The Ultimate Guitar Amp?
All Tube, True Stereo, Single Cabinet
Guitar Amp with Very Well Optimized Adjustable Distortion Generation for the
Bluesman, and five tone controls per channel for optimal tone with any guitar.
35 watts rms per channel.
by Robert B. Richards, AE
(c) 2008, Robert B. Richards, All rights reserved internationally.
(It's the one on the left)
For well over a decade there have been many sound effect
devices on the market that synthesize a great sounding stereo output. This was
true when I bought my Boss SE-70 digital effects processor back in about 1996.
Most guitarists I've known (me too) can't be bothered with dragging multiple
amps to a gig, so we stay "traditional" - monophonic. Since the guitar
itself is very probably monophonic, a stereo amp will still put the direct "core"
sound out in mono, but then the digital reverb, delay taps, chorusser, phlanger,
and etc. will radiate an ambience that is easier to hear and connect with, easier
to be carried away by.
When it comes to the question of what sound effects
an amp should have, it's arguable that the days of built in tremolo and reverb
are a bit obsolete. Every guitarist I know of has accumulated many footpedals
and/or bigger effects processors. They all have their favorites, and probably
don't need or want the main amp cabinet to have somebody else's idea of what
looks good on paper or in an advertisment. I'm talking about reverbs, chorusers,
flangers, delay effects, hybrid effects, etc. So I decided that this amp should
do a more limited part of the total process very
well (tone, distortion and be stereo), and leave the rest up to the external
footpedals or digital processors. There's a lot to be said for keeping things
simple. This amp has a stereo preamp with five tone controls per channel (four
passive Baxandahl (James) and a 6 position switchable high pass shelf), an input
gain, a master volume, a stereo/mono switch, a pilot light and an On/Off switch.
No other clutter on the front panel. The poweramp is 70watt stereo (35w/ch.),
and there's two 10 inch vintage reissue Jensen speaker drivers for the best
of bluesy compression, cone breakup and bubbly fat distortions. I tried to make
an amp that Eric Clapton would be thrilled with.
This is the only all tube, true stereo guitar amp, with
very good tone controls and of truly useful power levels, that I know of. And
it's all in a vintage tweed style cabinet that is about the size of a Fender
Twinamp or Twin Reverb.
Although it goes
quite loud clean, it's also very optimized for bringing the very best distortion
sounds out of tubes. Harmonic distortion spectrum shape and intermodulation-modulation
is optimized throughout, and choice of speaker driver was very specific for
This is a no feedback
design in both the preamp and poweramp sections, so well voiced distortion is
there, but none of the drawbacks of feedback are there (slewing related distortions,
loss of second harmonic distortion product, harmonic distortion products that
extend substantially out from the fundamental, phase margin instability, recirculating
I.M. generation, etc.).
The very slightly
outward angled speaker panels create a better sense of stereo spread when you're
back 10 feet or more, without the cabinet getting bulky and awkward to carry.
can be added or used instead if desired.
A second set of
preamp outputs can be footswitched on and off. This can be nice when you want
to switch on a P.A. system connection or additional guitar amps at each end
of the stage for soloing.
Shell is solid pine, all bracing is
Speaker panels are half inch plywood,
to minimize weight and have good strength and not be at all warped. Any warp
at all is no good..
The front is 3/4 inch plywood. My own
old world design, as far as I know. Low-key vintage tweed look. Not horribly
distracting, just hopefully artistically inspiring.
Out where the audience is, there will
be noticeably more stereo separation due to the slightly angled driver panels.
Tweeding the front.
The Stereo Poweramp:
I tried to keep it simple for the sake
of reliability. Since there's no negative feedback, I needed fewer stages.
I added a separate power supply for bias
on the E34Ls. Otherwise I'd have to derive the approximately 20 volts bias (subtract)
from the B+, which would reduce the final power out substantially. You want
the amp circuit to see zero ohms from DC to infra-red, when it looks at the
power supply, so it's frequency response at the speaker is somewhat consistant
(this would substantially affect phase margin if it were a feedback circuit).
