps may see them fail. The
original Marshall had B+ to OPT CT from one C
after the diodes, and the screen supply was filtered by choke and
another C, and this is the most commonly used
set up. I found the choke has enough inductance and low enough
resistance to be used to filter all the B+ current
to both anode and screen supplies. I also added R10 50r in series
with Si diodes to limit peak charge currents to C10.
My arrangement reduces the working B+ to +440V at C8//C9 and OPT
CT, down from what would have
been +460V in original amp. I found the B+ surged to +475V seconds
after turn on, and sagged to +440V after
about 15 seconds when EL34 began to conduct full Idc. The B+ at CT
is thus well filtered with low 100Hz ripple
so that rail noise is not in series with each 1/2 of OPT primary
during AB operation, and thus output remains
clean and musical, especially in triode mode where 100Hz diode
rail noise is more audible.
(6) To accommodate the pair of 12AU7 which have much higher Iadc
than the pair of 12AX7, the resistors R6,
R8 have been reduced to 3k3 from the original 10k0. There is still
very adequate filtering of B+ rails at input stages
and separation between stages and there is no sign of
motorboating. I added R3 1k8 + C2 22uF for decoupling B+
to the added V2+V3 12AU7.
(7) The negative bias Vdc is generated using a resistance divider
R4, R7, R9 similar to original but with much lower
R values and high V ratings for C5, C6. There is -92Vdc produced
to be supplied to the bias balancing network on
SHEET 2. Notice there is no need to regulate the negative bias
rail. Regulation NOT wanted because if B+ is higher
than normal, you want the negative bias B- rail to also be higher
thus making bias applied to tubes more negative to
try to keep the heat in the tubes more constant. The balance of
biasing will not be much affected.
(8) R1 + R2 + C1 form an R divider to bias the 6.3Vac heater
winding to +65Vdc. This may seem strange, but
12AX7 and 12AU7 have 90V ratings for V between heater element and
cathode. I have 12AU7 cathode followers
and maximum cathode Ek of V4 is +175Vdc. There is a chance of
leakage or short circuit from heater to cathode.
If heaters are at +65Vdc, then the Ek-h is reduced to +110V, and
most tubes survive 20 years OK. I have seen
many amps where the cathode of a a follower driving a tone stack
is at +200V, with Vk-h = 200V, way above the rating.
The best amps have a second 6.3Vac winding which is used only for
the cathode followers. But like many things
abolished by bean counters in amp companies, the extra 6.3Vac wind
was first to be exterminated. Amp makers
owned shares in tube making companies so that dividends increased
when poor circuit design caused tubes to fail
more often. Everything in this world which can be corrupted, is
Note. In general, Jim Marshall used very well rated power and
output transformers and choke, and I see there would
be no problems using 2 x KT88/6550 with mods mentioned above.
However, the choice of OPT load ratios are definitely a cause of
concern with 3k2 : 4r, 8r, 16r. If the S2 speaker Z
select switch is set for 8r0, and speaker is 6r5, like so many
"8r0", then 6r5 reduces anode load to 2k6, and in class
AB the tube load = 650r, but with the added winding resistance of
both P and S windings, about 200r, so
tube load = 850r. If peak current = 0.32A, and Ea = 400V with
sustained signal, then tube Pda = 40.3W.
With a continuous sine wave at clipping, audio Po = 34W, but the
tubes will soon overheat. To prevent this ever
happening, the S2 switch should be set to 4r0, not 8r0. A speaker
with Z = 6r5 will then produce a primary load
of 5k2. Class AB load for each tube = 1k3 plus Rw 200r, so total
1k5, and peak Ia is reduced to 0.24A and Pda
will be 30W, and the tubes will survive longer. The audio Po is
slightly higher at 36W for the higher ohm load.
So why would anyone ever use a 3k2 primary load for EL34 set up
with B+ = 450V?
The EL34 have low limits on peak Ia, and the 3k2 is better load
for 6550 / KT88 which would give a maximum of
60W at sine wave clipping, with Pda just reaching the 42W rated
Back to Index page