I don't recommend chokes. They were useful back in the days when the biggest
cap you could get was 25-50 uF. Now two 500uF J.J.
brand (or equiv.) electrolytics with about 20 ohms between them and a 0.1 polypropylene
across them makes more sense to me (guess I need to update this schemo a bit).
If you use chokes and caps, you are creating a 2 pole filter. (If you don't
know what you are doing with multi-pole filters, you could be creating a resonant
circuit. Something that rings significantly every time a rectifier diode switches
on or off - which they do 120 times a second.)
It's definately worth the time to lay
everything out and see what fits where.
There's really only barely enough room
in most places, to fit what you probably want to have.
I used all white dry-transfer lettering
for the nomenclature. Fabricating, painting and lettering the chassis is a lot
of work. I always put two coats of satin clear over the lettering.
shit, now I gotta wire the dam thing...
I couldn't find a chassis that was a reasonable
size and shape, so I bolted two 8 X 12 X 2 inch Hammond chassis together. I
cut two big holes on the mating surfaces with a saber saw, so power and gnd
wires could go where they need to. Three large bolts hold them together on the
common surface, and the power transformer stradles the two chassis, thereby
adding strength. Perfecto.
What isn't obvious is the 25 black wires
returning to the ground star center lugstrip. It's near the lower-middle area.
You want to have a separate ground wire returning to the star center for each
section of circuitry, and have the star center generally near the main 500uF
filter cap(s), not in the direct charging current path for the big filter caps
(take a tap from there to what you will use as the actual star center "signal
ground" bolt). Avoid having the star center ground very near where the
third prong of the AC plug (the earth ground pin) ties to the chassis. This
made a BIG difference in hum level.
The orange things are bias testpoints.
There's a ground testpoint too, but you can't see it in the photo. Each tube
has it's own bias pot right next to it. If you disconnect the speaker cables
while the amp is in use, the built in load resistors automatically switch in,
to protect the circuit from potential self destruction. I decided to add a Stereo/Mono
switch in the lower right corner, for more flexibility. I could have both poweramp
channels in this cabinet do one preamp output and a second amp on the other
side of the stage doing the other preamp output (the other amp being optionally
foot switchable). That's one of several ways the system as a whole could be
Waiting for the preamp chassis to be
completed and installed.
The Stereo Preamp Circuit :
This is where a big part of the magic happens.
I wanted to have the 6 position highpass switch ahead of the 12AX7 distortion
stage so I could vary the charactor or frequency balance of the distortion,
and then have the regular tone controls after the distorter in case I want to
roll off the higher order distortion products, create more of a reverse fletcher
munson curve or just deal with whatever peculiarities a given instrument might
have. The EF86 turned in a whopping high gain of 237, I double checked it. I
chose to go big on the cathode bypass cap so sub-harmonics and I.M. difference
products could effectively modulate the distortion generation in the next stage.
Part of the distortion "casserole". The 1 meg resistor feeding the
Input Gain pot is so the range of the pot is how I like it. The control will
be at about its mid position when the distortion is just getting thick with
a normal Stratocaster guitar. It also reduces I.M. distortion in the front end
(EF86)by lightening the load on it. Some I.M. can be good, too much is not good.
I want a good clean sound option too.
The transition into distortion is pretty
gradual. I like the way it sounds. The passive tone sections have about a 12
dB loss per. The 12AX7 gain/buffer distortion generating stage has a measured
gain of 71 or 37dB. The final output buffer is a self biased follower, which
will have less distortion than the direct coupled type in the 12AX7 stage. I
used 6SN7 octal tubes for the tone buffer and output buffer because anyone knows
that octals are very sexy. 6SN7's are very linear so don't mess things up. They
have been worshipped by certain hard core audiophiles for the pleasantness of
their sound. They also provide a lower and more stable output impedance, which
is definitely an issue when driving long cables. I wanted this to be able to
drive 20 foot cables going to external guitar amps put else where in the jam
room, when the footswitch enables that function (good for lead solo). 12AU7's
might have been fine here. I later decided to put 50X attenuators on the foot
switchable outputs (not shown here), since the additional guitar amps have way
too much gain to handle that preamp output level (10K in series and a 200ohm
to gnd.). Slightly less attenuation would have been better.
Although as shown, the circuit sounds pretty
darn good, there were some final tweaks done to the distortion section
(the 12AX7) that changed the nature of the distortion sound a little bit. The
actual values and topology are very similar to the 1959 Fender Bassman or 1957
The footswitch connector could be wired
for a double footswitch, if a second function was desired. The jack is stereo
and the extra connection is wired as a redundant ground connection here. I may
eventually modify this so the second footswitch bumps up the pre-distortion
gain by maybe 3-5 dB (could make that adjustable). I'm using a Marshall double
footswitch that I bought years ago for another project that got dismantled.
I've actually changed the B+ filter cap
arrangement since I drew this. Now I have two 100uF caps in place of the one
470uF cap, with about a 200 ohm 5W resistor between them. It works much better
(much less ripple) than the single bigger cap. It's hard to find exactly the
transformer you want. Custom's are too spendy for me. The shortcoming is the
current rating of the 5 volt winding, and height of the B+ windings. When you
convert the 5V winding to DC, you approximately double the current draw because
of the loading of the large filter cap, so the winding could only handle the
filiments of the pentodes and 12AX7's. The 6SN7's have AC on their filiments.
The hum is quite low for a guitar amp. How I grounded the star center of the
circuit to the chassis (in each chassis), and where I tied the green wire (earth
ground) to the chassis made a huge difference in hum level. That's something
that doesn't show on a schematic. You need a Ouija Board to get that right.
shit... now I've gotta wire the dam thing...
Parts on the left, tools on the right.
I had to remove all the pots and rotary switch from the one channel in order
to wire the first channel. It's pretty tight in there (4 X 3 X 17 inch Hammond
You want to fully test
the first channel before you wire the second channel. Changing parts later in
such a tight space will be major surgury - not fun.
Below is part density before adding the
2nd channel tone circuit components and wiring.
I guess I like challenges.
You want to actually put it in the cabinet
and use it for at least a week, do pink noise cal'd mic tests and etc., before
wiring that 2nd channel.
Below it's all done - both channels.
Deep inside the forest is a secret garden...
or so they say.
Too late for changes now.
Oops no, I changed my mind, the Hi-treble
control had too little effect in the range that the speakers reproduce.
Later: Oops no. Now it's too sqeaky sounding.
I went too far. I'll change it again.
Modelling the circuits with a Duncan
program or equivelent is a great idea, but there's no substitute for real-world
Never did find a secret garden.
I almost forgot to ground my pentode shock-plate.
When it wasn't grounded, touching it would cause a small but noticable increase
in hum. The shock plate, made of heavy 1/8 inch brass on very fancy soft rubber
grommets, keeps the amp from getting microphonic feedback when the gain is set
to maximum (it actually works too). It was pointed out to me that you don't
want to use tube shield covers on the somewhat microphonic front end pentodes
because the spring inside them will resonate when stimulated physically by the
speakers. The EF806S (J.J.'s version of the EF86) which I'm using, already has
an internal shield. I think the shield compatible sockets are nice though since
they shield the pin area of the tube better.
Above, you can barely see the footswitch
jack between the filter cap and the fuse. The IEC AC connector on the short
cord is for the poweramp. That way you can have the ON/OFF switch on the preamp
turn on and off both the preamp and the poweramp.
Above you can see the relay board for the
footswitch mounted right on the footswitch jack. The orange thing is the relay.
The big filter cap inside the chassis is for the 6.3 volt DC for the
front end tube filiments and pilot light.
Preamp Waveforms and discussion:
Below is the EF86 being slightly overdriven.
Very smooth asymetric distortion .
This particular distortion is not likely
to happen much without an external gain pedal, which might be a nice addition
later. The Input Gain pot being right after this stage makes that possible.
Because of the way the screen grid of the pentode is wired, there is some modulation
of the attack and decay, which I believe is an important part of the sound.
Tone burst tests will show this. I haven't gotten around to taking pictures
of this yet.
After the EF86 pentode, hi-pass shelf switch
and input Gain control, the signal goes to the 12AX7 distortion generating stage.
Below is the 12AX7 "Full-Wave Saturator"
circuit distortion waveforms:
To my knowledge, the 12AX7 gain stage direct
coupled to a follower buffer was first used in the 2nd or 3rd Fender Twinamp
(the '57 e-type, for example, which appears to be Eric Clapton's favorite these
days), and in most of the early Bassman circuits (the '59 for ex., which was
essentially copied in the early Marshall JTM 45). Certain other Fenders as well,
but not all. Twin Reverbs and most Bandmasters didn't have this. Showman's either.
It's arguable that any "blues" amp should have this, regardless of
what else it has. It distorts gradually and sounds good (not recommended for
Hi-Fi). I call it a "full wave saturator" because both half cycles
get clipped by a tube saturating, even though the positive half cycle saturation
Left photo is 12AX7 gain stage plate just
short of clipping, right is same just in clipping.
Below left is the follower output when
it just starts to have the "flat spot" on the negative half cycle.
Middle is with a little more input drive. Right is same with even more drive.
The follower starts distorting noticeably
before the gain stage would actually clip. The positive half cycle gets lopped
off when the direct coupled follower starts drawing grid current, and the negative
half cycle gets the flat spot when driving a load that is even slightly capacitive
(30Kohm in series and .01uF to ground), such as a tone control circuit.
Pink Noise Tests:
The test setup:
Not an ideal setup at all, but maybe revealing
enough if you know how to interpret the results.
So you get the preamp to look good on the
bench, waveforms and such, but what does this "e-phase" version sound
like in use, and how does it measure acoustically with pink noise and a calibrated
mic? Again, now is the time to do any tweaking of frequency response, tone circuit,
12AX7 distortion mechanism, and whatever else, before finishing wiring that
Amp about a foot out from wall in my living
room, mic at about 5 ft. Not exactly "Hi-Fi". Tone controls not centered.
I usually have my amp about a foot out
from a wall, since my place is small and it's in the way anywhere else. To see
how many of the frequency response anomolies are primarily a function of the
acoustics created by the wall, I then pulled the amp out and rotated it so there's
no wall near its rear or front, and mic'd it far and close. I was hearing the
possibly resonant peak around 1kHZ, and it was bugging me when I played either
my metal string acoustic (Taylor) or my Strat. I wanted to measure just how
much of it was my room and how much was the amp. The wavelength of about 1kHZ
is about 1 foot, so maybe putting some acoustic absorbtive material inside the
cabinet would take the hurt out of that some.
Amp away from walls and mic 5 ft. out.
Tone not at mid.
Same as above but close mic'd. Tone not
Then I played with the tone controls and
close mic'd that.
Tone controls all centered, shelf off,
close mic'd, minimal walls.
This is not a perfect test setup (Behringer
Ultra-Curve Pro RTA and Pink noise with Behringer cal'd mic on stand), but close
enough if you know how to interpret the results in my opinion and pretty important
(much better than nothing).
All tone controls at minimum position.
All tone controls at max position.
The speaker drivers I used are the 40 watt
version of the Jensen vintage reissue P10R, called the P10Q. I wanted to have
the sound be as close to the P10R as I could, since I loved how they sounded
in another amp I built. The Jensen 100watt Neodymium driver didn't sound nearly
as good (very poor high frequency extension - very "honky" sounding).
I tried those, and then sent them back. I really like the sound of the P10Q.
It pretty much has the same high end extension of the P10R, with a more practical
After Acoustic Padding is added:
I've glued some 1/4 inch thick wool felt
to the inside of the cabinet wherever there was a place to put it, in the top
area and sides. I wanted to be relatively thorough so I might learn something
from it. After playing for a whilewith all tone controls midranged, I turned
up the lower midrange tone control to where I know it to be mostly flat with
the calibrated mic and pink noise (3 o'clock). I played it like that for a while
as well. Playing my strat through it with the insulation, it sounds very good
to my ear. Although the difference is small, and generally doesn't show up on
the very improperly done calibrated mic tests, I think it helped some but not
a big change. Here's some pics:
The added insulation (the amp is upside
down). The poweramp covers pretty much the whole bottom, so no insulation there.
Some days it seems like my whole world
Close mic'd, tone controls all centered,
cabinet away from wall with insulation (above). The 170HZ peak is probably about
the room (due to the wavelength (about 7ft.) and absence in other graphs), rather
than the amp.
Same thing but no insulation. I'd only
trust this graph comparison in the mid frequencies due to variations in the
angle of the amp in the room and exact location of the mic relative to the speaker.
From say 300HZ - 4kHZ maybe. Not a big difference that shows with this test
method, but I think I hear a significant difference when playing through it.
I'm pretty sure I like it better with the insulation. It may have lowered the
Q on whatever resonances there are in the cabinet, that might have been active
around 400HZ - 7kHZ. I mention those frequencies because they are the ones who's
wavelengths fits inside the cabinet dimensions in various ways, and can therefore
As above but amp near wall; about a foot
out, mic at about 5 ft. Carpeted floor.
I would think it could still benifit from
an attenuation of the frequencies around 500 - 1.2kHZ, but the test with no
wall and mic at 5 ft. suggests that the amp is not that bad and it's the nearby
wall that is actually pumping up the 500 - 1.2kHZ frequency region. Looking
at the wavelength of those frequencies (6 inches to 2 feet), I think it's believable.
Those frequencies can be effectively attenuated somewhat by turning up the lower
and upper midrange tone controls a bit.
Tone controls set by ear to sound better
with strat (close mic'd). This graph suggests that the low-Q slight notch should
be around 1.3kHZ.
I'm not sure what caused the notch at about
2kHZ, but since that's above the highest note on the guitar (somewhere around
800HZ), I'm not real worried about it. I think that section of frequency, 1kHZ
- 5kHZ, has a lot to do with the overall sound of the amp, since those are the
frequencies that the ear is most sensitive to. The 2kHZ notch probably has more
to do with the fact that two physically displaced speakers are feeding one mic.
I tried to have a Deco look.
Poweramp in the bottom, preamp in the
top. This amp, with 35 watts per channel, goes very dam loud. The distortion
circuit also sounds very good at very low levels (when the poweramp and
speakers aren't at all overdriven).
A musician wants an amp that is intuitive,
easy to adjust, and doesn't distract him or her away from the creative part
of their brain. Achieving that on a stereo amp is perhaps even more challenging.
You've got Input Gain, a high-pass Shelf
6-position rotary switch, James style tone; Lo, Mid-Lo, Mid-Hi, and Hi variable
slope, then a Master volume and a Stereo/Mono switch. Just what is needed and
no more. The signal flows from left to right through the knobs. As intuitive
as it gets.
I feel pretty good about what I ended
up with. The distortion is bubbly and thick but not whimpy, not at all fingernails
on the blackboard like many amps I've heard (possibly all transistor amps and
some tube amps). It has lots of "bluesy charactor". It makes you feel
good. It's probably the best sounding amp I've ever played through. It also
sounds exceptional when adjusted for no distortion and used with either an electric
guitar or an acoustic guitar (perhaps especially with my nylon string acoustic
guitar). Every guitar has it's own frequency response "signiture".
They vary a lot. You want to have the tone control flexibility to bring out
the best in any instrument, whatever that is.
Warning to amp
It's been horribly labor intensive to build
(all point to point wiring), and parts alone were over $1500. .Also, despite
a strong effort to keep the weight down, it ended up being 68 lbs (a '57 Fender
Twinamp e-type reissue is spec'd at 52lbs). I guess that's why nobody makes
all in one stereo tube amps of this power rating. Be warned if you're a builder.
If I were to go into production with this, I'd definately use circuit boards
and switching power supplies (to reduce the weight by almost 10 lbs. and make
assembly much more cost effective). I had decided to just go buy a Fender Blues
Deville for around $700, until I went and played through one. It was very bright
and just not pleasant to listen to. I felt that I had no choice but to research,
design and build an amp of my own. My amp sounds WAY better to my ear.
This amp doesn't do everything, but what
it does do, it does pretty dam well.